Equine endometrial organoids preserve tissue structure and cycle-stage transcriptional identity†

To characterize the developmental trajectory of EEO, we assessed their morphology, growth dynamics, and viability during early passages (P0-P2). Organoids were successfully established from all mares ( n  = 5) in both estrus ([P 4 ]  3.1 ng/mL). Histological (H&E) and phase-contrast imaging revealed similar structural characteristics between estrus- and diestrus-derived organoids ( Figure 2A–H ). EEO-P2 predominantly formed spherical structures composed of an outer layer of simple cuboidal to low columnar epithelial cells lining a central cavity (cystic-like structure). The epithelial cells exhibited a lacy to vacuolated eosinophilic cytoplasm, cytoplasmic blebbing, and round to oval nuclei with finely stippled chromatin and one to four acidophilic nucleoli and were arranged predominantly in a single layer supported by a thin basement membrane. Occasionally, the epithelial lining projected into the central cavity, suggesting apical-in orientation. The lumina of the EEO variably contained pale, lacy to homogenous eosinophilic fluid. Some organoids displayed a disorganized, pseudostratified epithelial lining with increased cytoplasmic vacuolation. The central cavity of these organoids contained sloughed, hypereosinophilic and karyorrhectic epithelial cells. Across all epithelial populations, moderate anisocytosis and anisokaryosis were observed, with sporadic karyomegaly and frequent mitotic figures, including scattered bizarre mitoses ( Figure 2A–B ). EEO-P2 viability was comparable between estrus- and diestrus-derived cultures, averaging approximately 70% ( Figure 2I–J ; Supplementary Figure S1 ). Growth rate, defined by the time required to reach passage readiness (based on organoid size and number), did not differ between estrus- and diestrus-derived organoids ( Figure 2K ). Diameter ( Figure 2L ) and cell number ( Figure 2M ) were also not significantly different between Estrus-EEO-P2 and Diestrus-EEO-P2. Morphology, growth, and viability assessment of EEO (P0-P2). (A–B) Histology images of passage 2 organoids (EEO-P2) derived from mares in estrus and diestrus, respectively, highlighting their cystic nature. (C–H) Phase-contrast images of EEO at P0, P1, and P2, derived from mares in estrus and diestrus, illustrating their cystic architecture. (EEO = equine endometrial organoids, P0 = passage 0, P1 = passage 1, P2 = passage 2; Scale bar = 50 μm) (I–J) Fluorescence images of passage 2 organoids stained with Hoechst (blue) and PI (red) to assess viability. Both estrus- and diestrus-derived EEO-P2 had an average cell viability of 70% (Scale bar = 50 μm). (K) Growth rate of EEO, defined as the number of days required for the next passage. Estrus-EEO reached passage readiness in a median of 5 days (4–5; min–max) or 4.6 ± 0.55 days (mean ± SD) for P1 and in a median of 5 days (4–5; min–max) or 4.8 ± 0.45 days (mean ± SD) for P2. Diestrus-EEO reached passage readiness in a median of 5 days (4–7; min–max) or 5 ± 1.22 days (mean ± SD) for P1 and in a median of 5 days (4–6; min–max) or 4.8 ± 0.84 days (mean ± SD) for P2 (P1: P  = 0.41, P2: P  = 0.50). Each dot represents measurement from a plate of organoids derived from one mare group. (EEO = equine endometrial organoids, P1 = passage 1, P2 = passage 2). (L) Diameter (μm) of passage 2 organoids: the mean diameter of Estrus-EEO-P2 was 148.74 ± 50.49 (mean ± SD; μm) with a median of 131.94 (74.64–295.99; min–max). The mean diameter of Diestrus-EEO-P2 was 154.87 ± 54.42 (mean ± SD; μm) with a median of 151.25 (55.51–325.21; min–max) ( P  = 0.54). Each dot represents an individual organoid measurement. (M) Number of cells per passage 2 organoids: the mean cell number of Estrus-EEO-P2 was 46 ± 16 (mean ± SD) with a median of 41 (24–88; min–max). The mean cell number of Diestrus-EEO-P2 was 50 ± 19 (mean ± SD) with a median of 44 (22–106; min–max). ( P  = 0.21) Each dot represents an individual organoid measurement. We analyzed the protein expression of key markers indicative of cell type and reproductive cycle stage in END and EEO-P2 using IHC ( Figure 3 ). Cells integrating the Estrus-EEO-P2 and Diestrus-EEO-P2 dually expressed cytokeratin and vimentin, but lacked detectable levels of desmin, platelet endothelial cell adhesion molecule 1 (PECAM-1), and smooth muscle actin (SMA) ( Figure 3A ). Regarding reproductive cycle-related markers, epithelial cells composing both Diestrus-EEO-P2 and Estrus-EEO-P2 had weak cytoplasmic immunolabeling for estrogen receptor α (ERα / ESR1 ) ( Figure 3B ). Intranuclear progesterone receptor (PGR) immunolabeling was frequent in Estrus-EEO-P2, but weak in Diestrus-EEO-P2 ( Figure 3B ). The OXTR expression was not detected in EEO-P2 ( Figure 3B ). Immunohistochemistry of END and EEO-P2 derived from mares in estrus and diestrus. (A) Organoids stained for cytokeratin, desmin, PECAM-1, SMA, and vimentin to assess cell identity. (B) Endometrial tissue and organoids stained for ERα, OXTR, and PGR to evaluate receptor expression. (PECAM-1 = platelet endothelial cell adhesion molecule, SMA = smooth muscle actin, ERα = estrogen receptor α, OXTR = oxytocin receptor, PGR = progesterone receptor, END = endometrial tissue, EEO-P2 = passage 2 equine endometrial organoids. Scale bar = 3A: 100 μm; 3B END: 50 μm; 3B EEO-P2: 100 μm with insets at higher magnification) Transmission electron microscopy was performed on EEO-P15 derived from mares in estrus and diestrus to investigate the ultrastructural organization and potential reproductive cycle stage-related differences ( Figure 4 ). Both apical-in and apical-out organoids were observed ( Fig. 4A–C ). Cystic organoids displayed variability, with some consisting of a single layer of cells and other of multiple layers. The majority of cells exhibited a columnar shape with basally located nuclei, though cuboidal and squamous-like cells were also present. The cells were anchored to each other apically by tight junctions and desmosomes. Key cellular components such as mitochondria, Golgi apparatus, ribosomes, endoplasmic reticula, glycogen, vacuoles and secretory granules were observed. Additionally, we compared key parameters between organoids derived from estrus and diestrus tissues to explore potential differences linked to the reproductive cycle stage. The length and number of microvilli, and number of secretory granules per cell were significantly different between estrus- and diestrus-derived organoids ( Fig. 4H , I, and   K ). The number of mitochondria per cell was not different between estrus- and diestrus-derived organoids ( Figure 4J ). Transmission electron microscopy EEO-P15. (A) Graphical representation of apical-in (microvilli oriented toward the lumen) and apical-out (microvilli oriented outwards) organoids. (◉ representing the lumen). (B) Toluidine blue staining of a cystic organoid with nuclei oriented towards the outer surface of cells and epithelial lining projecting into the central cavity, indicative of an apical-in orientation. (Scale bar = 50 μm; ◉ representing the lumen). (C) Toluidine blue staining of a cystic organoid with nuclei oriented toward the inner center and epithelial lining projected outward, indicative of an apical-out orientation. (Scale bar = 50 μm; ◉ representing the lumen). (D–E) Estrus-EEO-P15. (Scale bar = 5 μm). (F–G) Diestrus-EEO-P15. (Scale bar = 5 μm). (H–K) Microvilli number and length (μm), number of mitochondria and secretory granules. The length (estrus: 0.40 ± 0.09, 0.40 (0.20–0.60); diestrus: 0.37 ± 0.11, 0.35 (0.10–0.70); μm; mean ± SD, median (min–max); P  = 0.0178) and number (estrus: 13 ± 6, 11 (3–29); diestrus: 15 ± 8, 14 (2–34); mean ± SD; median (min–max); P  = 0.0185) of microvilli, and number of secretory granules (estrus: 12 ± 14, 10 (0–33); diestrus: 6 ± 15, 0 (0–16); mean ± SD, median (min–max); P  = 0.0001) per cell were significantly different between estrus- and diestrus-derived organoids. The number of mitochondria per cell was not different between estrus- and diestrus-derived organoids (estrus: 16 ± 8, 15 (2–37); diestrus: 14 ± 8, 13 (0–34); mean ± SD, median (min – max); P  = 0.05). Each dot represents the measurement from a single cell within an organoid. (D = desmosome, G = glycogen, M = mitochondria, MV = microvilli, N = nucleus, S = secretory granules, TJ = tight junction; EEO-P15 = passage 15 equine endometrial organoids) A schematic overview of all the groups and comparisons is provided in Figure 5A . One organoid sample (Estrus-EEO-P2 group) failed during library generation (technical issues with loading the sample). The average number of raw reads generated from each sample was 30,479,140 ± 6,150,676 (SD) ( n  = 25), with an average of 92% of reads having a quality exceeding Q30 Phred score ( Supplementary Table S1 ). No differences were detected in the amount of data generated or quality, across all groups ( P  > 0.05, Supplementary Table S1 ). The data were subsequently aligned to the equine reference genome, EquCab3.0, with an average mapping rate of 93.24% ( Supplementary Table S1 ). A total of 11,950 genes were expressed in all the samples, 96.16% being protein-coding genes. Bulk transcriptomic analyses on endometrium and derived organoids. (A) Schematic representation of sample groups and comparisons. At the first level (I.), samples are categorized into three main groups: END, P2 endometrial organoids (EEO-P2), and P15 endometrial organoids (EEO-P15), with estrus and diestrus samples combined. At the second level (II.), each group is further subdivided based on reproductive cycle stage, resulting in six specific subgroups: Estrus-END, Diestrus-END, Estrus-EEO-P2, Diestrus-EEO-P2, Estrus-EEO-P15, and Diestrus-EEO-P15. Comparisons were made both within and between these groups to assess transcriptomic differences across developmental stages and estrous cycle stages. (B) PCA of all samples: 1 organoid sample is clustered with END. (C) Hierarchical clustering of all samples based on the top 1,000 most variable genes. Both PCA and hierarchical clustering of global gene expression demonstrated distinct clustering, highlighting a clear separation between END, EEO-P2, and EEO-P15 (with the exception of one organoid sample) ( Figure 5B and C ). The first principal component (PC1), accounting for 82% of the variance, reflected differences between in vivo (END) and in vitro (EEO-P2 and EEO-P15) systems, while the second principal component (PC2), representing 8% of the variance, showed cycle-related transcriptional differences within the EEO-P2 samples ( Figure 5B ). We performed comparative transcriptome analysis between EEO-P2 and their respective tissue of origin (END) to characterize the molecular differences between in vitro organoids and in vivo endometrium. Gene expression overlap analysis identified 12,590 genes (75.7%) shared between EEO-P2 (Estrus-EEO-P2 and Diestrus-EEO-P2 combined) and END (Estrus-END and Diestrus-END combined), with 246 genes (1.5%) exclusively expressed in organoids, and 3,790 genes (22.8%) uniquely expressed in END ( Figure 6A , Supplementary Table S2 ). Among the shared genes were key markers of endometrial epithelial cells ( FOXA2 , MUC1 , EPCAM ), stromal cells ( VIM , CD44 ), as well as important hormone receptors ( PGR , ESR1 , OXTR ) and secretory genes ( P19 , ACP5, SCGB1A1 ). For the genes exclusively expressed in EEO-P2, no relevant pathways were enriched. Genes uniquely expressed in END were enriched in biological pathways related to immune system processes (interleukins, chemokines, toll-like receptor), and vasculature development ( TIE1 , ANGPTL3 , ANGPT4 , ANGPT1 ), among others, and within the KEGG pathways oxytocin signaling pathway ( CD38 ) was enriched ( Supplementary Table S3 ). Differential gene expression of endometrium and derived organoids. (A) Venn diagram of genes expressed by END and passage 2 equine endometrial organoids (EEO-P2) demonstrating 75.7% commonly expressed genes under which markers for epithelium, stroma, endothelium, hormone receptors, and secretion. (B) Upset plot of differentially expressed genes (DEGs) comparing END with passage 2 organoids (EEO-P2). Horizontal bars indicate the number of genes upregulated in Estrus-EEO-P2 vs. Estrus-END, in Estrus-END vs. Estrus-EEO-P2, in Diestrus-EEO-P2 vs. Diestrus-END, and in Diestrus-END vs. Diestrus-EEO-P2. Vertical bars indicate the number of DEGs shared among sample type, as indicated by black dots. (C) Number of DEGs for the following comparisons: Estrus-END vs. Diestrus-END, Estrus-EEO-P2 vs. Diestrus-EEO-P2, Estrus-EEO-P15 vs. Diestrus-EEO-P15. (D) Gene expression overlap analysis of the DEGs from Estrus-EEO-P2 vs. Estrus-EEO-P15 and Diestrus-EEO-P2 vs. Diestrus-EEO-P15 comparisons highlighting the influence of passage. (E) Gene expression overlap analysis of the DEGs from END vs. EEO-P2 and DEGs from END vs. EEO-P15 highlighting the influence of passage of organoids. To better understand the reason of the pronounced distinction between EEO-P2 and END, we performed paired transcriptomic analysis comparing Estrus-EEO-P2 with Estrus-END and Diestrus-EEO-P2 with Diestrus-END, minimizing potential confounding effects of the reproductive cycle stage. In the Estrus-EEO-P2 versus Estrus-END comparison, we identified 6,037 differentially expressed genes (DEGs; FDR < 0.01), including 2002 upregulated and 4035 downregulated in Estrus-EEO-P2 ( Supplementary Table S4 ). Similarly, the Diestrus-EEO-P2 versus Diestrus-END comparison revealed 6,454 DEGs (FDR < 0.01), with 3,582 upregulated and 2,872 downregulated in Diestrus-EEO-P2 compared to Diestrus-END ( Supplementary Table S4 ). Notably, we observed that 1,331 genes were upregulated in both Estrus-EEO-P2 vs. Estrus-END and Diestrus-EEO-P2 vs. Diestrus-END comparisons, indicating consistent differences across cycle stages ( Figure 6B ). These genes were related to cell cycle and cell proliferation ( E2F1/2 , PCNA , RRM2 , CCNB1 , CDK1 , AURKA , BUB1 , TPX2 , KIF11 ), metabolic processes ( FABP5 , PKM , SLC7A5 , RPS6KA3 , ATP5IF1 , DHFR ), and reproductive process ( LIF , PTGS2 , FGF9 , AREG , CSF2 ), among others. Similarly, 1,847 genes were downregulated ( Figure 6B ) and were related to cell differentiation ( MSX1 , SOX17 , HOXA11 , FOXJ1 ), immune system processes ( CD19 , 1L1B , CCR1 , LAT , BTK ), and vasculature development ( TIE1 , ANGPT4 , VEGFB, FLT4 ), among the relevant pathways ( Figure 6B ) ( Supplementary Table S5 ). To generate a reference for the influence of the reproductive cycle stage on endometrial gene expression, we first compared transcriptome profiles of END from mares in estrus (Estrus-END) and diestrus (Diestrus-END). As described in Section 3.4.1, PCA revealed clear clustering of END samples, regardless of the reproductive cycle stage ( Figure 5B ). Differential gene expression analysis between Estrus-END and Diestrus-END identified 656 DEGs (FDR < 0.01) ( Figure 6C ), with 428 genes upregulated, and 228 genes downregulated in diestrus compared to estrus ( Supplementary Table S4 ). Gene ontology and pathway analysis of the upregulated genes in Diestrus-END revealed enrichment in biological processes related to transport ( SLC1A1 , SLC11A2 , AP OA1, NPC2 ), metabolic processes ( HK2, ST3GAL1, GFPT1 ), and response to extracellular stimuli ( PTGES ), while no relevant pathways were identified for the downregulated genes ( Supplementary Table S6 ). To determine whether the reproductive cycle stage of the donor mare influences the transcriptome of derived organoids, we conducted paired transcriptomic analysis comparing organoids derived from mares in estrus (Estrus-EEO) and diestrus (Diestrus-EEO). Differential gene expression analysis revealed 1,705 DEGs (FDR < 0.01) ( Figure 6C ) between Estrus-EEO-P2 and Diestrus-EEO-P2, with 1,289 upregulated and 416 downregulated in Diestrus-EEO-P2 ( Supplementary Table S4 ). Genes upregulated in Diestrus-EEO-P2 were enriched in pathways related to complement activation ( C7 ). Upregulated genes in Estrus-EEO-P2 were associated with processes including cell proliferation ( CDK2 , EGFR ), metabolic processes ( SLC2A6 , FUT1 , SCARB1 ), and secretion ( NIBAN2 ), among others ( Supplementary Table S7 ). When comparing DEG lists from Estrus-END vs. Diestrus-END and Estrus-EEO-P2 vs. Diestrus-EEO-P2, we noted 36 genes (such as AQP9 , HK2 , P19 ) that were significant in both comparisons, suggesting partial molecular carryover of cycle-stage signatures into organoids ( Figure 6C ). In addition, we conducted paired transcriptomic analysis comparing Estrus-EEO-P15 with Diestrus-EEO-P15. We identified 189 DEGs (FDR < 0.01) ( Figure 6C ), including 110 upregulated and 79 downregulated in Diestrus-EEO-P15 compared to Estrus-EEO-P15 ( Supplementary Table S4 ). No relevant pathways were identified for the upregulated genes in Diestrus-EEO-P15, while the downregulated genes were associated with regulation of hormone levels ( HSD17B1 , SULT2B1 ) ( Supplementary Table S8 ). Similarly, intersecting DEG lists from Diestrus-END vs. Estrus-END and Diestrus-EEO-P15 vs. Estrus-EEO-P15 revealed 12 genes ( DISP3, PRLH, LOC100049926, CELF5, FAM240A, LGR6 ) in common ( Figure 6C ). To assess transcriptomic stability of EEO across passages, we compared the transcriptome profiles of early (P2) and later (P15) passage organoids. We identified 569 DEGs (FDR < 0.01) when comparing Estrus-EEO-P2 with Estrus-EEO-P15, comprising 173 upregulated and 396 downregulated in Estrus-EEO-P15 ( Supplementary Table S4 ). Similarly, when we compared Diestrus-EEO-P2 with Diestrus-EEO-P15, we identified 487 DEGs (FDR < 0.01) including 153 upregulated and 334 downregulated in Diestrus-EEO-P15 ( Supplementary Table S4 ). Genes upregulated in P2 compared to P15, in both estrus and diestrus, were enriched in pathways related to cell differentiation ( FOXN4 , WNT5A , WNT16, GATA4 ), cell population proliferation ( FGF7 , FGF10 , FGF1 9, EGF, HGF ), and cellular response ( PIK3R1, PGR, TLR7, OLMFL1 ) among others, while downregulated genes are enriched in pathways related to nucleoside biosynthesis ( DUT ) ( Supplementary Tables S9 and S10 ). To determine whether the observed transcriptomic changes were passage-dependent rather than influenced by the reproductive cycle stage, we performed an overlap analysis of the DEGs from Estrus-EEO-P2 vs. Estrus-EEO-P15 and Diestrus-EEO-P2 vs. Diestrus-EEO-P15 comparisons. By intersecting DEG lists from Estrus-EEO-P2 vs. Estrus-EEO-P15 and Diestrus-EEO-P2 vs. Diestrus-EEO-P15, we identified 131 genes that were differentially expressed in both comparisons. These shared DEGs suggest that a subset of passage-related transcriptional changes is consistent across cycle stages ( Figure 6D ). We performed a comparative transcriptome analysis between EEO-P15 and their respective tissue of origin (END) and identified 10,947 DEGs (FDR < 0.01) ( Supplementary Table S4 ). Intersecting DEG lists from END vs. EEO-P2 ( Supplementary Table S4 ) and END vs. EEO-P15 showed 8,095 genes in common, representing the majority of differences between organoids and tissue. This indicates that most organoid–tissue differences were stable across passages ( Figure 6E ). We performed scRNAseq on endometrium collected from one mare during estrus and diestrus (Estrus-END and Diestrus-END), as well as on corresponding endometrial organoids at passage 18 (Estrus-EEO-P18 and Diestrus-EEO-P18), to investigate cellular heterogeneity and the influence of the reproductive cycle. After QC and filtering, we integrated the four datasets to construct a cellular atlas comprising 17,399 cells. Uniform Manifold Approximation and Projection (UMAP) plot revealed substantial overlap between END and EEO-P18 samples, with a higher concentration of EEO more centrally ( Figures 7A ). Low expression of CASP3 confirmed the viability and quality of the cells used for downstream analyses ( Figure 7B ). Based on canonical markers, we broadly identified epithelial ( EPCAM , CDH1 , FOXA2 , MUC1 ), stromal ( VIM , CD44 ), and endothelial ( PECAM1 ) cell populations ( Figure 7B ). Furthermore, key secretory genes ( ACP5 , P19 , SCGB1A1 ) and endocrine genes ( OXTR , ESR1 , PTGS2 ) were detected, with homogeneous expression patterns across groups, except for P19 , which showed more restricted expression ( Figure 7B ). Single-cell transcriptomic analysis on endometrium and derived organoids. (A) UMAP of endometrium collected during estrus and diestrus (Estrus-END, Diestrus-END) and generated passage 18 organoids (Estrus-EEO-P18, Diestrus-EEO-P18) clustered by group. (B) Feature plots showing expression levels and distribution of marker genes for cell death, epithelial, stromal, and endothelial cells, endometrial secretion, and hormone function (x-axis: umap_1, y-axis: umap_2). Unsupervised clustering identified 12 distinct cell clusters across the integrated dataset ( Figure 8A ) ( Supplementary Table S11 ). Cell identities were manually annotated based on expression of known marker genes ( Figure 8B ) ( Supplementary Table S12 ) [ 66–71 ]. These clusters were grouped into six major cell types: (1) epithelial (including secretory glandular, proliferative glandular, ciliated, and proliferative subsets), (2) stromal, (3) endothelial, (4) smooth muscle, (5) progenitor-like, and (6) immune cells. Regardless of reproductive cycle stages, most cell types were more abundant in END compared to EEO-P18, with the exception of progenitor-like cells and proliferative epithelial cells, which were enriched in the organoids ( Figure 8C ; Supplementary Table S13 ). Within END samples, secretory glandular epithelial cells 1, stromal cells 1, and proliferative epithelial cells were more abundant in Diestrus-END than in Estrus-END. In contrast, luminal epithelial cells, secretory glandular epithelial cells 2, ciliated epithelial cells, antigen-presenting cells, proliferative glandular epithelial cells, and smooth muscle cells were less abundant in diestrus ( Figure 8C ; Supplementary Table S13 ). In the organoids, progenitor-like cells were more prevalent in Diestrus-EEO-P18 compared to Estrus-EEO-P18, whereas secretory glandular epithelial cells 1, ciliated and proliferative epithelial cells, antigen-presenting cells and smooth muscle cells were more prevalent in the estrus derived organoids ( Figure 8C ; Supplementary Table S13 ). Single-cell transcriptomic analysis on endometrium and derived organoids: cell type annotation. (A) UMAP of END and passage 18 endometrial organoids clustered by cell type. (B) Heatmap with enriched gene markers for cell type annotation. (C) A stacked bar plot showing the contribution of each group (END = endometrial tissue, EEO-P18 = passage 18 equine endometrial organoids) to a specific cell cluster.

Five healthy mixed-breed mares, aged 8–15 years, were selected for the study. The mares were housed in outdoor paddocks with ad libitum access to hay, trace minerals, and water. The reproductive cycle stage of the mares was assessed by transrectal palpation and ultrasonography. Endometrial biopsies were collected from each mare during both estrus (defined as follicle diameter > 30 mm, presence of uterine edema, and absence of corpus luteum) and diestrus (5–7 days after confirmed ovulation, presence of corpus luteum and absence of uterine edema) [ 40 ]. To confirm the stage of the estrous cycle, blood samples were collected from the jugular vein in serum tubes for progesterone concentration (P 4 ) determination [ 40 ]. Blood samples, collected on the days of tissue collection, were centrifuged at 2000  g at 4°C for 10 min, and serum was stored at −20°C until analyzed. P 4 concentrations were measured by enzyme immunoassay using the R4359 anti-P 4 antibody, following previously described procedures [ 41 ]. All animal work performed at UCDavis was in accordance with Institutional Animal Care and Use Committee (IACUC, #24117). All the materials and reagents used in this study, along with their suppliers and catalog numbers, are listed in Supplementary File S1 . Endometrial tissue samples were collected using established, standard clinical procedures [ 42 ]. Two tissue samples were obtained from the basis of the uterine horn during both estrus and diestrus. To reduce the risk of contamination, the collected tissue was rinsed with Dulbecco Modified Eagle Medium (DMEM). Upon collection, one tissue sample was immediately placed in transport medium ( Supplementary File S2 ) and transported to the laboratory on ice for immediate processing into single-cell suspensions. The second tissue sample was divided into two equal parts. One part was fixed in 10% buffered formalin for histological evaluation and immunohistochemistry (IHC), while the second part was placed in RNA later , kept overnight at 4°C, and subsequently stored at −80°C for RNA sequencing. Endometrial organoids were generated following established protocols with slight modifications [ 10 , 13 ]. Specifically, the tissue samples were washed with DMEM, transferred to a Petri dish, and cut into 1–2  \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{upgreek} \usepackage{mathrsfs} \setlength{\oddsidemargin}{-69pt} \begin{document} ${\mathrm{mm}}^3$\end{document} pieces using sterile scissors. The tissue fragments were transferred to a gentleMACS C-tube and dissociated into single-cell suspension using the Multi Tissue Dissociation Kit 1 and gentleMACS Dissociator according to the manufacturer’s instructions. Tissue fragments (0.5 g) were incubated twice in the enzyme mix (Multi Tissue Dissociation Kit 1), for 30 min each, with continuous rotation at 37°C. Before and after each incubation, the samples were placed in the gentleMACS Dissociator device, running the program Multi_B at room temperature. The dissociated mixture was resuspended with 15 mL DMEM, filtered through a 70 μm strainer, and centrifuged for 7 min at 300  g at room temperature. The supernatant was removed, and the pellet was suspended in ice-cold growth factor-reduced Matrigel Matrix. Matrigel was thawed overnight and diluted with ice-cold phosphate-buffered saline (PBS) at a 3:2 (Matrigel:PBS) ratio before use. For plating, 40 μL Matrigel/cell suspension droplets (~3.9 x 10 6 cells/droplet) were prepared and placed in the center of pre-warmed 24-well culture plates. The plates were incubated at 37°C for 10–15 min to allow the Matrigel to polymerize. Each Matrigel dome was then overlayed with 700 μL Expansion Medium (ExM, composition provided in Supplementary File S2 ) [ 9 , 10 , 13 ]. The plates were incubated at 39°C, 6.2% CO 2, and high relative humidity (80%–100%). Every other day, 300 μL of the culture medium was removed and replaced with fresh ExM. Part of the single-cell suspension not used for organoid generation was used for single-cell RNA sequencing (scRNAseq) and the other part was cryopreserved in a freezing solution (10% Dimethyl sulfoxide (DMSO) + 90% Advanced DMEM/F12), stored at −80°C overnight using the Mr. Frosty freezing container, and subsequently transferred to liquid nitrogen for biobanking purposes (not part of this study). Organoids were passaged every 4–7 days, depending on their concentration (50 organoids per dome) and size. To passage, the Matrigel domes containing organoids, without removing the ExM, were harvested by scraping the bottom of the well with a 1000 μL pipette tip and transferred into 2 mL microcentrifuge tubes (2 wells per 2 mL tube). The tubes were centrifuged at 600  g for 6 min at 4°C. Then, the pellet (containing the organoids) was mechanically disrupted by aspirating and ejecting the pellet through a 25-gauge needle three to four times (depending on the size of the organoids). After a second centrifugation (600  g , 6 min, 4°C), the pellet was resuspended in Matrigel and cultured as described above. At each passage, the resulting cell suspension was distributed over twice the number of Matrigel domes from which it was harvested (n domes were distributed over n x 2 domes). Media changes (300 μL) were performed every other day until the following passage. A graphical representation of the organoid generation protocol is provided in Figure 1A . EEO generation and study design. (A) EEO generation (EEO = equine endometrial organoids, ExM = Expansion Medium). (B) Study design. Endometrial tissue from mares ( n  = 5) in estrus and diestrus was processed for organoid culture, bulk RNAseq ( n  = 10), scRNAseq ( n  = 2; ∽4,300 cells each), histology ( n  = 6), and IHC ( n  = 6). Organoids (P0–P2; EEO generation) were assessed for morphology, growth, and viability. Passage 2 organoids (EEO-P2) were processed for histology ( n  = 10), IHC ( n  = 6), and RNAseq ( n  = 10). A subset of organoids from three mares was further cultured. Passage 15 organoids (EEO-P15) were processed for RNAseq ( n  = 6) and TEM ( n  = 6). Organoids from one mare were further cultured. Passage 18 organoids (EEO-P18) were processed for scRNAseq ( n  = 2; ∽4,300 cells each). (END = endometrial tissue, EEO = equine endometrial organoids, P0 = passage 0, P2 = passage 2, P15 = passage 15, P18 = passage 18, IHC = immunohistochemistry, TEM = transmission electron microscopy, RNAseq = bulk RNA sequencing, scRNAseq = single-cell RNA sequencing, n = number of mares) Endometrial tissue was collected from five mares, with biopsies obtained from each mare during both estrus ( n  = 5) and diestrus ( n  = 5). This paired design allowed each mare to serve as her own control for estrus versus diestrus comparisons. Samples were processed for organoid culture, histology, IHC, and bulk- and single-cell RNA sequencing (RNAseq and scRNAseq, respectively). Cultures and RNAseq were generated for all mares in both cycle stages; histology and IHC were performed on tissue from three mares, and scRNAseq on tissue from one mare, in both estrus and diestrus. Organoids derived from seeding the single-cell suspension were classified as passage-0 (P0). After two subsequent passages, they were referred to as P2 organoids. To ensure sufficient organoid numbers for downstream analyses, including RNAseq, histology, IHC, and viability assessment, organoids were allowed several days to regenerate and proliferate post-passage. RNAseq, histology, and viability were assessed in EEO-P2 derived from all five mares during both estrus and diestrus, whereas IHC was performed on EEO-P2 from three mares in both cycle stages. Organoid harvesting followed the standard protocol used for passaging. The remaining organoids were cryopreserved in a solution containing 5% DMSO, 40% Fetal Bovine Serum, and 55% Advanced DMEM/F12. Cryopreserved samples were kept at −80°C for 24 h using the Mr. Frosty freezing container before being transferred to liquid nitrogen for long-term storage. P2 organoids from three mares (estrus and diestrus) were thawed by retrieving the cryovial from liquid nitrogen storage and immediately adding room temperature Advanced DMEM/F12. Thawing was performed entirely at room temperature without the use of a heating bath. Organoids were then cultured in ExM until P15, after which they were harvested for RNAseq and transmission electron microscopy (TEM). Additionally, organoids derived from one mare (estrus and diestrus) were further cultured until P18 and used for scRNAseq. A schematic representation of the study design is provided in Figure 1B . Organoid morphology was assessed in cultures derived from mares in estrus and diestrus using phase-contrast microscopy (ZEN 3.5 blue software). Images were captured on days one through four of passages 0, 1, and 2 (P0, P1, and P2). If organoids were not passaged by day four, additional images were captured on days five and six. Morphological evaluation focused on density, whereby cystic structures were defined as a cell layer surrounding a central lumen/cavity [ 9 , 43 ]. The growth rate of the endometrial organoids was determined by the number of days until the subsequent passage. Organoids were passaged when their number and size had approximately doubled compared to the number and size after 24 h of seeding or passaging. Organoid size was measured using ZEN 3.5 blue edition software, and the number of organoids was estimated by counting them in a field of view with a magnification of 50X. The growth rate was compared among estrus and diestrus using the Wilcoxon test since data were not normally distributed using JMP® 18 (SAS Institute Inc., Cary, NC). The data are presented as median (min–max) based on the non-parametric test used, along with mean ± standard deviation (SD) for ease of interpretation. A P -value <0.05 was considered statistically significant. Organoid (EEO-P2) viability was assessed using Hoechst 33342 and propidium iodide (PI) fluorescent staining, under a fluorescence microscope, as previously described [ 44 ]. Organoids were randomly selected in triplicate from each mare both in estrus and diestrus. Live and dead cells were counted using ImageJ (version 1.54m), by blinded operators, and the viability (ratio live/dead cells) was calculated. The viability was compared among estrus and diestrus using the Wilcoxon test since data were not normally distributed. The data are presented as median (min–max) based on the non-parametric test used, along with mean ± standard deviation (SD) for ease of interpretation. A P -value <0.05 considered statistically significant. P2 organoids (EEO-P2), harvested as described for passaging, were fixed in 4% paraformaldehyde for 9 min at room temperature, embedded in HistoGel, whereafter they were stored in 70% ethanol until further processing. Endometrial tissue (stored in formalin) and embedded organoids were processed through graded alcohols and xylene, then embedded in paraffin (FFPE), sectioned at approximately 5 μm intervals until tissue/organoids were observed. The slides were then stained with Hematoxylin and Eosin (H&E) using standard procedures [ 45 ]. Average size (diameter) and cell number were determined on randomly selected organoids ( n  = 11) from all five mares in estrus and diestrus, resulting in a total number of 110 organoids. The organoid size and cell number were compared between estrus and diestrus using the Wilcoxon test (not normally distributed). The data are presented as median (min–max) based on the non-parametric test used, along with mean ± standard deviation (SD) for ease of interpretation. A P -value <0.05 was considered statistically significant. Additionally, morphological features were blindly assessed by a veterinary pathologist. For IHC, 4 μm sections of FFPE organoids and END were mounted on positively charged Superfrost® Plus slides and subjected to IHC using the automated BOND-MAX and the Polymer Refine Detection kit. Following automated deparaffinization, heat-induced epitope retrieval was performed using a ready-to-use citrate-based buffer (pH 6.0) or an EDTA-based buffer (pH 9.0) at 100°C for 20 min before incubation with primary antibodies ( Supplementary File S3 ). Sections were then incubated with the primary antibodies (dilutions listed in Supplementary File S3 ) for 30 min at room temperature, followed by either a polymer-labeled goat anti-rabbit or anti-mouse IgG coupled with horseradish peroxidase for 8 min at room temperature. 3,3′-diaminobenzidine tetrahydrochloride was used as the chromogen (10 min), and counterstaining was performed with hematoxylin. Slides were mounted with a permanent mounting medium. P15 organoids (EEO-P15), harvested at room temperature as described for passaging, were fixed in Karnovsky Fixative until further processing. Organoids were processed as previously described [ 46 , 47 ]. In brief, organoids were embedded in HistoGel and postfixed in 2.5% glutaraldehyde in 0.1 M sodium cacodylate buffer, followed by a second post-fixation step in 1% osmium tetroxide in 0.1 M sodium cacodylate buffer. Samples were then washed three times in double-distilled water followed by dehydration through a graded ethyl alcohol series (25%–100%). Next, dehydrated samples were infiltrated with 1:1 ethanol:resin for 1 h, transferred to a 1:3 ethanol:resin mixture for an additional hour. Samples were then infiltrated with 100% resin, embedded, and polymerized at 58°C for 48 h. Following polymerization of the resin, blocks were trimmed and sectioned at 60–70 nm using an ultramicrotome. Ultrathin sections were collected on a 100-mesh nickel grid and contrasted with 5% uranyl acetate for 2 min, followed by ready-to-use lead citrate for 6 min. Samples were visualized using a JEOL 1400 transmission electron microscope. Images were obtained using an AMT Capture Engine Version 7.00 camera and software. Images were randomly selected from the estrus and diestrus organoids and were processed and analyzed using ImageJ. The presence and orientation of microvilli were assessed, distinguishing between those directed towards the lumen (apical-in organoids) and those oriented outward (apical-out organoids). In addition, microvilli length and density were measured and counted. Cellular morphology was evaluated by measuring cell height and width, classifying cells as columnar (height-to-width ratio of 2:1), cuboidal (1:1), or squamous (1:2). Furthermore, intracellular components were examined, including the number of secretory granules, mitochondria, as well as the amount of intracellular glycogen. These parameters were quantified and compared between estrus- and diestrus-derived organoids to characterize potential differences in cellular organization and metabolic activity across reproductive cycle stages. The number and length of microvilli, number of mitochondria, and number of secretory granules were compared among estrus and diestrus using the Wilcoxon test. The data are presented as median (min–max) based on the non-parametric test used, along with mean ± standard deviation (SD) for ease of interpretation. A P -value <0.05 was considered statistically significant. Total RNA was isolated from P2 organoids (Diestrus-EEO-P2, n  = 5; Estrus-EEO-P2, n  = 5; derived from endometrium from five mares, each sampled in both diestrus and estrus, respectively), P15 organoids (Diestrus-EEO-P15, n  = 3; Estrus-EEO-P15, n  = 3; derived from three mares, in both diestrus and estrus) and their respective endometrium sample (Diestrus-END, n  = 5; Estrus-END, n  = 5; endometrium from mares in diestrus respectively estrus). Using the same mares in both estrus and diestrus allowed each animal to serve as its own control. The organoids were harvested as described for passaging, resuspended in RNA later , kept at 4°C overnight, and subsequently stored at −80°C until extraction. For the RNA extraction, the RNeasy® Mini Kit was used, with small modifications [ 48 ]. In short, the samples were transferred to 2 mL microtubes with ceramic beads, filled with QIAzol® Lysis Reagent. The tubes were shaken for homogenization using Bead Ruptor 4 Mill Homogenizer. Chloroform was used to promote phase separation, followed by 70% ethanol to precipitate the nucleic acids. The resulting fluid was transferred to the RNeasy® Mini Spin Column, followed by the manufacturer’s instruction. Isolated RNA sample quality was assessed by TapeStation using the High Sensitivity D5000 Screen Tape Assay and quantified by Qubit 2.0 RNA HS assay. All samples had an RNA integrity number > 8.5. Paramagnetic beads coupled with oligo d(T)25 were combined with total RNA to isolate poly(A) + transcripts based on NEBNext® Poly(A) mRNA Magnetic Isolation Module manual. Prior to first strand synthesis, samples were randomly primed (5 ´ d(N6) 3 ´ [N = A,C,G,T]) and fragmented based on manufacturer’s recommendations. The first strand was synthesized with the Protoscript II Reverse Transcriptase with a longer extension period, approximately 30 min at 42°C. All remaining steps for library construction were used according to the NEBNext® UltraTMII Directional RNA Library Prep Kit for Illumina. The final library quantity was assessed by Qubit 2.0 and quality was assessed by TapeStation HSD1000 ScreenTape. Illumina® 8-nt dual-indices were used. Equimolar pooling of libraries was performed based on quality control (QC) values and sequenced on an Illumina® [NovaSeq S4 2X150] with a read length configuration of 150 bp paired-end reads for 40 million paired-end reads per sample. The RNAseq data from this study is available at the Sequence Read Archive database with the Accession # PRJNA1284319. Raw reads were trimmed for quality and adapter with Trim-Galore v0.4.0 and then mapped to EquCab3.0 (ENSEMBL release 94 [GCA_002863925.1]) [ 49 , 50 ] using STAR v2.7.2a [ 51 , 52 ]. Differentially expressed genes (DEGs) were analyzed for the following comparisons: (1) P2 organoids derived from endometrium of mares in estrus vs. the originating endometrium samples (Estrus-EEO-P2 vs. Estrus-END), (2) P2 organoids derived from endometrium of mares in diestrus vs. the originating endometrium samples (Diestrus-EEO-P2 vs. Diestrus-END), (3) endometrium from mares in estrus vs. diestrus (Estrus-END vs. Diestrus-END), (4) endometrial organoids derived from mares in estrus vs. diestrus (Estrus-EEO-P2 vs. Diestrus-EEO-P2 and Estrus-EEO-P15 vs. Diestrus-EEO-P15), (5) P2 organoids vs. P15 organoids for estrus and diestrus (Estrus-EEO-P2 vs. Estrus-EEO-P15 and Diestrus-EEO-P2 vs. Diestrus-EEO-P15), (6) P2 organoids vs. the originating endometrium samples (EEO-P2 vs. END), and (7) P15 organoids vs. the originating endometrium samples (EEO-P15 vs. END). DEGs were evaluated using DESeq2 v1.44.0 in R, using paired analysis, with a false discovery rate (FDR) <0.01 [ 53 ]. Where indicated, overlap between comparisons represent descriptive intersections of DEG lists (based on FDR < 0.01), and were used to highlight genes consistently significant across independent pairwise contrasts. To investigate the biological functions of DEGs, ShinyGO version 0.82 was used to functionally annotate genes based on gene ontology (GO) [ 54 ] and KEGG [ 55 , 56 ]. Single-cell suspensions derived from the tissue from one mare, both during estrus and diestrus, were centrifuged (600  g , 6 min, room temperature), and the resulting cell pellet was resuspended in PBS containing 0.1% bovine serum albumin (BSA). Organoids (EEO-P18), derived from the same mare, were harvested and transferred to a Pyrex 9 depression glass spot plate. Approximately 100 organoids were manually collected with a 300 μm Flexipet pipette under a stereoscope. Subsequently, the organoids were washed in three 500 μL droplets of PBS at room temperature before being transferred to a 2 mL microcentrifuge tube and resuspended in PBS containing 0.1% BSA. The organoids were mechanically dissociated to a single-cell suspension by aspirating and forcefully ejecting the pellet through a 25-gauge needle. The cell concentration and viability were assessed using the Countess 3 FL Automated Cell Counter with Trypan Blue staining. Single cells (average concentration of 8 x 10 5 , live cell rate of 87%) were loaded in the HIVE CLX Collector device (Honeycomb Biotechnologies) according to the manufacturer’s instructions. Following cell capture, HIVE collectors were frozen at −20°C overnight and stored at −80°C until shipment for library preparation and sequencing. The transcriptome recovery and library preparation were performed according to the manufacturer’s instructions (HIVE CLX Transcriptome Recovery & Library Preparation User Protocol). In brief, the HIVE CLX collectors were thawed and washed with CLX Storage Wash Solution. RNA was captured onto beads by sequential addition of lysis buffer and CLX Hybridization Buffer, followed by bead recovery. The captured RNA was used for first-strand cDNA synthesis, followed by RNA strand degradation and second-strand synthesis using random primers to generate double-stranded cDNA containing cell-specific barcodes. The library was sequenced on the Illumina NovaSeq X platform. BCL files were demultiplexed and converted to FASTQ format using bcl2fastq v.2.20.0.422 (Illumina). Raw FASTQ files were processed with BeeNetPLUS v2.0 (Honeycomb Biotechnologies) to generate gene expression count matrices [ 57 ]. The number of expected cell-containing barcodes was set to 5000. Count matrices were imported into Seurat v5.0.1 [ 58–62 ] for downstream analysis running under R version v4.2.0 [ 63 ] and Rstudio v2025.01.999 [ 64 ]. Doublet detection was performed using DoubletFinder v2.0 package [ 65 ]. Data were normalized and later integrated using Seurat’s NormalizeData function canonical correlation analysis dimensional reduction. Features and anchors for integration were chosen using Seurat’s SelectIntegrationFeatures and FindIntegrationAnchors functions, respectively. Data were centered and scaled using Seurat’s ScaleData function followed by principal component analysis (PCA) and nearest-neighbor graph construction. Differential gene expression analysis was conducted using Seurat’s FindMarkers function. DESeq2 v1.46.0 [ 53 ] was used to apply an empirical Bayes method for statistical inference. Genes with adjusted P -values ≤0.05 were considered significantly differentially expressed. In addition, cell proportions were compared (END vs. EEO-P18 and Estrus vs. Diestrus) using N-1 chi squared test.

In this study, we successfully generated 3D structures derived from endometrium from healthy mares in both estrus and diestrus. These organoids demonstrated expansion potential up to P18 (4 months), and maintained viability after freeze-thawing, underscoring their self-renewing and self-organizing capacities. Hematoxylin and Eosin and phase-contrast live imaging revealed cystic structures which align with human, murine, and equine endometrial organoids [ 9–11 , 13 , 14 , 72 ]. Immunohistochemistry confirmed the presence of epithelial cells (cytokeratin-positive) in our organoids. Notably, the co-expression of cytokeratin and vimentin indicates an epithelial to mesenchymal transition phenotype (EMT). EMT is a physiological process involved in endometrial remodeling and implantation, and it also contributes to the regenerative capacity of organoids by supporting cellular plasticity and self-renewal [ 73 ]. The first objective was to assess the relevance of EEO as a model by comparing them to native END using ultrastructural and molecular analyses. Transmission electron microscopy analyses of organoids showed that most cells exhibited a columnar shape, resembling the in vivo equine endometrium [ 38 , 39 , 74 ]. We identified both apical-in and apical-out organoids. Traditionally, organoids exhibit an apical-in polarity, resembling the enclosed lumen of the organ [ 9–11 , 13 , 75 ]. However, this polarity restricts access to the apical/luminal surface, posing challenges for applications such as host–microbe interaction studies and fetal-maternal interface research in the endometrium. Recent studies have demonstrated that an alternative culture method, namely suspension-based culture, generates apical-out organoids, facilitating exposure of the luminal surface [ 23 , 24 , 76 ]. While culture method appears to be the primary determinant of organoid polarity, variation within each culture condition may still occur. In the present study, the presence of both apical-in and apical-out EEO suggests that some organoids may have escaped from the Matrigel dome, leading to polarity inversion and a mixed population. Further investigation is needed to clarify the mechanisms controlling polarity in EEO. Additionally, we observed tight junctions and desmosomes in both polarity types, indicating that the epithelial barrier remained intact. This suggests that EEO retain functional epithelial integrity, supporting their suitability for modeling processes such as pathogen invasion, where barrier function plays a critical role. We also observed abundant microvilli, which increase epithelial surface area and may enhance absorptive or secretory functions. Unlike mouse and human endometrial organoids [ 9 ], we did not observe ciliated cells. Additional studies have shown that cilia formation can be induced by estrogen treatment, mimicking the endometrial proliferative phase [ 12 , 77 ]. The high P 4 concentration in the culture media (average [P 4 ] of 7.4 ng/mL) could explain the absence of cilia in the present EEO [ 17 , 21 ]. This high P 4 level is primarily attributed to the use of supplements B27 minus vitamin A and N2 ( Supplementary File S2 ). These supplements were originally used to support the growth and maintenance of mainly neural cells and tissues but have since been widely adopted in other culture systems and routinely used in organoid cultures across species and tissue types, including endometrial organoids [ 9–12 , 18 , 78–82 ]. While these components have become the standard in the organoid research field, their hormonal content may influence the organoid properties. For future organoid studies, it is essential to consider the presence of P 4 in the culture media and its potential impact on the molecular and phenotypic characteristics of the organoids. To further investigate the organoid model, we characterized and compared the gene expression profiles of EEO-P2 and END. Overlap analysis revealed that 75.7% of expressed genes were shared between EEO-P2 and END, while 1.5% were exclusively expressed in organoids, and 22.8% were uniquely expressed in END. Among the shared genes, key endometrial markers, including FOXA2 , MUC1 , PGR , ESR1 , OXTR , P19 , ACP5 , and SCGB1A1 were identified [ 83 , 84 ]. The larger number of unique genes in tissue suggests greater complexity compared to the organoids. PCA, hierarchical clustering, and differential gene expression analysis further highlighted distinctions between END and EEO-P2. Our analysis revealed that 6,037 and 6,454 genes had different expression levels between Estrus-EEO-P2 versus Estrus-END and Diestrus-EEO-P2 versus Diestrus-END, respectively. Genes upregulated in EEO-P2 compared to END were enriched in pathways related to metabolic processes, cell cycle, cell proliferation, and reproductive processes. Organoids are cultured in conditions optimized for stemness and proliferation (e.g., EGF, FGF, B27 supplement, nicotinamide), promoting their self-renewing capacity. In addition, continuous passaging of organoids selects for highly proliferative cells enabling long-term expansion. In turn, high cell proliferation demands increased metabolic activity to support cell division, explaining enrichment in pathways related to energy production, nucleotide biosynthesis, and amino acid metabolism, among others. Enrichment of epithelial and stromal cells in organoids compared to the more complex native tissue could result in upregulation of genes involved in reproductive processes. Conversely, genes downregulated in EEO-P2 compared to END were enriched in pathways associated with immune system processes, vasculature development, and cell differentiation, among others. These findings align with the absence of immune cells [ 85 ], and vasculature [ 86 ] in organoids, and in addition, culture conditions favor cells with a more undifferentiated, proliferative state [ 7 ]. While these differences highlight the reduced complexity of EEO compared to native tissue, they also emphasize their value as simplified, controllable models for studying endometrial (patho)physiology. All together, these results indicate that the in vitro culture conditions affect the transcriptional profile of the organoids, while maintaining key endometrial profiles which are aligned with the results obtained in gastro-intestinal and liver organoids [ 35 , 87–89 ]. Although the culture medium contained high P 4 concentration, organoids retained transcriptional differences reflecting the donor’s reproductive cycle stage. This suggests that intrinsic, stage-specific programming of the endometrial epithelium persists in vitro despite exogenous hormone exposure. Nevertheless, we acknowledge that sustained culture in a high-P 4 environment could attenuate or mask certain hormone-dependent differences over time, potentially contributing to convergence of transcriptomic profiles between estrus- and diestrus-derived organoids. The similar numbers of DEGs observed when comparing estrus endometrium with estrus-derived organoids and diestrus endometrium with diestrus-derived organoids support the interpretation that medium composition did not fully override the original cycle-stage signature. Although based on only one sample per group, our single-cell transcriptomic analysis provides a unique high-resolution cellular atlas of the equine endometrium and corresponding endometrial organoids, shedding light on their cellular composition. The substantial overlap in UMAP space between native endometrium and long-term cultured organoids supports the notion that EEO retain key cellular features of their tissue of origin, even after extensive passaging. Major cell types, including various epithelial subsets, and stromal cells, were preserved in the EEO. In contrast, endothelial, smooth muscle, and antigen-presenting cells were underrepresented, consistent with our IHC findings and the established notion that organoids derived from tissues typically lack non-epithelial lineages. The relative enrichment of progenitor-like and proliferative epithelial cells in EEO suggests a proliferative bias inherent to the organoid culture system, potentially driven by the growth-promoting conditions used [ 70 , 90 , 91 ]. Nonetheless, genes involved in innate immune responses (e.g., TLR , CXCL8 , PTGS2 , LTF , DEFB ) were enriched in defined cell clusters in EEO, supporting their applicability for infection modeling. Likewise, the expression of genes implicated in fetal-maternal interaction (e.g., LIF , PTGR1 , PTGS2 , FGF9 , IGFBP3 ) highlights the relevance of EEO for studying reproductive and implantation-related processes [ 1 , 92 ]. Given the strong hormonal regulation of endometrial function and morphology, our second objective was to determine how the reproductive cycle stage of donor mares affected EEO [ 38 , 39 ]. Many endometrial organoid studies do not specify the donor’s cycle stage [ 9 , 24 , 71 , 77 , 93–95 ]. While EEO have previously been generated from mares in estrus and the transitional period, their cycle-dependent characteristics remain largely underexplored [ 13 , 14 ]. In humans, organoids have been successfully derived from both secretory and proliferative endometrium, though subsequent analyses do not distinguish between them [ 10 , 11 , 21 , 70 ]. Unlike a previous murine study that reported differences in organoid-forming efficiency across estrous cycle phases [ 9 ], we observed no significant differences in morphology, viability, growth rate, size, or cell number between estrus- and diestrus-derived EEO (P0–P2). However, TEM analysis revealed subtle ultrastructural differences: estrus-derived EEO exhibited fewer but longer microvilli and a higher number of secretory granules, consistent with the secretory activity of the endometrium during estrus [ 74 ]. Immunohistochemistry analysis further demonstrated the expression of ERα, and PGR in both estrus- and diestrus END. In EEO-P2, ERα was expressed regardless of the cycle stage, while PGR was detected primarily in estrus-derived organoids and weakly in diestrus-derived ones. OXTR was not detected in either organoid group. The absence of OXTR and the cycle-dependent detection of PGR, despite their gene expression, suggests post-transcriptional regulation or that culture conditions may not support their protein expression. A previous study reported ERα and PGR expression in estrus-derived EEO by IHC [ 13 ]. In human endometrial organoids, most cells exhibit a weak expression of ERα, while PGR is typically absent under basic media conditions (no hormonal treatment) [ 9 , 10 , 96 ]. Additionally, endometrial organoids derived from mice in estrus have been shown to express ERα as well [ 9 ]. These findings suggest species- and estrous cycle stage–dependent differences in hormone receptor expression, as well as a potential influence of steroid concentrations in the culture media. Bulk transcriptomic analyses identified 656 DEGs between Estrus-END and Diestrus-END, 1,705 DEGs between Estrus-EEO-P2 and Diestrus-EEO-P2, and 189 DEGs between Estrus-EEO-P15 and Diestrus-EEO-P15. The organoids derived from mares in estrus and diestrus exhibited clear expression differences at P2, as reflected by the higher number of DEGs compared to END and EEO-P15. Genes upregulated in Estrus-EEO-P2 were associated with cell proliferation, metabolic processes, secretion, and immune response, suggesting a role in uterine clearance. In contrast, genes upregulated in Diestrus-EEO-P2 were linked to complement activation, a process that may reflect endometrial preparation for a potential preimplantation embryo. The cycle-dependent transcriptional signature diminished in later passages, as indicated by the reduced number of DEGs at P15. This suggests an adaptation to prolonged in vitro culture, where in vivo reproductive-related features, systemically regulated by hormones, are gradually lost, while cell-intrinsic traits persist. Similar findings have been reported in intestinal organoids, where compartment-specific transcriptional profiles were maintained, but features associated with age, developmental stage, microbiota, and sex were lost in vitro compared to the native tissue [ 35 , 87 , 97 ]. This distinction underscores the role of epigenetic programs in maintaining local tissue identity, while systemically driven features rely on extrinsic cues such as hormonal signaling, microbiota interactions, and metabolic factors, which are absent in the organoid culture system [ 98 ]. The presence of exogenous P 4 during culture, as observed in our study, or alternatively, the addition of estrogens without P 4 , may enhance estrous cycle stage-specific phenotypes over the culture period. The effects of modifying the culture medium on cellular phenotypes, as well as on molecular and epigenetic markers, warrant further investigation. Further scRNAseq analysis revealed hormone-dependent differences in epithelial cell subtype distribution, emphasizing the influence of the estrous cycle on endometrial cellular dynamics. Specifically, proliferative epithelial and secretory glandular epithelial cells 1 were more abundant in Diestrus-END than in Estrus-END. In contrast, luminal epithelial cells, secretory glandular epithelial cells 2, ciliated epithelial cells, antigen presenting cells, proliferative glandular epithelial cells, and smooth muscle cells were more abundant in estrus. This aligns with the estrus phase being characterized by high estrogen levels, which promote epithelial proliferation, increased immune surveillance, and enhanced luminal receptivity. Ciliated cells, in particular are known to increase during estrus, contributing to the clearance of debris and sperm from the uterine lumen. Similarly, in the organoids, progenitor-like cells were more prominent in Diestrus-EEO-P18, whereas secretory glandular epithelial cells 1, ciliated and proliferative epithelial cells, antigen presenting cells, and smooth muscle cells were more prevalent in the estrus derived organoids. This finding suggests that despite reduced hormonal influence in later passages, some cycle-dependent transcriptional and cellular features are retained in the EEO model in the presence of high P 4 in the medium. This highlights the robustness of the EEO model and its potential for studying hormone-driven epithelial dynamics and immune-endometrial interactions in a controlled setting. One of the key advantages of organoids is their ability to sustain long-term culture while preserving key tissue-specific characteristics [ 9–11 , 18 ]. Genomic and transcriptomic stability across passages has been demonstrated in various organoid models [ 10 , 11 , 18 ], though this has not been thoroughly assessed for EEO. The previous study cultured EEO for only three passages, leaving open questions about their long-term cultural capability and stability [ 13 ]. Consequently, our third objective was to assess transcriptomic stability of long-term EEO cultures. By comparing transcriptomic profiles between EEO-P2 and EEO-P15, we identified 569 and 487 DEGs in organoids derived from mares in estrus and diestrus, respectively. Furthermore, 8,095 DEGs were shared between END vs. EEO-P2 and END vs. EEO-P15, suggesting moderate transcriptomic expression drift over time. While adaptation to the culture environment is expected, EEO retained key endometrial markers ( FOXA2, MUC1, ESR1, ACP5 ), reinforcing their potential as a robust in vitro model for equine reproductive studies. Ideally, transcriptomic profiling (bulk and single-cell level) would have been performed on matched samples across more passage points to better capture the dynamics of transcriptional changes. Future studies will benefit from advances in sequencing technologies that require fewer input cells and are increasingly cost-effective, making it more feasible to include larger sample numbers and more frequent time points. In conclusion, EEO recapitulate key morphological, ultrastructural, and transcriptional features of the native endometrium, establishing their utility as a promising in vitro model for equine reproductive biology. Despite reduced cellular complexity compared to native tissue, EEO retain essential endometrial markers and partially preserve cycle-dependent features, particularly in early passages. These characteristics gradually diminish over time, likely due to the culture conditions and absence of systemic cues such as hormonal fluctuations. While refinements to the organoid models are ongoing, EEO already offer a versatile and relevant platform for studying epithelial dynamics, barrier integrity, and host–microbe interactions in the equine endometrium. As such, EEO hold great potential for advancing our understanding of endometrial physiology and disease in the horse.

The endometrium, the inner lining of the uterus, plays a central role in mammalian reproduction by orchestrating processes such as sperm transport, reproductive cycle, embryo development, and uterine defense against ascending infections. Diseases affecting the endometrium can severely impair fertility, underscoring the need to understand its underlying mechanisms [ 1 , 2 ]. However, in vivo studies face challenges, including inter-individual variability, high costs, ethical concerns, and the incomplete representation of human and veterinary conditions when using animal models. Over the past decade, organoids have emerged as a powerful tool for studying organ biology, disease mechanisms, and drug responses, while reducing reliance on in vivo studies [ 3–5 ]. These three-dimensional, self-organizing, and self-renewing multicellular structures, derived from stem cells, closely recapitulate key functional, morphological, and molecular features of their tissue of origin [ 6–8 ]. Endometrial organoids have been established in multiple species from both healthy [ 9–17 ] and diseased [ 10 , 18–20 ] tissues. Furthermore, these organoids have been reported to preserve the structural characteristics, secretory properties, and hormonal responsiveness of their original tissue. Efforts are ongoing to improve their physiological relevance by incorporating multiple cell types [ 21 , 22 ], inducing vascularization [ 23 ], and adjusting polarity [ 24 ], with the goal of using endometrial organoids to investigate intrauterine infections [ 25–27 ], feto-maternal interactions [ 23 , 28 ], and other reproductive processes. Equine endometrial organoids (EEO) have been successfully generated from both fresh and frozen-thawed endometrial biopsies [ 13 , 14 ]; however, their biological characteristics remain poorly defined. Addressing questions regarding their similarity to the tissue of origin is critical to fully comprehend their potential as reliable models for equine reproductive research [ 6 , 29 ]. While EEO exhibit functional and structural integrity [ 13 , 14 ], their molecular signatures have not been thoroughly documented. While functional similarity is critical to establishing physiological relevance, transcriptomic characterization offers complementary insights into the molecular pathways and cellular states underpinning organoid behavior. Together, these approaches provide a more comprehensive evaluation of the model’s fidelity and guide interpretation of experimental outcomes. Comparative transcriptome analyses between EEO and their native tissue, a key step in validating organoids as reliable models [ 30–33 ], are still missing. Organoids inherently lack components of the in vivo environment, such as immune cell recruitment, vascularization, and systemic hormonal signaling. Moreover, their development and differentiation are influenced by in vitro culture conditions, including growth factors, signaling molecules, and extracellular matrix composition [ 31 , 34 , 35 ]. These factors can profoundly impact their gene expression profile, making it essential to evaluate the gene expression pattern of EEO. We hypothesized that EEO retain core molecular features of their tissue of origin, including epithelial identity and reproductive cycle-associated gene expression, but that culture conditions induce alterations in genes related to cellular metabolism, immune function, and other processes. To address this premise, we aimed to analyze EEO gene expression at both bulk and single-cell levels and compare it to their native endometrial tissue (END), allowing assessment of their molecular fidelity and suitability for specific research applications. Furthermore, the dynamic nature of the endometrium, driven by hormones, prompts consideration of the potential influence of the donor’s reproductive cycle stage on organoid properties. In mares, the endometrium undergoes significant morphological (proliferative and secretory) and functional changes between the follicular (estrus) and luteal (diestrus) phases of the estrous cycle, including differences in epithelial cell height, ciliation, and receptor expression [ 36–39 ]. A prior study in mice demonstrated higher organoid-forming efficiency from END collected during estrus compared to other cycle stages [ 9 ], suggesting that the hormonal milieu of the donor animal at the time of tissue collection may shape organoid characteristics. However, this phenomenon remains scarcely examined. We hypothesized that the reproductive cycle stage of native tissue influences the molecular and morphological properties of EEO. To investigate this, we characterized the transcriptomic profiles of END collected during estrus and diestrus and assessed whether similar cycle-related differences are reflected in the structural, molecular, and developmental traits of the derived organoids. Finally, the stability of organoids during long-term culture and freezing/cryopreservation is another important factor for their utility as in vitro models. While endometrial organoids from other species have demonstrated genetic and transcriptomic stability over successive passages [ 10 , 18 ], the extent to which EEO maintain these properties remains unknown. Here, we hypothesized that EEO maintain the molecular stability during extended culture. Therefore, we aimed to assess the gene expression profiles of EEO across early and late passages to evaluate transcriptomic stability over time. Overall, to establish EEO as a robust and reliable model, we aimed to perform comprehensive molecular characterization, understand the influence of the donor estrous cycle stage, and ensure transcriptomic stability over time. By addressing these gaps, EEO could become a powerful tool for studying uterine physiology, intrauterine infections, and feto-maternal interactions, ultimately advancing equine reproductive research.