The human endometrium undergoes ~450 cycles of proliferation, differentiation, breakdown, shedding, and repair across a woman's reproductive lifespan. The endometrium is composed of two layers. The basalis is adjacent to the muscular myometrium and is not shed during menstruation, and it is from this layer that the upper layer of the endometrium, the functionalis, arises during each menstrual cycle (1). The functionalis undergoes the most structural changes throughout the menstrual cycle in response to ovarian-derived steroids 17β estradiol and progesterone (2).

The estradiol-dominant proliferative phase begins on approximately day 4 of an average cycle (3), stimulating proliferation of the glandular epithelium, the vasculature and stroma. Estradiol primes the endometrium for the structural changes that it will undergo during the secretory phase by inducing estrogen-dependent expression of the progesterone receptor (4). During the secretory phase, progesterone is secreted by the corpus luteum following ovulation. Epithelial cell proliferation decreases, stromal cells undergo cellular enlargement to become pre-decidual cells. During the mid-secretory phase decidualization of pre-decidual cells occurs under the luminal epithelium and around spiral arterioles. By the late secretory phase, the decidua is infiltrated by T cells, uterine natural killer cells and macrophages (5).

In the absence of an implanted blastocyst, the corpus luteum regresses resulting in a rapid decrease in ovarian-derived steroid production. Progesterone withdrawal initiates menstruation, a cascade of events that results in the piecemeal shedding (6) of the functionalis and expulsion of tissue via the vagina. Whilst outward bleeding may last for up to 5 days in some women, repair processes have been initiated from day 1. Scanning electron microscopy (SEM) studies show that re-epithelialization of the endometrium occurs within 48 h and in a piecemeal fashion (7). The endometrium is unique in that it displays unparalleled tissue remodeling following menstruation, resulting in a scar-free tissue (6). Furthermore, this process occurs in a steroid hormone-depleted environment, as evidenced in animal models of endometrial repair (8, 9).

Re-epithelialization of the endometrium is thought to occur by two mechanisms. The first was proposed by Novak and Te Linde in 1924, who suggested the new luminal epithelium arises from the residual basalis glands (10). SEM studies of menstrual phase endometrium show epithelial extensions attached to basal glands (7). The second mechanism is that of mesenchymal to epithelial transition (MET) where stromal cells in the basalis undergo cellular transformation to become new luminal epithelial cells. Three studies have reported low epithelial cell proliferation during post-menstrual repair, along with isolated epithelial cells on the surface of the endometrium, unassociated with the glandular epithelium (11–13). The role of MET has also been investigated in mouse models of menstruation/post-partum repair, which would indicate that the stroma does contribute in some part to the luminal epithelium (14–16) but it is likely the glandular epithelium is the main contributor to the new luminal surface.

Regeneration of the endometrium following repair is an estrogen-dependent process, whereby the endometrium grows from a post-menstrual depth of 0.5 to 7–8 mm during the mid-proliferative phase (17). This highly regenerative capacity is likely driven by stem/progenitor cell populations that reside in the basalis.

In this review we will focus on stem/progenitor populations that are likely involved in menstruation, repair, and early regeneration of the tissue as well as those populations which may contribute to menstrual/regeneration disorders such as endometriosis and Asherman's syndrome.

Human endometrial epithelial progenitors were first identified as rare clonogenic cells comprising 0.22% of the epithelial cell adhesion molecule positive (EpCAM⁺) epithelial cell population from hysterectomy tissue which includes the basalis (21). Subsequently, the stem cell properties of self-renewal, high proliferative potential, and differentiation into large gland like structures in 3D cultures were demonstrated in vitro for individual large endometrial clonogenic epithelial cells (22). Specific markers of basalis epithelial cells were then identified; AXIN2 (23, 24), SSEA-1 and nuclear SOX9 (nSOX9) (25). The first specific surface marker enriching for clonogenic epithelial cells, N-cadherin encoded by CDH2, was identified using an unbiased gene profiling approach comparing EpCAM⁺ endometrial epithelial cells from pre- and post-menopausal women (26). A potential epithelial hierarchy was also identified, based on the location (niche) of the N-cadherin⁺ cells in the bases of the branching glands in the basalis adjacent to the myometrium. N-cadherin⁺ SSEA-1⁺ nSOX9⁺ epithelial cells were proximal to N-cadherin⁺SSEA-1⁻ cells and N-cadherin⁻ SSEA-1⁺ nSOX9⁺ (Figure 1) more proximal again to occupy an ill-defined functionalis-basalis junction. The majority of the functionalis comprised epithelial cells negative for N-cadherin, SSEA-1 and nSOX9. However, the luminal epithelium is SSEA-1⁺ nSOX9⁺, most likely due to rapid re-epithelializing of the raw surface by these cells migrating from the remaining gland stumps during menstruation (Figure 1), indicating their role in endometrial repair (27, 28). The ALDH1A1 isoform of ALDH co-localizes with 78% of N-cadherin⁺ epithelial cells by immunofluorescence and confocal microscopy (29), suggesting a role for retinoic acid signaling in the progenitor function of N-cadherin⁺ epithelial cells. EpCAM⁺ N-cadherin⁺, EpCAM⁺ SSEA-1⁺ and the very rare EpCAM⁺ N-cadherin⁺ SSEA-1⁺ epithelial cells have been detected and quantified in menstrual fluid (30, 31), indicating that small populations of these cells are shed during menstruation (see section Role of Stem/Progenitor Cells in Menstrual Disorders/Regeneration Disorders).

Atrophic post-menopausal endometrial epithelium also contains N-cadherin⁺ epithelial cells in the bases of atrophic glands adjacent to the myometrium (26). A similar endometrial epithelial hierarchy has been identified in post-menopausal women taking estrogen replacement therapy in a fully regenerated endometrium with a basalis and functionalis. Atrophic post-menopausal endometrial epithelium also contains nuclear AXIN2⁺ epithelial cells (23).

New information on endometrial cell lineages using next generation sequencing technologies at the single cell level is rapidly being generated. Since most studies have been undertaken on endometrial biopsies (32–34), gene expression signatures for basalis epithelial cells have not always been available. However, a Visium spatial transcriptomics study of several cadaveric full thickness uterine tissues has captured signatures of luminal, glandular, and basalis epithelium and revealed that SOX9-expressing epithelial cells with a cell-cycling profile are widely distributed in proliferative-stage endometrium (34). Although, this Visium spatial analysis was unable to determine if SOX9 was expressed in the nucleus or cytoplasm, it is possible that nSOX9⁺SSEA-1⁺ epithelial cells may extend further into the functionalis than first observed and behave as transit amplifying cells that contribute to the rapidly expanding glandular epithelium during endometrial regeneration. However, the SOX9⁻ expressing basalis epithelial cells have a non-cycling gene expression profile indicating their quiescence, as shown many years ago in tritiated thymidine incorporation ex vivo into endometrial tissue (2). Mouse endometrial epithelial progenitor cells were first identified as quiescent label retaining cells (LRC) by pulse-chase experiments using bromo-deoxyuridine (BrdU), a DNA synthesis label, to detect rarely dividing cells which retain the label (35, 36). Mouse endometrial epithelial LRC were identified in the luminal epithelium and did not express nuclear ERα (35), but were the first cells to proliferate on estrogen replacement of ovariectomised BrdU-labeled LRC mice, thereby driving endometrial luminal and glandular regeneration (37).

More recently, a single cell pulse-chase lineage tracing study using Cre-loxP-Keratin19 reporter system was used to identify mouse endometrial epithelial stem cells and their niche (38). The epithelial stem cells which generated EpCAM⁺ FoxA2⁻ luminal epithelial cells and EpCAM⁺ FoxA2⁺ glandular epithelial cells were located in the intersection zone of the luminal and glandular epithelium. They had capacity to repair the luminal epithelium and regenerate the glandular component over numerous estrus cycles and following pregnancy. Other lineage tracing studies have identified Lgr5⁺-expressing cells at the tips of the glands invaginating into the uterine mesenchyme in neonatal mice (39) and Axin2-expressing epithelial cells deep in the gland bases of adult mice which regenerated endometrial glands during estrus cycling (24). It appears that there are several stem/progenitor populations in mouse endometrium that are responsible for endometrial repair and regeneration, however the hierarchy of these stem/progenitor cells is yet to be determined.

Most postnatal tissues, whether highly regenerative or not, contain a population of mesenchymal stem cells (MSC), including human endometrium (eMSC) (40). MSC were first identified in bone marrow aspirates as clonogenic cells with a fibroblastic morphology (CFU-F) with capacity to differentiate into multiple mesodermal lineages (41). MSC were later defined by the International Society for Cell & Gene Therapy (ISCT) as plastic adherent stromal cells that differentiated into adipocytes, chondrocytes and osteocytes in vitro and had a characteristic surface marker phenotype distinguishing them from haemopoietic stem cells (42). However, stromal fibroblasts also fulfill the ISCT criteria, which does not distinguish clonogenic MSC with adult stem cell properties (self-renewal, proliferative potential and differentiation in vivo) (43). More recently, functional and morphological studies have identified numerous tissues with perivascular MSC with adult stem cell function (44, 45), including human endometrium (46, 47). In this review we will focus on perivascular endometrial MSC (eMSC) rather than endometrial stromal fibroblasts. For a more detailed discussion on the differences between perivascular eMSC and endometrial stromal fibroblasts in endometrial tissue and menstrual fluid see Bozorgmehr et al., (48).

The endometrium has a substantial vascularized stroma which regenerates during the proliferative stage of each menstrual cycle, and is likely mediated by eMSC. eMSC were first identified as clonogenic stromal cells (1.25% of stromal cells) (21, 49), which fulfill the ISCT criteria and also undergo self-renewal and in vitro differentiation to multiple mesodermal lineages (22). Their perivascular niche was discovered when specific surface markers were identified that enriched for the clonogenic endometrial stromal cells, first as pericytes co-expressing CD140b (PDGFRβ) and CD146 (46) and as SUSD2⁺ perivascular cells (47).

SUSD2⁺ eMSC can also generate endometrial stromal tissue in vivo when transplanted underneath the kidney capsule (47). In this model, SUSD2⁺ eMSC generate vimentin⁺ fibroblasts and induce the migration of endothelial cells that promote angiogenesis (47). The pro-angiogenic activity of eMSC has also been shown in vitro (50). Taken together, these data suggest that eMSC may be responsible for stromal vascular regeneration of the endometrium during the proliferative phase, mediated by both growth and differentiation and also by paracrine effects promoting angiogenesis.

These perivascular eMSC were identified around small and large blood vessels in both the basalis as expected and also in the functionalis, which indicated they would be shed in menstrual fluid (see section Role of Stem/Progenitor Cells in Menstrual Disorders/Regeneration Disorders). Gene expression profiling has demonstrated that CD140b⁺CD146⁺ eMSC rapidly lose their gene signature in culture and adopt the CD140b⁺CD146⁻ stromal fibroblast signature (51). This suggests eMSC differentiate into stromal fibroblasts, further contributing to the confusion between MSC and stromal fibroblasts (52). However, this differentiation has not been identified in vivo. SUSD2⁺ and SUSD2⁻ endometrial cells differentiate into decidual cells with similar but distinct gene profiles and both contribute to the formation of the maternal placenta (53–55).

Other markers of perivascular eMSC include NG2, Stro-1, EphA3, W8B2, and CD271 and CD34 which are found in the adventitia of blood vessels rather than a pericyte or medial location [reviewed in (48, 56)]. However, the CD34 population failed to regenerate human endometrium in functional studies (57). The perivascular niche suggests that eMSC likely contribute to vascular and stromal regeneration each menstrual cycle.

Endometrial thickness can be restored in post-menopausal women via oral estrogen therapy, suggesting that endometrial stem/progenitor populations remain quiescent after menopause but when exposed to exogenous estrogen rapidly respond to regenerate both glands and stromal vascular tissue (26). Like N-cadherin⁺ eEPC, SUDS2⁺ eMSC can be isolated from post-menopausal (PM) endometrium (58). They have a lower cloning efficiency than in pre-menopausal endometrium, but are detected in similar numbers (58).

Single cell RNA sequencing of human endometrium has identified a small smooth muscle population expressing SUSD2, CD146 (MCAM), and CD140b (PDGFRB) (32). In mouse endometrium, scRNAseq of Pdgfrb-BAC-eGFP reporter mice also identified a perivascular population of Pdgfrb⁺ MCAM⁺ cells with a perivascular gene profile and perivascular niche in vivo, that was distinct from 3 novel endometrial fibroblast populations also identified (59). These studies confirm our biological findings and provide further insight into the role of perivascular eMSC in endometrial regeneration.

Bone marrow derived stem cells (BMDSC) have been suggested as an exogenous source of stem cells in the endometrium (60). In humans, it has been reported that BMDSC contribute up to 48% of the epithelium and 52% of the stroma (60). Most of the supporting data has been generated from mouse models, which have shown BMDSC contribute to epithelial, stromal, and endothelial lineages (61–64). However, a 2018 study disputed their contribution. Using two different transgenic fluorescent tagged mouse lines to repopulate the bone marrow of irradiated recipient mice and sophisticated imaging and microscopy, Ong et al. demonstrated that BMDSC did not contribute to epithelial or stromal lineages (65). Instead they highlighted that intra-epithelial and stromal-derived cells were CD45⁺ leukocytes and that bone marrow-derived macrophages failed to immunostain with CD45 (65). This highlighted the limitations in the previous body of work relying solely on the identification of CD45⁺ cells. BMDSC contribution to other body organs has been similarly controversial but the body of evidence suggests their plasticity or ability to transdifferentiate does not occur for similar technical issues (66). For the purpose of this review we will be focusing on endometrial-derived eMSC and eEPC.

Menstrual fluid, discharged via the vagina following declining circulating progesterone levels, is a complex fluid containing shed endometrial tissue (31, 48), secreted proteins (67, 68), immune cells, peripheral blood components, vaginal epithelial cells, clots, and mucous (69). Shed menstrual fluid cells mainly comprise CD45⁺ leukocytes (>90%), with the remainder being endometrial cells (31, 69). Menstrual fluid endometrial cells include stromal cells, for example fibroblasts (69), SUSD2⁺ eMSC (70), and epithelial cells such as N-cadherin⁺ and SSEA-1⁺ eEPC (31). A similar proportion of these stem/progenitor cells is identified in menstrual fluid compared to endometrium (31, 70). Both SUSD2⁺ eMSC and N-cadherin⁺ eEPC have also been detected and their abundance reported in menstrual blood collected from the uterine cavity during surgery before efflux into the vagina (30). The proportion of clonogenic units in menstrual fluid endometrial cells is comparable for epithelial clones (0.31% for menstrual fluid and 0.22% for eutopic tissue) and somewhat lower for stromal clones (0.22% for menstrual fluid and 1.25% for eutopic tissue) than in endometrial tissue (31).

Menstrual fluid also contains cells described broadly as menstrual stem cells (MenSC) that fulfill the ISCT criteria which are easily isolated based on their adherence to plastic. However, they are likely heterogeneous due to non-specific and non-standardized isolation procedures (48, 71, 72). MenSC express markers for MSC, with the exception of STRO-1 (70, 73). MenSC likely comprise a combination of eMSC and predominantly endometrial stromal fibroblasts. A comparison of MenSC and eMSC has been extensively reviewed elsewhere, which highlights the need for standard isolation and characterization of MenSC (48).

Viable eMSC can be reliably isolated from menstrual fluid, their proportions studied, their clonogenicity assessed (31), and their behavior characterized in vitro (70). Together this indicates that menstrual eMSC are a reliable source for research that characterizes the biology of eMSC. Menstrual eMSC do tend to be more apoptotic and necrotic than other sources of MSC, however they can be prevented from undergoing apoptosis and senescence by A8301 treatment (70), thus indicating menstrual eMSC are also a viable option for cellular therapy (48).

The study of human primary endometrial epithelial biology has previously been limited to 2D culture such as (1) epithelial monolayers, which either become overgrown by stromal cells or senesce (74, 75), (2) colony forming unit assays, seeded at very low density and generally terminal experiments that do not permit long-term culture or functional assays (21, 49), or (3) serial subcloning which permits up to four passages for large colonies derived from CFU-F (22). However, our recent evidence that the endometrium contains N-cadherin⁺ and putative SSEA-1⁺ eEPC (26, 76) capable of forming gland-like structures in 3D culture maintained in ECM, together with organoid technology (77, 78), has enabled generation of human endometrial organoids (EMO) using defined serum-free culture conditions (79, 80) (Figure 2, yellow box).

EMO have since been derived from biopsies, hysterectomy tissue, endometrial cancer, endometrial hyperplasia, endometriosis lesions from various locations, and placental decidual tissue (79–81). They can be expanded for >6 months in culture, cryopreserved and generated from cryopreserved gland fragments (82) and show responsiveness to estrogen and progesterone (79, 83). The gene expression profile of EMO has been studied at both the bulk and single cell level (79, 80, 83, 84), however many of these studies are limited by their use of endometrial biopsies instead of hysterectomy tissue, which contain the full hierarchy of eEPC, including those located in the rhizome-like glandular structures of the basalis (85, 86). Changes in EMO cell fate in response to hormones (83) and inhibition of key developmental signaling molecules (NOTCH) (87) enhance our understanding of endometrial epithelial cell fate trajectory and its role in development and disease—an area that has previously been very challenging to study. EMO also show promise for drug screening, demonstrating sensitivity to specific compounds (88).

EMOs are derived from bulk endometrial epithelial fragments and therefore comprise a heterogenous population. They potentially include contamination of surrounding stroma, which recede from culture after multiple passages, but may influence organoid formation and contribute to variation in patient-derived organoid lines. While the ability to generate single cell EMO (scEMO) from existing EMO cultures has been demonstrated (79, 80, 84), the ability of naïve single endometrial cells to form EMO has been underexplored. Generating scEMO from FACS-sorted epithelial subpopulations has potential for investigating the roles of epithelial progenitor cells in endometrial regeneration.

Organoids can be generated from shed endometrial tissue in menstrual fluid (MFO) (89, 90) (Figure 2, red box).While MFO are less abundant than EMO, they appear to represent EMO in proliferation rates, responses to hormones, and gene expression profiles (89, 90). They can also be derived from disease states including endometriosis and adenomyosis. They can also be derived from girls and young women without the need for an endometrial biopsy, enabling the study of early disease mechanisms, precision medicine, and diagnosis (91).

While EMO and MFO represent major advancements in studying menstrual biology, these systems largely support epithelial cultures in isolation, and lack the stromal, mesenchymal, vasculature, and immune cells important for a functioning endometrium. Biomaterial engineering has generated synthetic hydrogels (92, 93) that permit co-culture of endometrial epithelial and stromal cells and these are being rapidly applied to EMO (Figure 2, orange box) (81, 94, 95). They are also being applied to the cells/tissues required to support the endometrium such as the vasculature (92) and immune cells [reviewed in (96)]. Coculture in defined hydrogel systems has enabled development of organ-on-chip systems for disease modeling and low-cost drug screening for other organs and disease states (97)—their potential for endometrium, endometriosis, and adenomyosis are exciting prospects to be explored (96). Organ-on-chip and micro-physiological systems (Figure 2, orange box) have multiple advantages, including infinite tunability (cell input, matrix composition, hormonal, and nutrient delivery), scalability, enabling coupling for modeling the complexity of endometrial regeneration.

Whilst single cell and co-cultured 3D organoids overcome some of the issues of using 2D in vitro models to study endometrial dynamics, endometrial repair/regeneration is a multicellular/multizonal, tightly controlled process which has previously been difficult to replicate in a dish. 3D thin tissue slice cultures provide a culture system that maintains endometrial structure (98). These tissue slice approaches have shown that the tissues can respond to estrogen and progesterone over 21 days in vitro (Figure 2, blue box). Histology of the slices indicates that zone-specific changes in vitro mimic in vivo hormone responses (98). Whilst stem/progenitor populations were not assessed in this study, the authors did show non-specific transduction of lacZ via adenovirus-mediated gene delivery and therefore the model has promise for studying stem/progenitor populations and interactions in an “in vivo-like” system, however 3D endometrial slice cultures are limited to terminal experiments.

Historically, the location and expression profiles of endometrial stem/progenitor cells have been presented in 2D via standard immunohistochemistry/immunofluorescence of thin tissue sections. The introduction of tissue clearing, whereby a tissue sample is rendered optically transparent via solvent- or aqueous-based solutions, has enabled a more detailed morphological examination of the endometrium in 3D. Samples are fixed, permeabilized, and then cleared so that light scattering and absorption caused by cellular contents are minimized. Optimal tissue clearing maintains native tissue architecture and preserves fluorescent proteins (99) (Figure 2, green box). Two recent studies have reconstructed the 3D morphology of full thickness endometrium and have shown that basalis glands form a rhizozome-like plexus structure horizontally across the endometrial basalis using tissue clearing (85) or genetic lineage tracing (86), challenging the long held view of vertical glands extending from blind-ended glands in the basalis. This plexus structure remains during menstruation, from which branched glands arise and vertically penetrate the functionalis. 2D studies clearly indicate that a hierarchy of epithelial/stem progenitor cell types with specific markers exists extending from the basalis to the luminal epithelium. Now these new technologies are available it will be exciting to see whether this epithelial hierarchy can be reconstructed in 3D, whilst also investigating the relationship between MSCs and the endometrial vasculature across all cycle phases.

Asherman's syndrome (AS) is characterized by intrauterine adhesions/scarring and loss of a functional endometrium. Adhesions can be caused via surgical scraping/cleaning of the uterus or via uterine infection in a setting of low circulating estrogen e.g., post-partum and pregnancy termination. This trauma, to the endometrial basalis, causes loss of the germinal compartment for regenerating the endometrium. It has been proposed that trauma damages stem/progenitor populations and their surrounding stem cell niche in the basalis, preventing regeneration of the functionalis (40, 129, 130).

The use of endometrium-derived stem cells for treatment of Asherman's is still very much in its infancy. At the time of writing, few studies have investigated using menstrual fluid as a potential therapeutic option. Menstrual blood derived eMSC form spheroids that, when injected into the uteri of rats with induced AS, can improve fertility rates (131). In a small study of 7 human patients, autologous transfer of cultured MenSC resulted in an increase in endometrial thickness in 5 patients, four of which were able to undergo embryo transfer and two patients conceived successfully (132). Given menstrual fluid is a plentiful, easily available resource, research into autologous MenSC transfer would be worthy of further investigation.