Endometriosis and adenomyosis unveiled through single-cell glasses

The analyses of human endometrium and the endometrial disorders - endometriosis and adenomyosis - at single cell resolution, described herein, complement and extend bulk tissue and isolated cell transcriptomics, epigenomics, and metabolomics, as well as blood and peritoneal fluid analyses, and in vitro cellular and tissue organoid/assembloid models 1 – 4 , 78 – 82 . Single cell transcriptomics is taking the field of “endometromics” to new heights – demonstrating commonalities and differences among cells of the same lineage and in different locations within a given tissue and across tissue types, inferred cell-cell communications, and insights into inflammation as a driver of endometriosis and adenomyosis, as well as the consequences of endometrial epithelial somatic mutations (shared in both disorders) on individual cell gene expression and a unique epithelial cell subtype associated with the risk of ovarian cancer in patients with endometriosis. The recent discovery linking Fusobacterium in ovarian endometriosis in women and antibiotic therapy reducing lesions in a mouse model of endometriosis and peritoneal colonization with Fusobacterium has set the stage for well controlled clinical trials for possible antibiotic therapies to reduce disease burden and for symptom management in patients. Also, the associated dysbiosis of the gut microbiome in mouse models opens opportunities for nutritional interventions in patients that could reduce the disease burden and ameliorate symptoms 56 – 58 . These new directions warrant further analysis of the circulating metabolome and single cell analyses of endometriosis in the setting of associated tissue microbiota. So, what will we see beyond the looking glass and into the future? The advances noted above will no doubt serve as launchpads for new and novel biomarker discovery and noninvasive diagnostic development and ultimately prognostic indicators of disease and treatment outcomes 83 , as well as individualized, patient-tailored therapeutics for symptoms associated with both disorders. Mapping human reproductive tissues and disorders with single cell genomics is certainly coming into its own 84 , and it is anticipated to benefit from the ever-growing single cell atlases (e.g., www.humancellatlas.org ; www.reproductivecellatlas.org ) that are providing immensely rich, publicly available datasets. Indeed, current and future studies on endometriosis and adenomyosis will greatly benefit from the abundant single cell data on fundamental cell types and subtypes in other reproductive and non-reproductive tissues and diseases. This is likely, as cells in different tissues and disorders share some fundamental codes that identify their cell type, and that there are tissue-specific codes and tissue-specialized or altered codes (e.g., in the presence of inflammation or hormonal and other medical therapies) 72 , 83 , 85 . Data integration and comparative analyses of these codes and features across a multitude of endometrial and uterine disorders and their co-morbidities, as well as other tissues and diseases has the possibility to break the silos of discipline-specific study and capture the breadth of genomic features integrated with holistic features of patient metadata to render individualized health care and well-being across the life span and across diverse populations. Moreover, the gut microbiome-brain axis has been recognized as a key regulator of health and disease 86 , and, recently, dysbioses of the gut 87 and of the female reproductive tract 88 have been reported in patients with endometriosis and gastrointestinal (GI) symptoms. Notably, circulating- and lesion-derived estrogens are related to these symptoms and with pelvic pain 89 . Additionally, dysbiotic gut bacteria (the “estrobolome”) that convert conjugated estrogens to active estrogens and other bioactive metabolites are postulated to communicate with the brain regarding pain perception and GI symptoms and may also stimulate disease 87 , 88 , 90 . Studies on adenomyosis-associated altered microbiota in the endometrium and vagina are more limited compared to endometriosis-associated dysbiosis 91 – 93 . However, they are exciting as the paradigm of altered female reproductive tract microbiota in both these estrogen-dependent disorders is rapidly evolving. Identifying cell populations and sub-populations through single cell analyses in these various uterine and gut compartments and understanding mechanisms underlying their role in endometriosis-associated symptoms could open a novel approach to non-hormonal antibiotic and nutritional interventions for symptom relief, in well-designed clinical studies 87 , 94 . Data integration across compartments and the circulating metabolome, proteome, and immunome are additional players in the path forward. We have much to do to enable teams and new investigators to join this unprecedented effort in science and medicine beyond the looking glass – as technology progresses, so will biology, and then so will technology, and so will biology 95 .

While single cell RNA sequencing and related innovative technologies and single cell atlases are providing unprecedented insights into human physiology and pathophysiology 71 , 72 , there are some unique challenges and opportunities in studying the steroid-hormone responsive human uterine compartments and their associated disorders. All research on human tissues, obtained under approved IRB protocols and written informed consent, requires accurate information about the study cohorts (cases and controls) to minimize confounders and maximize data accuracy and interpretation. Key are well annotated, de-identified patient metadata - e.g., demographics (age, race/ethnicity, BMI), pregnancy history, family history, personal habits, indication for surgical procedure, medical and surgical histories, current hormonal and nonhormonal medications, supplements, co-morbidities, the surgical procedure in which tissue samples were acquired, histologic confirmation that samples contain relevant tissues/cells (e.g., endometrium, myometrium, EMJ, endometriosis, adenomyosis, fibrosis, other), and hormonal status at the time of sampling, documented in a comprehensive database. Acquiring such information and processing samples for sequencing and data management demand teams of experienced study directors, clinical, surgical, laboratory, and pathology colleagues and services, genomics and sequencing cores, laboratory staff, clinical research coordinators, data entry staff, biostatisticians, and computational scientists. Some challenges and opportunities regarding studies specifically on the female reproductive tract are described below. The hormonal status of cases and controls whose tissues are under study is critical in studies on endometrium, endometriosis, and adenomyosis, as these tissues are estrogen- and progesterone-responsive. The context of the steroid hormone milieu (cycle phase, hormonal medications, pre- and post-menopausal status) is essential for data analysis and interpretation. If patients are not using hormonal preparations, then menstrual cycle phase determination must distinguish the estrogen-dominant proliferative phase and the sub-phases of the progesterone-dominant secretory phase 44 . Many patients, however, with symptomatic endometriosis and/or adenomyosis undergo surgery for pain and/or heavy menstrual bleeding for which combined contraceptive steroids, progestins, gonadotropin releasing hormone analogues, and/or aromatase inhibitors have been administered 5 , 12 . While synthetic progestins and progesterone signal through PR, they also elicit unique signaling pathways and transcriptomes in endometrial cells when administered systemically and in in vitro models 39 – 43 , which can impact some or most cell features discovered by scRNAseq analysis. It is also important to have hormonally similar controls without disease so hormonal status is not a confounder. Cohort studies restricted to a single menstrual cycle phase avoid hormonal confounders due to similar hormonal environments of the samples in situ at the time of acquisition. Also, precise timing is key, as the endometrial and endometriosis and/or adenomyosis lesion transcriptomes may change within a given cycle phase (e.g., early vs late secretory phase and, strikingly, at the opening of the implantation window in normal patients without disease 19 ). A few scRNAseq studies have focused exclusively on one cycle phase versus another e.g., window of implantation (WOI) mid-secretory phase versus proliferative phase 31 , or timed menstrual effluent sampling 32 , which reduces confounding. Notably, studies comparing hormone treatments with no treatment among cases or controls are highly valuable to understand the efficacy of therapies on disease cell pathophysiology and could direct individualized cell-responsive therapeutics in the future. Lesion type, location and co-morbidities are additional key components for context and data interpretation. Age of the lesions is largely unknown, and thus comparing a lesion “type” within the same patient or between patients may be affected by different temporal histories of the lesions. Challenges also include the high prevalence of co-existing common gynecologic disorders (uterine fibroids, uterine polyps) in both cases and controls and defining the control group (no endometriosis or adenomyosis and without or with other gynecologic disorders). Standard operating procedures for biospecimen (tissue, blood, peritoneal fluid, other) collection, processing, storage, and distribution, as developed by Sheldon et al 73 and the World Endometriosis Research Foundation 74 – 77 , choice of sequencing technologies (Chapter 1), and pursuit of multiple computational strategies for comprehensive data analysis and validation of findings are key to success in working with these tissues and looking through single cell glasses. While numbers of cells sequenced enrich the phenotypic features of individual cell types, the numbers of subjects recruited is, by comparison, low, and notable is the limited diversity of the cohorts. There is a preponderance of White subjects recruited for scRNA sequencing of endometriosis and adenomyosis 4 ( Table 1 ), opening a great opportunity to expand studies to include more diverse populations and comparative analyses of data across ethnicities in the future.

Uterine compartments relevant to adenomyosis pathogenesis include the endometrial-myometrial junction (EMJ) [basalis and lower functionalis endometrium + sub-endometrial inner myometrium], all of Mullerian origin and displaying cycling estrogen receptors (ER) and progesterone receptors (PR) 59 – 61 ( Figure 9 ). Few molecular and genomic studies of the EMJ in adenomyosis uteri have been conducted, likely as access to basalis and inner myometrium requires hysterectomy versus the functionalis, which can be readily obtained by office biopsy. Studies to date in adenomyosis patients using immunohistochemistry and qPCR reveal abnormal endometrial functionalis epithelial aromatase expression (thus high E 2 milieu), elevated stromal fibroblast IL-6, IL-8, IL-17 secretion, TH17-Treg imbalance, altered uNK receptors, and aberrant P 4 -signaling 5 , 26 , 62 – 64 – very similar to endometrium of patients with endometriosis except for IL-17 and TH17-Tregulatory cell imbalance. We conducted bulk RNA seq of endometrial functionalis 65 and a study on separate compartments of EMJ inner myometrium and endometrium from patients with diffuse adenomyosis versus controls in the proliferative phase 66 . ECM remodeling, inflammation, IL-17 signaling, and mTOR signaling were enriched in endometrium, and inflammation, pain, and neuronal signaling pathways in myometrium, consistent with clinical symptoms of this disorder. Two recent publications have analyzed adenomyosis lesions 67 , 68 . They are important beginnings to understanding the pathogenesis and pathophysiology of adenomyosis, building upon bulk transcriptomic data, which could be further mined in deconvolution studies to validate new and likely forthcoming additional single cell data. Liu et al 67 sequenced 42,292 cells derived from a control endometrium of a patient with uterine fibroids and eutopic (functionalis) endometrium and matched ectopic lesion from a patient with adenomyosis (type and menstrual cycle phase undefined). The data revealed seven distinct cell types. Comparison of the data from the lesion with those from eutopic matched endometrium and control endometrium revealed processes involved in progression of normal→endometrial→adenomyosis epithelial phenotypes. These included motility, proliferation, angiogenesis, inflammation, and cancer, consistent with the theory of collective cell migration and invasion into the myometrium of endometrial-derived cells. The novel data set revealed vascular mimicry (important in tumor formation) and epithelial transformation to endothelial cells. It is anticipated that the novel epithelial→endothelial transition phenotype will likely be mined in the future and expanded studies of adenomyosis and perhaps other gynecologic disorders and may serve as a unique cell type for drug targeting. Bulun and colleagues 68 reported results of single cell analyses of adenomyosis lesions and matched eutopic endometrium and myometrium from 3 subjects, all with diffuse adenomyosis and in the proliferative phase (total of 9 samples and 66,000 cells sequenced). They identified 11 cell types in eutopic endometrium of patients with adenomyosis ( Figure 10 A ). Fibroblast-like cells comprised a large fraction (36%), and pseudotime analysis revealed they originate from pericyte progenitors that differentiate to ER+ and PR+ endometrial stromal fibroblasts. Notably, the pseudotime trajectory was found to end with ciliated epithelial cells, suggesting mesenchymal- epithelial transition at play in this tissue. In adenomyosis lesions, they identified 13 cell types including 2 (smooth muscle cells and neurons) not found in the endometrium of the same patients ( Figure 10 B ). The adenomyosis lesions contained even more fibroblast-like cells compared to endometrium (50% versus 36%, respectively). In contrast to matched endometrium, the fibroblast-like cells did not originate from pericyte progenitors and differentiated to ECM-expressing fibroblasts and smooth muscle cells (SMC). Also, myometrial SMC were found to derive from fibroblast-like clusters. Moreover, the WNT inhibitor, secreted frizzled related protein (SFRP) family expressed in fibroblast-like clusters were similar across tissues - endometrium, adenomyosis and myometrium 68 . Regarding the epithelial cells, P 4 -resistance was found in ciliated and unciliated cells in adenomyosis tissue versus matched eutopic endometrium ( Figure 10 C ) – this is a key finding regarding treatment efficiencies. Further data analysis revealed that the canonical WNT/SFRP pathway signaling was uniquely activated in adenomyosis unciliated and ciliated epithelium and paracrine signaling fibroblast-like mesenchymal: epithelial interaction networks ( Figure 10D ), opening new avenues for drug discovery for this disorder. The epithelial→endothelial transition reported by Liu et al 67 was not detected in this study and may be due to different adenomyosis lesion types, cycle phase, patient age, and other variables. Whether the findings of these two studies involving patients with diffuse adenomyosis also apply to other forms of adenomyosis ( Figure 1 ) is unknown and is an opportunity to investigate cell features wherein novel therapeutics could be derived. Moreover, recent data suggest that the endometrial and vaginal microbiomes differ in patients with versus without adenomyosis 69 , 70 , raising the possibility of adjunctive antibiotic therapies for symptomatic patients with disease. Additionally, further study of the local and systemic immune system cells and their subtypes as well as in affected uterine tissues at single cell resolution are warranted.

Endometriosis and adenomyosis are common estrogen-driven uterine disorders derived mainly from the basalis endometrium, characterized by “ectopic” endometrial-like cells outside their normal location. A key feature distinguishing them is the location of the ectopic lesions. For endometriosis they are outside the uterus - mostly in the pelvis and invading pelvic organs 1 – 4 , and for adenomyosis they are within the uterine myometrium 5 – 8 . Endometriosis and adenomyosis disease lesions comprise similar phenotypic endometrial epithelium and mesenchymal cell types and display characteristics of inflammation, neoneuroangiogenesis, and fibrosis 1 – 9 . However, whether their cellular features are equivalent in different anatomic niches is unknown and is of importance for disease stratification in view of the vastly heterogeneous lesion architectural structures and for developing tailored, personalized, patient-specific treatments and prognostic disease outcome indicators. These complex “sister” conditions often occur concomitantly 10 and result in similar symptoms of pain and uterine bleeding treated medically (mostly hormonal) with variable efficacies and are associated with infertility and similar poor reproductive outcomes 1 , 4 – 13 . They also greatly impact quality of life of those affected 14 – 18 . Given the devastating impact of these disorders and limited success of effective therapies and disease prevention and cure, a transformative approach to disease phenotyping and novel, personalized therapeutics is essential to improve the lives of millions of patients with endometriosis and adenomyosis across the globe. Single cell technologies and precision medicine approaches are early in their application to these disorders and are beginning to provide insights into commonalities/differences of cell types in endometriosis, adenomyosis, and eutopic endometrium, and the role of their unique anatomical niches influencing cell behavior, disease establishment and pathophysiology. Moreover, single cell data and associated signaling pathways are anticipated to lead to novel drug candidate discovery to control symptoms, minimize disease burden resulting ultimately in cure, identifying diagnostic biomarkers, and informing disease classification and prognostics for disease courses. The time is now to move the needle on both disorders through single cell “glasses”! Regarding endometrium, an atlas of normal, cycling endometrium has been derived at single cell resolution by several groups 19 – 22 (also see Chapter 3). These studies underscore heterogeneity of specific cell types and altered features in different hormonal states across the menstrual cycle and provide a key backdrop to studies at the single cell level of endometrium of patients with endometriosis and adenomyosis, endometriosis lesions, and adenomyosis lesions. Herein, we begin with an overview of clinical characteristics of endometriosis and adenomyosis (sister disorders but not identical twins), followed by a summary of current advances in scRNA sequencing data of endometriosis and adenomyosis disease lesions, corresponding eutopic endometrium, control endometrium, and disease-specific and common cell features across tissues. We conclude with a review of challenges of conducting research on human tissues, especially those that are steroid hormone responsive, and how these data inform looking to the future through single cell glasses.

Subsequent to the single cell cartography of human endometrium published by Wang et al 19 (see Chapter 3), several groups have recently reported eutopic endometrial scRNAseq data of patients with endometriosis, endometriosis lesions and control eutopic endometrium 20 , 31 – 36 ( Table 1 ). Types of samples, inclusion/exclusion criteria, choice of controls, study designs, numbers of cells sequenced, and number of reads (i.e., base pairs sequenced) are key to data integrity and interpretation and to comparative analyses across tissues and studies. They have provided insights into origins of ectopic disease, the heterogeneity of cell types/subtypes, unique clusters and signatures of endometrium and ectopic lesions informing mechanisms and pathways involved in cellular dysfunctions relevant to pain and fertility compromise, novel cell-cell communications, relationships between lesions and eutopic tissue/cells, and biomarker discovery. Across these studies, samples were obtained in different hormonal milieu (menstrual cycle phase, exogenous hormones), and between 55,000 to 378,000 cells were sequenced ( Table 1 ). As cycle phase and exogenous hormone use are key drivers of endometrial gene expression 37 – 44 , comparison of data across studies can be challenging and mixed hormonal milieu are sometimes experienced within the same study in both cases and/or controls. Below we summarize results of studies focused exclusively on eutopic endometrium of patients with endometriosis relevant to endometrial function and dysfunction and biomarker discovery, followed by studies focused on ovarian endometriomas, superficial, and/or deep infiltrating disease, and eutopic endometrium, relevant to disease heterogeneity, pathophysiology, pathogenesis, and in the setting of progestins – commonly prescribed for symptom management of patients with endometriosis and dysmenorrhea and chronic pelvic pain. Endometrial biopsies of cases with stage I/II disease (revised American Society for Reproductive Medicine (rASRM) scoring system) 45 and infertility (n=3 WOI, n=3 early proliferative phase) versus n=7 fertile controls without disease (3 WOI, 3 proliferative, 1 late secretory phase) were studied by Huang et al 34 . A total of 128,243 cells were sequenced, and 30 type/cell type clusters were identified. Notably, a unique epithelial cell cluster in WOI endometrium of endometriosis patients was identified that lacked expression of two key genes associated with implantation: PAEP (progesterone-associated endometrial protein) and CXCL14 (chemokine (CXC motif) ligand 14) ( Figure 2 ). PAEP, also known as glycodelin A and placental protein 14, is involved in immune response regulation, promoting opening of the WOI by lowering maternal immune responses to an implanting embryo 46 . CXCL14 is involved in activating the innate immune response stimulating chemotaxis of uterine natural killer cells (uNK) and other monocytes to epithelial glands for implantation 46 . Other findings included absence of normal variation of endometrial uNK and T cell numbers across the cycle, and lower anti-inflammatory IL-10 and more pro-inflammatory cytokines expressed in endometrial immune cells in the WOI versus in the proliferative phase in cases and the opposite in controls. Notably, 11 ligand–receptor pairs were upregulated between endometrial immune and epithelial cells during the WOI of cases. Thus, lower epithelial receptivity markers, abnormal immune cell frequencies, a pro- inflammatory WOI, and new insights into cell-cell communications between epithelium and immune cells support an important role of the immune system and an adverse environment for embryo implantation for fertility and potentially for poor pregnancy outcomes in patients with stage I/II endometriosis who do achieve pregnancy versus fertile controls. In a recent study, our group sequenced eutopic endometrium (237,620 cells) from 29 subjects with endometriosis (stages I/II and III/IV) and 15 controls without disease (7 healthy controls, 8 with symptomatic uterine fibroids), across the menstrual cycle 36 . We identified 5 main cell groups: stromal fibroblasts, smooth muscle, endothelial, epithelial, and immune. In the latter, further analysis revealed 18 sub-populations including uterine NK cells, T cell subsets, monocytes, macrophages, and dendritic cells. Epithelial subpopulations comprised glandular, luminal and ciliated cells, with abnormal expression of CXCL14, PAEP, CCL20 , MMP7, and SFRP4 genes in cases, regardless of phase. Endometrium in controls with FIGO Type 2 uterine fibroids revealed overexpression of WFDC2 and DAXX genes in cases. In immune cells, fibroblasts, and epithelial cells, cytokine production, stress signal and pro-inflammatory pathways in cases were noted, similar to Huang et al 34 findings, and enriched pathways in epithelial and ciliated cells related to migration and cell motility were higher in cases. Cell communication analysis revealed a complex network of inflammatory pathways, with significant involvement of classical monocytes and NK cells. Finally, genes implicated in ovarian cancer were upregulated in endometrium of cases compared to controls, confirming known association of endometriosis with specific subtypes of ovarian cancer 12 . Thus, in the eutopic endometrium in patients with endometriosis, there are numerous alterations in specific cell types across the cycle versus controls without disease and disease-free controls with uterine fibroids 36 . These can predispose to impaired fertility and tissue dyshomeostasis. A recent scRNAseq study 32 of endometrial tissue in menstrual efflux (not biopsies) of n=11 cases with laparoscopic evidence of endometriosis (undefined stage) (11,924 cells) and n=9 healthy controls with no disease or symptoms (14,327 cells) revealed 18 cell clusters. These included abundant uNKs, stromal fibroblast, epithelial, B, T, and myeloid cell clusters, and a small plasmacytoid dendritic cell (pDC) cluster. Further analysis separated data by cases vs controls and revealed a marked decrease in uNK1 and uNK2 cells and significantly enriched B cells in cases ( Figure 3A ). Log2 odds ratio (OR) analysis confirmed uNK1/2 significantly enriched in controls and B cells greatly enriched in cases ( Figure 3B ). Moreover, menstrual endometrium scRNA seq revealed decreased decidualization markers in stromal fibroblasts (reflecting P 4 -resistance) and pro-inflammatory stromal cells, as observed by others in FACS-sorted cell analyses and in vitro cell culture studies 6 , 47 . Overall, this study demonstrates that menstrual endometrium (commonly referred to as a “liquid biopsy of the endometrium”) reflects secretory endometrium abnormalities as possible biomarkers of disease. The proposed disease model is shown in Figure 3C wherein defective endometrial stromal fibroblast decidualization is driven by multiple factors including inflammation, chronic endometritis, stress, and/or P 4 -resistance – which all have clinical relevance for pregnancy establishment and success, as well as tissue homeostasis. This, in turn, may direct fibroblast differentiation to a chronic inflammatory phenotype and senescence, which may also impair decidualization. Reduced decidualization may also compromise the infiltration and proliferation of uNK cells, likely to be important for senescent cell removal and important for an optimal setting for embryo nidation 32 . In their landmark spatial-temporal single cell transcriptomic study 48 , García-Alonso et al 20 analyzed endometrium (3 functionalis and 6 full thickness (functionalis and basalis) tissues) across the cycle and identified 14 unique cell clusters that localized to corresponding endometrial compartments and niches. The proposed endometrial epithelial stem-cell niche resides within the basalis layer and contains SOX9+ cells and other “stem” markers (e.g., LGR5, SSEA1, β-catenin, N-cadherin) 49 , 50 , and Garcia-Alonso et al 20 found proliferative SOX9+LGR5+ cells mapped to the basalis stem cell niche. Epithelial stem cells are believed to play a role in endometriosis pathogenesis, and by leveraging bulk (microarray) transcriptomic data of superficial peritoneal lesions (9 red, 9 white, 11 black) ( GSE141549 ), this team identified enrichment of SOX9+LGR5+ markers in peritoneal lesions in the proliferative phase, compared to normal endometrium (n=42) and peritoneum (n=12) 20 . A proliferative phenotype of endometriosis lesion epithelial cells is consistent with P 4 -resistance noted in endometriosis 4 , 6 . Moreover, dysfunctional epithelium as a major driver of endometrial diseases, including endometriosis, and discovery of SOX9+LGR5+ stem-like populations dominant in endometriosis peritoneal lesions are foundational observations about pathogenesis and pathophysiology of superficial endometriosis disease lesions deriving from the basalis layer epithelium. A comprehensive scRNAseq analysis of >370,000 individual cells from 8 ovarian endometriomas, 28 superficial and deep disease lesions, 10 eutopic endometrium, 4 unaffected ovary, and 4 endometriosis-free peritoneum samples, has resulted in a cell atlas of endometrial-type epithelial cells, stromal cells, and microenvironment cell populations across tissue types 35 . Samples were derived from 17 cases (9 only endometriosis; 8 endometriosis and adenomyosis, uterine fibroids, and/or uterine polyps). 14 were cycling (7 proliferative and 7 secretory phases), and 3 were on contraceptive steroids ( Table 1 ). Controls (n=4) were postmenopausal (3) or perimenopausal (1), without endometriosis but with uterine fibroids +/− adenomyosis +/− uterine polyps, and all were on hormonal replacement regimens. Histologic and macroscopic features of specimens from one patient, shown in Figure 4A , underscore the heterogeneity of the lesions by cell abundance, cell types, and tissue architecture. The major cell types identified from scRNAseq of all cells from n=49 samples comprising the 5 major tissue classes (endometrioma, peritoneal endometriosis (deep and superficial), unaffected peritoneum, eutopic endometrium, unaffected ovary) included epithelial, smooth muscle, mesenchymal, endothelial, mast, myeloid, B, and T/NKT cells, and erythrocytes. Correlations based on cluster frequency across all specimens profiled by scRNAseq are shown in Figure 4B and reveal 3 main clusters: 1) mainly endometriosis specimens; 2) mix of endometriomas (5 of the 8 endometrioma specimens) and eutopic endometrium (5 of the 10 eutopic samples; 3) only controls: unaffected ovary and 5 of the 10 eutopic endometrium samples 35 . Endometrioma, endometrium, superficial and deep endometriosis epithelia revealed distinct cell subtypes and signaling pathways, and endometriomas displayed markedly different epithelial differential gene expression and signaling pathways compared to deep and superficial peritoneal disease and eutopic endometrium ( Figure 4C ). Interestingly, endometriomas displayed immune cell and complement activation and were enriched in B cells and plasma cells, suggesting possible infection in endometriomas and a unique role for B cells which have received limited attention in endometriosis pathophysiology 51 – 53 . Peritoneal disease was enriched in mast cells and T/NK-T cells engaging novel immune targets, and some histologically negative mesothelium surprisingly had disease signatures, although depth of lesion evaluation for endometriosis tissue may vary among pathologists. Endometrial-type epithelial cells and fibroblasts exhibited differential gene expression associated with deep or superficial disease status ( Figure 5 ). Interestingly, deep infiltrating disease epithelium displayed upregulation of nerve growth factor, suggesting a role of the epithelium in promoting innervation of this disease type which is particularly painful 12 . While somatic mutations have been reported in endometrium and endometriosis lesions, mainly involving KRAS activating mutations and ARID1A loss of function mutations 54 , 55 , Fonseca et al 35 confirmed these somatic mutations in various lesions ( Figure 6A ) and demonstrated in vivo transcriptional consequences of these mutations at the single cell level ( Figure 6B ). ARID1A mutation was associated with pro-lymphangiogenic stroma ( Figure 6C ) and adjacent mesothelium pro-inflammatory features and ciliated epithelial signatures - consistent with ovarian oncogenic potential. It is well-known that endometrioid ovarian cancer (EOC) and clear cell ovarian cancer (CCOC) arise in association with endometriosis, suggesting endometrial-type epithelial cells may be precursors for these tumors. A multi-subject single cell deconvolution analysis of the single cell data ( Figure 7 ) revealed strong enrichment of signatures for ciliated endometrial-type epithelial cells compared with the other keratin-positive clusters. In contrast, high grade serous ovarian cancer, an epithelial ovarian cancer type, not believed to originate from endometriosis epithelial cells, showed no pattern for preferential enrichment of any epithelial cluster. Endometrium and all three lesion types versus control endometrium displayed different cell/molecular signatures across tissues, consistent with restructuring/transcriptional reprogramming in lesions. Epithelial and stroma differed by site with striking differential gene expression in endometriomas and peritoneal lesions suggesting these are distinct disease entities. Superficial peritoneal and deep infiltrative endometriosis epithelium and fibroblasts had some differential gene expression and pathways, but similar structural genes to each other, suggesting they are part of the same continuum and distinct from endometrioma. While this study is a tour de force of endometriosis lesion and eutopic endometrium single cell transcriptomic analysis giving major insights into cell- and tissue-specific features and functions, including oncogenic potential of endometrioma epithelial cells, it underscores some of the challenges in conducting these types of studies. These include the presence of co-existing gynecologic disorders that can confound data interpretation; cells sequenced comprise a minor population in most samples (e.g., <1% of cells profiled), and controls and normal tissues are challenging to obtain e.g., controls were peri- or post-menopausal on hormone replacement therapies and had other uterine disorders. Nonetheless, these data are important and provide foundational information for comparison with other studies on eutopic and ectopic lesions in endometriosis patients and controls without disease. Sequencing 55,000 single cells from 3 endometriomas, 3 matched eutopic endometrium, and 3 endometrial samples from controls without disease in the proliferative phase, Ma et al 31 found 9 cell type clusters: fibroblasts, macrophages/monocytes, neutrophils, and T, NK, epithelial, endothelial, and mast cells, and unknown. Thirteen fibroblast subtype clusters were identified with significant differences in subtype composition among ectopic endometrium, eutopic endometrium, and normal endometrium. Fibroblast subtypes from normal proliferative phase endometrium displayed enriched functions of DNA replication, repair, growth regulation, and metabolic processes. However, endometrioma fibroblasts comprised 4 subtypes: 1) cytokine, inflammatory response; 2) FGF stimulation, immune response; 3) extracellular matrix (ECM) organization, cell adhesion; 4) angiogenesis, hypoxia response, and epithelial-mesenchymal transition (EMT), all of which were enriched in MAPK, TNF, IL-17, TGF-β signaling pathways. Fibroblasts from eutopic endometrium of cases displayed many pathways similar to normal tissue, but some features suggested a “transition state” from normal endometrium to endometrioma, involving TGF-β, MAPK, Rho-Rack, NF-kB, JAK-STAT, and EMT. A developmental trajectory was postulated based on these fibroblast features, with endometrioma fibroblasts derived from endometrial fibroblasts 31 , much as the epithelial cell population has been proposed, based on shared epithelial somatic mutations in these two tissues 54 . Thus, endometrioma and paired eutopic endometrium fibroblasts differ, and the single cell data support their key roles in disease pathogenesis. Analysis of immune cells revealed they differed significantly in normal endometrium, endometrioma, and paired eutopic tissue ( Figure 8A ). Specifically, T cell and uNK cell frequencies were lower, uNK cells were more active, and macrophages (Mφ) were enriched and had features of tissue remodeling versus eutopic endometrium 31 , similar to data from Fonseca et al 35 . Thus, in the proliferative phase, fibroblasts and immune cell subpopulations contribute to a pro-inflammatory, angiogenic environment in endometriomas. Moreover cell-to-cell communication/interactome analysis revealed that immune cells stimulate fibroblast adhesion and growth in endometriomas ( Figure 8B ), underscoring the power of single cell analyses to inform the pathogenesis and pathophysiology of endometriomas and cell-cell communications. Furthermore, recent observations indicated that Fusobacterium 56 and dysbiosis of gut 57 , 58 altered systemic and potentially local immune functions, contributing to the development of endometriosis. Notably, while transcriptomic alterations of specific populations and sub- populations of the endometrium, lesions, and immune cells (local and peripheral) under these conditions awaits further study, antibiotic and nutritional therapies for dysbiosis is emerging as a new paradigm for treating endometriosis development and symptom management. As progestins are a mainstay of medical therapy for endometriosis-related pain 12 , analysis of lesions and endometrium of subjects taking progestins is highly valuable in understanding their effects on cell types and tissue features. Tan et al 33 performed scRNAseq on 122,000 cells from 14 individuals, selective cell localization by imaging mass cytometry, and validations in endometrial epithelial organoid cultures. Peritoneal and/or ovarian disease and endometrium from 19 cases (rASRM stage II-IV) included 14 on norethindrone (NETA) + ethinyl estradiol (EthE 2 ), 2 on drosperinone + EthE 2 , 1 on levonorgestrel + EthE 2 , 1 on only NETA, 1 menstrual phase, and 8 controls with no endometriosis or inflammatory conditions: 4 on NETA + EthE 2 , 2 in proliferative phase, 1 on medroxyprogesterone acetate, 1 on norgestromin + EthE 2 . The peritoneal lesions had similar cell compositions as eutopic endometrium but dysregulated innate immune and vascular components - vastly different from ovarian endometriosis lesions. The authors identified a unique perivascular mural cell specific to the peritoneal lesions with features promoting neo- angiogenesis and immune cell trafficking. They also identified a progenitor-like epithelial subpopulation, different from the SOX9+LGR5+ epithelial subpopulation found by Garcia-Alfonso et al 20 . Analysis of peritoneal lesion myeloid populations of immunomodulatory Mφ and DCs revealed a coordinated immunotolerant phenotype in the disease microenvironment, promoting immunosurveillance escape and thus conducive to lesion establishment. Overall, the data demonstrate that immune and vascular components of peritoneal endometriosis favor neo-angiogenesis and an immune tolerant niche in the peritoneal cavity, beneficial for lesion establishment and growth. These are remarkable findings given that most subjects were on hormone suppressive therapies for pain [progestin (+ estrogen)] and raises the issue of how efficacious these therapies are on specific cells and the role of estrogenic components in disease management. This valuable single cell atlas of disease from patients largely on hormonal treatments offers the opportunity for comparisons of cell types and features in peritoneal lesions and endometriomas in the setting of other hormonal treatments and/or in various phases of the cycle to interpret efficacies of selective therapies on cell-specific pathways and cell-cell communications.