PCOS endometrium-derived epithelial organoids as a novel model to study endometrial dysfunction

PCOS is the most common endocrine disorder in women, affecting one woman out of eight ( Stener-Victorin et al. , 2024 ). It is diagnosed by the presence of at least two out of three of the ‘Rotterdam Criteria’, that is, irregular menstrual cycles, hyperandrogenism, and polycystic ovarian morphology ( Stener-Victorin et al. , 2024 ). The syndrome is characterized by metabolic (obesity and insulin resistance) and reproductive (infertility and pregnancy complications) impairments, as well as psychological distress (e.g. depression, anxiety, and negative body image) ( Joham et al. , 2022 ). PCOS is the most common cause of anovulatory infertility; however, in addition to ovarian dysfunction, endometrial abnormalities have also been reported in PCOS ( Piltonen, 2016 ; Palomba et al. , 2021 ). Several genes related to endometrial function—such as steroid hormone receptors for estrogen, progesterone, and androgen ( ESR , PGR , AR ), endometrial receptivity markers such as HOXA10 , IGFBP1 , and MUC1 , and inflammation-related markers including IL6 and CCL2 —are differentially expressed in endometrial epithelial cells (EECs) and/or stromal cells (ESCs) from women with PCOS. As normal endometrial function is key for placentation, endometrial dysfunction (abnormal gene expression, improper response to hormones, etc.) plays a central role in the development of placentation-related morbidities such as miscarriage, pre-eclampsia, and premature delivery. Indeed, women with PCOS are at an increased risk for the development of these reproductive morbidities as well as for endometrial cancer, with the latter further implying abnormal endometrial function ( Barry et al. , 2014 ; Bahri Khomami et al. , 2019 ; Lim et al. , 2019 ). Excessive body weight is a common feature in the PCOS population, with up to 60% of the women with PCOS being either overweight (BMI > 25 kg/m 2 ) or obese (BMI > 30 kg/m 2 ) ( Ollila et al. , 2016 ). Excessive body weight is also an independent contributor to endometrial dysfunction ( Smith and Smith, 2016 ; Barber et al. , 2019 ). For example, decidualization—a progesterone-driven phenomenon in ESCs that is imperative for well-coordinated embryo implantation and subsequent placentation—is impaired in primary ESCs from women with obesity ( Hawkins and Matzuk, 2008 ; Schulte et al. , 2016 ). In line with this, obesity also increases the risk for pregnancy complications like miscarriage and preeclampsia ( Catalano and Shankar, 2017 ). While several studies have demonstrated endometrial dysfunction in PCOS and/or obese endometrium, these studies have typically focused on the stromal compartment (e.g. decidualization of ESCs). Data regarding the impact of PCOS and/or obesity on EEC function are scarce. To improve our understanding of PCOS-related endometrial dysfunction, and specifically the role of the epithelial compartment, there is a great need for a research model that accurately represents the disease and enables the investigation of EECs during the implantation process. Recently, 3-dimensional in vitro organoid models have demonstrated their superiority compared to primary 2-dimensional cell cultures, as they strongly mimic the in vivo (patho)biology of the original tissue, and at the same time allow long-term expandability without loss of phenotype ( Maenhoudt et al. , 2022 ). Organoids are 3-dimensional, self-organizing multi-cellular structures created in vitro from stem cells. They are typically generated by embedding tissue (stem) cells in Matrigel or other kinds of matrix to serve as a substitute for the extracellular matrix (ECM), and cultured in medium that includes essential factors needed for maintaining stem cells in vivo ( Kretzschmar and Clevers, 2016 ). We have previously established patient-derived endometrium epithelial organoids (EEOs) from several endometrial pathologies other than PCOS, including endometriosis and endometrial cancer, and have demonstrated that these organoids maintain disease-associated traits and patient heterogeneity even after long-term expansive culture ( Boretto et al. , 2019 ). In addition, by applying a specific hormonal exposure protocol, EEOs have been shown to display hormone responsiveness that is comparable to that of the naturally cycling endometrium ( Boretto et al. , 2017 ). In this study, we aimed to establish and characterize EEOs from PCOS patient-derived endometrial biopsies. We show for the first time that PCOS EEOs recapitulate the inflammatory (gene expression) characteristics of PCOS endometrium and have a differential response to (ovarian) sex hormones, which is in line with previously described in vivo findings ( Savaris et al. , 2011 ; Piltonen et al. , 2013 ). Additionally, we show that PCOS EEOs can be established from two different subtypes of women with PCOS—that is, overweight/obese and lean—and we uncover specific features for each group. This new in vitro model provides a valuable research tool to investigate PCOS-related endometrial dysfunction in further detail.

To establish EEOs, endometrial biopsies were collected from a cohort of women who were overweight/obese (O-cohort) who were diagnosed with PCOS (meeting all three Rotterdam criteria) (O-PCOS), and from BMI-matched women without PCOS symptoms (O-Ctrl) ( Table 1 ). Organoid formation could be observed as soon as 2 days after tissue fragment/cell seeding, after which the organoids expanded rapidly ( Fig. 1b ). The epithelial character of the O-EEOs was validated by a positive immunostaining of CDH1 ( Fig. 1c ). Next, O-EEOs were either cultured in hormone-free medium (baseline state) or exposed to different combinations of hormones to mimic the proliferative and secretory phase of the menstrual cycle (i.e. β-estradiol (E2 exposure) or a combination of E2, progesterone, cAMP, and WNT inhibitor XAV-939 (EPCX exposure), respectively). To investigate the effects of a hyperandrogenic environment (as present in women with PCOS), hormonal exposure took place in the presence or absence of DHT ( Fig. 1a ). Bulk RNA-sequencing (RNA-seq) analysis revealed that E2 exposure of O-Ctrl EEOs caused increased expression of estrogen-related genes such as ESR1 and GREB1 , as well as genes related to estrogen-induced ECM remodeling ( MMP26 , OLFM4 ) ( Pilka et al. , 2003 ; Dassen et al. , 2010 ; Tong et al. , 2019 ) ( Fig. 1d ). Furthermore, EPCX-exposed O-Ctrl EEOs demonstrated a gene signature similar to secretory phase endometrial (epithelial) cells, including the upregulated expression of receptivity-related genes leukemia inhibitory factor ( LIF ), progestagen-associated endometrial protein ( PAEP ), IGFBP1 , MUC1 , and PGR ( Press et al. , 1988 ; Jeschke et al. , 2009 ; Bhagwat et al. , 2013 ), while the expression of WNT- ( WNT7A , WNT7B , WNT8B , and WNT10A) and estrogen-related targets ( ESRRB and ESRP2 ) was reduced ( Fig. 1e ). Notably, comparison of EPCX- versus E2-treated control EEOs highlighted downregulation of PGR upon EPCX treatment ( Supplementary Fig. S1c ), which was validated via RT-qPCR ( Supplementary Fig. S1d ). In addition, while being nearly absent at baseline, PR protein expression was highly upregulated upon E2 exposure in both experimental groups ( Supplementary Fig. S1e and f ). These findings reinforce that EEOs (here derived from control overweight/obese women) are physiologically responsive to sex hormone exposure, thereby further supporting previous findings on (healthy) endometrial organoids ( Turco et al. , 2017 ; Boretto et al. , 2019 ). Next, the transcriptomes of O-PCOS and O-Ctrl EEOs at baseline state were compared using bulk RNA-seq analysis. GO enrichment analysis revealed that pathways related to inflammatory response and reactive oxygen species (ROS) regulation were significantly increased in O-PCOS EEOs, while pathways related to the establishment of apical/basal cell polarity and ECM organization were reduced compared to O-Ctrl EEOs ( Fig. 1f ). 21 DEGs were identified between O-Ctrl and O-PCOS EEOs ( Fig. 1g ). In line with the GO analysis, genes with increased expression in O-PCOS EEOs were related to inflammation. Specifically, ICAM1 , an adhesion molecule implicated in embryo attachment and induced by inflammatory signals ( Luo et al. , 2023 ), and OSMR , which encodes the receptor for the cytokine oncostatin-M ( Wolf et al. , 2023 ), were both upregulated. In addition, XDH , whose enzymatic activity generates ROS ( Battelli et al. , 2016 ) was increased. Meanwhile, genes related to ECM function ( FBN2 and GPC6 ) and cell polarity ( GOLGA7B , STPG3 , MEGF6 , and FSCN1 ) were downregulated. The gene profile of O-PCOS EEOs was further investigated by focusing on inflammation-related DEGs (O-PCOS vs O-Ctrl) with a P -value <0.05 from the RNA-seq dataset. The expression of several markers for inflammation was increased in O-PCOS EEOs, including C3, C6 , CCL2 , and VNN1 ( Supplementary Fig. S1g ). Additionally, based on the pathway enrichment related to ROS in O-PCOS EEOs, genes related to mitochondrial function were also assessed in greater detail. It was found that the expression level of several mitochondrial genes—such as MT-CO2 , MT-RNR1 , and MT-ND2 —was reduced ( P  < 0.05) in O-PCOS versus O-Ctrl EEOs at baseline ( Supplementary Fig. S1g ). The protein products of these genes are integral to mitochondrial function, with some (e.g. MT-CO2 , MT-ND2 ) forming essential subunits of respiratory chain complexes I to IV, and others (e.g. MT-RNR1 ) contributing to mitochondrial ribosome structure and function ( Li et al. , 2006 ; Barbarino et al. , 2016 ; Stroud et al. , 2016 ). Further, RNA-seq findings were validated using RT-qPCR which revealed increased expression of OSMR and ICAM1 and reduced expression of GPC6 in O-PCOS EEOs (baseline state) ( Fig. 1h ). In addition, RT-qPCR gene expression analysis of several other DEGs revealed trends (up or down) comparable to the RNA-seq findings, although not significant ( Supplementary Fig. S1h ). Increased OSMR expression was validated at the protein level by immunostaining, which demonstrated elevated OSMR expression in O-PCOS EEOs versus O-Ctrl EEOs ( Fig. 1i and j ). Finally, the transcriptomic profiles of O-PCOS EEOs at baseline were compared to available microarray data of fluorescence-activated cell-sorted (FACS) EECs from obese women with and without PCOS ( Piltonen et al. , 2013 ). Several of the findings were in line with findings from EECs from obese women with PCOS—for example FBN2 , ICAM1 , and CCL2 were consistently dysregulated in both PCOS EEOs and isolated EECs ( Supplementary Data File S1 ). Overall, these data show that O-PCOS EEOs have increased inflammatory and reduced mitochondrial gene expression, and transcriptional profiles that overlap with the gene expression profiles of isolated EECs from obese women with PCOS. Next, the effects of E2 and EPCX exposure on O-PCOS and O-Ctrl EEOs transcriptomes were assessed (see Fig. 1a ). GO enrichment analysis of E2-treated O-EEOs revealed that pathways related to peptidase and tyrosine kinase activity were enriched in O-PCOS EEOs, whereas pathways related to hormone metabolism were reduced ( Fig. 2a ). Specifically, among the 20 DEGs, ALKAL2 and PRSS1 (related to tyrosine kinase activity) were significantly increased, while the expression level of UGT2B7 (related to estrogen metabolism, specifically, encoding the enzyme catalyzing the glucuronidation of endogenous estrogen; Thibaudeau et al. , 2006 ) was significantly reduced in O-PCOS EEOs ( Fig. 2b ). O-PCOS endometrium epithelial organoids (EEOs) show aberrant responses to sex hormone exposure. ( a, c ) Gene ontology (GO) enriched pathways in overweight/obese PCOS (O-PCOS) versus control (O-Ctrl) EEOs after β-estradiol (E2) exposure (a) or exposure to E2, progesterone, cyclic adenosine monophosphate and XAV-939 (EPCX) (c). Green and blue bars represent up- and downregulated pathways, respectively. ( b, d ) Volcano plot of differentially expressed genes (DEGs; adjusted P -value ( P .adj.)<0.05) between O-PCOS and O-Ctrl EEOs after E2 (b) or EPCX (d) exposure. Log2(FC), Log2(fold change). Gene in bold was validated using RT-qPCR. Green and blue dots indicate significantly up- and downregulated expression, respectively; gray dots denote non-significant changes. ( e ) Heatmap of average Log2 Fragments Per Kilobase Million (FPKM) of receptivity-related genes in O-PCOS and O-Ctrl EEOs after E2 or EPCX exposure normalized to baseline state. The asterisk indicates significant difference between O-PCOS and O-Ctrl EEOs within the respective hormone treatment group. Dark and light green colors denote high and low expression, respectively. ( f ) RT-qPCR gene expression analysis of PRSS1 and PZP after E2 exposure, and leukemia inhibitory factor ( LIF ), progestagen-associated endometrial protein ( PAEP ), and PGR after EPCX exposure as relative to HPRT1 and GAPDH expression in O-PCOS (purple) and O-Ctrl (blue) EEOs. N = 4 independent EEO cultures from different donors per group. EEOs from low passage ( P  ≤ 5) were used. Student’s t -test or Mann–Whitney test; *= P  < 0.05; **= P  < 0.01. Data are presented as mean±SEM. The analysis of EPCX-exposed O-PCOS EEOs revealed pathway enrichment and upregulated DEGs related to leukotriene biosynthesis processes ( GGT2P ), co-factor catabolic process ( HP ), and fatty acid derivative biosynthetic process ( GGTA1P ) compared to O-Ctrl EEOs. In contrast, significantly reduced pathways and associated downregulated DEGs included heparan sulfate proteoglycan binding ( GPC6 ), metallocarboxypeptidase activity ( CPXM1 ), and s-acetyltransferase activity ( GOLGA7B ) ( Fig. 2c and d ). Further sub-analysis of receptivity-related genes (using a cut-off of P  < 0.05) revealed that the expression levels of LIF and PAEP were reduced in O-PCOS EEOs after EPCX exposure. In contrast, the gene expression levels of the adhesion molecule SPP1 and chemokine CXCL14 were increased in the O-PCOS group ( Fig. 2e ). The increased PRSS1 expression in E2-exposed O-PCOS as compared to O-Ctrl EEOs was confirmed using RT-qPCR as well as the decreased expression of PZP (selected from RNA-seq data using P  < 0.05 cut-off) ( Fig. 2f ). Comparable to the RNA-seq findings, RT-qPCR analysis revealed a decreased response of LIF expression in EPCX-exposed O-PCOS EEOs, while PAEP showed a downward trend although non-significant and PGR expression did not differ between the groups ( Fig. 2f ). Overall, O-PCOS EEOs showed abnormal expression levels of receptivity-related markers following E2 or EPCX exposure when compared to O-Ctrl EEOs. As a next step, the general morphology of the O-PCOS EEOs was assessed as compared to O-Ctrl EEOs. Overall, organoids from both groups were cystic and did not show a divergent shape ( Fig. 3a ). However, after eight days of culture, the size (diameter) of O-PCOS EEOs was significantly smaller compared to that of O-Ctrl EEOs at baseline condition. This difference was not present in E2- or EPCX-exposed O-EEOs ( Fig. 3a and b and Supplementary Fig. S2a ), or in EEOs at Day 2 ( Supplementary Fig. S2b ). We hypothesized that the reduced organoid size in O-PCOS could be due to increased cell apoptosis (which may be in line with altered expression of mitochondrial genes; see Supplementary Fig. S1g ) or reduced proliferation levels. However, no differences were identified in the expression of CC3, a common marker for apoptosis, between O-PCOS and O-Ctrl EEOs, neither at protein nor at mRNA level ( Fig. 3c , Supplementary Fig. S2c and d ). The gene expression levels of CASP1, CASP7 , and LCN2 (all involved in apoptosis) were also not different between the two groups ( Supplementary Fig. S2d ). Similarly, the proliferation rate of O-EEOs (as assessed by Ki-67 immunostaining and RT-qPCR) did not differ between Ctrl and PCOS groups ( Fig. 3c , Supplementary Fig. S2c and d ). O-PCOS endometrium epithelial organoids (EEOs) exhibit smaller size at baseline. ( a ) Representative brightfield images of overweight/obese PCOS (O-PCOS) and control (O-Ctrl) EEOs after 8 days of culture without hormones (baseline), or exposure to β-estradiol (E2), or exposure to E2, progesterone, cyclic adenosine monophosphate, and XAV-939 (EPCX). Scalebar = 200 µm. ( b ) Quantification of mean diameter from O-PCOS (purple) and O-Ctrl (blue) EEOs after 8 days in culture. Each symbol represents a different EEO sample, with the same samples shown across different hormone treatments. ( c ) Representative images of immunofluorescent staining for CC3 (green), Ki-67 (green), and immunohistochemistry staining of ERα in O-PCOS and O-Ctrl EEOs. Blue indicates DAPI-stained nuclei, scalebar = 50 µm. N = 4 independent EEO cultures from different donors per group. EEOs from low passage ( P  ≤ 5) were used. Two-way ANOVA; **= P  < 0.01. Data are presented as mean±SEM. In addition, estrogen receptor protein (ERα) expression at baseline was assessed, as it is known to mediate EEC proliferation ( Marquardt et al. , 2019 ). In line with the absence of a difference in proliferation rate between both groups, no difference in protein (ERα) or mRNA ( ESR1 ) expression was found between O-PCOS and O-Ctrl EEOs ( Fig. 3c , Supplementary Fig. S2c and d ). In summary, O-PCOS EEOs displayed a smaller size compared to O-Ctrl EEOs at baseline, but no differences were identified regarding levels of cell apoptosis or proliferation. Since hyperandrogenism is a key feature of PCOS, the effect of DHT exposure on O-PCOS and O-Ctrl EEOs transcriptomes was investigated. Intriguingly, RNA-seq analysis revealed that a 6-day DHT exposure with or without simultaneous E2 or EPCX exposure did not alter the transcriptome of O-Ctrl or O-PCOS EEOs, as no DEGs were identified apart from one. Only, CCDC74A , which encodes a microtubule-binding protein involved in chromosome alignment ( Zhou et al. , 2019 ) was differently expressed between E2-treated O-PCOS EEOs with and without DHT exposure ( Fig. 4a ). To further explore the lack of DHT-induced differences in gene expression, the gene expression levels of AR were determined with RT-qPCR, which revealed very low AR expression in O-PCOS or O-Ctrl EEOs, as opposed to the high expression levels of ESR1 ( Fig. 4b ). In support of this finding, the low expression level of AR in EECs was confirmed using a publicly available single-cell RNA-seq dataset ( Wang et al. , 2020 ): while ESR1 is highly expressed in endometrial epithelia, AR is predominantly expressed in endometrial stroma but not in EECs ( Fig. 4c ). In addition, immunostaining revealed only very few AR-positive cells in EEOs (0.22% ± 0.13 in Ctrl EEOs, 0.62% ± 0.46 in PCOS EEOs (mean ± SEM), P  = 0.827), while AR-positive ESCs were clearly present in endometrial sections throughout different phases of the menstrual cycle ( Fig. 4d and e ). Dihydrotestosterone exposure does not affect O-EEO transcriptome. ( a ) Number of differentially expressed genes (DEGs; adjusted P -value ( P .adj.)<0.05) between overweight/obese endometrial epithelial organoids (O-EEOs) cultured without or with dihydrotestosterone (D). EEOs were cultured hormone-free (baseline, B), or with β-estradiol (E), or with β-estradiol, progesterone, cAMP, and XAV-939 (EPCX, X). Gray bars show total DEGs per comparison; green and blue indicate significantly up- and downregulated DEGs, respectively. P, PCOS; C, control. ( b ) RT-qPCR gene expression analysis of androgen receptor ( AR ) and ESR1 at baseline in overweight/obese PCOS (O-PCOS) (purple) and control (O-Ctrl) (blue) EEOs as relative to HPRT1 and GAPDH . ( c ) Expression of AR and ESR1 in the different endometrial cell types as shown using dot plot (dot size: percentage of expressing cells in the indicated clusters (Y-axis); dot color scale: average expression level). Data generated from single-cell datasets of Wang et al. (2020) . ( d, e ) Representative images of immunostaining of AR in O-EEOs at baseline (d) and in O-Ctrl and O-PCOS endometrial tissue from proliferative (PE) and mid-secretory (MSE) phase (e). Arrow and asterisk indicate endometrial stromal cells (ESCs) and epithelial cells (EECs), respectively. Scalebar = 50 µm; 20× and 60× indicate magnification levels of 20- and 60-fold, respectively. N = 4 independent EEO cultures from different donors per group. EEOs from low passage ( P  ≤ 5) were used. Student’s t -test or Mann–Whitney test. Data are presented as mean±SEM. To sum up, similarly as in EECs, AR expression in O-EEOs is very low to almost absent, which likely explains the lack of effects of DHT exposure on the O-EEOs transcriptome. Although the majority of women with PCOS are overweight or obese, a subset of affected individuals has a BMI within the normal range. To explore the role of obesity in the observed O-PCOS EEO phenotype, we enriched our study sample set with EEOs derived from lean women with PCOS (L-PCOS) and BMI-matched controls (L-Ctrl) (median BMI = 20.5 and 22.9 kg/m 2 , respectively) ( Supplementary Table S1 ). Similarly to O-PCOS EEOs, the size of L-PCOS EEOs was significantly smaller than that of L-Ctrl EEOs, but unlike the O-PCOS cohort, the reduced size of L-PCOS EEOs was also detected in hormone-exposed groups (EPCX, P  < 0.01; E2, P  < 0.01; Fig. 5a and b and Supplementary Fig. S2e ). Similarly to the O-EEO cohort, no difference in size was found at Day 2 ( Supplementary Fig. S2f ). Of note, comparing L-and O-Ctrl EEOs revealed that organoid size was reduced when derived from overweight/obese individuals across different treatments, but such difference was not observed in the PCOS EEOs cohort ( Supplementary Fig. S2g and h ). L-PCOS endometrium epithelial organoids (EEOs) also show increased inflammatory gene expression. ( a ) Representative brightfield images of EEOs from lean PCOS (L-PCOS) and control (L-Ctrl) groups after 8 days of culture without hormones (baseline), or exposure to β-estradiol (E2), or exposure to E2, progesterone, cyclic adenosine monophosphate, and XAV-939 (EPCX). Scalebar = 200 µm. ( b ) Quantification of average EEO diameter of L-PCOS (purple) and L-Ctrl (blue) EEOs after 8 days of culture. Each symbol represents a different EEO sample, with the same samples shown across different hormone treatments. ( c ) RT-qPCR gene expression analysis of androgen receptor ( AR ) and ESR1 at baseline in L-PCOS (purple) and L-Ctrl (blue) EEOs as relative to GAPDH . ( d, e, f ) RT-qPCR gene expression analysis of OSMR, ICAM1 , and GPC6 at baseline (d), PRSS1 and PZP after E2 exposure (e), and leukemia inhibitory factor ( LIF ), progestagen-associated endometrial protein ( PAEP ), and PGR after EPCX exposure (f) as relative to GAPDH . L-PCOS (purple) and L-Ctrl (blue). ( g, i ) Representative images of immunostaining of PR at baseline and after E2-exposure (g); and of OSMR at baseline (i) in L-EEOs. Green indicates PR or OSMR protein, blue indicates DAPI-stained nuclei. Scalebar = 100 µm. ( h, j ) Proportion of PR- (h) and OSMR- (j) positive cells in L-PCOS (purple) and L-Ctrl (blue) EEOs. N = 3 (for size measurements), 4 (for RT-qPCR), or 3–4 (for immunostainings) independent EEO cultures from different donors per group. EEOs from low passage ( P  ≤ 5) were used. Two-way ANOVA (b). Student’s t -test or Mann–Whitney test (c–f); *= P  < 0.05; **= P  < 0.01. Data are presented as mean±SEM. Further, AR expression was also very low in lean EEOs (L-EEOs) as opposed to higher ESR1 expression ( Fig. 5c ), thereby mirroring findings from the O-EEOs. The upregulation of inflammatory genes OSMR and ICAM1 was also observed in L-PCOS EEOs compared to their Ctrl EEOs (at baseline) ( Fig. 5d ), which is similar to findings in the O-cohort ( Fig. 1h ). On the other hand, the decreased expression of GPC6 was not observed in the L-cohort ( Fig. 5d vs Fig. 1h ). In addition, E2 did not significantly alter the expression of PRSS1 and PZP , which is in contrast with observations in the O-cohort ( Fig. 5e vs Fig. 2f ), and may be due to their extremely low expression in L-EEOs. PAEP and LIF expression following EPCX exposure showed no significant change in L-PCOS EEOs, whereas PGR expression showed a significantly lower response, which was different from O-PCOS EEOs ( Fig. 5f vs Fig. 2f ). Lastly, immunostaining for PR and OSMR mirrored the O-cohort results: baseline L-EEO exhibited low PR expression, which increased following E2 treatment ( Fig. 5g and h ), and baseline L-PCOS EEOs showed significantly higher OSMR expression compared to L-Ctrl EEOs ( Fig. 5i and j ). Taken together, the inflammatory gene profile and reduced organoid size at baseline found in O-PCOS EEOs is also present in L-PCOS EEOs, but gene expression responses to hormonal exposure appear to differ between the L- and O-cohorts.

The Ethical Committees of Northern Ostrobothnia Hospital District (65/2017) and UZ/KU Leuven ( S65570 ) approved this study. Informed consent forms were obtained prior to participation in accordance with the Declaration of Helsinki. Endometrial biopsies from women with PCOS (3/3 Rotterdam criteria fulfilled for overweight/obese cohort (O-cohort), minimal 2/3 criteria fulfilled for lean cohort (L-cohort), as diagnosed by physicians in Oulu University Hospital or UZ Leuven) and non-PCOS controls were collected at the Oulu University Hospital and UZ Leuven. The Rotterdam criteria are as follows: (i) ovulatory dysfunction (menstrual cycle length  35 days, or  4, or free androgen index  > 4.5); (iii) polycystic ovarian morphology on ultrasound ( > 20 follicles per ovary, or ovarian volume  > 10 ml) ( Teede et al. , 2023 ). The women were between 24 and 39 years old. All samples, except for one sample per experimental group, were derived from proliferative phase endometrium. Previous studies have shown that the phase of the menstrual cycle has no effect on the organoid phenotype ( Boretto et al. , 2017 ; Cindrova-Davies et al. , 2021 ). The endometrial biopsies were collected from the volunteers using a plastic suction curette (Pipelle, Cooper Surgical, USA). The control group included regularly cycling (25–35 days) women without PCOS symptoms. The exclusion criteria for both groups included pregnancy, breastfeeding for 3 months prior to sampling, use of hormones or hormone-containing products for 2 months prior to sampling, and smoking. Fasting blood samples were collected for the O-cohort during the proliferative phase or for anovulatory cases on any suitable day (University of Oulu). Fasting blood samples from the L-cohort were taken at the time of biopsy collection at a random timepoint of the menstrual cycle (UZ/KU Leuven). For the O-cohort, anti-Müllerian hormone (AMH), LH, FSH, sex hormone-binding globulin (SHBG), testosterone, fasting glucose, and insulin were measured as described in ( Lee et al ., 2024 ). For the L-cohort, glucose was measured using the Cobas hexokinase test (Roche Diagnostics, Switzerland). Insulin, AMH, LH, FSH, SHBG, and testosterone were measured using Elecsys assays (Roche Diagnostics). Patient characteristics are presented in Table 1 (O-cohort) and Supplementary Table S1 (L-cohort). Overweight/obese cohort participant characteristics. Data are presented as median (interquartile range (IQR)). P  < 0.05 depicted in bold. O-Ctrl, overweight/obese control women; O-PCOS, overweight/obese women with PCOS; HOMA-IR, homeostatic model assessment for insulin resistance; AMH, anti-Müllerian hormone; E2, estradiol; P4, progesterone; T, testosterone; SHBG, sex hormone-binding globulin; FAI, free androgen index. As summarized in Supplementary Fig. S1a , endometrial Pipelle samples were rinsed with Ca 2 +  /Mg 2 +  -free phosphate-buffered saline (PBS, Gibco, USA) followed by 30 min of digestion with collagenase (6.4 mg/ml Collagenase Type I (Sigma-Aldrich, USA) and 100 U/ml hyaluronidase (Merck, Germany) in Hanks Buffered Salt Solution (Gibco, with Ca 2 +  and Mg 2 +  )) and mechanical trituration. Epithelial fragments were collected via passage of the cell suspension through a cell strainer of minimum 40 µm, collection of the fragments on the cell strainer, and optional purification of the fragments by 2-h culture in stromal cell media (in which the stromal cells adhere to the vessel bottom) (see Piltonen et al. , 2015 ). Next, the epithelial fragments were either cryopreserved in liquid nitrogen in 10% dimethyl sulfoxide (DMSO, Sigma-Aldrich) in Keratinocyte serum-free medium (SFM) (1% fetal bovine serum (FBS, charcoal stripped, BioIVT, USA), 0.2% defined keratinocyte SFM supplement (ThermoFisher Scientific, USA) in defined keratinocyte SFM basal medium (Gibco)) or used immediately for organoid establishment by mechanically triturating the fragments with a 200-µl pipette tip and centrifugation at 400  g for 5 min. Next, the pellet was resuspended in 70% Matrigel (growth factor-reduced, Corning, USA, catalog no. 356231)/30% Dulbecco’s Modified Eagle Medium F12 (DMEM/F12, Gibco) supplemented with 1:500 rock inhibitor (RI, Y0503, Merck). This step was followed by plating 20 µl drops on prewarmed 48-well plates. After a 30-min inverted incubation, organoid medium was added (see Supplementary Table S2 ) ( Boretto et al. , 2019 ). In this study, organoids were generated from both cryopreserved epithelial fragments (after thawing in 10% FBS in DMEM/F12), and from fresh epithelial fragments, with no phenotypic differences being observed, as previously demonstrated ( Boretto et al. , 2017 ; Heidari-Khoei et al. , 2022 ). Organoids were cultured at 37°C with 5% CO 2 and passaged every 7–10 days by dissociating the Matrigel drop with ice-cold DMEM/F12 followed by disintegrating the organoids with TrypLE (Gibco) and mechanical trituration with 1:1000 RI. The obtained cell suspension was either re-plated (1:4 expansion ratio) or cryopreserved in cryo-medium (90% FBS, 10% DMSO) for subsequent use. Organoids from low passage ( P  ≤ 5) were utilized for all experiments. Organoids were passaged ( P  ≤ 5) and 2 days later, they were exposed to different combinations of hormones, i.e. no hormones (baseline), or 6 days in 10 nM β-estradiol (E2, Sigma-Aldrich), or 2 days in 10 nM E2, followed by 4 days in a combination of 10 nM E2, 1 µM progesterone (Sigma-Aldrich), 0.25 mM cyclic adenosine monophosphate (cAMP, Tocris, UK), and 10 µM Wnt/β-catenin signaling (WNT) inhibitor XAV-939 (Tocris), all in the presence or absence of 100 nM dihydrotestosterone (DHT, Sigma-Aldrich) ( Fig. 1a ). The dose of 100 nM DHT was selected based on its activity on Ishikawa cells (a human endometrial epithelial cell line) ( Cermik et al. , 2003 ) and on PCOS endometrial tissue ( Li et al. , 2015 ; Hu et al. , 2021 ). The WNT inhibitor XAV-939 was added to mimic the secretory-phase hormonal environment as WNT signaling is known to be downregulated during this phase of the cycle ( Y. Wang et al. , 2010 ). The medium was refreshed every 2 days, and all organoids were collected 8 days after passaging for further analysis. Establishment and characterization of O-PCOS and O-Ctrl endometrium epithelial organoids (EEOs). ( a ) Hormone exposure protocol. EEOs were cultured for 8 days without hormones (baseline), or exposed to 10 nM β-estradiol (E2 exposure) for 6 days, or exposed to 10 nM E2 for 2 days, followed by 4-day exposure to a combination of 10 nM E2, 1 µM progesterone (P4), 0.25 mM cyclic adenosine monophosphate (cAMP), and 10 µM XAV-939 (EPCX exposure), all in the presence/absence of 100 nM dihydrotestosterone (DHT). ( b ) Representative brightfield images of overweight/obese PCOS (O-PCOS) or control (O-Ctrl) EEOs during 8-day culture without hormone exposure; scalebar = 200 µm. ( c ) Representative images of hematoxylin and eosin (H&E) staining and immunostaining for CDH1 (red) in EEOs and in endometrial (endom.) tissue as positive control. Blue indicates DAPI-stained nuclei, scalebar = 50 µm. ( d, e ) Volcano plots of differentially expressed genes (DEGs; adjusted P -value ( P .adj.)<0.05) in O-Ctrl EEOs after E2 (d) or EPCX exposure, (e) compared to non-exposed (baseline) Ctrl EEOs. Log2(FC), Log2(fold change). Green and blue dots indicate significantly up- and downregulated expression, respectively; gray dots denote non-significant changes. ( f ) Gene ontology (GO) enriched pathways ( P .adj.<0.05) in O-PCOS versus O-Ctrl EEOs at baseline. Green and blue bars represent up- and downregulated pathways, respectively. ( g ) Volcano plot of DEGs ( P .adj.<0.05) between O-PCOS and O-Ctrl EEOs after 8-day culture at baseline. Genes in bold were validated using RT-qPCR (see Fig. 1h ). ( h ) RT-qPCR gene expression analysis of OSMR, ICAM1 , and GPC6 in O-PCOS (purple) and O-Ctrl (blue) EEOs at baseline as relative to HPRT1 and GAPDH expression. ( i ) Representative images of immunostaining of OSMR (green) in O-EEOs of Ctrl and PCOS at baseline. Blue indicates DAPI-stained nuclei, scalebar = 100 µm. ( j ) Proportion of OSMR-positive cells in O-PCOS (purple) and O-Ctrl (blue) O-EEOs. N = 4 independent EEO cultures from different donors per group. EEOs from low passage ( P  ≤ 5) were used. Student’s t -test or Mann–Whitney test; *= P  < 0.05; ***= P <0.001. Data are presented as mean±SEM. At Days 2 and 8 after passaging, organoids were imaged using brightfield microscopy (MZ6 or DFC425 (Leica, Germany) cameras, or with Axiovert 40 CFL (Zeiss, Germany)) and organoid size and morphology were assessed by measuring the diameter of 100 randomly chosen organoids per sample. The geometric average of the 100 measurements per sample was used to assess differences between PCOS and control (Ctrl) EEOs. Organoids from similar passage numbers were used to compare groups. All EEOs were from low passage ( P  ≤ 5). Organoids were removed from the Matrigel with TrypLE, fixed for 1 h in paraformaldehyde (PFA, 4% in PBS), and embedded in agarose gel (2% in PBS) followed by paraffin-embedding and sectioning. Endometrial tissue was fixed overnight (ON) in PFA followed by paraffin-embedding and sectioning. After 1 h of baking at 55°C, 5 µm sections were de-paraffinized and re-hydrated in xylene, absolute ethanol, and 96% ethanol. Sections were subjected to hematoxylin and eosin (H&E) staining or immunostaining for estrogen receptor alpha (ERα) (1/200, Abcam, UK, ab16660) and AR (1/100, Abcam, ab133273) using the EnVision FLEX +  kit (Agilent, USA). Heat-induced epitope retrieval in citrate buffer (pH 6) was performed in the microwave (2 min at 800 W and 10 min at 150 W) and blocking was done for 30 min in 10% normal goat serum (NGS, Gibco) and 0.1% Triton X-100 (Sigma-Aldrich) in PBS before primary and secondary antibody incubation. After primary antibody (ON,  + 4°C) and secondary antibody exposure (diaminobenzidine, 10 min, room temperature), slides were counterstained in hematoxylin (Sigma-Aldrich), followed by dehydration, mounting (Micromount, Leica), and imaging (20 × , Zeiss Axio Imager). For immunofluorescent staining of Ki-67 (1/100, Abcam, ab16667), CDH1 (1/200, Abcam, ab231303), and CC3 (1/400, Cell Signaling Technology, USA, 9661), slides were blocked for 30 min before primary and secondary antibody incubation using 10% NGS, 0.1% Tween 20 (Sigma-Aldrich), 0.01% Triton X-100, 1% bovine serum albumin (Sigma-Aldrich), and 0.2 M glycine (Sigma-Aldrich) in PBS. For immunofluorescent staining of PR (1/300, Proteintech, USA, 25871-1-AP), and Oncostatin M Receptor (OSMR) (1/100, Proteintech, 10982-1-AP), slides were blocked for 1 h before primary and secondary antibody exposure using 5% donkey serum (Sigma-Aldrich) and 0.1% Tween 20 in Tris-buffered saline. Next, primary antibody (ON,  + 4°C) and secondary antibody (1:800, 1 h, Alexa Fluor 488 or 546, ThermoFisher) incubation was done, followed by 5-min incubation with Hoechst (1:1000, ThermoFisher) and mounting (Immu-Mount, Fisher Scientific). Imaging was conducted using the Stellaris 8 Dive confocal microscope at 20 ×  (Leica). Percentages of OSMR-, CC3-, Ki-67-, ERα-, and PR-positive cells were determined from immunostainings using ImageJ software. Automatic thresholding was applied to identify and quantify immune-positive cells. Total cell counts were obtained from nuclear Hoechst staining. For the ERα and PR antibodies, expression of the target proteins was validated in endometrial tissue samples from women with and without PCOS, collected from both the proliferative and secretory phase of the menstrual cycle ( Supplementary Fig. S1b ). Organoids were removed from the Matrigel with TrypLE followed by RNA extraction and DNase treatment using the RNeasy Mini kit (Qiagen, Germany) according to the manufacturer’s instructions. Quality and quantity checks were done with Dropsense 16 (Trinean, Belgium). The RNA was submitted to Novogene Europe (UK) for further processing. Poly-T oligo-attached magnetic beads were used for mRNA purification followed by fragmentation and the first- and second-strand cDNA synthesis using random hexamer primers and dUTP or dTTP, respectively. Library check was done using Qubit real-time PCR for quantification and bioanalyzer was used for size distribution detection. After cluster generation, the library preparations were sequenced using Illumina NovaSeq 6000 platform (Illumina, USA) to generate paired-end reads. Raw reads (fastq format) were processed by Novogene’s in-house scripts to remove reads containing adapters, reads containing poly-N, and low-quality reads from raw data. Cleaned reads were mapped to the reference genome (Homo_sapiens_Ensemble_94) using Hisat2 v2.0.5. FeatureCounts v1.5.0-p3 was used to count reads mapped to each gene followed by calculation of the expected number of fragments per kilobase of transcript per million mapped (FPKM). Differential expression analysis was performed using DESeq2 R package (1.20.0). The resulting P -values were adjusted using Benjamini and Hochberg’s approach for controlling the false discovery rate ( P .adj.). Unless stated otherwise, genes with P .adj.<0.05 were assigned as differentially expressed (DEGs). The ClusterProfiler R package was used to test DEGs enrichment in Gene Ontology (GO) pathways. GO terms or pathways with P .adj.<0.05 were considered significantly enriched. The lists of DEGs and enriched GO pathways are provided in Supplementary Data Files S1 and S2 , respectively. For reverse-transcription quantitative PCR (RT-qPCR), cDNA was synthesized using the RevertAid First Strand cDNA Synthesis Kit (Thermo Scientific) or Superscript III First-Strand Synthesis SuperMix (ThermoFisher Scientific) followed by qPCR using triplicate cDNA samples on the StepOneTM Real-Time PCR system (Applied Biosystems, USA). Forward and reverse primers were designed using PrimerBank and/or PrimerBlast. The primers are listed in Supplementary Table S3 . HPRT1 and/or GAPDH were used as housekeeping gene(s). Relative expression levels are shown as 2  − dCt (mean ± SEM) in which dCt = Ct target gene  − Ct housekeeping gene . Data analysis was performed using GraphPad Prism 10. A minimum of three replicates per experiment was used. Normality was assessed using Shapiro–Wilk’s normality test after which differences between two groups were determined using Student’s t -test or the Mann–Whitney test (in the case of a non-normal distribution). Alternatively, differences between more than two groups were determined by one-way ANOVA, two-way ANOVA, or a non-parametric alternative. The RT-qPCR data were tested for outliers using the following method: datapoints Q3 + (1.5 × IQR) were considered as outliers, where the interquartile range (IQR) = Q3 − Q1. Differences were considered statistically significant at P  < 0.05. Data are presented as mean ± SEM, unless stated otherwise.

The incidence of subfertility in women with PCOS is 15-fold higher compared to women without PCOS, independent of BMI ( Stener-Victorin et al. , 2024 ). Although anovulation is the main contributor to PCOS-related adverse fertility outcomes, the role of the endometrium remains underexplored, particularly in its contribution to an abnormal environment for embryo implantation, impaired maternal-embryo interactions, defective placental development, and increased risks of miscarriage and pregnancy complications in women with PCOS. Furthermore, details of the molecular and functional abnormalities, including the altered expression of genes critical for endometrial receptivity, are still lacking. To improve the reproductive outcomes for women with PCOS, there is a great need for in vitro research models that faithfully recapitulate the disease and enable investigation of the specific molecular pathways that may be affected. Our study is the first to show the establishment of EEOs starting from PCOS patient-derived endometrial biopsies. Interestingly, we found that PCOS EEOs show reduced size, increased inflammatory phenotype, and an aberrant expression of receptivity-related genes compared to healthy endometrium-derived EEOs. Regarding the correlation between dysregulated hormonal signaling and PCOS-related endometrial impairment, the PCOS endometrium has been previously shown to exhibit altered expression of sex hormone receptors (ERα, PR) and their nuclear receptor co-regulators (AIB1/SRC-3 and ARA70) during the mid-secretory phase, indicating increased sensitivity to estradiol and decreased sensitivity to progesterone ( Quezada et al. , 2006 ; Hulchiy et al. , 2016 ). Additionally, our prior research demonstrated that decidualized ESCs in PCOS patients had altered responses to androgens ( Khatun et al. , 2021 ) and ESCs derived from PCOS patients exhibited a reduced response to progesterone ( Lavogina et al. , 2021 ), supporting a disruption in hormonal signaling within the PCOS endometrium. However, beyond abnormal hormonal signaling, there are other intrinsic abnormalities in the endometrium of PCOS patients. Plenty of clinical studies have shown that even when ovulation is pharmacologically restored, PCOS patients still experience a higher rate of miscarriage compared to those undergoing assisted reproduction for other reasons ( Matorras et al. , 2023 ). Besides, microarray analysis of the whole endometrium during the window of implantation in PCOS patients has revealed distinct differences in gene expression compared to non-PCOS women ( Qiao et al. , 2008 ; Rashid et al. , 2020 ). These differences involve genes critical for membrane function ( CXADR , TRPM6 ), cell adhesion ( TM4SF4 , CTNNA2 , ITGB4 ), and invasion ( MMP26 , TFPI2 ), and cytoskeletal dynamics ( ANK3 , PKP2 ), reflecting inherent abnormalities in endometrial biology of women with PCOS. Only a few studies have investigated the epithelial compartment of PCOS endometrium ( Piltonen et al. , 2013 ), and many studies lack appropriate BMI-matching, which is imperative in PCOS research. Therefore, the first sample cohort for this study was created from women diagnosed with PCOS who were obese or overweight, along with BMI-matched controls. Given that up to 60% of women with PCOS are overweight or obese, studying this weight group is highly relevant ( Barber et al. , 2019 ). All patients presented with severe PCOS phenotypes, and represented all three Rotterdam criteria and oligo- or amenorrhea. To investigate the role of excessive body weight in the PCOS endometrium phenotype, and given that lean women with PCOS also experience subfertility and adverse pregnancy outcomes ( Bahri Khomami et al. , 2019 ), we also established EEOs from women with PCOS with BMI in the normal range and their BMI-matched controls. We searched the literature for previous transcriptomic studies focusing on PCOS endometrium. However, at present, only one study has compared the transcriptome of EECs of women with PCOS and BMI-matched controls. Piltonen et al. (2013) performed a microarray analysis of EECs isolated by FACS derived from proliferative-phase endometrial biopsies from obese women with PCOS and BMI-matched controls. Interestingly, several dysregulated genes were shared between primary PCOS EECs and PCOS EEOs, including FBN2 , ICAM1 , and CCL2 , thereby demonstrating the translational relevance of the PCOS EEO model. Moreover, our findings also show similarities with studies using organoid models from other endometrial diseases, including endometriosis and endometrial cancer ( Boretto et al. , 2019 ). Epithelial receptivity-related markers PAEP and LIF were also downregulated in EEOs derived from endometrial cancer. Additionally, several genes related to androgen signaling and carcinogenesis (including RD5A2 and NWD1 ; Henderson and Feigelson, 2000 ; Correa et al. , 2014 ), and EEC adhesion and proliferation (including CRISP3 ; Evans et al. , 2015 )) were differentially expressed in both O-PCOS EEOs and EEOs derived from ectopic endometrium of endometriosis patients ( Boretto et al. , 2019 ). The finding of common DEGs between PCOS EEOs and EEOs from other endometrial disorders further underscores dysregulated endometrial function in PCOS, specifically in the epithelium, and possibly suggests the involvement of common molecular pathways across these conditions. One of the major findings of the transcriptome analysis was the increased expression of inflammation-related genes in PCOS EEOs at baseline, both in O- and L-cohorts. This similar finding in both BMI cohorts supports that this feature is related to PCOS status, rather than BMI. Systemic inflammation as well as reproductive tissue-related inflammation (i.e. in the ovaries, endometrium, and placenta) has been widely described in women with PCOS ( Piltonen, 2016 ; Arffman et al. , 2019 ; Velez et al. , 2021 ). Among the genes that were upregulated in PCOS EEOs, ICAM1 has also been shown to be increased in PCOS epithelium ( Piltonen et al. , 2013 ). Additionally, VNN1 —encoding an enzyme that hydrolyzes pantetheine to produce pantothenic acid and cysteamine, both of which are implicated in the regulation of oxidative stress and inflammation ( Yu et al. , 2024 )—has been shown to be differentially expressed in the endometrium of infertile obese women with PCOS ( Van Diepen et al. , 2016 ; Salamun et al. , 2022 ). Thus, our PCOS organoid model recapitulates inflammatory (gene expression) characteristics described in PCOS endometrium. Nevertheless, it should be noted that future studies are needed to further corroborate this finding. One of the features that contribute to inflammation may be altered mitochondrial function, as mitochondria-derived constituents can promote inflammation via several mechanisms ( Marchi et al. , 2023 ). Interestingly, mitochondrial defects have previously been reported in reproductive tissues of women with PCOS ( Siemers et al. , 2023 ). In line with this, the PCOS EEO model showed dysregulated expression of several mitochondria-related genes (including decreased MT-CO2, MT-RNR1 , and MT-ND2 expression) as well as enrichment of the GO pathway related to ROS regulation (specifically, XDH was upregulated). Thus, our data support that mitochondrial function may be dysregulated in O-PCOS EECs, which may contribute to the inflammatory phenotype, both of which are recapitulated in the organoid model. The inflammatory phenotype may also be connected to steroid hormone imbalance. While estrogen is a promoter of inflammation, progesterone is known for its anti-inflammatory effects ( García-Gómez et al. , 2019 ). Several endometrial diseases are associated with progesterone resistance, which refers to the diminished responsiveness of target tissues to available progesterone ( Chrousos et al. , 1986 ), for example in endometriosis ( McKinnon et al. , 2018 ). Progesterone signaling is typically evaluated by examining progesterone-responsive genes such as PGR and LIF ( Savaris et al. , 2011 ). Previous studies, including our own work, have suggested an impaired progesterone response in the endometrium of women with PCOS based on microarray analysis and wide-scale proteomic and phosphoproteomic screens ( Savaris et al. , 2011 ; Piltonen et al. , 2015 ; Lavogina et al. , 2021 ). Although we observed reduced PGR expression in response to hormonal (EPCX) exposure in L-PCOS EEOs (suggesting an altered progesterone-related response), no difference in PGR protein expression was found in baseline or E2-treated EEOs across BMI groups, as assessed by immunostaining. Whether the lean PCOS phenotype is more associated with, and affected by, altered progesterone signaling as compared to the overweight/obese phenotype should be further assessed with a larger cohort. Interestingly, a previous study identified ‘metabolic’ and ‘reproductive’ subgroups based on clinical characteristics among women with PCOS, also supported by genetic findings ( Dapas et al. , 2020 ). Whether the O-PCOS and L-PCOS cohorts represent the ‘metabolic’ and ‘reproductive’ subgroups, respectively, remains to be assessed. Regarding epithelial receptivity markers following EPCX exposure, the expression of the key receptivity genes PAEP and LIF ( Maenhoudt et al. , 2022 ) was found to be lower in O-PCOS EEOs than O-Ctrl EEOs in the sequencing data. This pattern was not observed in the L-cohort, thereby supporting that obesity may be a contributing factor in the impaired endometrial receptivity in women with PCOS. In line with this, an improved endometrial receptivity profile and increased likelihood of spontaneous conception have been reported in obese infertile women with PCOS following a weight-loss program ( Bergant et al. , 2022 ). Additionally, it has been shown that the pregnancy success rate is decreased, and the risk of preterm birth is increased, in obese versus normal-weight women with PCOS ( De Frène et al. , 2014 ; Haas et al. , 2023 ), again supporting the idea of distinct phenotypic subtypes in the PCOS population. We have previously shown abnormal receptivity-related gene expression in primary PCOS ESCs ( Khatun et al. , 2021 ), but a recent study that utilized targeted gene expression profiling for endometrial receptivity testing could not identify any differences in the expression of 68 endometrial receptivity genes between ovulatory women with PCOS and healthy controls in mid-secretory phase ( Meltsov et al. , 2023 ). The women included in the latter study were not obese; thus, the findings are in line with our results in the L-cohort. Intriguingly, we observed a smaller size of PCOS EEOs when compared to their respective controls in both BMI cohorts. The difference in size was not present at Day 2 and only became evident after 8 days of culture. A possible explanation could be that the difference was too small to be observed at the earlier timepoint, but this remains to be further investigated. In addition, the obesity factor further restricted EEO size, as evidenced by the reduced size of O-Ctrl EEOs versus L-Ctrl EEOs. Similarly, a smaller size has been reported for intestinal organoids from obese versus lean populations, which can potentially be attributed to impaired stem cell differentiation in the organoids of the obese group ( Farhadipour et al. , 2023 ). The smaller size of the PCOS EEOs was not due to increased apoptosis or decreased proliferation rate. It could be hypothesized that the increased inflammatory nature along with altered mitochondrial gene expression might affect the size of the PCOS EEOs, warranting further research. Hyperandrogenism is known as one of the hallmarks of PCOS and has been shown to disrupt normal expression of receptivity-related genes and found to affect ESC steroid hormone responsiveness and mitochondrial function ( Yusuf et al. , 2023 ). A previous study showed that exposure of an in vitro endometrial tissue model to high testosterone concentrations (3 vs 0.8 nM in the control group) in the presence of physiological estrogen concentrations caused increased cell proliferation and differential gene expression related to proliferation, migration, and nuclear receptor transcription ( Wiwatpanit et al. , 2020 ). However, here, we did not observe any effect on global gene expression (through bulk RNA-seq analysis) in PCOS or Ctrl EEOs following a 6-day exposure to 100 nM of the non-aromatizable androgen DHT. This lack of effect is in line with the low to almost absent AR expression in EEOs, which accords with the low expression in EECs and high expression in ESCs, as evidenced by publicly available single-cell RNA-seq data ( Wang et al. , 2020 ). It must be noted that the endometrial model of Wiwatpanit et al. (2020) not only included EECs but also ESCs. As we have previously shown that ESCs do respond to DHT exposure and that this response is aberrant in PCOS-derived ESCs compared to controls ( Khatun et al. , 2021 ), the presence of the stromal component could explain the detection of an effect of testosterone (i.e. through the ESCs) in that study ( Wiwatpanit et al. , 2020 ). Moreover, testosterone—in contrast to DHT—can be aromatized to estrogen via CYP19A1, which may cause confounding effects ( Gibson et al. , 2016 ). Taken together, future research on the effects of a hyperandrogenic environment should utilize a co-culture (assembloid) system that includes both (organoid-derived) EEC and ESC cell populations (see, e.g. Rawlings et al. , 2021 ). Several limitations have to be considered here. Our study is the first proof-of-concept study to introduce a new model for future PCOS research. Given its exploratory nature, the sample size was relatively small, which, along with patient heterogeneity, may have contributed to the revealing of relatively subtle rather than more pronounced differences. By sharing the establishment and first characterizations of this new model, we aim to provide a foundational framework for future investigations, enabling deeper exploration and elucidation of the physiological relevance of our initial observations. Additionally, the study focused solely on the epithelial compartment without presence of ESCs. Future studies utilizing co-culture models in combination with optimized hydrogels are expected to better mimic the in vivo tissue environment and provide a more holistic understanding ( Gnecco et al. , 2023 ). Moreover, while we detected increased expression of several markers linked to an inflammatory phenotype in EEOs, the immune cell population is lacking in the established EEO model. Therefore, the findings should be taken with some caution and should be further validated using co-culture models containing an immune cell component. Another consideration is the use of both frozen and fresh tissue materials for the establishment of the EEOs. We ( Boretto et al. , 2017 ) and others ( Heidari-Khoei et al. , 2022 ) have previously shown that the use of organoids from fresh versus cryopreserved tissues does not affect the organoid phenotype, and neither does the phase of the menstrual cycle during which the endometrial biopsy is taken ( Boretto et al. , 2017 ; Cindrova-Davies et al. , 2021 ). Importantly, the possibility of using both fresh and frozen tissues for PCOS EEO establishment is highly relevant for biobanking purposes. Several notable strengths of this study include the involvement of two well-defined populations of overweight/obese and lean women with PCOS and BMI-matched healthy controls. Strict BMI matching is particularly crucial since BMI plays a significant role in the PCOS phenotype. While no significant differences in patient characteristics were observed between the control and PCOS groups in the L-cohort (diagnosis based on oligo-anovulation and clinical hyperandrogenism), differences in EEO phenotype were still evident. It can be speculated that future studies involving stronger lean PCOS phenotypes, including biochemical hyperandrogenism, could uncover more pronounced phenotypic differences with the healthy lean controls. Lastly, our study focused on the epithelial component of the endometrium in PCOS, which is understudied. Our findings propose potential research links to increased risk of pregnancy complications in women with PCOS, such as the role of inflammation on PCOS endometrium, which may contribute to aberrant placentation and subsequent placental function ( Bahri Khomami et al. , 2024 ). Furthermore, the results might provide hints to understanding the increased risk of endometrial cancer among women with PCOS, thereby suggesting that inflammation and abnormal levels of ROS and cellular stress in the EECs associated with PCOS may contribute to the onset of endometrial cancer. In summary, the newly developed endometrium organoid model for PCOS is a valuable tool to investigate endometrial epithelium-related subfertility in women with PCOS and can be employed in future co-culture studies and in research focusing on improving endometrial health and fertility outcomes for affected women.