Written informed consent was obtained from each patient and the study was approved by the Human Research Ethics Committee at the Royal Women’s Hospital (03066B) and Melbourne IVF (74/19-MIVF).

All women (aged 26-42 years) had regular menstrual cycles (28-32 days), were not using intrauterine contraceptives for at least 3 months before surgery. Fertile women had proven parity (≥1 parous pregnancy) and infertile women had primary unexplained infertility defined as one year of failed conception with no apparent identifiable factor. Tissue samples were collected by curettage and examined by experienced gynecological pathologists to confirm the cycle stage and absence of apparent endometrial dysfunction. Each endometrial biopsy was cut into two pieces, one fixed in 10% formalin and the other one stored in ice-cold DMEM and processed for organoids culture within 1 h after collection.

The endometrial adenocarcinoma cell line Ishikawa was provided by Dr M. Nishida (Tsukuba University, Tochigi, Japan). The HTR8/SVneo trophoblast cell line (CRL-3271) was purchased from the ATCC and cultured as in the manufacturer’s instructions. Human trophoblast progenitor cells (HTPCs) are a kind gift of Professor Susan Fisher and Dr Olga Genbacev (University of California, San Francisco) and cultured as previously described (21). HTPCs are developed from human embryonic stem cells have a similar gene expression profile to human day 5 trophectoderm (21).

Collected endometrial tissue was finely incised and digested with collagenase III (727 µg/mL, CLS-3, Worhtington; NJ, USA) and Dnase I (25 µg/mL, 11284932001; Sigma, MO, USA). Epithelial glands and luminal epithelium were collected, washed and resuspended in ice-cold Matrigel (356231, Corning; NY, USA). 25 µL drops of Matrigel suspension were plated into a 12 well plate (3 drops/well) and cultured under defined expansion medium (ExM) (11). Organoids were passaged after 7-10 days of culture and passage 2-10 were used for this study. The ExM favours epithelial cell growth and stromal cells are lost after 1-2 passages (18). To model the proliferative and receptive phase of endometrial epithelium, organoids were treated with hormones as previously described (11). Briefly, organoids were passaged and cultured under ExM for 4 days to reform the organoid structure. ExM was then supplemented with 10 nM E2. After 2 days treatment, organoids were divided into two groups: і) 10 nM E2 or ii) 10 nM E2+1µM MPA+1µM cAMP. After 4 days of treatments, IOF and EOF (the media organoids were cultured in) were collected for analysis as described below. Human endometrial organoids are comprised of both luminal and glandular epithelial cells. This has been confirmed by immunofluorescence staining of luminal epithelial marker acetylated α-tubulin and glandular epithelial marker FOXA2 (11, 18). Recent single cell analysis has further identified that after E2+MPA+cAMP treatment, 20.8% of the organoid cells are ciliated cells which indicate luminal epithelium (18).

After hormone treatments, IOF and EOF were collected as previously described (16). The cultured ExM medium containing EOF was collected and centrifuged at 500 g for 5 min to remove cell debris. For IOF collection from both fertile (fertile IOF) and infertile organoids (infertile IOF), the remaining Matrigel domes were rinsed with ice-cold PBS and then incubated with Cell Recovery Solution (354253, Corning) at 4°C for 45 min. After incubation, organoids were collected and centrifuged at 270 g for 10 min, washed with ice-cold PBS and gently vortexed for 5 min to release IOF into 375 µL desired buffers based on downstream applications. The remaining organoid cells were collected and subjected to protein extraction and quantification by the bicinchoninic acid assay or RNA isolation using TRI Reagent.

Spheroids were generated using HTR8/SVneo trophoblast cells or HTPCs to mimic blastocyst adhesion. Briefly, 2000 HTR8/SVneo cells or 2500 HTPCs were cultured into each well of a U shape low attachment 96-well plate for 2 days to form a spheroid of size similar to a human blastocyst. Spheroids were then collected and added to the endometrial epithelial monolayer to mimic blastocyst adhesion. To test if incubation of spheroids with IOF (E2+MPA+cAMP treated) changed their adhesive capacity, HTR8/SVneo trophoblast cells were first used to optimize the IOF volume condition for incubation. In brief, 2000 HTR8/SVneo cells were plated into each well of a U shape low attachment 96-well plate containing 150 µL culture medium as previously described (9) with the addition of 1%, 5% or 10% (v/v) of fertile IOF. The remaining organoid cells were also collected as above and protein quantified to ensure equal volumes of IOF added were generated from a similar population of organoids. After 2 days of culture, trophoblast spheroids were collected and transferred to Ishikawa monolayers for the spheroid adhesion assay (9). Culture medium was removed after 4 h incubation and each well was gently washed once with PBS to remove non-adherent spheroids. The remaining spheroids were then counted, and the percentage attachment was expressed as a percentage of the original spheroid number.

After optimization, 10% (v/v) IOF was used for the spheroid adhesion assay using HTPC-derived spheroids where the Ishikawa cells were replaced with receptive phase organoid-derived monolayers. To generate the monolayer, each well of the 96 well plate was precoated with a thin layer of Matrigel/DMEM/F12 (1:1). Organoids (after at least 4 days of culture) were broke up by pipetting up and down several hundred times before being transferred to Matrigel coated wells. Organoid cells were cultured in DMEM/F12 with 10% fetal calf serum (Invitrogen, CA, USA) and once cells outgrew and reached 100% confluency, HTPC spheroid adhesion was tested as described above. As controls for IOF treatment, HTR8/SVneo trophoblast cells or HTPCs were treated with cell culture medium (medium control) or 10% (v/v) fertile EOF and spheroid adhesion was determined and compared with IOF treated spheroids.

Organoid-derived epithelial cells on Matrigel/DMEM/F12 were cultured in adherence to 70% confluence and transfected with miR-29c or scrambled control for 24 h using Lipofectamine RNAiMAX reagent and Opti-MEM (Thermo, MA, USA) according to the manufacturer’s instructions. Transfection medium was then replaced with fresh culture medium and cells were cultured for 48 h before downstream functional analyses.

Human endometrial tissues were fixed in 10% formalin, embedded in paraffin and sectioned at 4 μm thickness. Sections were then dewaxed, rehydrated, and antigen retrieved (microwaving in 10 mM sodium citrate for 5 min). Endogenous peroxidase activity was blocked with 3% hydrogen peroxide for 15 min at room temperature. Following washing in Tris-buffered saline (TBS) and non-immune blocking in 10% goat serum and 2% human serum in TBS, sections were incubated with primary antibodies overnight at 4°C (details summarized in Supplemental Table 1 ). An isotype control was included in every slide in which the non-immune antibody of the same IgG isotype was substituted for each primary antibody at the same concentration. Positive signaling was revealed via the avidin-biotin-diaminobenzidine system. Sections were counterstained with hematoxylin to indicate cell nuclei. Organoids were embedded in 5% agar and processed as described for endometrial tissues. For immunocytochemistry staining, organoid cells were fixed in 4% paraformaldehyde and permeabilized by incubation in 0.1% Triton X-100, cells were then treated as described for endometrial tissues.

Total RNA was isolated using TRI Reagent (T9424, Sigma) and treated with the TURBO DNA-free kit (AM1907, Thermo) to remove genomic DNA contamination. For gene targets detection, 300 ng total RNA was converted to cDNA using the Thermo SuperScript III First-Strand Synthesis kit. qPCR was performed using SYBR Green Master Mix (4367659, Thermo) on the Applied Biosystems ViiA7 system as follows: 95 °C for 10 min and 40 cycles of 95 °C for 15 s followed by 60 °C for 1 min. Gene expression was normalized to 18S. miR-29c qPCR was conducted using Taqman miR reverse transcription kit and Universal Master Mix II (4366596 and 4440040, Thermo) according to the manufacturer’s instructions. miR-29c expression was normalized to U6. Relative expression levels of both genes and miR-29c were calculated using the comparative cycle threshold method (ΔΔCt). Primer sequences used are summarized in Supplemental Table 2 .

Secreted proteins in IOF were concentrated via SpeedVac evaporation and quantified via bicinchoninic acid assay. Samples (25 µg each) were processed for proteomics analysis using single-pot, solid-phase-enhanced preparation as described previously (22). Mass-spectrometry results were analyzed using MaxQuant (Version 2.0.1.0). Peptides were matched against the UniProt Homo sapiens proteome database (updated June 20, 2021) for protein identification (23). The cleavage enzyme was set to Trypsin with a maximum of two missed cleavages and modifications selected were Oxidation (M) (variable) and Carbamidomethyl (C) (fixed). Proteins labeled as “only identified by site” and “reverse” were removed from the data set (24). The remaining list of proteins was filtered using criteria “Razor & unique peptides >1” and “Unique peptides >0” (24). Proteins were also quantified by MaxQuant’s built in Label-free quantification program and intensity-based absolute quantification was identified for each protein. Functional annotation and clustering of predicted proteins were performed using DAVID. Functional clusters that satisfied the criteria of “enrichment score >1” and “P<0.05” were considered enriched (25).

IOF proteins were precipitated by addition of 5 times volume of ice-cold acetone and overnight incubation at -20°C. Protein pellets collected by centrifugation were air dried and resuspended in 5% sodium dodecyl sulfate buffer (containing 20 mM EDTA, 140 mM NaCl, 100 mM Tris pH 8.0 and proteinase inhibitor) for protein extraction. Proteins were quantified using a bicinchoninic acid protein assay kit and equal amounts of protein were resolved on 4–15% precast polyacrylamide gel and transferred to polyvinylidene difluoride membranes. Membranes were blocked with 5% skim milk in TBS for 1 h and incubated with primary antibody prepared in 5% skim milk ( Supplemental Table 1 ). After washing three times with TBS-Tween 0.1% (v/v), membranes were incubated with appropriate horse radish peroxidase (HRP)-conjugated secondary antibodies. After additional washes in TBS-Tween 0.1% (v/v), labeled proteins were detected using a chemiluminescence kit. For quantification, appropriate bands were assessed by densitometry.

A minimum of three biological replicates were performed for all experiments (indicated by n in figure legends). For endometrial organoids, all the datapoints on the plots were representative of different organoid lines derived from different patients. Statistical analysis was carried out using PRISM 8.0 and student’s t-test or one-way ANOVA as appropriate. All data were checked for normal distribution using the Shapiro Wilk test and non-parametric tests were used for data that did not pass normal distribution test. Graphical data were presented as the mean ± SEM. P<0.05 was considered statistically significant.

We established organoids from both fertile and infertile endometrium. For both groups, the glandular and luminal fragments formed organoid structures within 1-2 passages with no obvious differences ( Figure 1A ). The spatial localization pattern of the tight junction protein Cadherin 6 (CDH6) and epithelial cell marker E-cadherin in organoids and matched endometrium from the same women demonstrated that both factors were similar between organoids and endometrial glands ( Figure 1B ).

Treatment of fertile organoids with estrogen or estrogen and progesterone model changes in the proliferative and receptive phases of the endometrial epithelium accordingly (11, 12, 18). To determine whether infertile organoids exhibit defects in response to hormone treatments, we examined the expression of select receptivity markers that show difference between estradiol (E2) only or E2+medroxyprogesterone acetate (MPA, a stable progestogen) +cAMP ( Figure 2A ) (11, 12, 18). Among six receptivity markers examined in fertile organoids, the expression of Hydroxysteroid 17-Beta dehydrogenase 2 (HSD17B2), Glutathione peroxidase 3 (GPX3), FOXO1, MUC1 and Paired box 8 (PAX8) was significantly increased while PGR was significantly reduced (P<0.05) after E2+MPA+cAMP treatment, compared to E2 only ( Figure 2B ). By contrast no significant differences were identified for the expression of all genes investigated in infertile organoids between E2 only and E2+MPA+cAMP treatments ( Figure 2C ).

We generated endometrial epithelial cell monolayers using fertile organoids to investigate cell adhesive capacity in vitro. To validate our model we compared the effect of miR-29c overexpression on cell adhesion with our previous work using primary HEEC monolayers (10). Light microscopy first revealed that the endometrial epithelial cells outgrew similarly from endometrial glands and organoids ( Figure 3A ) to form a confluent cell monolayer using the same culture medium. Immunocytochemistry staining of organoid-derived epithelial cell cultures identified that E-cadherin was localized to the cell membrane and cytoplasm ( Figure 3B ). Immunocytochemistry staining of the basal membrane marker COL4A1 confirmed that the endometrial epithelial monolayers derived from the organoids stained basally ( Figure 3C ).

Functionally we identified that miR-29c overexpression in organoid-derived monolayers significantly compromised their adhesive capacity to HTR8/SVneo spheroids (blastocyst surrogates) compared to control ( Figure 3D ). Among four potential miR-29c targets examined in our previous study using primary HEECs, only COL4A1 expression is reduced after miR-29c overexpression (10). Similarly, in organoid-derived monolayers, miR-29c overexpression significantly reduced (P<0.05) only COL4A1 expression ( Figure 3E ). No significant differences were identified for the expression of Cell division cycle 42 (CDC42), Murine double-minute 2 homolog (MDM2) and Integrin subunit beta 1 (ITGB1) compared to control ( Figure 3F ).

Endometrial glands and organoids exhibited similar apical-basal polarity (12, 14). To identify the effect of organoid apical secretions on blastocyst adhesion, organoids from both fertile and infertile groups were treated with E2+MPA+cAMP to model the receptive phase and IOF was collected using a recently optimized methodology (16). We confirmed that after lightly vortexing organoids were mildly disrupted to release IOF ( Figure 4A ). To optimize the concentration of IOF to investigate its effect on endometrial epithelial cell adhesive capacity, fertile IOF at 1%, 5% and 10% (v/v) was prepared and used to treat HTR8/SVneo trophoblast spheroids and their adhesive capacity on Ishikawa monolayers measured ( Figure 4B ). Only treatment with 10% (v/v) IOF significantly increased HTR8/SVneo spheroid adhesive capacity (P<0.05) to Ishikawa monolayers compared to medium control ( Figure 4C ).

10% fertile or infertile IOF was therefore used in the subsequent experiments. HTR8/SVneo spheroids and Ishikawa cells were replaced by HTPC spheroids and fertile organoid-derived monolayer, respectively. Incubation of HTPC spheroids with 10% (v/v) fertile IOF significantly increased their adhesion to fertile organoid-derived monolayers compared to medium control and 10% extra-organoid fluid (EOF, representing organoid basal secretions) control (P<0.001, Figure 4D ) respectively. By contrast incubation of HTPC spheroids with 10% (v/v) infertile IOF significantly reduced their adhesion to fertile organoid-derived monolayer compared to 10% (v/v) fertile IOF (P<0.0001), medium control (P<0.01) and 10% (v/v) fertile EOF control (P<0.01) respectively ( Figure 4D ). No significant difference in adhesion was identified between medium control and 10% (v/v) fertile EOF control treatment groups ( Figure 4D ).

In view of the ability of IOF to alter the adhesive capacity of trophoblast spheroids, we next sought to investigate proteomic profile of fertile and infertile IOF. Organoids from both groups were treated with E2+MPA+cAMP and IOF collected for proteomic analysis. The analysis identified a total of 1150 proteins in fertile and infertile IOF. Using a threshold of 1.5-fold change and a false discovery rate of <0.05, we identified 131 proteins were decreased and 19 proteins increased in infertile IOF, compared to fertile IOF ( Figures 5A, B ; Supplemental Tables 3 and 4 ). 131 proteins that were reduced in the infertile IOF were further interrogated for enriched KEGG pathway and Gene Ontology (GO) (biological process, cellular component and molecular function) using DAVID Bioinformatics Resources (version 2021). Analysis of the top 15 KEGG pathways identified enrichment of pathways associated with cell membrane movement (endocytosis and regulation of actin cytoskeleton), infections (Pathogenic Escherichia coli infection) and protein metabolism (proteasome, cysteine and methionine metabolism and biosynthesis of amino acids) ( Figure 5C ; Supplemental Table 5 ). Similarly, analysis of the top 15 GO molecular function categories returned enriched terms related to cell membrane movement and protein binding (actin-binding, structural constituent of cytoskeleton, protein domain specific binding and G-protein beta/gamma-subunit complex binding) ( Figure 5D ; Supplemental Table 6 ). These enriched pathways and molecular functions have been consistently identified in day 16-21 normal human endometrial fluid samples as enriched categories (8), suggesting the physiological relevance of IOF. Common proteins between IOF and endometrial fluid were underlined in Supplemental Table 3 . The most dominant GO cellular component category was extracellular exosome, with 74% of the decreased proteins (97/131) in infertile IOF mapped to this category ( Figure 5E ; Supplemental Table 6 ).

We confirmed the differential secretion of MUC5AC between fertile and infertile groups by immunoblotting. Similar to the proteomics data, MUC5AC was significantly increased in infertile IOF compared to fertile IOF (P<0.05, Figure 6A ). Confirmation of the proteomics data at the mRNA level was also investigated. The organoid cells after IOF collection were subjected to qPCR for the top six proteins that were differentially produced between fertile and infertile IOF ( Supplemental Tables 3 and 4 ). Among the top decreased proteins in infertile IOF, the expression of Small RNA binding exonuclease protection factor la (SSB), Myosin heavy chain 10 (MYH10), Heat shock protein family A (Hsp70) member (HSPA)9 and WD repeat domain 61 (WDR61) was significantly decreased in infertile organoid cells compared to fertile organoid cells respectively ( Figure 6B ). No significant changes were identified in the expression of X-ray repair cross complementing 6 (XRCC6) and HSPA2 ( Figure 6B ). Among the top six proteins that showed a significant increase in infertile IOF, the expression of Mucin 5AC (MUC5AC) was significantly increased in infertile compared to fertile organoid cells (P<0.05), while no significant differences were identified in the expression of Lymphocyte cytosolic protein 1 (LCP1), RNASE1, Cathepsin H (CTSH), BPI fold containing family b member 1 (BPIFB1) and Periplakin (PPL) between groups ( Figure 6C ).