ABSTRACT
Uniquely among adult tissues, the human endometrium undergoes cyclical shedding, scar-free repair and regeneration during a woman's reproductive life. Therefore, it presents an outstanding model for study of such processes. This Review examines what is known of endometrial repair and regeneration following menstruation and parturition, including comparisons with wound repair and the influence of menstrual fluid components. We also discuss the contribution of endometrial stem/progenitor cells to endometrial regeneration, including the importance of the stem cell niche and stem cell-derived extracellular vesicles. Finally, we comment on the value of endometrial epithelial organoids to extend our understanding of endometrial development and regeneration, as well as therapeutic applications.
Introduction
Recent advances in regenerative medicine offer great potential for treatment and possible cures for a number of challenging diseases, such as abnormal uterine bleeding, endometriosis and Asherman's syndrome. The regeneration field has used new technologies including stem/progenitor cell therapies, three dimensional (3D) bioprinting, nanofiber applications and gene editing. However, most human adult tissues do not naturally regenerate after damage and, therefore, human models are not generally available to study molecular and cellular mechanisms underpinning natural regeneration. The endometrium, the mucosal lining of the uterus, is unique in this context. It is highly dynamic and experiences continuous significant cell turnover during the reproductive years. This extraordinary tissue is shed and then fully repaired at menstruation, and subsequently regenerated; a cyclical process that occurs each month during a woman's reproductive life (∼400 cycles). The endometrium is also fully restored following parturition (childbirth). Importantly, the endometrium repairs without scarring (a feature otherwise only present in fetal wounds and some adult mucosa) and then regenerates the full thickness of tissue (∼8 mm) with functionality to enable subsequent successful embryo implantation and placentation. Thus, it is timely to examine the processes involved in this repair and regeneration, and to highlight lessons that may be applied to scar-free tissue repair and subsequent tissue regeneration within the framework of developmental sciences.
The endometrium
The endometrium comprises two layers, the outer functionalis (functional) and the underlying basalis (basal) layer proximal to the myometrium. The functionalis comprises an outer luminal epithelium with its apical surface facing the uterine cavity and abundant vertical glands contiguous with the luminal epithelium. These glands penetrate into a stromal compartment comprised of fibroblasts, vasculature (including unique spiral arterioles) and variable numbers of leukocytes within an interstitial extracellular matrix (Fig. 1A-E). All these cells can be identified clearly by classical cellular biomarkers. The basalis contains horizontal branching networks of glands (Tempest et al., 2020; Yamaguchi et al., 2021), stromal cells and vasculature, which provide the majority of the stem/progenitor cells essential for regeneration of the functionalis after menstruation (Fig. 1B). Immune cells are also present in the functionalis throughout the menstrual cycle (Box 1).
A variable number and types of immune cells are present in the functionalis throughout the menstrual cycle, but these are generally in low abundance until the late-secretory phase when there is a massive and highly selective influx. These immune cells play a key role in menstruation and repair (Fig. 1C). Indeed, immediately before menstruation, up to 50% of the total cell population are leukocytes; 6-15% of all nucleated cells in the stromal compartment of the functionalis endometrium are neutrophils, and the same abundance is observed for macrophages (CD16+) and uterine natural killer (uNK) cells (CD56+/CD16−) (Bulmer et al., 1991). Eosinophils, mast cells (both 3-5%) and T lymphocytes (1-2%) are less common (reviewed by Salamonsen and Lathbury, 2000). Immune cells in menstrual blood closely resemble those in the pre-menstrual uterine immune microenvironment (van der Molen et al., 2014). Importantly, these cells are phenotypically different from their counterparts in peripheral blood (PB), indicating the effects of the local microenvironment. For example, production of active elastase is much reduced and α1-anti-trypsin highly elevated in endometrial neutrophils compared with PB neutrophils (Salamonsen and Lathbury, 2000). Together, these leukocytes, many detected in activated form, establish an inflammatory cascade that results in tissue breakdown. The key role of immune cells in endometrial repair is indicated by reduced uNK cell numbers in late secretory tissue in women with heavy menstrual bleeding (Biswas Shivhare et al., 2015) Such complex phenotypic changes in a highly dynamic physiological setting in women severely limit investigation of their individual functions.
The menstrual cycle. (A) Phases of the menstrual cycle. The first day of bleeding is defined as day 1 of the cycle. The major phases are menses and repair of the luminal surface (which occur simultaneously), the regenerative or proliferative phase, and the phase of differentiation or secretory phase; this is further divided into early, mid- and late-secretory phases. The regenerative phase is driven essentially by rising circulating estrogen levels, whereas differentiation requires progesterone, in the presence of estrogen. Androgens play a role but are at much lower levels with little cyclicity. (B) Schematic of cross-section of the endometrium showing the basalis layer adjacent to the myometrium and the functionalis layer, which apposes the uterine cavity across a normalized menstrual cycle of 28 days. The functionalis comprises luminal epithelium, stroma, glands and vasculature, the glands terminate in the basalis apposing the myometrium, where the epithelial and some mesenchymal stem/progenitor cells are located. (C-E) Cartoons illustrate the cellular changes during the premenstrual, menstrual/repair and regenerative phases. During menses (C), falling progesterone levels stimulate the production of vasoactive factors, matrix degrading enzymes and chemoattractants in the uterus, followed by a massive influx of leukocytes, particularly monocytes, neutrophils and eosinophils. Tissue destruction results from the actions of their products. Repair of the endometrial surface, predominantly by re-epithelialization in the presence of factors in menstrual fluid (C), is followed by regeneration of the tissue, likely from clonogenic, self-renewing stem/progenitor cells (N-cadherin+) located in the basalis (D), which are not shed during menstruation. The SSEA-1+ cells are located around an ill-defined basalis-functionalis junction and likely migrate to re-epithelialize the raw endometrial surface during menstruation, giving rise to the SSEA-1+ luminal epithelial cells (D,E). The majority of glandular epithelial cells located in the functionalis are highly proliferative, but do not express N-cadherin, SSEA-1, nuclear SOX9 or nuclear AXIN2 and are shed at menstruation. Endometrial mesenchymal stem cells are located around blood vessels as pericytes and perivascular cells in both basalis and functionalis (E). The majority of side population (SP) cells are endothelial cells and perivascular mesenchymal stem cells (MSC).
The menstrual cycle. (A) Phases of the menstrual cycle. The first day of bleeding is defined as day 1 of the cycle. The major phases are menses and repair of the luminal surface (which occur simultaneously), the regenerative or proliferative phase, and the phase of differentiation or secretory phase; this is further divided into early, mid- and late-secretory phases. The regenerative phase is driven essentially by rising circulating estrogen levels, whereas differentiation requires progesterone, in the presence of estrogen. Androgens play a role but are at much lower levels with little cyclicity. (B) Schematic of cross-section of the endometrium showing the basalis layer adjacent to the myometrium and the functionalis layer, which apposes the uterine cavity across a normalized menstrual cycle of 28 days. The functionalis comprises luminal epithelium, stroma, glands and vasculature, the glands terminate in the basalis apposing the myometrium, where the epithelial and some mesenchymal stem/progenitor cells are located. (C-E) Cartoons illustrate the cellular changes during the premenstrual, menstrual/repair and regenerative phases. During menses (C), falling progesterone levels stimulate the production of vasoactive factors, matrix degrading enzymes and chemoattractants in the uterus, followed by a massive influx of leukocytes, particularly monocytes, neutrophils and eosinophils. Tissue destruction results from the actions of their products. Repair of the endometrial surface, predominantly by re-epithelialization in the presence of factors in menstrual fluid (C), is followed by regeneration of the tissue, likely from clonogenic, self-renewing stem/progenitor cells (N-cadherin+) located in the basalis (D), which are not shed during menstruation. The SSEA-1+ cells are located around an ill-defined basalis-functionalis junction and likely migrate to re-epithelialize the raw endometrial surface during menstruation, giving rise to the SSEA-1+ luminal epithelial cells (D,E). The majority of glandular epithelial cells located in the functionalis are highly proliferative, but do not express N-cadherin, SSEA-1, nuclear SOX9 or nuclear AXIN2 and are shed at menstruation. Endometrial mesenchymal stem cells are located around blood vessels as pericytes and perivascular cells in both basalis and functionalis (E). The majority of side population (SP) cells are endothelial cells and perivascular mesenchymal stem cells (MSC).
The menstrual cycle
The menstrual cycle is dated from the day on which bleeding is first apparent (day 1) with the ‘normal’ cycle of ∼28 days being divided into three major phases: menstrual, proliferative and secretory; the latter of which is further categorized into early-, mid- and late-secretory phases. The major drivers of this cycle are the circulating ovarian steroid hormones, estradiol-17β (estrogen, E) released from the developing ovarian follicles and progesterone (P) produced in the corpus luteum, which forms after ovulation (Fig. 1A). Low levels of androgens also contribute to endometrial repair and cell proliferation, albeit to a lesser extent (Gibson et al., 2020). The menstrual phase (∼days 1-5), which encompasses bleeding, tissue loss and repair (Fig. 1B,C), is followed by the E-dominant follicular or proliferative phase, during which the tissue thickness is regenerated with cellular proliferation in glands, stroma and vasculature (Fig. 1B,D). After ovulation (∼day 14), when both E and P levels rise, the secretory or luteal phase is initiated, with substantial differentiation of the major cell types occurring from the early- to mid-secretory phases (Fig. 1B,E). The glands become convoluted and highly secretory; ciliated and non-ciliated epithelial cells are observed in both glandular and luminal epithelium and there is an abrupt and uniform transcriptional activation, particularly in unciliated epithelial cells (Wang et al., 2020). These phenotypic and secretory changes in the glands include changes in polarity and adhesive capacity of the luminal epithelium to provide receptivity for embryo implantation. The stromal fibroblasts begin to differentiate into pre-decidual cells in a process known as decidualization that occurs in a wave, starting close to the vasculature. In a conception cycle, these provide the foundation for the maternal decidual component of the placenta. Unique blood vessels, known as spiral arterioles, also develop during this phase (Fig. 1E). Various leukocyte subpopulations infiltrate/proliferate with progression throughout the secretory phase and a massive increase in numbers in the late secretory phase. Menstruation is initiated in the absence of human chorionic gonadotrophin (an embryonic anti-luteolytic signal), resulting in the demise of the corpus luteum and subsequent fall in the levels of circulating estrogen and progesterone. The endometrium responds to these falling hormonal levels by pre-menstrual initiation of the menstrual cascade, with bleeding indicating the start of a new menstrual cycle (Fig. 1A).
Endometrial shedding: menstruation
During menstruation, the functionalis of the endometrium is shed in a piecemeal manner, with breakdown and rapid repair occurring simultaneously at adjacent sites (Garry et al., 2009; Ludwig and Spornitz, 1991) (Fig. 1B,C). Repair of the endometrial surface by re-epithelialization occurs as menstruation proceeds and is completed as bleeding ceases.
Menstruation can be viewed as a highly regulated inflammatory response to progesterone withdrawal; mice lacking the progesterone receptor (PR) have a highly inflamed uterus with considerable infiltration of inflammatory cells (Lydon et al., 1996). The first phase of menstrual events occurs in the decidualized endometrial stromal cells, which express PR and hence sense hormone withdrawal. The stromal cells initiate a sequence of interdependent inflammatory events including nuclear translocation of NF-κB, a transcription factor that regulates the innate and adaptive immune response. NF-κB signaling causes the progressive production of many inflammatory cytokine and chemokine mediators, along with increased prostaglandin synthetic enzymes and production of proinflammatory prostaglandins (Evans and Salamonsen, 2012). These inflammation responses recruit and activate leukocytes (predominantly granulocytes) into the endometrium (Box 1). Each cell type produces an array of proteolytic enzymes including matrix metalloproteinases, plasminogen activator family members, cytokines, chemokines and other molecules, with considerable interactions occurring that initiate self-activating cascades. For example, in vivo, endometrial-derived immune cells produce a wide range of enzymes important for other cell activation (e.g. degranulation of eosinophils induced by neutrophil elastase) or molecular processing, such as conversion of latent to active matrix metalloproteinases by elastase or cathepsin G. These combined actions result in degradation of the extracellular matrix (ECM) and tissue breakdown (Henriet et al., 2012; Salamonsen and Woolley, 1999) (Fig. 1C). As tissue shedding during menstruation is piecemeal, fragments of endometrial tissue can be found in menstrual fluid (MF) along with single endometrial cells, blood and ECM debris (Box 2). Finally, hypoxia is one of the key signals for menstruation and women with heavy menstrual bleeding have lower levels of hypoxia-inducible factor 1-alpha (HIF1a) (Maybin et al., 2018).
Components of menstrual fluid (MF), which bathes the denuded surface during menstruation, indicate key roles in stimulating rapid re-epithelialization. MF contains a mix of shed endometrial cells, including stem/progenitor cells, soluble factors and blood. MF is also rich in bioactive molecules, such as proteolytic enzymes derived from endometrial cells, activated leukocytes and microbial species (Molina et al., 2020). Soluble factors in degenerating human endometrium (Gaide Chevronnay et al., 2009) are released as a result of the tissue lysis along with released microvesicles and exosomes. Unbiased proteomic analyses comparing MF with matched peripheral blood plasma have identified 84 proteins exclusively present in MF and 186 significantly more abundant in MF (Evans et al., 2019). Among these are strong candidates for facilitating post-menstrual repair. Application of MF in vivo in a porcine skin wound repair model enhances re-epithelialization, whereas individual MF-specific proteins promote both migration and adhesion to fibronectin in human endometrial epithelial and keratinocyte wound models. Thus factors in MF advance the initial migratory phase of healing, a key difference from current skin wound repair treatments that stimulate epithelial proliferation (Evans et al., 2019). Vascular repair agents (including VEGF) will likely play roles in the initial repair of the vasculature and subsequent angiogenesis. MF also contains stromal fibroblasts (menstrual stem cells), fewer perivascular endometrial mesenchymal stem cells (eMSCs), even rarer epithelial progenitors and clonogenic endometrial cells, which are not present in peripheral blood (Bozorgmehr et al., 2020; Masuda et al., 2021). Perivascular eMSCs, and potentially stromal progenitors, can be purified from MF and provide a non-invasive source of stem/progenitor cells for clinical application.
Endometrial repair during menstruation
Endometrial repair is initiated almost immediately when tissue shedding starts, to ensure that the endometrial ‘wound’ is repaired as rapidly as it is shed. Repair is complete by the time bleeding stops (3-5 days). Importantly, post-menstrual endometrial repair is unique compared with wound healing in other adult tissues because it occurs without scarring; it involves the rapid re-epithelialization of the endometrial surface, which protects the underlying endometrium (Fig. 1C). By contrast, regeneration of the endometrial functionalis occurs only after this repair is complete.
Re-epithelialization starts with the migration of epithelial cells from the exposed endometrial glands (Fig. 1D) or any intact remaining epithelium bordering the denuded stromal tissue, in a manner akin to skin repair (Ferenczy, 1976; Ludwig et al., 1990). The epithelium of the exposed gland stumps are in the same region as SSEA-1+ (CD15+; FUT4+) epithelial cells immunolocalized in full-thickness endometrial tissue in the endometrial basalis (Valentijn et al., 2013), particularly around the basalis-functionalis junction (Nguyen et al., 2017). This suggests that SSEA-1+ cells may migrate across the denuded surface during repair to become the new luminal epithelium, which is also SSEA-1+ (Fig. 1D) (Cousins et al., 2021; Nguyen et al., 2017). SSEA-1+ cells have some features of progenitor cells, such as longer telomeres (Valentijn et al., 2013), and are thought to also contribute to gland regeneration during the proliferation phase (discussed below).
Unlike in women, the mouse endometrium does not spontaneously decidualize under P stimulation; decidualization requires a physical stimulus such as the presence of a blastocyst. Mouse models of menstruation, in which decidualization is induced before P withdrawal, have enabled functional insight into endometrial repair following menstruation (Brasted et al., 2003; Cousins et al., 2016b; Finn and Pope, 1984). In mice, re-epithelialization occurs very rapidly (as in women) and is totally independent of the effects of E (Kaitu'u-Lino et al., 2007), reflecting the situation in menstruating women in whom E levels are minimal. Furthermore, although luminal epithelial cell proliferation is readily detected during the very early phase of repair in the mouse model, subsequent mesenchymal-to-epithelial cell differentiation (MET) has been proposed to provide a source of new luminal epithelial cells and so promote restoration of an intact epithelial layer (Cousins et al., 2014; Patterson et al., 2013). Specifically, some stromal cells in close proximity to the resurfacing epithelium immunocolocalized cytokeratin and vimentin. Several MET genes, such as Cdh2 and Vim, are initially upregulated during initiation of menses in mice, followed by downregulation of stromal Wnt4 and upregulation of epithelial genes, such as Cdh1, Wnt7a and Krt18 during the re-epithelialization stage. Several MET genes (Snai1, 2, 3, Wt1, Twist) also show differential regulation during the menses/re-epithelialization process (Cousins et al., 2014). However, recent in vivo mesenchymal cell fate-tracing studies in embryonic and adult mice have found no evidence for MET during endometrial repair (Ghosh et al., 2020). Another potential source of the luminal epithelium in mice is a rare population of label-retaining putative epithelial progenitors (Cervello et al., 2007; Chan and Gargett, 2006), but as their proliferation is initiated by E (Chan et al., 2012) they may be more important for glandular development after repair. How closely these findings correspond to re-epithelialization events in women remains to be established. Epithelial-to-mesenchymal cell transition (EMT) and MET play important roles in human skin wound healing that form scars, but as endometrium repairs free of scarring, the mechanisms may be quite different (reviewed by Owusu-Akyaw et al., 2019).
Dynamic changes in the ECM, specifically in the degraded area, are required to facilitate migration of luminal and glandular epithelial cells across the wounded surface during the early stages of endometrial repair following menstruation (Evans et al., 2011). These include changes in basement-membrane proteins, integrins and cell-adhesion molecules. Functional studies that block the most important recognition site for ∼50% of all known integrins (the RGD motif) significantly and specifically inhibit the extent of re-epithelialization in a human cell-culture model. These studies demonstrate that locomotion on fibronectin is important for this epithelial cell migration, akin to the motile activity of skin keratinocytes migrating to cover a wound surface (Luparello et al., 2020). Concomitant with re-epithelialization is rapid repair of transverse endometrial arterioles in the sub-epithelial stroma and in spiral arterioles, which are severely damaged with endometrial shedding (Garry et al., 2009; Ludwig and Spornitz, 1991). Vascular endothelial growth factor (VEGF) and other angiogenic factors in MF (Box 2) most likely stimulate this repair, along with dehydroandrostenedione (DHEA, of adrenal origin) that also promotes angiogenesis (Gibson et al., 2018; Liu et al., 2008).
Immune cells also play an important role in wound repair and clearly these are abundant at the sites of endometrial repair (Box 1; Fig. 1C). However, given that menstruation involves simultaneous breakdown and repair it is difficult to tease out the specific involvement of individual cell types (Maybin et al., 2018). The mouse models of menstruation have provided useful information; neutrophil depletion reduces endometrial repair in the mouse model of menstruation and repair (Kaitu'u-Lino et al., 2007) and CXCL4+ macrophages are also implicated (Maybin et al., 2018). Both may differentiate into repair-specific phenotypes within the tissue (Fig. 1C). It is likely therefore that at least these two immune cell types are important for endometrial repair in women. Although differences between the mouse model and human make interpretation of data difficult, a new spiny mouse model has been found to closely represent the human in its female reproductive physiology. As the spiny mouse undergoes spontaneous menstruation (Bellofiore et al., 2017), this now offers an alternative mouse model to better resolve the source of luminal epithelium during menstrual repair and the mechanisms involved in vivo.
Together, the speed of repair, the rapid migration of epithelial cells over the denuded matrix and the fact that the damaged surface is bathed in MF rich in migration factors (Box 2) supports the idea that the restoration of the endometrial luminal epithelium is akin to wound healing, although the lack of scarring is unique to the endometrium. The lack of scarring may be due to the microenvironment of MF (Box 2) and the preservation of the basalis (Fig. 1C,D). When the basalis is damaged or lost [such as in intrauterine adhesions (IUA) and Asherman's syndrome, respectively] scarring ensues because the myometrium, which may be equivalent to the dermis in skin, comes in direct contact with any repairing epithelium. This extraordinarily rapid response is needed to protect the surface from infection and limit bleeding, together with vasoconstriction. The regeneration of the endometrium and restoration of its cellular components is an entirely separate process that occurs once the surface is covered and E levels start to rise.
Endometrial regeneration after menses: the proliferative phase
As discussed above, by the time menses has ceased the ‘wound’ is essentially covered by a new luminal epithelium (Fig. 1C). During the next 10 or so days, endometrial thickness and the full cohort of cellular structures, including glands, stroma, vasculature and ECM (both the interstitial matrix and basal lamina), is regenerated through massive cellular proliferation in the functional layer as it regrows (Fig. 1D).
Hormone signaling
E is a crucial factor in endometrial regeneration; indeed, mice lacking aromatase (necessary for E synthesis) have overt hypoplastic uteri that are poorly organized and possess a paucity of glandular structures (Britt et al., 2001). Furthermore, the estrogen receptor (ER) α is present in endometrial epithelial and stromal cells during the proliferative phase of the cycle, and rising E from the ovarian follicles during that time plays a crucial role in endometrial regeneration and growth (Hewitt et al., 2016).
In addition to E, androgens, which retain their blood levels when E and P levels decline (Fig. 1A), can retard re-epithelialization but are needed for regeneration during the proliferative phase (Cousins et al., 2016a; Simitsidellis et al., 2018). Androgens contribute to epithelial proliferation and gland formation – at least in the mouse uterus (Simitsidellis et al., 2019). However, because androgen receptors are primarily detected within the basalis layer (Cousins et al., 2016a; Simitsidellis et al., 2019), androgens most likely act on initiating glandular regeneration. The action of these hormones through both genomic (ERα) and non-genomic pathways via plasma membrane G-protein coupled estrogen receptors (GPER) are topics of current interest (Gibson et al., 2020).
Stem/progenitor cells contribute to endometrial regeneration
The regenerative ability of the endometrium indicates an important role for somatic stem or progenitor cells in the regenerative process of both epithelial (glandular) and stromal compartments. Somatic stem cells are undifferentiated cells capable of maintaining their own pool through self-renewal mechanisms. They have high proliferative potential and differentiate into one or more lineages of more specialized tissue cells, which have limited potential for cell division (reviewed by Potten and Loeffler, 1990).
The concept that somatic stem/progenitor cells exist in the basal layer of the endometrium and are responsible for the cyclical growth of the endometrium was proposed as early as 1989 (Padykula et al., 1989). Since the first evidence for the presence of stem cells as rare colony-forming units (CFUs) in the adult human endometrium (Chan et al., 2004) was generated, these cells have been characterized and studied in the context of regeneration. Indeed, the contribution of human endometrial stem/progenitor cells to endometrial regeneration was clearly demonstrated when unfractionated single-cell suspensions of human endometrial epithelial and stromal cells, xenografted in mice, generated glands and stroma in vivo. These cells, which likely include niche cells (Box 3), respond to E and P to generate endometrial tissue, and upon withdrawal of these hormones they form blood filled cysts reminiscent of menses (Masuda et al., 2007).
Each stem/progenitor cell type resides within a stem cell niche, the specific regulatory environment that controls their fate and provides a functional unit of tissue regeneration. One emerging concept is that mesenchymal (stromal fibroblasts) cells can secrete components of the epithelial progenitor cell niche, including extracellular matrix (ECM) components, extracellular vesicles (mainly exosomes) and paracrine factors. Activated mesenchymal stem cells (MSCs) can increase both their own pool and contribute to the differentiation of other cells, either by direct differentiation or attracting supporting cells to the niche (Sagaradze et al., 2020). Specific stem cell niches in the endometrium must co-locate with the identified stem/progenitor cells. N-cadherin+ progenitors reside in the bases of the glands in the basalis (Nguyen et al., 2017) suggesting that myometrial smooth muscle cells and CD90lo basalis stromal fibroblasts (Schwab et al., 2008) could be niche cells. Endometrial MSCs are perivascular cells, and with a likely role in stromal vascular regeneration, neighboring endothelial cells may be their niche cells as they rapidly sense tissue damage and upregulate adhesion molecules and inflammatory markers to initiate angiogenesis and tissue regeneration (Gargett, 2007). Alternatively, other perivascular cells could have stem cell niche roles. TGFβ family members are bound to the ECM until released by specific proteases, and may also contribute to this niche because TGFβ receptor inhibition in vitro appears to be crucial in restraining endometrial MSC differentiation (Lucciola et al., 2020). Moreover, TGFβ family members are widely distributed in the human endometrium (Jones et al., 2006). Mixed cellular organoids (assembloids) (Rawlings et al., 2021 preprint) provide an excellent model for the study of the endometrial cell niche (Diniz-da-Costa et al., 2021).
Several small populations of stem/progenitor cells identified in adult human endometrium have classic stem cell properties, such as clonogenicity, in vitro self-renewal and differentiation (Gargett et al., 2009). Such cell types include epithelial progenitors, endometrial mesenchymal stem cells (eMSCs) and side-population (SP) cells (reviewed by Bozorgmehr et al., 2020; Gargett et al., 2016; Masuda et al., 2015; Santamaria et al., 2018) (Table 1; Fig. 1D,E). Somatic stem/progenitor cells occupy specific niches (Box 3) and are resistant to damage signals. For example, endometrial stem/progenitor cells (see below) reside predominantly in the basal layer, which remains after menstruation or following parturition, although mesenchymal stem/progenitor cells (MSCs) are also present in the functional layer (Fig. 1D,E). To determine the role of endometrial stem/progenitor cells in endometrial regeneration, specific markers are required that enrich for these rare cells and show their functional properties (Gargett, 2007). Such markers also show the location and niche of the stem/progenitor cells, providing clues to their role in tissue regeneration (reviewed by Cousins et al., 2021).
Transcriptomic mapping at single-cell resolution has further extended our understanding of time-dependent transcriptomic signatures of each cell lineage in the functional endometrium, but has not yet included examination of the basalis layer from which regeneration is initiated (Wang et al., 2020). Recently, single-cell sequencing and Visium spatial transcriptomics of partial endometrial biopsies or the full thickness endometrium has defined the cell states of luminal, glandular and basalis epithelial cell lineages (Garcia-Alonso et al., 2021 preprint). This has provided new information on the role of SOX9+ epithelial cells, previously identified in the endometrial luminal and basalis epithelium (Fig. 1D; Table 1) (Valentijn et al., 2013). In this study, SOX9-expressing epithelial cells were widely distributed in proliferative-stage regenerating endometrium, with glandular cells in the functionalis possessing a cell cycling profile, indicating their role in the rapidly regenerating glandular epithelium (Garcia-Alonso et al., 2021 preprint). In contrast, SOX9-expressing cells of the recently repaired luminal epithelium were not cycling and also expressed the intestinal epithelial stem cell marker LGR5. In the basalis, SOX9-expressing epithelial cells are LGR5− and lack a cell cycling profile, in keeping with the quiescent state of basalis epithelial cells (Nguyen et al., 2017). However, this study did not identify N-cadherin-expressing epithelial cells in full-thickness endometrium samples, perhaps due to their very low abundance and/or their location deep in the bases of branched glands adjacent to myometrium.
Clinical observations in bone marrow (BM) transplant patients suggests that BM-derived stem cells (CD45+ cells) transdifferentiate into both epithelial and stromal endometrial cells and contribute to endometrial regeneration (Du and Taylor, 2007; Ikoma et al., 2009; Taylor, 2004). However, more recent data from a GFP-mouse BM-transplant model contradicts this concept, indicating that previously identified BM cells have been misidentified, probably due to suboptimal staining and imaging of the CD45+ macrophages (Ong et al., 2018). This paper has stimulated considerable debate, offering alternative interpretations of existing results, and consensus has not yet been achieved (Cervello et al., 2012; Deane et al., 2019; Santamaria et al., 2019).
Glandular regeneration
In the 1980s, the Gerschenson laboratory (Conti et al., 1984) showed in rabbits that E exerts its influence on quiescent cells in the basal part of the glands to stimulate their proliferation and migration towards the lumen. It is likely that E acts indirectly on the epithelial cells via stromal ERα, as shown by tissue recombinants of mouse uterine mesoderm and epithelium in various combinations from ERα knockout and wild-type mice (Cooke et al., 1997). More recently, in mice, a single-cell lineage-tracing pulse-chase study using a Cre-loxP-Keratin19 system has identified bipotent endometrial epithelial stem cells in the intersection zone of the luminal and glandular epithelium, defining the murine epithelial stem cell niche (Box 3). As Foxa2 is a specific marker of glandular cells, but not luminal epithelium, the founder cells of these mixed clones that generate both the EpCAM+Foxa2+ glandular and EpCAM+Foxa2− luminal epithelial cell types demonstrate their stem cell function in regenerating both cell types during physiological estrous cycling (Table 1) (Jin, 2019). Other genetic lineage-tracing studies of the uterus have identified Axin2-expressing cells deep in the glands of adult mice (Syed et al., 2020) and Lgr5+ epithelial cells in the tips of developing endometrial glands of neonatal mice (Seishima et al., 2019), as the glandular epithelial stem/progenitor cell populations (Table 1). More research is needed to understand the relative roles of these epithelial populations and their relative contributions to murine endometrial regeneration.
Human studies have also identified specific markers of endometrial basalis epithelial cells. The WNT signaling pathway regulator AXIN2 has been identified in a study to distinguish basalis and functionalis epithelial cell gene profiles (Nguyen et al., 2012), and SSEA-1 has been identified using a candidate marker approach (Table 1; Fig. 1D,E) (Valentijn et al., 2013). SSEA-1+ epithelial cells co-express SOX9 and have some properties of adult stem cells, such as longer telomeres, but it is unknown whether they are clonogenic. The first marker of human endometrial epithelial cells with stem/progenitor cell function is N-cadherin, which was identified using an unbiased gene microarray approach to reveal surface markers of basalis epithelium (Nguyen et al., 2017). In this study, N-cadherin+ cells were enriched for clonogenic epithelial cells compared with N-cadherin− cells, and underwent self-renewal in a serial cloning assay, with up to 29 population doublings. Furthermore, N-cadherin+ cells differentiated into large cytokeratin+ gland-like structures in 3D organoid cultures. In full-thickness endometrium from both pre- and post-menopausal women, N-cadherin+ immunolocalizes to the bases of basalis glands adjacent to the myometrium, identifying their in vivo niche (Box 3). An endometrial epithelial cell hierarchy has also been identified, with N-cadherin+SSEA-1−SOX9− as the most primitive clonogenic epithelial cells in the gland bases, which generate a very small population of N-cadherin+SSEA-1+SOX9+, followed by N-cadherin−SSEA-1+SOX9+ basalis epithelial cells that extend just beyond the basalis-functionalis junction (Table 1; Fig. 1D,E) (Cousins et al., 2021; Nguyen et al., 2017). Proximal to the SSEA-1+SOX9+ cells are N-cadherin−SSEA-1−SOX9− glandular epithelial cells, which extend to the N-cadherin−SSEA-1+SOX9+ luminal epithelium (Cousins et al., 2021; Nguyen et al., 2017; Valentijn et al., 2013). These N-cadherin−SSEA-1−SOX9− cells proliferate and differentiate in the proliferative and secretory phase, respectively. More recently, an epithelial stem cell lineage marker, aldehyde dehydrogenase 1 (ALDH1) isoform (ALDH1A1) has been shown to immunocolocalize with 78% of N-cadherin+ epithelial cells in the deep endometrial basalis glands (Ma et al., 2020), and their location indicates that they are unlikely to colocalize with SSEA-1+SOX9+ basalis epithelial cells. As ALDH1 catalyzes the production of retinoic acid, signaling via the retinoic acid pathway may be important in clonogenic N-cadherin+ cell function (Cousins et al., 2021). The role of these cells of this basalis epithelial cell hierarchy in regenerating endometrium has not yet been established.
Stromal and vascular regeneration
Regeneration of the stromal vascular component of the endometrium is likely to be mediated by an eMSC population. Use of the term MSCs has been criticized recently owing to poor characterization of the cells from many tissue sources and confusion between their stem cell and immunomodulatory functions (Sipp et al., 2018). The International Society for Cellular Therapies (ISCT) has defined the properties of MSC as plastic adherent, capable of in vitro differentiation into mesodermal lineages (adipogenic, osteogenic and chondrogenic) and expression of a distinct surface marker phenotype (CD73, CD90, CD105 and lacking CD14, CD34, CD45) (Dominici et al., 2006). However, it has become apparent that stromal fibroblasts also have these same ISCT properties (although they lack clonogenicity and self renewal). Therefore, the ISCT definition of MSC is now inadequate (reviewed by Bianco et al., 2013; Bozorgmehr et al., 2020; Gargett et al., 2016). In the present Review, MSC and fibroblasts are clearly defined.
MSC markers have been identified in human and mouse endometrium by their stem cell function using candidate markers and lineage-tracing approaches, respectively. Initially, clonogenic, highly proliferative endometrial stromal cells have been shown to be enriched in a small population of cells co-expressing PDGFRβ and CD146, which demonstrates typical ISCT MSC properties. CD146+PDGFRβ+ cells (i.e. eMSC) are located around blood vessels in close proximity to the endothelial cells (CD146+PDGFRβ−), suggesting a pericyte identity (Table 1; Fig. 1E) (Schwab and Gargett, 2007). A perivascular cell identity for human eMSC has been confirmed by expression of a single perivascular marker, SUSD2 (W5C5 clone), which has been used to purify highly proliferative, clonogenic cells with ISCT properties, located around vessels in both the basalis and functionalis layers (Table 1; Fig. 1E) (Masuda et al., 2012). These clonogenic SUSD2+ cells with MSC properties can regenerate endometrial stroma in vivo in a xenograft assay. PDGFRβ+CD146+ eMSC and PDGFRβ+CD146− endometrial fibroblasts have distinct gene expression profiles (Spitzer et al., 2012), indicating that they are distinct cell types, even though both fulfil the ISCT MSC criteria. Gene signature comparisons between freshly isolated and cultured human PDGFRβ+CD146+ eMSC versus PDGFRβ+CD146− endometrial fibroblasts has shown that the eMSC rapidly lose most of their gene expression signature during culture, changing to the fibroblast signature. This transition suggests that eMSC differentiate into fibroblasts (Barragan et al., 2016), indicating that perivascular eMSC are the stem/progenitor cells that differentiate into stromal fibroblasts in vivo (Fig. 1E). However, lineage tracing in human models is not possible to prove this hypothesis. In vitro, both SUSD2+ perivascular MSC and SUSD2− endometrial fibroblasts differentiate into cells expressing decidual transcription factors and markers, and become key cells of the maternal component of the placenta, although their gene and secreted cytokine profiles differ (Diniz-da-Costa et al., 2021; Murakami et al., 2014). CD34 has also been identified as a perivascular marker of larger vessels in human endometrium, with an adventitial rather than pericyte/smooth muscle cell (medial) location in blood vessels; however, these cultured cells have limited potential to regenerate human endometrium (Zhu et al., 2021).
Single-cell sequencing of human endometrial biopsies has confirmed these biological findings, identifying a small smooth muscle cell population. The majority of this population, which includes SUSD2+ cells, express MCAM (CD146), PDGFRB and a perivascular gene profile (Wang et al., 2020). Similarly, single-cell RNA-sequencing of the Pdgfrb-BAC-eGFP reporter mouse endometrium has identified a small perivascular population of Pdgfrb+MCAM+ cells with a pericyte/smooth muscle gene profile and perivascular niche in vivo, distinct from three novel endometrial fibroblast subpopulations (Kirkwood et al., 2021). Similarly, single-cell profiling is uncovering heterogeneity in the fibroblast populations between mural (perivascular) and fibroblast subtypes in other organs (Muhl et al., 2020).
Endometrial regeneration after parturition
Menstruation and parturition share commonly evolved physiological processes dependent on progesterone withdrawal (Thomas, 2019) resulting in inflammatory-like processes including leukocyte infiltration, release of degradative enzymes, vasoconstriction and breakdown of the vasculature. There is, however, a paucity of published data relating specifically to postpartum uterine involution following parturition. Ultrasonography indicates that the endometrium averages at 10.5 mm thick 24 h following a normal delivery, with an upper limit of 25 mm, and gradually reduces to 3.8 mm by 42 days postpartum (reviewed by Ucci et al., 2021). Because of the size of the wound, re-epithelialization is slower than during menstruation; coverage of the non-placental site of the endometrium is achieved by day 7 postpartum and the placental site by day 14 (Sharman, 1953). When assessed by ultrasound, a normal healing process appears to occur from days 7-17 along with an abundant shedding of lochia. From days 26-56, the endometrium appears to be inactive owing to low circulating E levels postpartum, presumably resembling basalis endometrium (Mulic-Lutvica et al., 2001). However, the molecular mechanisms that regulate such changes remain unknown. In postpartum women and rhesus monkeys, endometrial regeneration after parturition appears to arise from stem/progenitor cells in the basalis layer (Maruyama et al., 2010; Padykula et al., 1989). Based on their location, these cells may be the clonogenic epithelial progenitors that reside deep in the basalis glands or SSEA-1+ basalis epithelial cells (Nguyen et al., 2017) or the SUSD2+ or CD34+ eMSC in the basalis vessels (see above; Table 1). In recent mouse lineage-tracing models of pregnancy and parturition, Axin2+ epithelial cells deep in the bases of the glands are responsible for the eventual regeneration of the glands but they do not appear to have a role in regenerating the luminal epithelium (Syed et al., 2020). Foxa2+ glandular epithelial cells also do not contribute to the luminal epithelium (Jin, 2019), and so the epithelial stem/progenitor cells that generate the luminal epithelium after parturition remain unknown. There are currently no studies clearly evaluating the contribution of non-immune BM-derived cells on regeneration of mouse endometrium following parturition (Spooner et al., 2021).
In double transgenic mice in which uterine cells express Amhr2 (a mesenchymal marker) and yellow fluorescent protein (EYFP), cells of mesenchymal origin are clearly identified in epithelium following parturition (Huang et al., 2012; Patterson and Pru, 2013). Indeed, following parturition, the epithelial compartment is repopulated from both the original epithelium and the EYFP-labeled mesenchymal cells (Huang et al., 2012). Similarly, 72 h postpartum, cells of mesenchymal origin can be identified in the stroma, as well as in luminal and glandular epithelial compartments, suggesting that MET has occurred (Patterson and Pru, 2013). Long-term fate-mapping techniques have indicated that once MET has occurred, it persists during subsequent homeostasis (Spooner et al., 2021). In gestational mice, non-proliferating label-retaining stromal cells (stem cells) are detected predominantly at inter-implantation/placental loci, and many of these undergo proliferation immediately postpartum, transiently expressing total and active β-catenin, and returning to their quiescent state after postpartum repair (Cao et al., 2014; Spooner et al., 2021). Whether similar cellular events occur in women postpartum remains to be determined.
Postpartum endometrial regeneration is a prolonged and more extensive process than the regeneration that occurs following regular menstrual cycles due to the extensive remodeling required once the placenta has been delivered. This process may be regulated by the low E environment associated with postpartum and lactation. Low E also makes postpartum regeneration a vulnerable state for the endometrium, because any damage to the remaining basalis from postpartum infection could result in IUA and loss of the stem/progenitor cells, which could hinder future regeneration once cycling is re-established.
Regeneration of atrophic postmenopausal endometrium
Following menopause, the endometrial functionalis is no longer present due to the lack of circulating E. The remaining thin, atrophic endometrium comprises luminal epithelium. A few inactive glands and stroma can be stimulated to regenerate full thickness endometrium if women take E-only hormone replacement therapy for 6-8 weeks (Ulrich et al., 2014b). This observation suggests that endometrial stem/progenitor cells remain in a dormant state until E levels rise. Both N-cadherin+ and SSEA-1+ epithelial cells have been identified in atrophic post-menopausal endometrium and in regenerated endometrium of E-treated women, and show the same pattern as in pre-menopausal endometrium (Fig. 1D) (Nguyen et al., 2017). They also immunocolocalized with ERα. Nuclear AXIN2 was immunolocalized in post-menopausal and in basalis epithelium, but not in the functionalis of pre-menopausal women (Nguyen et al., 2012). As an important negative regulator of the WNT pathway, AXIN2 may have a key role in compartmentalizing endometrial epithelial regeneration in premenopausal endometrium (Gargett et al., 2012).
Clonogenic self-renewing SUSD2+ eMSC with classic ISCT properties are also present in postmenopausal endometrium, whether atrophic or from women treated with E (Ulrich et al., 2014b). SUSD2+ eMSC have been found in a perivascular location in atrophic and regenerated endometrium, despite not expressing ERα, and are capable of proliferation. This suggests that SUSD2+ eMSC survive E depletion and can proliferate in response to exogenous E via niche cells (endothelial or other perivascular cells) to regenerate the endometrial vascular stroma of postmenopausal women.
Emerging roles for stem cell-derived extracellular vesicles
In addition to their provision of cells needed in tissue regeneration, stem/progenitor cells also release a range of factors assumed to mediate their paracrine effects that influence other cellular components of the stem cell niche (Box 3) (Liang et al., 2014; Timmers et al., 2007). Increasing evidence supports that stem/progenitor cells shed extracellular vesicles (EVs) that may be important in regeneration by providing cell-cell communication.
EVs are tiny particles released from cells into the extracellular environment. They have a range of sizes and include exosomes (∼40-200 nm diameter), microvesicles (∼100-2000 nm), apoptotic bodies (200-4000 nm) (Xu et al., 2016) (Fig. 2A) and the most recently described exomeres that have a diameter of only 35 nm (Zhang et al., 2018), approaching the smallest possible diameter of 10-20 nm representative of phospholipid membranes (Huang et al., 2017) (Fig. 2A). The double-layer phospholipid membranes of all EVs have external cell-binding properties determined by the cell of origin. EVs contain ‘cargo’, comprising nucleic acids, lipids and proteins that are selectively packaged. For exosomes, synthesis occurs within the endosomal pathway, whereas microvesicles are formed by the outward budding and fission of plasma membrane lipid rafts or microdomains (Fig. 2A). Importantly, the cargo contained within EVs is protected from extracellular degradation by their membranes. EVs can traffic to local or distant target cells, to which they bind with high specificity and are then mostly internalized, to release their contents intracellularly and thus regulate the function of the recipient cells.
Extracellular vesicles. (A) Extracellular vesicles can be separated into three types: apoptotic bodies, microvesicles (large EVs) and exosomes (small EVs), which differ in size, origin and expression of surface and intracellular proteins and lipids. Adapted by D. Greening from Xu et al. (2016) . (B) Small EVs derived from stem/progenitor cells can induce a range of functions in their target cells: those shown are from a number of studies in different cell types and tissues. Functions are not yet defined for small EVs of stem/progenitor cell origin in the endometrium.
Extracellular vesicles. (A) Extracellular vesicles can be separated into three types: apoptotic bodies, microvesicles (large EVs) and exosomes (small EVs), which differ in size, origin and expression of surface and intracellular proteins and lipids. Adapted by D. Greening from Xu et al. (2016) . (B) Small EVs derived from stem/progenitor cells can induce a range of functions in their target cells: those shown are from a number of studies in different cell types and tissues. Functions are not yet defined for small EVs of stem/progenitor cell origin in the endometrium.
Actions of MSC-derived EVs
EVs from MSCs have anti-inflammatory, anti-apoptotic, pro-angiogenic and immunomodulatory effects similar to those observed in MSCs in various disease models (Fig. 2B) (reviewed by Zhang et al., 2020). Although exosomes from eMSC have not yet been reported, there is evidence that SUSD2+ eMSCs have the potential to produce exosomes; their gene expression profile showed over 100 exosomal marker genes, including 81 known exosomal cargo miRNAs, and many endosomal biogenesis and transport genes (Gurung et al., 2018). Pre-clinical models also indicate that EVs from various stem and progenitor cell models, when injected at the required site, can recapitulate the therapeutic effects of these cells (Balbi and Vassalli, 2020). For example, MSC-secreted exosomes reduce myocardial ischemia-reperfusion injury (Lai et al., 2010).
MSC-derived EVs can promote both collagen synthesis (Shabbir et al., 2015; Zhang et al., 2015) and angiogenesis (Shabbir et al., 2015) in cutaneous wound healing scenarios. Specifically, administration of MSC-exosomes to in vitro wound healing models accelerates re-epithelialization, reduces scar widths and promotes collagen maturity (Zhang et al., 2015). MSC-exosomes also stimulate proliferation, migration and tube formation of human umbilical vein endothelial cells. Several signaling pathways activated by the exosomes are important in wound healing, including AKT, ERK and STAT3. In addition, exosomes induce the production of growth factors, including hepatocyte growth factor (HGF), insulin-like growth factor (IGF1), nerve growth factor (NGF) and stromal-derived growth factor-1 (SDF1) (Shabbir et al., 2015). Similar effects on angiogenesis, both in vivo and in vitro, have also been shown for EVs from other sources (Bian et al., 2014; Kuriyama et al., 2020).
Mouse BM MSCs pre-treated with neonatal serum have enhanced wound healing activity (Qiu et al., 2020), suggesting that the source and in vivo environment of the exosome-secreting cells is likely to be important in determining exosome functionality. Importantly, the proteomes of exosomes released by human endometrial epithelial cells are altered according to hormonal treatment of the parent cells; estrogen (E-exosomes) or estrogen plus progesterone (EP-exosomes) representing the non-receptive proliferative phase or the receptive secretory phase of the cycle, respectively, have substantially different proteomes (Greening et al., 2016). Subsequently, trophectodermal cell uptake of the EP-exosomes (compared with E-exosomes) enhances expression of the trophectodermal adhesion markers phosphorylated focal adhesion kinase, fibronectin and EpCAM by trophoblasts, altering their adhesive capacity in functional assays. Given that menstrual stem/stromal cells (predominantly endometrial fibroblasts) and eMSCs are present in MF (Box 2), it will be useful to determine the proteome/full cargo of their EVs, which could well contribute to endometrial regeneration.
Immunomodulatory properties of EVs
A growing body of evidence indicates that EVs derived from stem/progenitor cells also regulate immune function (Xie et al., 2020). During menstrual cycling the endometrium contains variable numbers of immune cells – uterine natural killer (uNK) and monocyte-derived macrophages in particular (Box 1). Therefore, the role of EVs in immune regulation is relevant to understanding endometrial regeneration. In other systems, stem/progenitor cells and immune cells communicate with each other via EVs (Di Trapani et al., 2016; Silva et al., 2017). A feedback mechanism appears to exist, with inflammation-stimulated (licensed) MSC-EVs showing greater release of anti-inflammatory cytokines (Fig. 2B). In addition, there is selective and context-specific uptake of various EVs by activated leukocyte subpopulations, compared with naïve MSC-EVs (Harting et al., 2018). These anti-inflammatory properties are executed via the COX2/PGE2 pathway. Furthermore, MSC-EV-mediated immune-suppressive effects on NK cells are possibly mediated via membrane-bound TGFβ mediating downstream TGFβ/Smad2/3 signaling and hence suppression of NK cell cytotoxicity (Fan et al., 2019). MSC-EVs can also act on macrophages by polarizing the pro-inflammatory M1 to the anti-inflammatory M2 phenotype (Xie et al., 2020). Given that eMSC and MenSC have immunomodulatory properties in vitro and in vivo (Mukherjee et al., 2019; Shokri et al., 2019; Ulrich et al., 2014a; Yang et al., 2019), it can be postulated that eMSC-EVs can similarly interact with uNK cells and macrophages. Such interactions would inhibit immune cell production of cytokines and promote the immunotolerant environment important for regenerating endometrium.
There are many outstanding questions in the EV field relevant to endometrial regeneration (Margolis and Sadovsky, 2019). For example, does the diversity of EV size affect their biological functions (Martinez-Greene et al., 2021)? What EV-specific recognition code determines the target cells of the EVs in the endometrium? By what mechanisms is the bioactive EV cargo released within the target cells or does its binding alone stimulate intracellular events? Specific to endometrial stem/progenitor cells, are their EVs alone responsible for establishing their individual niches, or are soluble secreted factors also required? Given the rapid expansion of the EV field, it may not be long before many of these questions can be answered.
In vitro systems to model the human endometrium
Apart from ultrasound imaging, it is not possible to study normal human endometrium in vivo, including for regeneration studies across the proliferative phase of the menstrual cycle. Two-dimensional (2D) cell culture studies of individual cell types, either cell lines or cells derived from biopsy material, have provided some insight. The emphasis now is on experimentation using more physiological models, such as slices of endometrium in culture. Importantly, 3D culture of cells as organoids is more representative of their in vivo environment. Recent studies including more than one cell type within the organoid will provide more accurate representation of the cell-cell interactions within this 3D structure.
Slice culture system
A 3D human endometrial slice culture system comprised of full-thickness endometrial tissues embedded in collagen maintains cellular viability, tissue architecture and responds to E and P over 3 weeks (Muruganandan et al., 2020). This offers a novel system for analyzing stem/progenitor cell function in vitro. Examination of hormone action on the individual cells of the epithelial hierarchy using hysterectomy tissue slices including N-cadherin+SSEA-1−, N-cadherin−SSEA-1+ and AXIN2+ basalis epithelial cell proliferation in both pre- and postmenopausal endometrium will likely reveal the cellular kinetics of these putative stem/progenitor cells and their interrelationships. They also offer the study of the horizontal glandular structures that comprise the basalis epithelium (Tempest et al., 2020; Yamaguchi et al., 2021), how glands might sprout from them and which epithelial progenitor cell types of the putative hierarchy are involved.
Endometrial epithelial organoids
Organoids are defined as 3D in vitro tissue models that recapitulate many of the physiologically relevant properties and features of the in vivo tissue, in a manner lost in 2D cell cultures. Human endometrial epithelial organoids have been generated from primary endometrial cells derived from enzymatically dissociated endometrial tissue biopsies (Boretto et al., 2017; Fitzgerald et al., 2019; Turco et al., 2017) or MF (Cindrova-Davies et al., 2021) and grown in Matrigel domes in chemically defined medium that promotes epithelial self-organization into hollow spherical structures: organoids (Fig. 3). Such organoids maintain their genetic and phenotypic features throughout extensive passaging (Turco et al., 2017) and can be cryopreserved and biobanked for subsequent use. Endometrial organoids comprise gland-like structures that maintain secretory characteristics of their donor endometrium (Fig. 3). They include a number of different epithelial cell types (shown by single-cell RNA sequencing; Fitzgerald et al., 2019; Garcia-Alonso et al., 2021 preprint), such as those with characteristic secretory capacity and ciliated cells, both found in luminal epithelium and glands in the functionalis. Importantly, the epithelial cell polarity that occurs in vivo is retained, demonstrated by distinct compositions of intraorganoid and extraorganoid fluid (Simintiras et al., 2021). Such organoids are responsive to E and P, which induce effects reflective of the proliferative and secretory phases of the cycle (Boretto et al., 2017; Turco et al., 2017) and induce the development of ciliated cells (Haider et al., 2019). Epithelial cells derived from organoids can be grown into monolayers that retain characteristics of fresh primary cells, which otherwise can only be cultured briefly. Thus, organoids provide important new platforms for study of human endometrium including glandular regeneration. However, to achieve their full potential, organoids need to include endometrial stromal cells, as described recently (Murphy et al., 2019). In addition, cells of the vasculature, including and possibly generated from the eMSC, should be juxtaposed to the glandular structures and luminal epithelium. Co-culture with cells of the immune system as would be present in vivo at the cycle stage of interest (Box 1) would help to uncover the role of the immune cells in endometrial function. Indeed, recently described human endometrial ‘assembloids’ consisting of gland organoids and primary stromal cells that have been stimulated to decidualize, closely resemble mid-secretory phase endometrium (Rawlings et al., 2021 preprint). Such complex in vitro structures will enable new insights into endometrial regenerative processes. In the meantime, organoids, with their potential for biobanking, present opportunities for drug evaluation and screening for diseases that originate from endometrial epithelium (Boretto et al., 2019).
Endometrial organoids. Endometrial organoids, derived from primary endometrial epithelium, form gland-like structures in culture. These structures increase in number by budding off from each other in culture. Image taken from Cindrova-Davies et al. (2021). The epithelial cells, identified as ciliated and non-ciliated, take up a polarized phenotype with identifiable apical and basal surfaces: secretions into the glands can be seen. These cells positively immunostain for cytokeratin and estrogen receptor (ER) α, and progesterone receptor (PR) is inducible as it is in vivo. Co-culture of epithelial organoids with other cells of the endometrial functionalis will improve the in vitro representation of this complex environment.
Endometrial organoids. Endometrial organoids, derived from primary endometrial epithelium, form gland-like structures in culture. These structures increase in number by budding off from each other in culture. Image taken from Cindrova-Davies et al. (2021). The epithelial cells, identified as ciliated and non-ciliated, take up a polarized phenotype with identifiable apical and basal surfaces: secretions into the glands can be seen. These cells positively immunostain for cytokeratin and estrogen receptor (ER) α, and progesterone receptor (PR) is inducible as it is in vivo. Co-culture of epithelial organoids with other cells of the endometrial functionalis will improve the in vitro representation of this complex environment.
Conclusions and future perspectives
Accumulated knowledge of endometrial scar-free tissue repair and regeneration already provides potential for clinical applications. These can be enabled by non-invasive harvesting of MF-containing repair factors, EVs and stem/progenitor cells (Masuda et al., 2021), which provide a rich complex milieu that will support repair and regeneration (Box 2). In addition to the wound repairing properties of MF (Evans et al., 2019), platelet-rich plasma (from peripheral blood) enhances cellular processes important for tissue regeneration in both endometrial stromal fibroblasts and eMSC (Aghajanova et al., 2018). The bioactive components of such fluids may prove useful to promote endometrial repair and regeneration in clinical situations in which endometrial growth is compromised or in situations where scarring occurs, such as IUA and Asherman's syndrome, where there is a failure of adequate repair and regeneration.
Although MSC from a variety of sources are now being used (Tang et al., 2018), there are currently limitations. MSC spontaneously differentiate and undergo cellular senescence (indicated by telomere shortening) during expansion in vitro. Importantly, recent studies specific to eMSC expansion demonstrate that inhibition of the TGFβ receptor-signaling pathway by the pharmaceutical agent A83-01 restricts apoptosis and senescence and safeguards the functional properties of eMSCs. These include upregulation of pro-angiogenesis, anti-fibrosis and multipotency genes, as well as secretion of angiogenic and immunomodulatory factors (Gurung et al., 2018). Induction of retinoic acid target genes (Lucciola et al., 2020) also offers new opportunities for clinical translation. The feasibility of non-invasive harvest of autologous stem/progenitor cells from a woman's MF in considerable numbers offers a further advantage (Bozorgmehr et al., 2020; Masuda et al., 2021) and organoids can now be derived from these cells (Cindrova-Davies et al., 2021) for further study. However, the common limitations of MSC therapy, such as the low retention of transplanted cells (particularly when given intravenously), tumor growth (although this risk is negligible compared with pluripotent stem cell derivatives), off-target migration and short life in storage, remain challenges. Possible solutions will include direct delivery via targeted exosomes or fabrication of cell-mimicking microparticles.
There is much to be learned from the molecular and cellular processes underpinning the dynamic shedding, repair and regeneration of the endometrium. The processes involved include highly controlled inflammation, leukocyte trafficking, tissue breakdown, re-epithelialization and regeneration. Although steroid hormones, particularly E and P and, to a lesser extent, androgens, are the primary determinants of endometrial cyclicity, just how the events are locally controlled is still far from understood. New developments, such as the role of stem/progenitor cell EVs and knowledge gained using organoid culture, should be pursued to answer these questions. Although the cellular sources for regeneration are now largely defined, the molecular mechanisms that determine the fine structure of this dynamic tissue that is essential for reproduction remain largely unknown. New tools, such as organoid cultures, single-cell RNA-sequencing and chromatin landscape evaluation, will drive studies to uncover these molecular mechanisms.
Acknowledgements
Footnotes
Funding
Research at the Hudson Institute of Medical Research is funded in part by State Government of Victoria infrastructure support program. The authors acknowledge grants from the Australian National Health and Medical Research Council to C.E.G. (Investigator Fellowship #1173882) and L.A.S. (Project grant #1139489), and the provision of a Department of Education, Skills and Employment, Australian Government Research Training Program (RTP) scholarship to J.C.H.
References
Competing interests
The authors declare no competing or financial interests.