WO2020225085A1

WO2020225085A1 - Medium for growth of organoids

Info

Publication number: WO2020225085A1
Application number: PCT/EP2020/062012
Authority: WO
Inventors: Matteo BORETTO; Hugo VANKELECOM
Original assignee: Katholieke Universiteit Leuven
Priority date: 2019-05-03
Filing date: 2020-04-30
Publication date: 2020-11-12

Abstract

The invention relates to a cell culture medium and the use of such medium for the growth and maintenance of organoids from endometriosis tumor tissue.

Description

MEDIUM FOR GROWTH OF ORGANOIDS

FIELD OF THE INVENTION

The inventions relates to the growth of organoids from endometrial tissue.

The invention further relates to growth media for organoids.

INTRODUCTION

Endometrial diseases are the first gynaecological burden and primary ground of infertility. Endometriosis, caused by ectopic growth of endometrium-like tissue, affects 1 in 10 women of reproductive age. Aetiology and molecular pathogenesis are still unclear, and treatment remains unsatisfactory. Retrograde menstruation is the most suspected culprit, possibly thriving on a predisposing genetic background. Endometrial cancer (EC) is the 4^th most common type of cancer among women. Type I EC, which is oestrogen-dependent, represents the majority of cases but shows favourable prognosis (85% 5-year survival rate). Type II EC is oestrogen- independent and frequently high grade with poorer prognosis (25-60% 5-year survival rate) and high risk of metastasis. Clinical management follows progression stage (I to IV) and usually involves surgical resection with chemotherapy and/or adjuvant radiotherapy. However, the cancer often recurs. Cellular and molecular mechanisms underlying EC remain largely unknown, and therapy efficiency and overall survival rate have not substantially improved in the last decade. This standstill is mainly due to a lack of reliable preclinical study models. The few carcinoma-derived cell lines available poorly recapitulate the clinical heterogeneity of EC [Vollmer Endocr. (2003) Relat. Cancer 10, 23-42]. Type I low grade tumours do not successfully grow in vitro and their ability to establish patient-derived tumour xenografts (PDTX) is bleak [Depreeuw et a/. (2015) Gynecol. Oncol. 139, 118-126]. Genetic mouse models show aberrations outside the endometrium which is inconsistent with clinical EC [Vollmer Endocr. (2003) Relat. Cancer 10, 23-42; Contreras et at. (2010) D/s Model Mech 3, 181-193]. Finally, no study models are currently available for pre-cancerous endometrial hyperplasia, and animal models of endometriosis do not recapitulate nature and complexity of this human-typical disease [King et al. (2016) J. Pathol. 238, 185-196].

Boretto et al. (2017) Development 144, 1775-1786 and Turco et al. (2017) Nat. Cell Biol. 19, 568-577 recently developed 3D in vitro organoids from human endometrium which recapitulate key biological features of the tissue's epithelium. SUMMARY OF THE INVENTION

The present invention discloses organoid models for endometrial disease, ranging from endometriosis and hyperplasia to low and high grade cancer. The organoids show long-term expandability and genomic and transcriptomic stability, and recapitulate disease diversity by retaining type- and patient-specific characteristics. The present invention discloses an endometrial organoid biobank encompassing both healthy and pathological conditions, serving as new promising preclinical study models and drug screening and discovery tools.

Endometrial disorders present a major gynaecological burden. Current research models fail to recapitulate these diseases in nature and heterogeneity, thereby hampering scientific and clinical progress. Herein disclosed are long-term expandable organoids from a broad spectrum of endometrial pathologies. Organoids from endometriosis show disease-associated traits and cancer-linked mutations. Endometrial cancer-derived organoids show faithful capture of cancer subtype, recapitulate the tumour's mutational landscape and display patient-specific drug responses. Organoids could also be established from pre-cancerous pathologies encompassing endometrial hyperplasia and Lynch syndrome with maintenance of inherited gene mutations. Finally, endometrial disease organoids reproduce the original lesion when transplanted in vivo. Disclosed herein are multiple organoid models capturing endometrial disease diversity and providing new powerful research models and drug screening and discovery tools.

The compounds and methods of the present invention have the following advantageous properties and applications.

The invention allows improving IVF pregnancy success through valorisation of IVF embryos by endometrial organoids

The invention allows to develop a diagnostic test to assess the quality of an individual IVF embryo by scoring the gene expression imprint of the embryo secretome (as present in the 'blastocyst spent medium' or BSM) on the endometrium's epithelium (as avatared by organoids, comprehensively shown to represent authentic 'replica'). Diagnostic tests will predict which embryo has the best chances of implantation/pregnancy after IVF and embryo transfer, by implementing endometrial organoids (EMO; also hybrid EMO) as biosensors of embryo quality. The final clinical aim is to improve the success rates of IVF-routed pregnancy. The invention allows to improving infertility management through valorisation of the infertile patient's endometrium (avatared by organoids) by their response to high quality embryo secretome. The invention allows to decipher how the 'infertile' endometrium responds differently to BSM of IVF embryos that gave rise to successful pregnancy (high quality embryos) as compared to 'fertile' endometrium. This expected to lead to a diagnostic test which scores (statistically calculates) the individual patient's endometrium competence and probability for embryo receptivity ('fertility talent'). The target population includes patients that show sub- or infertility (defined as the failure to achieve pregnancy after at least 12 months of unprotected intercourse) and start IVF, for instance endometriosis patients (striking 10% of the women in reproductive age of which 50% suffer from sub-/infertility).

The underlying research exploring embryo impact on 'infertile' EMO (iEMO) is also expected to identify molecular causes of endometrial failure, thus further impacting the field of (un-)fertility and assisted reproduction (IVF). This exploration can provide hints toward supportive or curing therapy, and will reveal potential therapeutic targets and guide endometrium-targeted therapy to treat the sub/infertile patient (even in a patient-tailored manner; personalized medicine). Since organoids have been shown to provide a valuable drug screening platform, the iEMO can be applied to serve as a preclinical drug screening tool to battle infertility, also potentially in an individual patient setting. Moreover, the present invention allows to lead to new diagnostic biomarkers of endometrium pathology.

The invention is summarized in the following statements:

1. Use of a medium for the generation, growth or maintenance of organoids from endometriosis tissue, wherein the medium comprises:

- a chemically defined lipid concentrate,

- IGF-1 (insulin-like growth factor 1)

- HGF (hepatocyte growth factor),

- IL-6 (interleukin-6),

- RSPOl (R-spondin 1),

- nicotinamide,

- an activin receptor-like kinase (ALK) inhibitor, such as A83-01 ,

- a p38 MAP kinase inhibitor, such as inSB202190, and

- 17 beta-estradiol. 2. The use according to claim 1, wherein the medium does not comprise insulin, transferrin, selenium, FGF2 (Fibroblast Growth Factor 2), and FGF10 (Fibroblast Growth Factor 10).

3. The use according to claim 1 or 2 , wherein in said medium:

- the concentration of RSPOl is between 50 and 200 ng/ml between 50 and 150 ng/ml, between 125 and 75 ng/ml, between 90 and 110 ng/ml, or is 100 ng/ml,

- the concentration of nicotinamide is between 4 and 6 mM, between 4.5 and 5.5 mM or 5 mM,

- the concentration of A83-01 is between 0.2 and 0.3 mM, or 0.25 pM,

- the concentration of SB202190 is between 1 and 0.01 pM, between 0.5 and 0.05 pM, between 0.25 and 0.05 pM, or 0.1 pM,

- the concentration of 17beta-estradiol is between 5 and 15nM, between 7.5 and 12 nM, between 9 and 11 nM, or 10 nM.

4. The use according to any one of claims 1 to 3, wherein said medium comprises a conditioned medium of a cell line expressing RSPOl, at a concentration between 2.5 and 7.5 v/v % between 4 and 6 v/v %, or between 4.5 and 5.5 v/v %, or of 5 v/v %. (volume RSPOl comprising conditioned medium/ total volume medium for generation, growth or maintenance).

5. The use according to any one of claims 1 to 4, wherein in said medium:

- the chemically defined lipid concentrate is at a concentration of between 0.5 and 2.5% (v/v), between 0.75 and 1.25 % (v/v) or at 1 % (v/v),

- IG is in a concentration of between 30 and 50 ng/ml, between 35 and 45 ng/ml or 40 ng/ml,

- HGF is in a concentration of between 10 and 30 ng/ml, 15 and 25 ng/ml, 17 and 23 ng/ml or 20 ng/ml, and,

- 11-6 is at a concentration of between 4 and 6 ng/ml, between 4.5 and 5.5 ng /ml, or 5 ng/ml.

6. The use according to any one of claims 1 to 5, wherein the medium further comprises transforming growth factor (TGF) -alpha in a concentration of 1 to 20, 5 to 15, 7.5 to 12.5 or 10 ng/ml medium.

7. The use according to any one of claims 1 to 6, wherein the medium comprises a basal medium with L-Glutamine, comprises Noggin, L-alanyl-L-glutamine dipeptide (Glutamax), and a ROCK inhibitor such as Y27632, and supplements such as B27 and N2.

8. The use according to any one of claims 1 to 7, wherein the endometrium is a tumour tissue. 9. A medium suitable for the generation, growth or maintenance of organoids from endometriosis tissue, wherein the medium comprises:

- a chemically defined lipid concentrate,

- IGF-1 (insulin-like growth factor 1)

- HGF (hepatocyte growth factor),

- IL-6 (interleukin-6),

- RSPOl (R-spondin 1),

- nicotinamide,

- an activin receptor-like kinase (ALK) inhibitor, such as A83-01 ,

- a p38 MAP kinase inhibitor, such as inSB202190, and

- 17 beta-estradiol.

10. The medium according to claim 9, wherein the medium does not comprise insulin, transferrin, selenium, FGF2 (Fibroblast Growth Factor 2), and FGF10 (Fibroblast Growth Factor 10).

11. The medium according to claim 9 or 10 , wherein in said medium:

- the concentration of RSPOl is between 50 and 200 ng/ml between 50 and 150 ng/ml, between 125 and 75 ng/ml, between 90 and 110 ng/ml, or is 100 ng /,

- the concentration of A83-01 is between 0.2 and 0.3 mM, or 0.25 pM,

12. The medium according to any one of claim 9 to 11, wherein said medium comprises a conditioned medium of a cell line expressing RSPOl, at a concentration between 2.5 and 7.5 v/v % between 4 and 6 v/v %, or between 4.5 and 5.5 v/v %, or of 5 v/v %. (volume RSPOl comprising conditioned medium/ total volume medium for generation, growth or maintenance).

13. The medium according to any one of claims 9 to 12, wherein in said medium:

- the chemically defined lipid concentrate is at a concentration of between 0.5 and 2.5% (v/v), between 0.75 and 1.25 % (v/v) or at 1 %,

- IGF-1 is in a concentration of between 30 and 50 ng/ml, between 35 and 45 ng/ml or 40 ng/ml,

- HGF is in a concentration of between 10 and 30 ng/ml, 15 and 25 ng/ml, 17 and 23 ng/ml or 20 ng/ml, and, - 11-6 is at a concentration of between 4 and 6 ng/ml, between 4.5 and 5.5 ng /ml, or 5 ng/ml.

14. The medium according to any one of claims 9 to 13, wherein the medium further comprises transforming growth factor (TGF) -alpha in a concentration of 1 to 20, 5 to 15, 7.5 to 12.5 or 10 ng/ml medium.

15. The medium according to any one of claims 9 to 14, wherein the medium comprises a basal medium with L-Glutamine, comprises Noggin, L-alanyl-L- glutamine dipeptide (Glutamax), and a ROCK inhibitor such as Y27632, and supplements such as B27 and N2.

In specific embodiments of the medium and the use of the invention it is envisaged that Noggin and/or RSPOl are omitted from the medium.

DETAILED DESCRIPTION FIGURE LEGENDS

Figure 1 Long-term expandable organoids can be established from endometriosis, (a) Flowchart depicting organoid establishment from the spectrum of endometrial diseases as indicated, and downstream analyses, (b) Representative bright field pictures of organoid development from healthy (EM-O), ectopic (ECT-O) and eutopic (EUT-O) endometrium after seeding (passage 0, P0). Arrow indicates the specific tissue subunit monitored in time. Scale bar, 50 pm. (c) Formation efficiency of EM- 0, EUT-0 and ECT-0 (scatter plot). ****P<0.0001, ***P<0.001, *P<0.05 (n= 6). (d) Efficiency of organoid establishment (% of total biopsies seeded; r?>10) from the different endometrium conditions as indicated (HYP-O: organoids from hyperplastic endometrium), (e) Proliferation analysis of ECT-O: immunofluorescence and quantification of Ki67⁺ cells in organoids from 3 independent donors (mean + s.d.) (DAPI as nuclear stain), (f) Representative bright field pictures of different passages (P) demonstrating long-term expansion of EM-0 and ECT-O (representative structure magnified in inset), and from matched EUT-0 and ECT-O from an individual patient (right). Scale bars, 200 pm. (g) Representative aCGH plots of ECT-O from low (1-2 months) and high (4-6 months) passage number. SCNA are absent, a property maintained during long-term culture, (h) Analysis of medium factor requirement for growth of EM-0 (as compared to growth in standard organoid medium or SOM, pictured by line) and ECT-O (as compared to growth in SOM, pictured by dashed line). Organoids were cultured in the absence ("-") of the indicated compounds (individually omitted from SOM) for 7 days and the organoid number determined (mean + s.e.m.; r?=3). *P< 0.05 among EM-0 as compared to SOM, ^$P<0.05 among ECT-O as compared to SOM. Figure 2 Endometriotic organoids reproduce the primary lesion in vitro and in vivo. (a) Histological (H&E) analysis, immunohistochemical examination of endometrial markers and detection of mucin (PAS staining) in primary endometriotic peritoneal lesions and corresponding organoids (ECT-O), and immunofluorescence analysis of laminin in organoids. Scale bars, 50 pm. Arrows point to the invasive phenotype (H&E) or to mucus production (PAS). Boxed areas are magnified as indicated. Scale bar, 10 pm. (b) TEM picture revealing glandular-like morphology, stratified epithelium (*) and presence of microvilli and cilia (box) and of non-ciliated secretory cells with secretory vacuoles (dashed box) in ECT-O. Boxed areas are magnified as indicated, (c) Representative H&E analysis, immunohistochemical examination for endometrial markers and detection of mucin (PAS; arrow) in peritoneal engraftments of ECT-O two months after intraperitoneal injection (n= 4). Inset shows H&E staining of sham (PBS-injected) mice, not growing any lesion. Scale bars, 50 pm.

Figure 3 Transcriptomic analysis of endometriotic organoids reveals disease- and stage-specific genes, (a) Heatmap of all differentially expressed genes (log2 normalized counts) as identified by RNA-seq analysis of ECT-O, EUT-0 and EM-O. Colors range from blue (low expression) to red (high expression). The most important genes are indicated, (b-d) Gene expression analysis of WNT pathway targets (b), WNT pathway receptors (c) and endometrial markers (d) in stage I to IV ECT-O as normalized to GAPDH and expressed as fold change relative to EM-0 (mean + s.e.m.; n= 4). *P< 0.05. (e) Mutation matrix representing hits in EC-associated genes as detected by WES in ECT-O. For a complete list of genetic alterations, see ArrayExpress with accession number E-MTAB-7688.

Figure 4 Organoids from endometrial pre-cancer lesions display disease-associated phenotype and gene mutations, (a) Brightfield pictures of organoids (HYP-O) from low and high passage number, derived from three main endometrial hyperplasia types as indicated. Scale bar, 200 pm. (b) Proliferation analysis of HYP-O: immunofluorescence and quantification of Ki67⁺ cells in organoids from 3 independent donors (mean + s.d.). (c) H&E analysis and PanCK immunostaining of primary hyperplastic lesions and HYP-O. Scale bar, 50 pm.

(d) Representative brightfield pictures of organoids from endometrial biopsies of Lynch syndrome patients at low and high passage number. Scale bar, 200 pm. (e) Targeted sequencing results of Lynch syndrome patient endometrial biopsies (primary) and corresponding organoids. The 'HYP-0_7' patient harbours the heterozygote c. l494dupT (p.Ala499Cysfs*14) mutation in MSH2 (LRG_219) as identified in the primary endometrial biopsy, which is retained in the organoids (P3); the ΉUR-0_6' patient harbours the heterozygote c.3545_3546DdelGA (p.Argl l82Asnfs*5) mutation in MSH6 (LRG_218) as identified in the primary endometrial biopsy, which is retained in the organoids (P3).

Figure 5 EC-derived organoids capture the cancer's heterogeneity and mutational landscape, and reveal type-associated gene expression, (a) Efficiency of organoid establishment from cancer (EC-O) before (i.e. in SOM) and after medium optimization (% of total biopsies seeded; n= 22 and 27, respectively), (b) Representative brightfield pictures of organoids from low grade (I) EC at low and high passage number (scale bar, 200 pm; representative structure magnified in inset), and H8iE analysis (scale bar, 50 pm), (c) Proliferation analysis of EC-0 derived from different EC grades: quantification of Ki67⁺ cells (mean + s.d.) and Ki67 immunofluorescence analysis in a low grade EC-0 line, (d) H8iE and immunohistochemical analysis of endometrial (cancer) markers in low grade (I) EC and corresponding organoids (inset shows brightfield picture of the organoid culture). Scale bar, 50 pm. (e) Mutation matrix depicting hits in 21 of the most frequently mutated genes in EC as identified by WES in primary EC samples and derived EC-0 (early and/or late passage). For a complete list of genetic alterations, see ArrayExpress with accession E-MTAB-7688. Representative brightfield pictures of 3 EC-0 lines harbouring CTNNB1 mutations (or not), cultured in SOM with or without RSPOl (right). Scale bar, 200 pm. (f) Heatmap of differentially expressed genes in EM-O, HYP-O, EC-0 and primary tumours, as identified by RT-qPCR and presented as relative expression to GAPDH (log2 relative expression). Colors range from blue (low expression) to red (high expression). The most important genes are indicated.

Figure 6 Orthotopically engrafted EC-derived organoids reproduce the primary tumour phenotype, (a) H8dº analysis of the grafted uterine horns. No lesion is observed in vehicle-injected horn showing preserved tissue architecture which is completely lost in horns injected with Hec-IA cells and high grade (III) EC-O. The latter EC-0 also gave rise to peritoneal metastasis with comparable histology. The low grade (I) EC-0 generated localized tumours (box). Insets present magnifications of boxed areas. Scale bars, 300 pm for overviews and 50 pm for magnifications, (b) Immunohistochemical analysis of ER-alpha and PR shows tumour-specific characteristics. Grafts from Hec-IA cells and high grade (type II) EC-0 are hormone receptor-negative and from low grade (type I) EC-0 hormone receptor-positive. Scale bar, 50 pm. (c) Dose-response curves of EC-0 treated with the indicated drugs, as measured using the XTT cell viability assay (each curve is mean of r?=3; individual data points are shown). IC50 is indicated with dashed line for each organoid type (for numerical data, see Fig. 14e).

Figure 7 Organoids derived from (pre-)cancerous endometrium reveal specific ion channel functionality, (a) Percentage of cells responding to specific ion channel activators in EM-O, HYP-0 and EC-O, the latter subdivided in non-invasive (N-IN) and invasive (IN) phenotypes. r?= 2-6 organoid lines per group. *P<0.05, ***p<0_ooi compared to EM-O, ^$$$P<0.001 of EC-0 IN versus ECO N-IN. (b-d) Absolute Ca²⁺ amplitude upon stimulation with the specific ion channel activator as indicated (mean ± s.e.m. ; n= 2-6 independent measurements per group with a total minimum of 195 cells). ***P<0.001 compared to EM-O, ^$$$P<0.001 of EC-0 IN versus ECO N-IN . (e) Percentage of Ki67⁺ cells in 3 different EC-0 treated with a cocktail of ion channel inhibitors (or vehicle, "SOM"). *P<0.05 (n>4) .

Figure 8 Long-term expandable organoids can be established from endometriosis at all stages, (a) Representative brightfield pictures of the initial development of EM-0 and matched ECT-0 and EUT-0 after seeding (P0). Organoid initiation from ectopic endometrium is slower and less efficient. Scale bar, 200 pm. (b) Representative brightfield pictures showing clonal potential of ECT-0 formation. Organoids were harvested from the Matrigel drop, dissociated and seeded at single-cell density (10 cells per well). Arrow indicates a single cell that is monitored for organoid formation for 20 days. Boxed area is magnified. Scale bar, 50 pm (n= 3). (c) Passaging time of EM-0 and of matched EUT-O/ECT-O cultures (box plot). Cultures from EM-0 of 3 donors and from matched EUT-O/ECT-O of 3 other donors were monitored for 6 passages (d) Representative brightfield and H&E images of organoids derived from rASRM stage I to IV endometriosis as indicated. Scale bar, 200 pm for brightfield and 50 pm for H&E. (e) Representative aCGH plots of short-term (1-2 months; low passage) and long-term (4-6 months; high passage) cultured EM-0 and EUT-O. SCNA are absent, a property maintained during long-term culture, (f) Gene expression analysis of endometrial markers in ECT-0 and EM-0 after short-term and long-term culture, presented as ACt (Ct of gene - Ct of GAPDH ), indicating transcriptomic stability after extensive expansion (mean ± s.e.m. ; n- 3). (g) Rescue of IWP2- induced organoid growth inhibition represented as the percentage of organoids formed after 10 days with the indicated treatment as compared to SOM (left graphs), and gene expression analysis of canonical (middle) and non-canonical (right) WNT target genes in ECT-0 (top) (n= 5) and EM-0 (bottom) (n³ 4). *P<0.05, **P<0.01, p****<0.0001. (h) WNT ligand gene expression in ECT-0 as determined by RT-qPCR and represented as ACt (Ct of gene - Ct of GAPDH ) (scatter plot from 8 independent donors) (left) and as extracted from the RNA-seq dataset and presented as heatmap of transcript per million values (right) in which colors range from white (lowest) to red (highest).

Figure 9 Endometriotic organoids show disease characteristics, (a) H&E analysis of matched EUT-0 and ECT-O. Scale bar, 50 pm. (b) Representative pictures of immunohistochemical analysis of MMP2 and MMP7 in primary peritoneal endometriotic lesions and ECT-O. Scale bar, 50 pm. Gene expression of MMP2 and MMP7, as normalized to GAPDH and expressed as fold change relative to EM-0 (mean + s.e.m.; n= 4). (c) Representative picture of E-cadherin immunofluorescence staining of ECT-O showing the epithelial nature and correct positioning of tight junctions which supports the cells' polarization. Scale bar, 10 pm. (d) TEM pictures of matched EUT-0 and ECT-O from an individual patient. The EUT-0 are composed of a single-cell layer bordering a lumen, containing microvilli (magnified box) and ciliated cells (as revealed by acetylated (Ac) alpha-tubulin immunofluorescence), whereas a stratified, double-cell layer (*) is present in the ECT-O with extensive microvilli (magnified box) and ciliated cells. Scale bar, 10 pm. (e) ECT-O, subrenally transplanted according to the schedule, give rise to lesions with endometriotic features. Representative H&E images and immunohistochemical analysis for ERalpha and PR of the kidney capsule and ECT-O grafts is shown. Lower panels display H&E staining of the non-grafted kidney, lacking any lesion. Scale bar, 200 pm (n= 6). Figure 10 Transcriptomic analysis of endometriotic organoids reveals disease- specific pathways and genes, (a) PCA plot showing the distribution of EM-O, EUT-0 and ECT-O, based on the RNA-seq data, (b) Diagram of the differentially expressed genes in EM-O, EUT-0 and ECT-O as identified by RNA-seq analysis. Of the 277 differentially expressed genes between EM-O/EUT-O and ECT-O, 34 are specifically expressed in EM-O/EUT-O and 243 in ECT-O. Of the 35 differentially expressed genes between EM-0 and EUT-O, 15 are specifically expressed in EM-0 and 20 in ECT-O. Relevant genes are indicated, (c) Expression of WNT ligand and Hippo pathway target genes in stage I to IV ECT-O as normalized to GAPDH and expressed as fold change relative to EM-0 (mean + s.e.m.; r?=4). *P< 0.05. (d) Expression of invasion and inflammatory marker genes in stage I to IV ECT-O as normalized to GAPDH and expressed as fold change relative to EM-0 (mean + s.e.m.; r?=4). *P< 0.05. (e) Top divergent pathways between ECT-O and EM-0 as identified by KEGG pathway analysis using the 277 differentially expressed genes, (f) Top divergent biological terms between ECT-O and EM-0 as identified by DAVID GO enrichment using the 277 differentially expressed genes, (g) PCA plot showing the distribution of matched EUT- O and ECT-O, based on the RNA-seq data, (h) Heatmap of 75 differentially expressed genes (log2 normalized counts) as identified by RNA-seq analysis of 4 matched ECT- O and EUT-0 organoid lines. Colors range from blue (low expression) to red (high expression). LGR6 is among the top upregulated genes in the ECT-O.

Figure 11 Organoids from endometrial pre-cancer lesions display disease-associated phenotype, (a) Representative brightfield pictures of organoid development from hyperplastic endometrium (HYP-O) after seeding (P0). Overview (left; scale bar, 200 pm) and magnified organoid pictures (right; scale bar, 50 pm) are shown, (b) H&E analysis, immunohistochemical examination of ERalpha, PR and P53, and mucin detection (PAS) in primary biopsies and corresponding HYP-0 of different types of endometrial hyperplasia as indicated. H&E staining reveals glandular-like morphology with a well-defined lumen in the organoids of simple benign and complex atypical hyperplasia and a poorly-defined lumen in hyperplastic polyp. P53 expression, being present in simple benign hyperplasia and endometrial polyp but absent in complex atypical hyperplasia, is reproduced in the matching organoids. Mucus production is only detected in the lumen of the endometrial polyp and derived organoids (*). Scale bar, 50 pm. (c) TEM analysis reveals some stratified epithelium (*). Microvilli are present while cilia are not observed (magnified box), (d) aCGH plot indicates the absence of SCNA in both primary hyperplastic tissue and corresponding HYP-O. Figure 12 EC-derived organoids capture disease and genetic heterogeneity.

(a) Representative brightfield pictures of EC-0 development after seeding (P0). Scale bar, 50 pm. (b) Representative brightfield overview pictures of grade I and III EC-0 after seeding (P0). Scale bar, 200 pm. (c) aCGH plots of primary biopsy and corresponding organoids at P3 from a high grade serous EC sample, and brightfield picture of the organoid culture.

(dl and d2) Medium optimization for EC-0 growth as compared to HYP-O. Representative brightfield pictures (top) and bar graph (bottom) displaying the number of organoids formed after 14 days in the absence ("-") of the indicated compounds individually omitted from SOM (pictured by line for EC-0 and dashed line for HYP-O) (mean + s.e.m.; n= 3). *P< 0.05 among HYP-0 as compared to SOM. (e) Bar graph depicting EC-0 passageability in SOM and optimized culture medium (left) (r?=4). Passaging time of HYP-0 and grade I and II/III EC-0 (box plot; 3 independent donors), monitored for 6 passages (right), (f) Representative brightfield and H&E images of long-term expansion of high grade (III) EC-O. Magnified picture of representative organoid in inset. Scale bar, 200 pm for brightfield and 50 pm for H&E. (g) Bar graph showing the clonogenic potential for the different organoid types as indicated, (h) Representative brightfield pictures of the different EC-0 morphologies; grape bunch-like structures are typical for organoid cultures of poorly differentiated, clear cell tumours, (i) TEM pictures of grade I and III EC-O. Grade I EC-0 exhibit normal nuclei (magnified box) while high grade EC-0 show nuclear abnormalities (magnified in box), including tri-nucleated cells (right).

(j) Immunohistochemical analysis of EC markers and mucin detection (PAS) in primary tumours and corresponding EC-0 showing that the organoids reproduce the heterogeneity of the disease. Mucinous differentiation (PAS⁺), MSI/MSS, squamous differentiation (P63⁺) and hormone receptor status are preserved. Scale bars, 50 pm.

(k) Flowchart of the genomic screen performed on EC-0 and primary tumours. MSI tumours, typically copy number-stable but mutation-driven, were subjected to WES while MSS and TP53⁺ phenotypes, typically stable at mutational level but copy number unstable, were subjected to shallow-seq. Some samples were directly subjected to aCGH (dashed line). (I) Example of aCGH (left) and shallow-seq (right) plot of EC-0 and primary tumour, (m) Matrix summarizing the SCNA detected in primary tumour and corresponding EC-0 (early and/or late passage). EC-0_12 (see Fig. 12c) and EC-0_17 were overtaken by organoids from healthy cells ("EC ® EM- O") .

(n) Overview of the number of genetic aberrations as identified by WES in primary tumours (except for EC-0_3) and corresponding EC-O. (o) Immunofluorescence staining of b-catenin in low and high grade EC-O. Scale bar, 50 pm. Representative brightfield pictures of EC-0 lines harbouring CTNNB1 mutations (EC-0_3), or not (EC-0_6 and EC-0_7), cultured in SOM with or without XAV939 (right). Scale bar, 200 pm. (p) PCA plot showing sample distribution based on RT-qPCR gene expression analysis (see Fig. 5f).

Figure 13 EC-derived organoids reproduce the primary tumour phenotype after in vivo transplantation, (a) H&E analysis and immunohistochemical examination of endometrial (cancer) markers in primary samples of low (I) and high (III) grade cancer and corresponding EC-0 grown for two months as subcutaneous xenografts, showing that organoids recapitulate the primary lesion upon in vivo transplantation, with expression of patient-specific features. Scale bar, 50 pm. (b) Representative macroscopic pictures of the uterine horns after orthotopic transplantation of vehicle, Hec-IA cells, low and high grade EC-0 or HYP-O. Hec-IA cells and high grade (III, poorly differentiated) EC-0 form a huge tumour mass localized near the cervix (* indicates the non-affected part of the uterine horn) while no lesion is observed in the vehicle-injected mice. A solid metastasis is detected in the peritoneum after orthotopic transpla ntation of the high grade EC-O, while growth of the low grade (I) EC-0 remains localized. No visible growth is observed in the HYP-0 xenograft, (c) Ki67 immunostaining of the uterine grafts and metastasis. Diffuse proliferation is observed in the Hec-lA-derived lesion and the high grade EC-0 xenografts while low grade EC-0 xenografts show lower proliferative activity. Scale bar, 50 pm.

Figure 14 Ion channel expression and functionality in organoids from endometrial diseases, and organoid biobanking , (a) Gene expression of a selected panel of ion channels in EUT-0 and ECT-0 subdivided into rASRM stage I-II and III-IV, as normalized to the geometric mean of the housekeeping genes HPRT1 and PGK1 and expressed as fold change relative to EM-O. Bars show mean + s.e. m. of n= 2-5 orga noid lines per group. **P<0.01, ***P<0.001 compared to EM-O. (b) Percentage of responding cells in EM-0 and (non-subdivided) EUT-0 and ECT-0 to specific ion channel activators. n= 4-5 organoid lines per group, (c) Gene expression of a selected panel of ion channels in HYP-0 and EC-O, the latter subdivided into non-invasive (N- IN) and invasive (IN) phenotypes, as normalized to HPRT1 and PGK1 and expressed as fold change relative to EM-O. Bars show mean ± s.e.m. of n= 3-6 organoid lines per group. *P<0.05, **P< 0.01, ***P< 0.001 compared to EM-O.

(d) Endometrial (disease) organoid biobanking . A representative organoid line ID card is shown .

(e) IC50 of indicated drugs on EC-O, as determined from the dose-response curves presented in Fig. 6c.

Endometrial disorders are primary ground of infertility, yet little is known on their pathogenesis. Disclosed herein are organoids spanning a wide ra nge of endometrial diseases. These organoids capture disease heterogeneity, maintain key features of the primary tissue including genetic background, and reproduce the lesion after in vivo transplantation. Importantly, the organoids show strong expandability thereby overcoming the hurdle of limiting quantities of primary biopsies.

The present invention disclosed the establishment of organoids from endometriotic lesions of all clinical stages which uncovered altered signa lling pathways such as integrin, PI3K-AKT and WNT when compared to healthy endometrium-derived organoids. The present invention identifies EC-associated mutations in the organoids from high-stage endometriosis. Together with recent reports using clinical samples [Anglesio et al. (2017) N. Engl. J. Med. 376, 1835-1848; Enomoto et al. (2018) Cell Rep. 24, 1777-1789], the present invention indicates the involvement of cancer driver genes in the disease. The endometriosis organoid biobank will be valuable to decipher disease (and type-specific) pathogenesis, especially if in future studies epithelial and stromal compartments are (re-)combined [Seino et al. (2018) Stem Cell 22, 454-467; Stzepourginski et at. (2017) Proc. Natl. Acad. Sci. U.S.A. 114, E506-E513; Ohlund et at. (2017) J. Exp. Med. 214, 579-596], and to search for new drug targets as better alternative to current hormonal suppression therapy.

Following meticulous medium optimization, long-term expandable organoids could be developed from EC, capturing clinical heterogeneity. In contrast to the present invention, the few endometrial tumour-derived organoid-like aggregates reported before were not strongly characterized (essential given the risk that non-cancer organoids overtake the culture), and not long-term expanded or in vivo grafted [Turco et at. (2017) Nat. Cell Biol. 19, 568-577; Pauli et at. (2017) Cancer Discov. 7, 462-477; Girda. et al. (2017) Int. J. Gynecol. Cancer 27, 1701-1707]. The present invention discloses that SCNA and genetic landscape of the tumours was largely conserved in the organoids although occasional deviations were retrieved which may be due to the genetic background of the patient, the presence of non tumour cells in the DNA-extracted primary tissue and/or the culture conditions which may favour, or hamper, growth of specific mutant subclones (as also reported for other cancer types [Broutier et al. (2017) Nat. Med. 23, 1424-1435; Gao et al. (2014) Cell 159, 176-187), but which may also reflect the natural evolution of the cancer as occurring in vivo, as supported by the considerable number of new mutations detected in late-passage organoids from MSI Lynch syndrome tumour. The EC-0 show patient-specific drug responses, thereby providing conceptual evidence that the organoids are amenable to (personalized) drug screenings. The herein disclodes EC-derived organoid models provide unique tools to search for pathogenetic mechanisms and new drug targets.

The present invention allows to develop organoids from hyperplastic endometrium (including Lynch syndrome) with faithful reproduction of the disease genotype. Organoids will be valuable to search for molecular mechanisms underlying the hyperplastic phenotype and its progression toward cancer [Drost et al. (2017) Science 358, 234-238].

Taken together, the present invention discloses the start of an extended biobank (Fig. 14d) across healthy and pathological endometrium which will provide new powerful research models and drug screening and discovery tools.

Patient biopsies (Table 1) were dissociated, and the obtained fragments and cells seeded in optimized culture conditions to generate 3D organoids (passage 0, P0; Fig. la). Organoids were amplified for multiple passages and subjected to downstream analyses including validation against the original primary tissue. The organoid lines developed were cryopreserved and biobanked (Fig. la and Table 1).

Table 1 Endometrial organoid biobank

a Cultures were overtaken by healthy cells

Establishment of long-term expandable organoids from endometriotic lesions

In endometriosis patients, endometrium-like tissue grows at ectopic sites such as peritoneum and ovary. Endometriosis samples (Table 1) were cultured as previously defined for establishing organoids from healthy endometrium [Boretto et at. (2017) Development 144, 1775-1786] (with minor modifications).

Table 2 Standard organoid medium (SOM) for healthy, eutopic, ectopic and hyperplastic endometrium

a Only for organoid expansion

b Only for organoid formation or after dissociation for passaging

Organoids (referred to as 'ectopic organoids' or ECT-O; Fig. la) developed within 7- 14 days, although at lower number than organoids from healthy endometrium (EM-

O) (Fig. lb,c and Fig. 8a) and at lower efficiency (Fig. Id) which may be due to the harsher conditions needed to dissociate the endometriotic tissue, resulting in poorer yield and viability. ECT-0 could also originate from single cells (Fig. 8b). The ECT-0 displayed substantial proliferation (Fig. le) and could be long-term expanded comparable to EM-O, at present reaching more than 8 months (Fig. If and Fig. 8c). ECT-0 could be established from all ectopic implantation sites and clinical rASRM (revised staging system, American Society for Reproductive Medicine, Am. Soc. Reprod. Med. (1997) Fertil. Steril. 67, 817-821)stages (i.e. I to IV, defined according to location, number, infiltration and adhesions of the lesions; Fig. 8d and Table 1). Long-term expandable organoids were also generated from the eutopic endometrium of endometriosis patients, generating matched 'eutopic organoids' (EUT-O) and ECT-0 of individual patients (Fig. la-d,f, Fig. 8a, c and Table 1). Importantly, the ECT-0 displayed genomic stability during long-term expansion (Fig. lg) which was also confirmed for EM-0 and EUT-0 (Fig. 8e). Also transcriptomic stability was observed as analysed for several endometrial markers (Fig. 8f). Organoid lines were cryopreserved and could be grown again after thawing.

The present invention tests the requirement of multiple medium compounds for growth of ECT-0 (as compared to EM-O). The WNT signal amplifier RSPOl was needed for efficient organoid growth (Fig. lh). WNT pathway inhibition with IWP2 (in the absence of RSPOl) significantly reduced growth (Fig. lh and Fig. 8g), thus indicating endogenous WNT signalling. Organoid growth could be rescued by RSPOl, WNT3A and WNT7A but not by the non-canonical WNT ligand WNT5A (Fig. 8g). WNT5A and WNT7A showed to be the highest expressed WNT ligands in the ECT-0 cultures (Fig. 8h) (as also found before in EM-O). Expression analysis of WNT target genes further supported the involvement of the canonical pathway in the observed effects (Fig. 8g). Omission of EGF significantly impaired both ECT-0 and EM-0 expansion, while removal of Noggin (NOG) compromised the growth of ECT-0 (Fig. lh).

Endometriosis-derived organoids recapitulate disease phenotype in vitro and in vivo

In general, the ECT-0 contained a lumen-bordering cell layer that was thicker than in EM-0 and EUT-0 (Fig. 9a). Several ECT-0 displayed luminal invasion as also found in the primary lesion (Fig. 2a and Fig. 9a), which is in line with the expression of MMP (Fig. 9b). Endometriotic tissue exhibits ERa and PR expression which was retained in the corresponding organoids (Fig. 2a) while PanCK and E-cadherin confirmed their epithelial nature (Fig. 2a and Fig. 9c). The primary lesions secrete mucus in their lumen which was also recapitulated in ECT-0 (PAS staining; Fig. 2a), indicating the presence of secretory cells and the polarized nature of the organoids, further supported by the presence of laminin at the basolateral side (Fig. 2a) and of microvilli and cilia, directed toward the lumen (Fig. 2b and Fig. 9d). Transmission electron microscopy (TEM) also confirmed the stratified epithelium in ECT-0 which was not observed in EM-0 and EUT-0 (Fig. 2b and Fig. 9d). In addition, TEM showed the presence of non-ciliated secretory cells with secretory vacuoles (Fig. 2b, dashed box). The present invention investigates whether ECT-0 could in vivo generate grafts reproducing endometriotic features. After transplantation in the mouse renal capsule, ERa⁺ and PR⁺ lesions developed (Fig. 9e). Given the prevailing peritoneal location of endometriotic lesions, ECT-0 were injected into the peritoneal cavity. The organoids generated implants in the peritoneum which expressed endometriotic markers (Fig. 2c).

Transcriptomic and genetic analyses of endometriotic organoids reveal disease-associated traits

RNA-seq analysis of ECT-O, EUT-0 and EM-0 revealed type-specific differentially expressed genes (Fig. 3a, Fig. 10). Principal component analysis (PCA) showed a general clustering of EM-0 and EUT-O, and more variation with and within the ECT- O samples (Fig. 10a) which may, at least partly, be due to their different lesion type and rASRM stage (Table 1). Gene Ontology (GO) analysis identified altered pathways and biological terms in the ECT-0 (as compared to the EM-O; Fig. lOef). In particular, genes of ECM-receptor interaction (e.g. COL3A1, FN1, ITGA11, LAMA2 ), PI3K-AKT signalling pathway (e.g. HGF, IGF1, ITGB8, PDGFD, PRLR ), WNT pathway (e.g. CTNNA2, LEF1, LGR6, WNT11 ), hormonal response (e.g. HSD11B1, IGF1, LIFR, PGR , PRLR), Hippo signalling (e.g. CTGF, GDF7, SNAI2) and adhesion and invasion (e.g. ITGB8, MMP2, SNAI2, TIMP4) were altered in expression (Fig. 3a and Fig. 10). These adhesion/invasion factors are important for anchorage and growth of tissue at ectopic sites and have previously been associated with endometriosis [Pitsos & Kanakas (2009) Reproductive Sciences 16, 717-726; Eyster et al. (2002) Fertil. Sterii. 77, 38-42; Burney et al. (2007) Endocrinology 148, 3814-3826; Matsuzaki et al. (2004) Mol. Hum. Reprod. 10, 719-728; Weigel et al. (2012) Eur. J. Obstet. Gynecol. Reprod. Biol. 160, 74-78; Wu et a/. (2006) Endocrinology 147, 232-246. Moreover, HOXD8 and HOXA9, both playing a key role in development of the female reproductive tract [Burney et al. (2007) Endocrinology 148, 3814-3826; Wu et al. (2006) Endocrinology 147, 232-246], were upregulated in ECT-0 (Fig. 3a and Fig. 10b). Contrasting EUT-0 with EM-0 has the potential to identify pathway differences in endometrium of endometriosis patients and healthy women. For instance, downregulation of BAMBI and upregulation of GDF11 in EUT-0 (Fig. 3a and Fig. 10b) may suggest increased TGFP pathway activity in patient endometrium.

Because the rASRM stages I to IV were not equally represented in the RNA-seq analysis (Table 1), hence precluding stage-specific profiling, RT-qPCR analysis (as compared to EM-O) was performed on multiple organoid lines from individual endometriosis stages for a selection of genes (including RNA-seq hits). A prominent finding was the upregulation of LEF1, WNT11 and LGR6 (the latter particularly in higher stage III-IV, in which deep lesions are predominant) (Fig. 3b, c and Fig. 10c). In addition, significant downregulation of the progesterone-regulated gene PAEP was identified, and increased expression of LI FR and of the candidate endometrium stem cell marker SOX9, all in line with previous findings in endometriosis [Valentijn et al. (2013) Hum. Reprod. 28, 2695-2708] (Fig. 3d). MMP2 is significantly upregulated in stages II and IV (clinical mild to strong invasion of ovarian/deep lesions) while inflammatory cytokines IL-Ib and IL-8 are predominantly upregulated in stage I-II ECT-0 (Fig. lOd), in concord with endometriosis establishing as inflammatory disease [Gongalves et al. (2017) Cytokine 89, 229-234]. Comparing matched ECT-0 and EUT-0 from individual patients identified LGR6 as one of the top upregulated genes in the ectopic tissue organoids (Fig. 10g,h).

Whole exome sequencing (WES) was performed on a selected number of ECT-0 lines with either low or high rASRM staging (I and IV, respectively). A KRAS hotspot mutation (G12V) was identified in the stage IV ECT-0_13 line (Fig. 3e), which was confirmed in the corresponding RNA-seq data (by SNP variant calling). Additional aberrations were found in driver genes associated with EC in stage IV but not stage I ECT-O. Particularly, potential somatic mutations were identified in CTCF, EP300 and ZNF471 (Fig. 3e). This is in line with recent findings of mutations in EC-associated genes in endometriotic (ovarian and deep) lesions [Anglesio et al. (2017) N. Engl. J. Med. 376, 1835-1848; Enomoto et al. (2018) Cell Rep. 24, 1777-1789].

Organoids from endometrial pre-cancer pathologies recapitulate disease phenotype and genetics

Endometrial hyperplasia often precedes cancer development [Murali et al. (2014) Lancet. Oncol. 15, e268-78; Jeong et al. (2009) Oncogene 28, 31-40]. Using the culture conditions defined in Table 2, organoids from endometrial hyperplasia samples (HYP-O) were successfully established (70% efficiency; Fig. Id) starting from 3 different subtypes, i.e. simple benign, complex atypical and polyp (Fig. 4a, Fig. 11a and Table 1). The organoids showed considerable proliferative activity (Fig. 4b) and could be serially passaged for more than 6 months (Fig. 4a). The HYP-0 displayed a stratified PanCK⁺ epithelium lining a central lumen (Fig. 4c and Fig. l ib). Molecular and subcellular features of primary tissue (i.e. presence/absence of ERa, PR, P53, mucus, microvilli) were faithfully recapitulated in the HYP-0 (Fig. llb,c). No somatic copy number alterations (SCNA) were found in the primary biopsies, neither in the corresponding organoids (Fig. lid), suggesting genetic stability.

Lynch syndrome is an inherited predisposition to cancer, primarily developing in colon and endometrium. Patients typically harbour mutations in DNA repair genes [Lynch et al. (2015) Nat. Rev. Cancer 15, 181-194]. Long-term expandable organoids were derived from endometrium hyperplasia biopsies of Lynch syndrome patients (Fig. 4d). Targeted re-sequencing showed that the specific mutations affecting the DNA repair genes (i.e. MSH2 and MSH6 ) were retained in the corresponding organoids (Fig. 4e).

Endometrial cancer-derived organoids capture disease heterogeneity

Organoids were generated from low and high grade EC (Table 1). EC organoids (EC- O) developed in 7-20 days (Fig. 12a, b). However, efficiency was considerably lower (20%; Fig. 5a) than from the other endometrial conditions (Fig. Id). The organoids formed showed limited expansion potential. Moreover, in some of the lines, organoids from non-cancer endometrial cells as present in the original tumour biopsy overtook the culture (as previously also reported for other cancer types [Broutier et al. (2017) Nat. Med. 23, 1424-1435; Gao et al. (2014) Cell 159, 176-187], For instance, organoids from a high grade serous tumour (EC-0_12) containing multiple genomic aberrations, showed a normal genome after organoid establishment (Fig. 12c). Therefore, to achieve more efficient EC-0 development, the influence of multiple medium components was assessed. EC-0 formed better in the absence of p38i (Fig. 12dl and d2), whereas organoid growth from healthy endometrium (EM-O; Fig. lg) and hyperplasia (HYP-O; Fig. 12dl and d2) was negatively affected. Of note, reducing p38i concentration also improved the establishment of breast cancer organoids [Sachs et al. (2018) Cell 172, 373-386]. By lowering p38i concentration, and at the same time adding IGF1 [Aizen et al (2015) Mol. Cell. Endocrinol. 406, 27-39], HGF [Yoshida et al. (2002) J. Clin. Endocrinol. Metab. 87, 2376-2383] and lipids (all found beneficial for long-term expansion; Table 3), EC-0 could be established at higher efficiency (40%; Fig. 5a) and robustly propagated in culture for more than a year now (Fig. 5b and Fig. 12e,f). The EC-0 showed prominent proliferative activity and clonogenic capacity (Fig. 5c and Fig. 12g). Morphological heterogeneity was observed between the different organoid lines (as well as within individual organoid cultures; Fig. 12h). EC-0 from low grade/stage cancer generally displayed a glandular-like morphology with a well-to-moderately defined lumen (Fig. 5b, d, Fig. 12b, h), whereas EC-0 from high grade/stage cancer commonly appeared dense without visible lumen (Fig. 12b, f,h). Also grade-associated degrees of nuclear abnormalities were conserved in the organoids (Fig. 12i).

Table 3: Optimized culture medium for endometrial cancer organoid formation and expansion

a Not essential

b Only for organoid formation or dissociation at passaging The EC-0 recapitulated the primary tumour's (immuno-)histology with expression of EC-associated markers (Fig. 5d and Fig. 12j). Expression of ERa and PR distinguishes type I from type II EC⁹ which was conserved in the organoids (Fig. 5d and Fig. 12j). PTEN mutations depend on EC type [Risinger et a/. (1998) Clin. Cancer Res. 4, 3005- 3010]; PTEN expression patterns were retained in the organoids (Fig. 12j). Microsatellite instability (MSI) is a recurrent feature of type I EC with frequent mutations in MLH1 and MSH6 [Getz et a/. (2013) Nature 497, 67-73]. MSI phenotype was also preserved in the EC-0 (Fig. 12j). In summary, organoids could be generated from a wide variety of EC ranging from well to poorly differentiated types encompassing tumours with squamous (P63⁺) and mucinous (PAS⁺) differentiation (Fig. 12j and Table 4).

Table 4 characterisation of organoids.

EC-derived organoids recapitulate the mutational landscape of the primary tumour and reveal disease-specific gene expression

The genetic make-up of EC-0 and their matched primary tumours were compared. First, samples were subjected to either array CGH (aCGH) or low-coverage whole genome sequencing (i.e. shallow-seq, as performed on the microsatellite-stable (MSS) tumours; see flowchart in Fig. 12k). These analyses revealed that the large majority of SCNA in primary tumours were retained in the corresponding EC-O, although occasional gains or losses were observed (Fig. 121, m). MSI tumour types typically have stable copy number profiles but are characterized by a high mutation burden. Organoids derived from MSI tumours were subjected to WES (Fig. 12k). The majority of the genetic alterations in the primary tumour were retained in the organoids even after extensive in vitro expansion (Fig. 12n). Interestingly, a considerable number of new substitutions (i.e. 1564) were retrieved in EC-0_6 after long-term expansion (Fig. 12n), which was not unexpected given the Lynch syndrome background of this tumour with defective DNA repair machinery. In line with this, EC-0_6 showed considerable SCNA alterations after long-term expansion (Fig. 12m).

EC hotspot mutations were identified in primary MSI tumours and these hits were retained in the organoids even after long-term expansion (Fig. 5e). Particularly, hotspot mutations were found in FBXW7 (R465H) and ARID1A (R693X and R1335X) in EC-0_6, and in PTEN (R130G) in EC-0_16. Furthermore, genetic hits were identied in 21 of the most frequently mutated genes in EC, concordantly in tumour and corresponding organoids. In particular, frameshift (loss-of-function) mutations were identified in the tumour suppressor genes PTEN, CTCF and ARID1A, and somatic mutations in PIK3CA (R88Q and R108H), TP53 (G245S), ARID1A, POLE and FAT1 [McConechy et al. (2012) J. Pathol. 228, 20-30]. In addition, CTNNB1 mutations were found in the 2 low-grade EC-0 analysed (EC-0_2 and EC-0_3; Fig. 5e). CTNNB1 mutations in low-grade EC lead to constitutive activity of the WNT pathway [Getz et al. (2013) Nature 497, 67-73; Yeramian et al. (2013) Oncogene 32, 403-413; Salvesen et al. (2009) Proc. Natl. Acad. Sci. U.S.A.106, 4834-4839]. In line with this, low grade EC-0 showed nuclear b-catenin localization (Fig. 12o) and higher expression of WNT target genes ( AXIN2 , LEF1, CMYC, TCF4, LRP4, RNF43, LGR6 ) as compared to EM-0 and high-grade EC-0 (Fig. 5f). Accordingly, growth of CTNNB1- mutant organoid lines was not affected by RSPOl withdrawal (Fig. 5e) and less influenced by XAV939 (which inhibits WNT signalling by increasing b-catenin degradation) than CTNNB1 -wildtype organoid cultures (Fig. 12o).

Gains in chromosome 17 are regularly observed in EC and often involve amplification of ERBB2 [Getz et al. (2013) Nature 497, 67-73; Salvesen et al. (2009) Proc. Natl. Acad. Sci. U. S. A.106, 4834-4839]. Gains were also found in chromosome 17 in the present dataset (Fig. 12m) and observed increased ERBB2 expression in EC-0 relative to EM-0 and HYP-0 (Fig. 5f). Higher expression levels of IGF1, IGFR1, CEACAM1 and MMP2 were detected in EC-0 (Fig. 5f), in line with previous findings in EC [Salvesen et al. (2009) Proc. Natl. Acad. Sci. U. S. A.106, 4834-4839]. ESR1 and FOXA2 were consistently downregulated in high-grade EC-O, as reported previously in EC [Getz et at. (2013) Nature 497, 67-73] (Fig. 5f). The PI3K-AKT pathway is frequently hyperactivated in EC [Getz et a/. (2013) Nature 497, 67-73; Yeramian et at. (2013) Oncogene 32, 403-413]. In analogy, genetic alterations were found in the pathway's signalling mediators (PTEN, PIK3CA, AKT1 ) in several EC-0 lines (Fig. 5e) as well as STAT3, VEGF and HIFla overexpression (Fig. 5f). On the other hand, endometrial markers (like PAEP and LIF) were downregulated in EC-0 as compared to EM-0 (Fig. 5f). Higher expression of EMT-associated genes ( CXCR4 , TWIST1, ZEB1, N-CADH ) was detected in EC-0 versus EM-O, while HYP-0 appeared to represent an intermediate stage (Fig. 5f), as supported by clustering analysis (Fig. 5g and Fig. 12p).

EC-derived organoids recapitulate disease phenotype in vivo

EC-O, subcutaneously injected into NOD/SCID mice, generated a cell mass recapitulating histological and molecular features of the primary tumour (Fig. 13a). After orthotopic engraftment in uterine horn, high grade (III) EC-0 created a large, invasive, highly proliferative mass (comparable to the EC Hec-IA cell line) as well as peritoneal metastasis, whereas low grade EC-O-derived lesions remained localized with lower proliferation and HYP-0 generated no visible outgrowth (Fig. 6a and Fig. 13b, c). The orthotopic grafts also reproduced the hormone receptor phenotype of the original cancer tissue (Fig. 6b).

EC-derived organoids show patient-specific drug responses

To assess whether EC-0 can be harnessed for drug screening as demonstrated for other cancer organoids [Jeong et al. (2009) Oncogene 28, 31-40; Lynch et al. (2015) Nat. Rev. Cancer 15, 181-194; Broutier et al. (2017) Nat. Med. 23, 1424-1435; Van De Wetering (2015) Cell 161, 933-945], 4 standardly used chemotherapeutic compounds (paclitaxel, 5FU, carboplatin, doxorubicin) and the mTOR inhibitor everolimus were tested in EC-0 cultures of 5 independent donors. The organoids showed patient-specific drug responses (Fig. 6c and Fig. 14e). EC-0_2 was most sensitive to everolimus suggesting strong dependence on the PI3K-AKT pathway which is in line with mutations found in the pathway's signalling mediators (PTEN, PIK3CA, AKT1 ) (Fig. 5e).

Endometrial organoids show differential disease-associated expression of functional ion channels

Hennes et al. (2019) Sc/. Rep. 9, 1779 disclosed that EM-0 conserved the functional expression of specific ion channels as present in primary endometrial epithelial cells. To search for distinctions with diseased endometrium, functional ion channel expression were analysed in the organoids obtained here, as compared to EM-O. Regarding endometriosis, no significant differences were observed in gene expression and functionality of the mechanosensitive PIEZOl channel and transient receptor potential (TRP) channels (Fig. 14a, b), in line with a recent report using clinical biopsies [Persoons et a/. (2018) Int. J. Mol. Sci. 19, 2467]. In contrast, significant differences were found in the responses of (pre-)cancerous organoids to stimulation with GSK, OAG and THC, agonists of TRPV4, TRPC6 and TRPV2, respectively (Fig. 7a and Fig. 14c). HYP-0 showed higher numbers of responding cells and increased calcium influxes ( versus EM-O) in answer to GSK and THC stimulation (Fig. 7a,b,d). Furthermore, distinct responses to GSK and THC stimulation were found in EC-0 from invasive (> stage IB) versus non-invasive EC, suggesting a distinct role of TRPV2 and TRPV4 in different EC stages. Blocking TRPV4 (together with TRPM4, TRPM7 and TRPC6) using a cocktail of available specific inhibitors strongly reduced EC-0 formation efficiency in line with a significant decrease in proliferative activity (Fig. 7e).

EXAMPLES

EXAMPLE 1 METHODS

Organoid culturing from endometrial biopsies. Endometrial biopsies were obtained from patients with different endometrial conditions (see Table 1) after informed written consent, which was approved by the Ethical Committee Research UZ/KU Leuven (S59006; S59177). Biopsies were minced into small pieces and extensively rinsed in Ca²⁺/Mg²⁺-free PBS (Thermo Fisher Scientific). Tissue samples were dissociated using collagenase IV (1-2 mg/ml; Thermo Fisher Scientific) in the presence of Rock inhibitor (RI) (Y-27632, 10 mM; Merck Millipore) and mechanical trituration for 1-3 h. Subsequently, tissue was incubated for 15 min in TrypLE (lx; Thermo Fisher Scientific) supplemented with RI. Digestion was stopped by medium dilution (without serum) and after centrifugation, the pellet was resuspended in 70% Matrigel/30% DMEM/F12 (Thermo Fisher Scientific) supplemented with RI, and droplets deposited in 48-well plates. Organoids were cultured as previously described with minor medium modifications (Table 2) [Boretto et ai (2017) Development 144, 1775-1786]. For passaging (performed every 10-20 days), organoids were recovered by liquifying the Matrigel drop with ice-cold DMEM/F12 (without any enzymes, growth factors or serum) and mechanical pipetting to ensure maximum collection of organoids. Subsequently, the organoids were dissociated using TrypLE (in DMEM/F12 containing RI), mechanically triturated and the mixture centrifuged at 230 g. The obtained fragments and cells were resuspended in 70% Matrigel/30% DMEM/F12 supplemented with RI and 20 pi droplets were deposited in pre-warmed 48-well plates. Established organoids were amplified, cryopreserved for biobanking and subjected to downstream analyses (Fig. la). Unless otherwise stated, organoids of low passage number (P3-P6) were used for the experiments described.

To assess clonogenic capacity, organoids were dissociated into single cells with TrypLE supplemented with RI, filtered through a 40 pm cell strainer and resuspended in 70% Matrigel/30% DMEM/F12 supplemented with RI at 10-100 cells per well. The organoids formed were counted after 20-30 days.

Testing medium compounds and rescue of IWP2. Organoids were cultured for at least 3 passages in standard organoid medium (SOM), then dissociated with TrypLE supplemented with RI and seeded in 70% Matrigel/30% DMEM/F12 supplemented with RI at 1000 cells per well in 96-well plates. Cells were cultured in SOM from which individual compounds were omitted (3 technical replicates per compound). Number of sizeable organoids (>300 pm) formed was determined after 7-20 days and averaged for 3 independent patient donors.

Dissociated organoid cells (from 5 independent donors) were seeded as described above and cultured for 10 consecutive days in SOM (without RSPOl) and 1 mM IWP2 (Sigma Aldrich), to which RSPOl (Table 2), 200 ng/ml WNT3A (R8iD systems), 500 ng/ml WNT5A (R8iD systems) or 1 pg/ml WNT7A (R8iD systems) were added. Medium was refreshed every 2 days and number of organoids counted after 10 days. To assess expression of downstream target genes, mature organoids were cultured for 72 h in the specified conditions and extracted RNA subjected to RT-qPCR as described below. In some experiments, XAV939 (10 nM; Tocris Bioscience) was used as WNT pathway inhibitor.

In vivo transplantation. Organoids, expanded for 4-6 passages, were removed from Matrigel, trypsinized and resuspended in 50 mI of PBS for transplantation under the kidney capsule or in 200 mI for intraperitoneal injection. For subcutaneous transplantation, organoids were resuspended in 200 mI of 50% Matrigel/50% PBS. These transplantation experiments were performed in ovariectomized NOD-SCID mice in which an estradiol pellet was subcutaneously implanted at time of ovariectomy (as described in Boretto et a/. (2017) Development 144, 1775-1786). For uterine injection, organoids were resuspended in 20 mI of 100% Matrigel and injected in the exposed uterine horn of anesthetized NOD-SCID mice. The EC cell line Hec-IA, used as positive control, was obtained from ATCC (LGC Standards, Teddington, UK) and cultured in DMEM/F12 with 10% fetal bovine serum (FBS). Mouse experiments were approved by the KU Leuven Ethical Committee for experimental animals. After the indicated period of time, mice were euthanized, grafts (and potential metastasis) localized and tissues processed for (immuno- )histological analysis.

Immunohistochemical analysis. Examination was performed as previously described [Boretto et at. (2017) Development 144, 1775-1786]. In brief, paraffin sections were subjected to antigen retrieval and incubated overnight with primary antibody and subsequently with horseradish peroxidase (HRP-)conjugated secondary antibody (ImmPRESS HRP universal antibody peroxidase-conjugated horse anti mouse IgG/anti-rabbit IgG polymer detection kit; Vector Laboratories). Sections were counterstained with hematoxylin. Periodic Acid Schiff (PAS) staining was performed to visualize mucin according to manufacturer instructions (Sigma Aldrich) [Boretto et at. (2017) Development 144, 1775-1786].

Transmission electron microscopy. TEM analysis was performed as described in detail [Boretto et al. (2017) Development 144, 1775-1786]. In short, organoids were removed from Matrigel and sequentially fixed in glutaraldehyde and osmium tetroxide/potassium ferrocyanide. Organoids were incubated with tannic acid for 20 min and uranyl acetate overnight, followed by aspartate for 30 min. Samples were dehydrated and embedded in epoxy resin (Agar 100, Agar Scientific), and 70 nm sections were analysed using a JEM 1400 transmission electron microscope (JEOL) equipped with an Olympus SIS Quemesa l lMpxl camera. TEM was performed on >2 biological replicates.

Two-photon excitation microscopy. Organoids were fixed with 4% paraformaldehyde for 30 min at 37 °C, washed with PBS and incubated with Hoechst33342 (2.5 pg/ml in PBS; Sigma) for 1 h at 37 °C. After rinsing, Vybrant Dil cell-labeling solution (0.1 mM in PBS; Thermo Fisher Scientific) was applied and organoids incubated for 2 h. Z-stacks (up to 350 pm) were acquired with a confocal microscope (Leica TCS SP8 X) equipped with a Mai Tai DeepSee multiphoton laser (Spectra physics) and a 40x long-working distance objective (HC PL IRAPO 40x/1.10 W CORR). Emission light was detected with Hybrid Refractive light Detectors (HyD RLD, Leica).

Array CGH. Organoids were recovered from Matrigel and genomic DNA from organoids and primary tissues isolated using the Purelink Genomic Mini Kit (Invitrogen) according to manufacturer instructions. aCGH analysis was performed using the 8x 60K CytoSure ISCA v3 microarray (Oxford Gene Technology). Genomic DNA was labeled for 2 h with the CytoSure Labelling Kit. Samples were labeled with Cy5 and hybridized to Cy3-labeled sex-matched reference DNA. Hybridization was performed for at least 16 h at 65 °C in a rotator oven (SciGene). Arrays were rinsed using Agilent wash solutions with a Little Dipper Microarray Processor (SciGene) and scanned using an Agilent microarray scanner at 2 pm resolution, followed by calculation of signal intensities using Feature Extraction software (Agilent Technologies). Result visualization and data analysis were performed using CytoSure Interpret Software and circular binary segmentation algorithm. Quality control metrics were monitored with CytoSure Interpret software. aCGH was performed on >2 biological replicates per endometrial pathology.

Targeted sequencing. Genomic DNA was amplified using targeted primers (sequences are available on request) and sequenced bidirectionally using ABI BigDye Terminator Sequencing 3.1 on the ABI3730XL instrument (Thermo Fisher Scientific). Nucleotide numbering reflects the cDNA transcript with + 1 corresponding to the A of the ATG translation initiation codon in the reference sequence for MSH2 (LRG_218) and MSH6 (LRG_219).

Library preparation and genomic screening. The genomic screen was performed according to the flowchart depicted in Fig. 12k. Whole genome DNA libraries were created using the Illumina TruSeq DNA sample preparation kit V2 according to manufacturer instructions. The exome was captured with the Nimblegen SeqCap EZ Developer Library kit and TP53 and MSI libraries created by amplicon-based enrichment. All resulting libraries were sequenced on HiSeq4000 (Illumina) using a V3 flowcell generating 2 x 151 bp reads.

Raw sequencing reads of whole-exome libraries were mapped to the human reference genome (GRCh37/hgl9) using Burrows-Wheeler Aligner [Li & Durbin (2017) Bioinformatics 25, 1754-1760]. Samples were sequenced with average sequencing depth of 23x and 85% of the exome was covered over lOx. PCR duplicates (on average 15%) were removed with Picard (vl.43). Genome Analysis Toolkit 2 (GATK) [Van der Auwera et a/, in Current protocols in bioinformatics 11.10.1-11.10.33 (John Wiley & Sons, Inc., 2013)] was applied to perform base recalibration and local realignment around indels. Substitutions were called by GATK, whereas small indels were detected using Dindel (vl.01) [Albers et ai. (2011) Genome Res. 21, 961-973]. Only substitutions with a quality score >Q30 and indels with a quality score >50 and a minimum of 10 reads, and tumour allelic frequency of 10% for both, were considered. To limit the detection of SNPs, mutations were further filtered by removing the common mutations present in the ESP6500v2, the 1000 genome project 2015, the complete genomics 46 and the ExAC (Exome Aggregations Consortium) databases, as well as the Genome of the Netherlands and a database consisting of 100 germ-line samples earlier sequenced by the present applicant. The remaining mutations were annotated using ANNOVAR [Wang et al. (2010) Nucleic Acids Res. 38, el64-el64]. Raw sequencing reads of the TP53 libraries were analysed in a similar manner using a target interval limited to the TP53 region to reduce computational effort, while the MSI status was assessed as previously described [Zhao et al. (2014). Elife 3, e03032].

Raw sequencing reads from whole-genome libraries were mapped to GRCh37/hgl9, duplicates removed, and reads further analysed by QDNAseq [Scheinin et a/. (2014) Genome Res. 24, 2022-2032] resulting in read counts per bin of 50kb. Ascat algorithm was used for segmentation of these data.

RNA-sequencing. RNA was extracted from organoids with the RNeasy Mini Kit (Qiagen) according to manufacturer instructions. RNA quality was analysed using Agilent picochips on an Agilent BioAnalyzer 2100. Only samples with RIN>7.5 were subjected to RNA-seq. RNA was amplified with the Smart-Seq V4 kit (Takara Bio Europe) and libraries were prepared using the NEBNext Ultra DNA Library prep kit, followed by sequencing on NextSeq500 (Illumina). The index trimmed single-end 75bp reads were aligned to the human reference genome (hg38) using Hisat2 (v2.1.0) to generate bam files. Gene-level read count matrix was summarized from bam files using featureCounts (vl.5.3). The count matrix was imported into the R Bioconductor DESeq2 package (vl.18.1) for differential gene expression analysis with false discovery rate (FDR) <0.1. The differentially expressed genes obtained were then used to generate heatmaps by applying R package pheatmap (vl.0.8; https://cran.r-proiect.org/web/packaqes/pheatmap/) with scaling and clustering of only the rows. PCA was performed using plotPCA function from DESeq2 package with unsupervised transformed counts. Gene ontology (GO) analysis was done using Gene Ontology Consortium software (http://qeneontoloqv.org/) and DAVID (https://david.ncifcrf.gov/) while pathway analysis was performed using the KEGG PATHWAY Database (https://www.qenome.ip/keqq/pathwav.html).

SNP variant calling of RNA-seq data. Raw reads in fastq format were filtered for adapters with ea-utils fastq-mcf (vl.1.2). Adapter-free reads were aligned to the human reference genome (hgl9; Ensemble version 75) using the splice-aware package Tophat 2 [Kim et al. (2013) Genome Biol. 14, R36]. Resulting BAM files were merged per lane with Samtools (vl.5). RNA-seq variant calling was performed following the GATK's best practices on GATK 3.6 haplotype caller [Van der Auwera et al. in Current protocols in bioinformatics 11.10.1-11.10.33 (John Wiley & Sons, Inc., 2013)]. Resulting VCF files were merged with vcf-tools and indexed with tabix. Variant annotation was performed with ANNOVAR (against hgl9) [Wang et al. (2010) Nucleic Acids Res. 38, el64-el64].

Gene expression analysis by RT-qPCR. RNA was reverse-transcribed and subjected to SYBR Green- or TaqMan assay-based quantitative real-time PCR (qPCR) as previously described [Boretto et al. (2017) Development 144, 1775-1786; Hennes et al. (2019) Sc/. Rep. 9, 1779]. GAPDH, HPRT1 and/or PGK1 were used as housekeeping genes (HKG). Relative gene expression levels were calculated as ACt values (Ct 'gene' minus Ct 'HKG') and presented as Log (2^~^^{ct aene~ ct HKG}ixl000), or as fold change compared to EM-0 (reference) using the formula 2^~^^{ct sampLe~ &Gt rererence}\ Bar graphs were generated with GraphPad Prism (Version 7.03) while heatmaps were produced with ClustVis (web tool for visualizing clustering of multivariate data; BETA).

Drug screening. Tumour organoids were recovered from Matrigel and dissociated. 2000 cells were seeded in 96-well plates and allowed to form organoids for 10 days. Then, paclitaxel (Paclitaxel AB), 5FU (Fluracedyl), carboplatin (Carbosin), doxorubicin (Sigma Aldrich) or everolimus (Tocris) were added at various concentrations and viability was measured after 72 h with the XTT assay following manufacturer instructions (Invitrogen). Vehicle (DMSO) was used as negative control. Dose- response curves and IC50 values were calculated using Graphpad Prism 7.

Ion channel functionality. Analysis of ion channel functionality was performed by calcium microfluorimetry as previously described in detail [Hennes et a/. (2019) Sc/. Rep. 9, 1779]. In short, dissociated organoid cells were seeded on collagen-coated coverslips in a 12-well plate, incubated with 2 mM Fura-2 acetoxymethyl ester, and intracellular Ca²⁺ measured after exposure to specific ion channel activators, i.e. 50 pM A⁹-tetrahydrocannabinol (THC; kindly provided by Prof. Appendino, Universita del Piemonte Orientale, Italy) for TRPV2, 10 nM GSK016790A (GSK; Sigma-Aldrich) for TRPV4 and 100 pM l-oleoyl-2-acetyl-glycerol (OAG; Calbiochem) for TRPC6. Absolute Ca²⁺ concentrations and amplitudes were calculated and responding cells identified. Only cells that responded to the positive control ionomycin (2 pM; Thermo Fisher Scientific) at the end of the experiment were taken into account. Experiments were performed on 2-6 biological replicates, recording a minimum of 195 cells in total. To block ion channel activity, a cocktail of available specific inhibitors was added, i.e. flufanemic acid (10 mM; Sigma Aldrich), NS8593 (30 pM; Tocris), HC067047 (100 pM; Sigma Aldrich) and larixyl acetate (2 pM; Sigma Aldrich), blocking TRPM4, TRPM7, TRPV4 and TRPC6, respectively. Impact on organoid (EC-O) formation and proliferation (Ki67 immunostaining) was performed 72 h later.

Statistical analyses. Normal distribution was verified before performing statistical analyses. In case of normal distribution, multiple comparisons were analysed using 1- or 2-way analysis of variance (ANOVA) followed by Dunnett's test for Multiple Comparison (95% confidence intervals). When normal distribution was not achieved, multiple comparisons were analysed using non-parametric Kruskal-Wallis test with Dunn's post-test (95% confidence intervals). Statistical significance was defined as P<0.05. All statistical analyses were calculated using GraphPad Prism (Version 7.03). All experiments were performed with at least 3 independent donors per group. Data availability. RNA-seq data have been deposited in Gene Expression Omnibus (GEO) with accession number GSE118928. Raw sequencing reads of shallow-seq and WES have been deposited in the ArrayExpress database at EMBL-EBI under accession number E-MTAB-7687 and E-MTAB-7688, respectively.

Claims

1. A cell culture medium comprising :

- a chemically defined lipid concentrate,

- IGF-1 (insulin-like growth factor 1)

- HGF (hepatocyte growth factor),

- RSPOl (R-spondin 1),

- nicotinamide,

- an activin receptor-like kinase (ALK) inhibitor, such as A83-01 ,

- a p38 MAP kinase inhibitor, such as inSB202190, and

- 17 beta-estradiol.

2. The medium according to claim 1, further comprising- IL-6 (interleukin-6),

3. The medium according to claim 1 or 2, wherein the medium does not comprise insulin, transferrin, selenium, FGF2 (Fibroblast Growth Factor 2), and FGF10 (Fibroblast Growth Factor 10).

4. The medium according to any one of claim 1 to 3 , wherein in said medium:

- the concentration of RSPOl is between 90 and 110 ng/ml,

- the concentration of nicotinamide is between 4.5 and 5.5 mM,

- the concentration of A83-01 is between 0.2 and 0.3 mM,

- the concentration of SB202190 is between 0.05 and 0.25 pM,

- the concentration of 17beta-estradiol is between 7.5 and 12 nM.

5. The medium according to any one of claim 1 to 4, wherein said medium comprises a conditioned medium of a cell line expressing RSPOl, at a concentration between 4.5 and 5.5 v/v % (volume RSPOl comprising conditioned medium/ total volume medium for generation, growth or maintenance).

6. The medium according to any one of claims 1 to 5, wherein in said medium:

- the chemically defined lipid concentrate is at a concentration of between 0.5 and 2.5% (v/v), between 0.75 and 1.25 % (v/v) or at 1 %, - IGF-1 is in a concentration of between 35 and 45 ng/ml,

- HGF is in a concentration of between 15 and 25 ng/ml, and,

- 11-6 is at a concentration of between 4.5 and 5.5 ng /ml.

7. The medium according to any one of claims 1 to 6, wherein the medium further comprises transforming growth factor (TGF) -alpha in a concentration of 7.5 to 12.5 ng/ml medium.

8. The medium according to any one of claims 1 to 7, wherein the medium comprises a basal medium with L-Glutamine, Noggin, L-alanyl-L-glutamine dipeptide (Glutamax), and a ROCK inhibitor such as Y27632, and supplements such as B27 and N2.

9. The medium according to any one of claims 1 to 8, further comprises EGF (epidermal growth factor).

10. Use of a medium for the generation, growth or maintenance of organoids from endometrium tumour tissue., wherein the medium comprises:

- a chemically defined lipid concentrate,

- IGF-1 (insulin-like growth factor 1)

- HGF (hepatocyte growth factor),

- RSPOl (R-spondin 1),

- nicotinamide,

- an activin receptor-like kinase (ALK) inhibitor, such as A83-01 ,

- a p38 MAP kinase inhibitor, such as inSB202190, and

- 17 beta-estradiol.

11. The use according to claim 10 , wherein the medium further comprises IL-6 (Interleukin-6).

12. The use according to claim 10 or 11, wherein the medium does not comprise insulin, transferrin, selenium, FGF2 (Fibroblast Growth Factor 2), and FGF10 (Fibroblast Growth Factor 10).

13. The use according to any one of claims 10 to 12 , wherein in said medium:

- the concentration of RSPOl is between 90 and 110 ng/ml,

- the concentration of nicotinamide is between 4.5 and 5.5 mM ,

- the concentration of A83-01 is between 0.2 and 0.3 mM,

- the concentration of SB202190 is between 0.05 and 0.25 pM,

- the concentration of 17beta-estradiol is between 9 and 11 nM.

14. The use according to any one of claims 10 to 13, wherein said medium comprises a conditioned medium of a cell line expressing RSPOl, at a concentration between 4.5 and 5.5 v/v % (volume RSPOl comprising conditioned medium/ total volume medium for generation, growth or maintenance).

15. The use according to any one of claims 10 to 14wherein in said medium:

- the chemically defined lipid concentrate is at a concentration of between 0.75 and 1.25 % (v/v)

- IGF-1 is in a concentration of between 35 and 45 ng/ml ,

- HGF is in a concentration of betweenl5 and 25 ng/ml, 17, and,

- 11-6 is at a concentration of between 4.5 and 5.5 ng /ml.

16. The use according to any one of claims 10 to 15, wherein the medium further comprises transforming growth factor (TGF) -alpha in a concentration of 7.5 to 125 ng/ml medium.

17. The use according to any one of claims 10 to 16, wherein the medium comprises a basal medium with L-Glutamine, comprises Noggin, L-alanyl-L- glutamine dipeptide (Glutamax), and a ROCK inhibitor such as Y27632, and supplements such as B27 and N2.

18. The use according to any one of claims 10 to 17, wherein the medium further comprises EGF (Epidermal growth factor).