Organoids from pituitary as a novel research model toward pituitary stem cell exploration

Animals

GHCre^+/− mice (expressing Cre under control of the rat GH promoter; Luque et al. 2007) were crossed with homozygous iDTR^+/+ mice (harboring the diphtheria toxin receptor (DTR) gene in the Rosa26 locus preceded by a floxed STOP cassette; Buch et al. 2005). The double-transgenic GHCre^+/−;iDTR^+/− offspring (hemizygous for both transgenes, further referred to as GHCre/iDTR) expresses DTR in GH-producing cells, thereby rendering these cells susceptible to DT-induced apoptosis (Fu et al. 2012, Willems et al. 2016). These double-transgenic mice were also crossed with Lgr5lacZ reporter mice (as obtained from Dr Clevers, Hubrecht Institute, Utrecht, The Netherlands; Barker et al. 2007). Sox2eGFP mice, expressing enhanced GFP in SOX2⁺ cells (Ellis et al. 2004), were used to sort SOX2^eGFPpos and SOX2^eGFPneg cells. To obtain Sox2CreERT2^+/−;eYFP^fl/− double-transgenic mice, Sox2CreERT2^+/− mice (obtained from Dr Martinez-Barbera, UCL, London; Andoniadou et al. 2013) were crossed with eYFP^fl/fl reporter mice (harboring the enhanced yellow fluorescent protein/eYFP gene in the Rosa26 locus preceded by a floxed STOP cassette; obtained from Dr. Carmeliet, KU Leuven). NOD-SCID mice were provided by in-house breeding. Genotyping of all mouse models was performed as previously described (Barker et al. 2007, Fu et al. 2012, Roose et al. 2017). Mice were bred on a C57BL/6 background and kept at the university’s animal housing facility (KU Leuven, Belgium), providing constant temperature, humidity, day/night cycle and water and food ad libitum. Animal experiments were approved by the KU Leuven Ethical Committee.

Treatment of mice

Adult (8–12 weeks old) male GHCre/iDTR mice and Cre-negative control littermates (referred to as –/iDTR) were injected intra-peritoneally (i.p.) with DT (4 ng/g body weight, twice a day; List Biological Laboratories, Campbell, CA, USA) for three consecutive days as described before (Fu et al. 2012, Willems et al. 2016). Mice were sacrificed at day 11 following start of DT treatment (day of first injection = day 1).

Sox2CreERT2^+/−;eYFP^fl/− mice were injected i.p. with tamoxifen (Sigma-Aldrich; dissolved in corn oil; 0.15 mg/g body weight) for two consecutive days and were killed 2 days after the last injection.

Organoid culture

The pituitary was isolated from killed mice, and the AP was separated and dispersed into single cells using trypsin as previously described (Chen et al. 2005). AP cells (10,000 unsorted cells or 8000 FACS-sorted cells, see below) were embedded in a 30 µL drop composed of serum-free defined medium (SFDM; Thermo Fisher Scientific) and Growth Factor Reduced Matrigel (Corning) in a 30:70 ratio and seeded in 48-well plates (Corning) (Supplementary Fig. 1A, see section on supplementary data given at the end of this article). The jellified droplets were cultured in SFDM containing a cocktail of growth and regulatory factors. For all cultures, ‘pituitary organoid medium’ (PitOM, Table 1) was used unless otherwise stated. At seeding of the primary cells or replating of the organoid-derived fragments or cells (i.e. passaging), ROCK inhibitor (Y27632, 10 µM; Merck, Overijse, Belgium) was added to the medium. Cultures were kept at 37°C in a 1.9% CO₂ incubator and medium was refreshed every 2–3 days. For quantification, all organoid structures with a diameter ≥50 µm were counted. In a few experiments, AP cells were first cultured in free-floating sphere-forming conditions as previously described (Chen et al. 2005, 2009, Roose et al. 2017). After 6 days, the formed pituispheres were embedded in Matrigel for organoid culture as described above.

Expanding organoid cultures were passaged every 7–10 days (Supplementary Fig. 1A). Organoids were collected and incubated with TrypLE Express (Thermo Fisher Scientific) for 5 min (cystic organoids) or 10 min (dense organoids) at 37°C. After TrypLE inactivation, the suspension was centrifuged at 200 g , the pellet was resuspended in SFDM and the organoids were dispersed by vigorous pipetting yielding mostly single cells (dense organoids) or cells and organoid fragments (cystic organoids). After centrifugation, the pellet was resuspended in 70% Matrigel, replated and cultured as described earlier. AP cells from Sox2eGFP^+/− adult mice were sorted by FACS (Aria III, BD Biosciences, Erembodegem, Belgium) into SOX2^eGFPpos and SOX2^eGFPneg, cells which were used for organoid culturing or RNA isolation (see below). Brightfield and fluorescence pictures of organoid cultures were recorded using the Axiovert 40 CFL microscope (Zeiss). To assess WNT pathway involvement, IWP2 (2 µM; Sigma-Aldrich) and XAV939 (10 µM; Sigma-Aldrich) were added to the medium. Heparin-binding epidermal growth factor-like growth factor (HB-EGF; Peprotech, Rocky Hill, NJ, USA) and TGFB1 (R&D systems) were added at a concentration of 10 ng/mL and 5 ng/mL, respectively. GHCre/iDTR AP-derived (cystic) organoids were treated with 10 ng/mL DT or vehicle (PBS) in PitOM for 24 h. This in vitro DT treatment has before been shown to induce GH⁺ cell ablation in GHCre/iDTR AP cell cultures and aggregates (Luque et al. 2011 and data not shown). Unless otherwise stated, organoids of passage (P) 2–4 were used for the analyses described below. Of note, pituitary organoids could also be obtained from mouse strains other than C57BL/6 such as FVB and CD1 (data not shown).

Organoid differentiation

To test the differentiation capacity of pituitary organoids in vitro, organoids were passaged and first grown in organoid culture medium for 1 week. Then, organoids were replated (without dissociation) in Matrigel drops and cultured in SFDM (without growth factors) or SFDM + retinoic acid (RA, 1 µM; Sigma-Aldrich) for 4 days followed by SFDM + RA + dexamethasone (Dex, 50 nM; Sigma-Aldrich) for 2 days, a protocol previously reported to induce GH⁺ cell differentiation (Cheng et al. 2011). In an additional approach, formed organoids were cultured for 7 days in SFDM on Matrigel-coated coverslips as previously described for pituispheres (Chen et al. 2005, 2009, Roose et al. 2017).

To test the differentiation capacity in vivo, pituitary eYFP⁺ organoids (more specifically, dense organoids expanded from tamoxifen-injected Sox2CreERT2^+/−;eYFP^fl/− mice) or PBS (sham) were injected under the kidney capsule of 8- to 10-week-old male NOD-SCID mice. After 2, 4 or 6 weeks, mice were killed and the grafted kidney was dissected. The kidneys were photographed using the Discovery V12 stereo microscope (Zeiss) and were further processed for immunofluorescent analysis (see below) of SOX2 or the pituitary hormones GH, prolactin (PRL), adrenocorticotropic hormone (ACTH) and glycoprotein hormone α-subunit (AGSU; i.e. the common subunit of thyroid-stimulating hormone (TSH), luteinizing hormone (LH) and follicle-stimulating hormone (FSH)).

Histochemical analysis and immunostaining

Tissues and organoids were fixed in 4% PFA at room temperature for 3 h and 1 h, respectively, and embedded in paraffin using the Excelsior ES Tissue Processor (Thermo Fisher Scientific). Sections were subjected to hematoxylin and eosin (H&E) staining, RNA in situ hybridization or immunofluorescence/immunohistochemical staining. For analysis of the in vitro differentiation experiments, organoids were either analyzed as a whole or dispersed and dissociated cells spun onto cytospin slides (Roose et al. 2017). Immunostainings were performed after antigen retrieval with 10 mM citrate buffer (pH 6) and permeabilization with Triton X-100 (0.1% in PBS). Following incubation with primary and secondary antibodies (Supplementary Table 1), sections were covered with Vectashield (containing 4′,6-diamidino-2-phenylindole or DAPI; Vector Laboratories, Peterborough, UK) or DPX mountant (Sigma-Aldrich) after counterstaining with hematoxylin. RNA in situ hybridization was performed using the RNAscope 2.0 Fast Red Detection Kit (Advanced Cell Diagnostics, Newark, CA, USA), according to the manufacturer’s protocol. Immunofluorescence staining of SOX9 was done in 7 µm cryosections of organoids obtained using a CM1860 Leica cryostat (Leica Biosystems) after 20-min fixation with 4% PFA and embedding in optimal cutting temperature compound (OCT; Sakura, Torrance, CA, USA). Processed sections were analyzed with a Leica DM5500 epifluorescence microscope (Leica Microsystems) accessible through InfraMouse (KU Leuven-VIB) or an Imager.Z2 microscope and Zen software (Zeiss). Organoids cultured on coverslips were analyzed using a Zeiss LSM 880 confocal microscope. Recorded images were converted into pictures for figures with ImageJ (http://imagej.nih.gov/ij/).

Two-photon excitation microscopy

Organoids were fixed with 4% PFA for 30 min at 37°C and incubated with Hoechst33342 (2.5 µg/mL in PBS; Sigma-Aldrich) for 1 h at 37°C. Then, organoids and cell membranes were permeabilized with 0.1% Triton X-100 (in 1% bovine serum albumin/BSA in PBS). After blocking with 1% BSA/PBS, organoids were incubated with Phalloidin-Atto520 (Sigma-Aldrich) for 40 min at room temperature. Z-stacks (up to 350 µm) of the organoids were acquired with a confocal microscope (Leica TCS SP8 X; Leica Microsystems) equipped with a Mai Tai DeepSee multiphoton laser (Spectra Physics, Vienna, Austria) and a 20×/0.75 water immersion objective (HC PL APO CS2; Leica Microsystems). Hoechst33342 was excited at 700 nm and Phalloidin-Atto520 at 1040 nm by two-photon excitation, and emission detected using hybrid detectors and refractive light detectors (Leica Microsystems) set at the appropriate detection window.

X-gal staining

Organoids and pituitaries were fixed for 15 min at room temperature in a mixture of 5 mM ethylene glycol tetraacetic acid (EGTA), 2 mM MgCl₂ and 0.2% glutaraldehyde in PBS. Samples were rinsed twice for 5 min with PBS containing 2 mM MgCl₂, 0.01% sodium deoxycholate and 0.02% NP-40. Then, samples were covered and incubated overnight (pituitaries) or for 2 h (organoids), at 37°C with X-gal staining solution (1 mg/mL X-gal, 2 mM MgCl₂, 5 mM K₃Fe(CN)₆, 5 mM K₄Fe(CN)₆, 0.01% sodium deoxycholate and 0.02% NP-40 in PBS; all reagents from Sigma-Aldrich). After incubation, pituitaries were rinsed three times for 10 min, post-fixed overnight at 4°C and processed for paraffin embedding. Sections were counterstained with Eosin Y (Sigma-Aldrich).

Transmission electron microscopy

Organoids were fixed overnight in 2.5% glutaraldehyde (Grade I, Sigma-Aldrich) in 0.1 M sodium cacodylate buffer (pH 7.2) at 4°C. The fixed organoids were pelleted at 200 g and washed three times in 0.1 M cacodylate buffer. The pellets were resuspended in 2.0% low melting point agarose (Thermo Fisher Scientific) and centrifuged at 1000 g . After solidification, the pellets were cut up in small cubes, washed and post-fixed in 1% osmium tetroxide and 1.5% potassium ferrocyanide in 0.1 M cacodylate buffer, washed again and mordanted by 0.1% tannic acid for 20 min. Then, samples were stained overnight en bloc with uranyl acetate (Agar Scientific, Stansted, UK) followed by 30-min lead aspartate incubation with intermittent washing in cacodylate buffer, dehydrated in a graded ethanol series and propylene oxide and finally embedded in epoxy resin (Agar 100; Agar Scientific). The embedded samples were polymerized for 2 days at 60°C. Thin 70 nm sections were cut with a UCT ultramicrotome (Leica) and analyzed using the JEM1400 transmission electron microscope (JEOL, Zaventem, Belgium) equipped with an Olympus SIS Quemesa 11Mpxl camera.

Gene expression analysis by RT-qPCR

Total RNA was extracted from organoids or dissociated AP cells with the RNeasy Micro kit (Qiagen) and subjected to reverse transcription (RT) using the Superscript III First-Strand Synthesis Supermix (Thermo Fisher Scientific), all according to the manufacturers’ recommendations. Resultant cDNA samples were subjected to SYBR green-based quantitative real-time PCR (qPCR) using the StepOnePlus Real-Time PCR System (AB Applied Biosystems). All qPCR reactions were performed with Platinum SYBR Green qPCR SuperMix-UDG mastermix (Thermo Fisher Scientific). Forward and reverse primers (IDT Technologies, Haasrode, Belgium; Supplementary Table 2) were designed using PrimerBank (Spandidos et al. 2010) and PrimerBLAST (https://www.ncbi.nlm.nih.gov/tools/primer-blast/). Hypoxanthine phosphoribosyltransferase (Hprt) and β-actin (Actb) were tested and used as housekeeping genes. Relative gene expression levels were calculated as ∆Ct values (Ct ‘target’ minus Ct ‘average housekeeping genes’) and expressed as fold change (2^{−(∆∆Ct)}) in comparison to a reference as specified in the figures. For some of the experiments, amplicons were visualized using gel electrophoresis in a 2% agarose gel with SYBR Safe (Thermo Fischer Scientific).

RNA-sequencing analysis

RNA extracted from organoids or FACS-sorted (SOX2^eGFPpos) AP cells was checked for RNA Integrity Number (RIN) using Agilent Picochips on an Agilent BioAnalyzer 2100 (Agilent Technologies). Using good-quality samples (RIN >7.5), RNA-seq was performed in collaboration with the Nucleomics Core (VIB/KU Leuven). First, RNA samples were amplified using Smart-Seq v4 kit (Takara Bio Europe, St-Germain-en-Laye, France). RNA library preparation was executed with the NEBNext Ultra DNA Library prep kit, followed by sequencing on a NextSeq 500 instrument (both from Illumina). Raw reads were subjected to quality control. Sequence reads were pre-processed to remove adaptors and filter low-quality reads. The high-quality reads were aligned to the mouse reference genome using Tophat (v2.0.13). Statistic modeling was performed using the DESeq 1.32.0 package of Bioconductor (http://www.bioconductor.org). Sequencing data are available from the NCBI Gene Expression Omnibus (GEO, http://www.ncbi.nlm.nih.gov/geo/) through accession number GSE116654. Gene ontology (GO) analysis of differentially expressed genes (defined by a ratio ≥5 or ≤0.2 and normalized counts ≥10) was done using the Database for Annotation, Visualization and Integrated Discovery (DAVID, version 6.8; https://david.ncifcrf.gov/). Hierarchical clustering of selected gene lists was performed using Multiple experiment Viewer software, version 4.9.0 (http://mev.tm4.org/).

Statistical analysis

Statistical analysis was performed (for n ≥ 3) using GraphPad Prism (version 5.01) and Microsoft Excel software (2016). Unpaired two-tailed t-Student test was applied for comparison of two groups. In case of multiple comparisons (more than two groups), one-way analysis of variance was done followed by Dunnett test. Statistical significance was defined as P < 0.05.

Development of organoids from mouse pituitary

To achieve the development of organoids from pituitary, we embedded dissociated cells from adult mouse AP in a 3D Matrigel scaffold and added culture medium containing the generic organoid culture components EGF, Noggin, WNT3A and RSPO1 (Supplementary Fig. 1A). In addition, we included components from pathways known to play a role in pituitary development and/or stem cell maintenance and proliferation, such as fibroblast growth factors (FGF), insulin-like growth factor-1 (IGF1) and sonic hedgehog (SHH) (Boretto et al. 2017, Cox et al. 2017). By testing these components in a trial-and-error manner, we eventually defined a culture medium (further referred to as ‘pituitary organoid medium’, PitOM; Table 1) in which organoid structures were found to develop, typically in a time span of 2 weeks (passage 0, P0; Fig. 1A). Time-lapse recording supported that organoids developed from individual cells (Fig. 1A). Remarkably, two discernibly different organoid morphologies were obtained, i.e. dense (solid) organoids and cystic (hollow) organoids (i.e. with a lumen) (Fig. 1A). The independent cultures (n = 14 biological replicates) consisted either of only dense organoids (15% of the cultures), cystic organoids (30%) or a mixture of both (55% of the cultures with a higher number of cystic vs dense organoids per culture) (Supplementary Fig. 1B and C). After 2 weeks, the organoids were dissociated into single cells and re-seeded (Supplementary Fig. 1A). Organoids formed again, but only the dense organoids could be efficiently passaged (Fig. 1B and Supplementary Fig. 1B). The cystic organoid cultures were not expandable and the cystic organoids in the mixed cultures quickly disappeared; consequently, only dense organoid cultures were retained (Supplementary Fig. 1B). At present, the dense organoids could be expanded up to P41 (i.e. 11 months of culture) (Fig. 1B).

View Full Size

Development of SOX2⁺ cell-derived organoids from pituitary. (A) Formation of organoids starting from dissociated AP cells (passage 0, P0). Representative light microscopic pictures are shown at different time points (seeding at day 1, d1), with the arrow pointing to the same structure at the consecutive days of culture, starting from a single cell. Inset: magnification of boxed area. Lower right panel: example of a culture at d13 containing both dense (arrow) and cystic (arrowhead) organoids. Scale bar: 200 µm. (B) Passaging and expansion of pituitary organoids, shown at P3 (meaning about 1 month in culture; pictures at 1 day/d1 and 10 days/d10 after seeding), P10 (3 months) and P41 (11 months). Organoids reform at passaging in 7–10 days. Scale bar: 200 µm. (C) Percentage of SOX2⁺ cells in (dense) organoids (bar represents mean ± s.e.m.; n = 3) and double immunofluorescence staining of SOX2 (red) and E-cadherin (green) in paraffin-embedded organoid sections (composite overview and single-color magnified views). Arrowheads indicate SOX2⁻/E-cadherin⁻ cells. Nuclei are stained with DAPI (blue). Scale bar: 50 µm. (D) AP SOX2^eGFPpos and SOX2^eGFPneg cells from transgenic Sox2eGFP mice cultured in organoid-forming conditions (P0). Representative pictures are shown, including brightfield (left) and fluorescence image for eGFP expression (right). Scale bar: 200 µm. (E) Brightfield (left) and eGFP fluorescence images of SOX2^eGFPpos organoids at P2 on d2 and d6 after replating. Scale bar: 200 µm.

Citation: Journal of Endocrinology 240, 2; 10.1530/JOE-18-0462

• Download Figure
• Download figure as PowerPoint slide

Biological characterization of the pituitary organoids

Organoids typically originate from the stem cells of a tissue. The large majority of the pituitary organoid cells (≥85%) were found to express the pituitary stem cell marker SOX2 (Fig. 1C). These SOX2⁺ cells co-expressed E-cadherin, which in the pituitary is also present in the SOX2⁺ cells (Fauquier et al. 2008, Garcia-Lavandeira et al. 2009, Roose et al. 2017). The SOX2-negative cells were mainly located in the inner region of the organoids and did not co-express E-cadherin (Fig. 1C; arrowheads).

To investigate whether the organoids are derived from the SOX2-expressing pituitary cells, we isolated SOX2^eGFPpos cells by FACS from the AP of SOX2 reporter (Sox2eGFP) mice. Organoids developed in these cultures (starting from individual cells), while no organoids formed in the SOX2^eGFPneg cell cultures (Fig. 1D and Supplementary Fig. 1D). Moreover, a clear enrichment in organoid-forming cells was obtained in the SOX2^eGFPpos cell fraction (Supplementary Fig. 1E). Of note, only a small number of the seeded SOX2^eGFPpos cells gave rise to organoids (i.e. 0.14%; Supplementary Fig. 1E), which is in agreement with previous studies in which SOX2⁺ stem cell activity was measured by pituisphere and colony formation capacity (Fauquier et al. 2008, Chen et al. 2009, Gaston-Massuet et al. 2011) (see ‘Discussion’ section). The organoids retained the SOX2^eGFPpos phenotype during expansion (as assessed until P9 or 3 months in culture; Fig. 1E and data not shown). To investigate whether at passaging organoids also develop from individual cells and not by cell aggregation, we mixed organoid-derived fluorescent (eYFP⁺) cells (see below) with non-fluorescent (WT) cells. Only entirely fluorescent or non-fluorescent organoids were obtained (Supplementary Fig. 1F and G), thereby providing strong support for a clonal origin of the organoids.

To explore whether the pituitary organoids obtained could give rise to the gland’s differentiated (hormonal) cells, several in vitro approaches were tested that may induce pituitary hormonal cell differentiation, including the addition of FBS (Garcia-Lavandeira et al. 2009), culture on Matrigel-coated surface (as effective for pituispheres; Fauquier et al. 2008, Chen et al. 2009) and treatment with retinoic acid (RA) and dexamethasone (Cheng et al. 2011). However, hormone expression was not detected after applying these protocols (Supplementary Fig. 2A and data not shown). Omission of all (stem cell) growth factors from the organoid medium did also not result in hormonal differentiation (Supplementary Fig. 2A and data not shown). It has been found that for certain tissues like the pancreas, the stem cell organoids have to be exposed to an in vivo environment to achieve differentiation (Huch et al. 2013a ). Therefore, we transplanted expanded (dense) organoid cultures under the kidney capsule of NOD/SCID mice. To track the organoid cells, we developed eYFP⁺ organoids from a SOX2⁺ lineage tracing mouse model (i.e. Sox2CreERT2^+/−;eYFP^fl/−; see ‘Materials and methods’ section) displaying expression of eYFP in the SOX2⁺ cells which remains expressed in the descendant cells (Fig. 2A). The eYFP⁺ organoids (not expressing hormones; data not shown) were subrenally grafted and analyzed 2, 4 and 6 weeks later (Fig. 2A). An eYFP⁺ graft could be clearly detected at all time points (Fig. 2B). After 2 weeks, a few PRL-immunopositive cells were observed while no GH, ACTH or AGSU (the common subunit of TSH, LH and FSH) could be detected in the graft (Fig. 2C and Supplementary Fig. 2B). After 4 weeks, various cells were immunopositive for PRL, GH and AGSU, whereas after 6 weeks, no hormone immunoreactivity was detected anymore (Fig. 2C and Supplementary Fig. 2B). Thus, differentiation to hormone-expressing cells occurred in vivo although at limited degree. Many cells of the graft still expressed SOX2 at the time points analyzed (Fig. 2C).

View Full Size

Differentiation capacity of pituitary organoids after in vivo transplantation. (A) Schematic representation of in vivo transplantation analysis. Sox2CreERT2^+/−;eYFP^fl/− mice were injected with tamoxifen (TAM) as indicated, and eYFP⁺ (dense) organoids were formed from the dispersed AP cells and expanded. Organoids were grafted under the kidney capsule and analyzed 2, 4 and 6 weeks later. (B) Grafted kidneys show a fluorescent eYFP⁺ streak of cells (dotted line), while in the control kidney (sham) background signal was only present in the surrounding adipose tissue (arrowhead). Scale bar: 1 mm. (C) Immunofluorescence staining of graft paraffin sections reveals some PRL⁺ (red)/eYFP⁺ (green) cells 2 weeks after grafting, as well as many SOX2⁺ (red)/eYFP⁺ (green) cells (arrows). After 4 weeks, several eYFP⁺ (green) cells stain for GH or PRL (red) while multiple eYFP⁺ (green) cells still express SOX2 (red) (arrows). After 6 weeks, no hormone⁺ cells were observed (anymore), while SOX2⁺(red)/eYFP⁺ (green) cells remain present (arrows). Sham-grafted kidneys show no hormone- or SOX2-immunofluorescent signals. Nuclei are stained with DAPI (blue). Scale bar: 50 µm. eYFP, enhanced yellow fluorescent protein; GH, growth hormone; PRL, prolactin.

Citation: Journal of Endocrinology 240, 2; 10.1530/JOE-18-0462

• Download Figure
• Download figure as PowerPoint slide

Different organoid phenotypes from normal and damaged pituitary

Recently, we reported that the pituitary stem cells are activated upon injury in the gland as inflicted by DT-mediated GH⁺ cell ablation using the GHCre/iDTR mouse model (Fu et al. 2012, Willems et al. 2016). Hence, we here addressed the question whether organoid culturing from damaged and normal (undamaged) pituitary would show differences. AP cells from DT-treated GHCre/iDTR and control (–/iDTR) mice were seeded in the organoid-forming culture conditions as established above. A higher number of organoid structures were formed in the cell cultures from injured GHCre/iDTR pituitary than from undamaged –/iDTR gland (Fig. 3A and B). Intriguingly, the majority (80%) of the cultures (n = 13 biological replicates) of the damaged pituitary were exclusively composed of cystic organoids while the remaining part contained both cystic and dense organoids, with the cystic organoids present in large excess (Supplementary Fig. 3A and B). Time-lapse recording also supported a clonal origin of these cystic organoids (Supplementary Fig. 3C). The more abundant organoid formation from damaged (GHCre/iDTR) vs normal pituitary (Fig. 3A and B) was not due to the possible presence of DTR (alias HB-EGF) in the medium as could have been shed from ablated, disintegrating GH⁺ cells. Although exogenously added HB-EGF could compensate for EGF in whose absence no organoid growth occurred, endogenous HB-EGF levels – if present – were not sufficient (Supplementary Fig. 3D). Moreover, adding HB-EGF to PitOM did not affect the abundance of cystic organoids (neither of dense organoids) (Supplementary Fig. 3E). It should further be noted that the AP was collected for organoid culture initiation 1 week after stopping the DT injection which makes it unlikely that large amounts of HB-EGF were still present in the organoid-forming AP cell cultures since the GH⁺ cell ablation process occurs promptly upon DT treatment (from the first day) and is finalized 3–4 days later (Fu et al. 2012).

View Full Size

Organoid formation and expansion from GHCre/iDTR and –/iDTR pituitary. (A) Formation of organoids starting from dissociated AP cells of DT-injected –/iDTR (control) and GHCre/iDTR mice (P0). Representative light-microscopic pictures at seeding (d1) and at d15 are shown. Whereas predominantly dense organoids are formed in the –/iDTR AP cell cultures, GHCre/iDTR AP cell cultures give primarily rise to cystic organoids. Scale bar: 200 µm. (B) Total number of organoids (i.e. dense + cystic) formed in –/iDTR and GHCre/iDTR AP cell cultures. Bars represent mean ± s.e.m. (n = 3). **P < 0.01. (C) Passaging and expansion of organoids from –/iDTR (control) and GHCre/iDTR pituitary, shown at P1 (d1 and d8) and P5 (d1 and d8). Organoids regrow to their original type, being either dense or cystic. After a number of passages, cystic organoids start to expand less (as shown for P5), while dense organoids keep expanding efficiently. Representative light-microscopic pictures are shown. Scale bar: 200 µm. Inset: H&E staining of dense (–/iDTR) and cystic (GHCre/iDTR) organoids. Scale bar: 50 µm. (D) Ki67 immunofluorescence staining (green) of paraffin sections from –/iDTR dense organoids and GHCre/iDTR cystic organoids, showing higher proliferative activity in dense organoids as quantified (right). Bars represent mean ± s.e.m. (n = 3). **P < 0.01. Scale bar: 50 µm.

Citation: Journal of Endocrinology 240, 2; 10.1530/JOE-18-0462

• Download Figure
• Download figure as PowerPoint slide

The organoids from the injured GHCre/iDTR AP could be passaged as well as the organoids from the control (–/iDTR) pituitary (Fig. 3C), the latter in agreement with the findings above (Fig. 1B and Supplementary Fig. 1B). Whereas the –/iDTR-derived cultures eventually turned into complete dense organoid cultures (Fig. 3C and Supplementary Movie 1) as also found above, the GHCre/iDTR cystic organoid cultures remained cystic (Fig. 3C and Supplementary Movie 2), while a part of the mixed GHCre/iDTR cultures finally turned into dense organoids (Supplementary Fig. 3A). Whereas the dense organoids from the control pituitary were expandable for many months (Fig. 3C and above), cystic organoids stopped growing and were eventually lost after 4–8 passages (Fig. 3C and data not shown). Along the same line, dense organoids retained higher proliferative activity during passaging (as analyzed by Ki67 immunostaining) than cystic organoids (Fig. 3D), while the cell cycle inhibitors Cdkn1b/p27 and Cdkn1c/p57 were higher expressed in the cystic organoids (Supplementary Fig. 3F).

We next compared the molecular phenotype of the two organoid forms. Most cells of both dense and cystic organoids were immunopositive for the pituitary stem cell markers SOX2 and E-cadherin (Fig. 4A and above). Given the emergence of SOX2⁺/GH⁺ cells in the pituitary after damage (Fu et al. 2012, Willems et al. 2016), we wondered whether the cystic organoids from the DT-treated GHCre/iDTR mice were derived from these putative, activated progenitor cells. However, GH was not found in the cystic (neither the dense) organoids (Supplementary Fig. 3G). Moreover, the cystic organoids were not abolished by in vitro treatment with DT (Supplementary Fig. 3H), indicating that the cystic organoids did not express the GH promoter-induced DTR which would be expected if the originating SOX2⁺ cells expressed GH. The absence of DTR in the cystic organoids was further confirmed by RT-qPCR analysis (Supplementary Fig. 3I).

View Full Size

Cystic organoids more closely resemble the pituitary phenotype. (A) Immunofluorescence staining of SOX2 (red), E-cadherin (green) in paraffin sections and SOX9 (red) in cryosections of dense organoids as derived from undamaged –/iDTR pituitary and of cystic organoids as established from injured GHCre/iDTR pituitary. Arrows indicate SOX9⁺ cells. Nuclei are stained with DAPI or Hoechst (blue). Scale bar: 50 µm. (B) RT-qPCR expression analysis of early pituitary markers in dense (–/iDTR) and cystic (GHCre/iDTR) pituitary organoids (relative to expression in control AP set as 1 and indicated by dotted line). Bars represent mean ± s.e.m. (n = 3). *P < 0.05; **P < 0.01. (C) RT-qPCR expression analysis of Lhx3 in dense (–/iDTR) and cystic (GHCre/iDTR) pituitary organoids, cultured in PitOM (including nicotinamide) or PitOM without nicotinamide (relative to expression in control AP set as 1 and indicated by dotted line). Bars represent mean ± s.e.m. (n = 3). **P < 0.01. ND, not detectable. (D) Ultrastructural (TEM) images of dense (–/iDTR) and cystic (GHCre/iDTR) pituitary organoids showing the absence of secretory granules. Boxed areas are magnified in the lower panels. Scale bar: 2 µm. (E) RT-qPCR expression analysis of pituitary hormones in cystic (GHCre/iDTR) organoids, cultured in PitOM (including nicotinamide) or PitOM without nicotinamide (relative to expression in control AP). Bars represent mean ± s.e.m. (n = 3). *P < 0.05. Cga, glycoprotein hormone α-subunit; Pomc, proopiomelanocortin (precursor of ACTH). (F) Immunofluorescent staining of ACTH (red) and AGSU (green) in cystic (GHCre/iDTR) organoids, first grown in PitOM (including nicotinamide) or PitOM without nicotinamide, and then cultured on Matrigel-coated coverslips. Cells immunopositive for ACTH or AGSU are indicated with arrows. Scale bar: 50 µm.

Citation: Journal of Endocrinology 240, 2; 10.1530/JOE-18-0462

• Download Figure
• Download figure as PowerPoint slide

Gene expression of additional proposed pituitary stem cell markers (Sox9, Cxadr, Efnb2; Fauquier et al. 2008, Chen et al. 2009, Chen et al. 2013, Andoniadou et al. 2013, Rizzoti et al. 2013, Cox et al. 2017) was also detected in both organoid types (Supplementary Fig. 4A). Cxadr showed the most pronounced enrichment vs AP tissue, whereas Sox9 was not enriched. In line with the latter observation, only a limited number of cells in the organoids (of both types) were found to express SOX9 at the protein level, and SOX9⁺ cells were not detected in every organoid or organoid section (Fig. 4A and Supplementary Fig. 4B). Expression of transcription factors involved in the earliest phases of pituitary embryogenesis (i.e. Pitx1, Pitx2, Eya1, Pax6, Six1, Six6; Vankelecom 2010, 2012) was found in both dense and cystic organoids, however showing different profiles (Fig. 4B). Whereas the dense organoids (from –/iDTR pituitary) exhibited significantly higher expression of Pitx1 (also enriched as compared to the whole AP, Fig. 4B), cystic organoids (from GHCre/iDTR pituitary) showed significantly upregulated expression of Eya1, Six1 and Six6 (Fig. 4B), the latter representing a direct downstream target of SOX2 in the developing pituitary and required for proper proliferation and differentiation of embryonic pituitary progenitor cells (Goldsmith et al. 2016). Expression of Lhx3 and Prop1, transcription factors active in embryonic (but also adult) pituitary cells, was only detectable in cystic organoids although at low level when compared to adult AP (Fig. 4C and Supplementary Fig. 4C). Gene expression of pituitary hormones was detected in both organoid types, but again at very low levels as compared to AP tissue (Supplementary Fig. 4D). In line with the apparent more pronounced pituitary phenotype of cystic organoids (i.e. higher expression of most embryonic transcription factors), expression of hormone genes, in particular Cga and Pomc (encoding αGSU and ACTH, respectively), was found higher in cystic than dense organoids (Supplementary Fig. 4D). However, the very low mRNA levels are not translated into detectable protein given the absence of hormone immunoreactivity (in both organoid types) as reported above (Fig. 2 and Supplementary Figs 2, 3G, and data not shown). In accordance, transmission electron microscopy (TEM) did not find secretory granules in the cells of both organoid types, thereby further documenting their immature phenotype (Fig. 4D). Secretory granules are typically observed in the hormone-producing cells of the AP but are not present in the putative stem cells of the gland (Supplementary Fig. 4E). Intriguingly, removing nicotinamide from the organoid culture medium PitOM resulted in a significant increase in Lhx3 expression in the cystic organoids, even to levels comparable to the adult AP, while Prop1 levels were not significantly upregulated (Supplementary Fig. 4C). Their expression remained absent in the dense organoids after nicotinamide withdrawal (Fig. 4C and Supplementary Fig. 4C). Removal of nicotinamide from PitOM also induced higher expression (significantly or trend) of hormone genes in the cystic organoids, but not in the dense organoids (Fig. 4E and data not shown). To assess hormonal differentiation capacity of the cystic organoids, we cultured the organoids on Matrigel-coated coverslips (Fauquier et al. 2008, Chen et al. 2009), which failed to induce in vitro differentiation of dense organoids (see above). Since mRNA levels of Pomc and Cga were highest in cystic organoids (Fig. 4E and Supplementary Fig. 4D), we analyzed specifically these hormones at the protein level. ACTH⁺ cells were detected in the cystic organoids (0.1 ± 0.07% and 1.3 ± 0.4% of the cells after pre-growth in PitOM with nicotinamide and without nicotinamide, respectively; n = 3, P = 0.10) (Fig. 4F). AGSU⁺ cells were also observed in both conditions (0.5 ± 0.3% and 0.6 ± 0.4%, respectively; n = 3, P = 0.86). In agreement with the presence of ACTH⁺ and AGSU⁺ cells in the cystic organoids after applying the differentiation protocol, gene expression of the receptors for the cognate hypothalamic regulatory hormones (i.e. Crhr1 and Gnrhr) was also detected (Supplementary Fig. 4F). Finally, we transplanted cystic organoids under the kidney capsule of NOD/SCID mice. However, we could not detect a graft as analyzed 4 weeks later (data not shown), likely due to the insufficient number of cystic organoids transplanted given their more restricted expandability.

Taken together, the organoids as developed from damaged pituitary (which harbors activated stem cells) differ from the organoids established from basal, undamaged pituitary and more closely resemble the (immature) pituitary phenotype.

Dense and cystic organoids exhibit a distinct epithelial phenotype

To further decipher the molecular differences between the dense and cystic pituitary organoid types, RNA-sequencing (RNA-seq) analysis was performed. GO examination revealed keratinization as one of the top enriched biological processes in the dense vs cystic organoids (Fig. 5A). Keratin (Krt) gene expression phenotype clearly differed between the cystic and dense organoids (Fig. 5B). Interestingly, the Krt expression profile in the cystic organoids clustered with the expression in the SOX2⁺ stem cells (as isolated from both injured and non-injured Sox2eGFP pituitary), whereas the expression pattern in the dense organoids was clearly different (Fig. 5B). In accordance, also the expression of the embryogenesis-related factors in the cystic organoids correlated with the expression in the SOX2⁺ cells (Supplementary Fig. 4G). The latter RNA-seq data further validated the differential expression between dense and cystic organoids as found above by RT-qPCR (Fig. 4B). Regarding the Krt expression profile, a clear dichotomy between cystic and dense organoids was observed for Krt8/Krt18 and Krt5/Krt14 (Fig. 5B, boxed) as confirmed by RT-qPCR (Supplementary Fig. 5A). Krt8 (encoding for cytokeratin 8, CK8) and Krt18 (CK18), typical markers of simple non-squamous (columnar or cuboidal) epithelium, were higher expressed in the cystic vs dense organoids (Fig. 5B and Supplementary Fig. 5A). Krt5 (CK5) and Krt14 (CK14), typical markers of stratified squamous epithelium, were strongly upregulated in the dense vs cystic organoids. Also other markers of squamous epithelium (Trp63, Lor, Tgm1; Candi et al. 2006) were found upregulated in the dense vs cystic organoids (Supplementary Fig. 5B). We further analyzed the dichotomic keratins at the protein level and found in the cystic organoids intense immunopositivity for CK8/CK18 but not for CK5/CK14, whereas the dense organoids strongly expressed CK5 and CK14 and showed only low immunoreactivity for CK8/CK18 (Fig. 5C).

View Full Size

Dense and cystic pituitary organoids show a different epithelium identity. (A) Overview of a number of enriched biological processes (and their significance level, denoted as the −log10 of the corresponding P-value) in dense organoids versus cystic organoids, as identified by GO analysis. (B) Heatmap from RNA-seq data showing the differential expression of keratin (Krt) genes (from high expression in yellow to low expression in blue) in dense (–/iDTR) and cystic (GHCre/iDTR) organoids as compared to pituitary SOX2⁺ and SOX2⁻ cells (isolated as SOX2eGFPpos and SOX2eGFPneg cells by FACS from –/iDTR undamaged or GHCre/iDTR damaged pituitary). (C) Immunofluorescence staining of CK8/CK18 (green), CK5 (green) and CK14 (red) in paraffin sections of cystic and dense pituitary organoids. Nuclei are stained with DAPI (blue). Scale bar: 50 µm. (D) Ultrastructural (TEM) images of cystic and dense pituitary organoids. Upper panels: TEM analysis of cystic organoids showing tight cell junctions (arrows) and microvilli facing the lumen (arrowheads). Lower panels: TEM analysis of dense organoids showing desmosomal cell junctions (arrows) and signs of keratinization in the center (arrowheads). Scale bar: 10 µm in overview picture, 1 µm in other pictures. C, center (lumen in cystic organoids).

Citation: Journal of Endocrinology 240, 2; 10.1530/JOE-18-0462

• Download Figure
• Download figure as PowerPoint slide

These different epithelial phenotypes were further supported by whole-organoid two-photon microscopy scanning (Supplementary Movies 1 and 2) and by TEM (Fig. 5D). Cystic organoids were composed of tightly connected cells with polarized character and presence of microvilli selectively at the luminal side of the organoids, all typical of simple columnar/cuboidal epithelium (Fig. 5D). In contrast, in the dense organoids, cell membranes were not tightly connected but contained many desmosomes, commonly found in squamous epithelium (Fig. 5D). Moreover, TEM revealed keratin depositions in the center of the dense organoids, in line with the upregulated keratinization process as found by GO analysis, and as further confirmed by pancytokeratin immunostaining (Supplementary Fig. 5C).

Application of the organoid model as a new tool toward pituitary stem cell exploration

Organoids are valuable models to study tissue stem cell biology (Clevers 2016). Therefore, we addressed the question whether pituitary organoids could provide a model to explore and reveal new aspects of the putative pituitary stem cells, in particular regarding identity, behavior and regulation.

Pituitary organoids reveal keratin and Prrx stem cell marker expression and behavior

Obtaining strikingly divergent cystic and dense organoids from pituitary raised the question whether different stem cell (sub-)populations or phenotypes/states exist in the pituitary gland. We used the prominent dichotomy of keratin expression in the two organoid types to approach this question by analyzing the expression of CK8/CK18 and CK5/CK14 in the pituitary in situ. CK8/CK18 immunoreactivity highly colocalized with SOX2⁺ cells in the proposed stem cell niches (marginal zone and parenchymal clusters) of normal and damaged pituitary (Fig. 6A). In contrast, CK5 and CK14 immunoreactivity was not or only scarcely detected in both pituitary conditions (Fig. 6B).

View Full Size

Expression of distinct keratins in the pituitary SOX2⁺ cell compartment and of Prrx1/2 in organoids and pituitary after damage. (A) Immunofluorescent staining of SOX2 (red) and CK8/CK18 (green) in –/iDTR and GHCre/iDTR pituitary after DT injection (d11). CK8/CK18⁺ signal is found in SOX2⁺ cells in both the marginal zone of the AP (arrows) and in parenchymal clusters (arrowheads), as shown in overview and magnified pictures (lower panel). AP, anterior pituitary; CK, cytokeratin; IL, intermediate lobe; PP, posterior pituitary. Scale bar: 50 µm. (B) Immunofluorescent staining of SOX2 (red) and CK5 or CK14 (green) in –/iDTR and GHCre/iDTR pituitary after DT injection (d11). CK5⁺ and CK14⁺ signals were not, or only scarcely detected. (C) RT-qPCR expression analysis of Prrx1 and Prrx2 in dense (–/iDTR) and cystic (GHCre/iDTR) pituitary organoids (relative to expression in control AP set as 1 and indicated by dotted line). *P < 0.05; **P < 0.01. (D) RT-qPCR expression analysis of Prrx1 and Prrx2 in the AP from GHCre/iDTR and control –/iDTR mice (relative to expression in –/iDTR AP set as 1). Bars represent mean ± s.e.m. (n = 3). **P < 0.01.

Citation: Journal of Endocrinology 240, 2; 10.1530/JOE-18-0462

• Download Figure
• Download figure as PowerPoint slide

We hypothesize that the CK8/CK18/SOX2⁺ cells, once activated as occurring after the GH⁺ cell ablation damage (Fu et al. 2012, Willems et al. 2016), readily form the large cystic organoids as observed, while in the undamaged resting pituitary, the majority of these CK8/CK18/SOX2⁺ cells are quiescent with only some sporadic cells (stochastically) activated in a sufficient way to form (small) cystic organoids (Fig. 1A). That activation of the pituitary stem cells may play a role in the cystic organoid formation, is further supported by the following findings. First, abundant organoids, all displaying a cystic phenotype, were obtained from neonatal pituitary (Supplementary Fig. 6A) whose stem cell compartment is more activated (Gremeaux et al. 2012). Second, organoids that were generated from pituispheres established from undamaged adult AP were of cystic phenotype (Supplementary Fig. 6B). Pituisphere formation involves (proliferative) activation of the stem cells by the specific culture conditions used (Chen et al. 2005, 2009, Fauquier et al. 2008, Garcia-Lavandeira et al. 2009, Fu et al. 2012). How the CK5/CK14/SOX2⁺ dense organoids originate is at present not clear given the high scarcity of CK5 and CK14 immunoreactivity in the adult AP, although it should be noted that Krt5 and Krt14 mRNA was clearly detectable in the SOX2⁺ cell population and found enriched when compared to the SOX2⁻ cells (Fig. 5B). The Krt5/Krt14-expressing SOX2⁺ cells may start to translate the mRNA at organoid formation. Another possibility is that CK8/CK18/SOX2⁺ cells switch epithelial phenotype (to CK5/CK14/SOX2⁺) in vitro in the context of the organoid-forming conditions. We occasionally observed cystic organoids that transformed into dense organoids when other dense organoids were present in the vicinity, thereby suggesting an inductive (paracrine) effect of the dense organoid to cause the cystic forms to switch epithelial phenotype (Supplementary Fig. 6C). Accordingly, mixed CK8/CK18⁺/CK5⁺ organoids were sporadically seen (Supplementary Fig. 6D).

PRRX1 and PRRX2 have recently been advanced as plausible stem cell markers in the pituitary (Vankelecom 2010, Higuchi et al. 2014, Cox et al. 2017). We detected strong upregulation of Prrx1 and Prrx2 in the damaged pituitary-derived cystic organoids as compared to the dense organoids (Fig. 6C; also observed in the RNA-seq analysis, data not shown). Taking this finding back to the pituitary in vivo, we found that Prrx1 and Prxx2 gene expression levels were upregulated in the AP after damage (Fig. 6D), thus suggesting a role in stem cell activation upon injury. This finding highlights the power of the pituitary organoid models, and in particular their comparison, to unravel the behavior of stem cell-associated markers in the pituitary as occurring during damage and regeneration.

Pituitary organoids reveal peculiarities of the stem cell-regulatory WNT/LGR pathway

The WNT/LGR pathway is known to play a key role in stem cell regulation in many tissues, including their self-renewal/maintenance and activation upon injury (Barker et al. 2007, Huch et al. 2013b ). To start investigating the role of WNT/LGR in pituitary stem cell regulation, we used our organoid system and tested the influence of this pathway in pituitary organoid initiation (P0) and expansion from both normal and damaged gland. Omission of WNT3A did not affect organoid development (P0), neither in the control (–/iDTR) nor in the GHCre/iDTR conditions (Fig. 7A and Supplementary Fig. 7A). In contrast, removal of RSPO1 reduced organoid development in the GHCre/iDTR culture, while not overtly changing the organoid formation efficiency in the –/iDTR condition (Fig. 7A and Supplementary Fig. 7A). Addition of the WNT pathway inhibitor IWP2 (which blocks endogenous WNT ligand palmitoylation as needed for secretion) significantly reduced organoid formation efficiency in the GHCre/iDTR condition (Fig. 7A and Supplementary Fig. 7A). Also, addition of another WNT inhibitor, XAV939 (which stabilizes the canonical WNT-inhibitory protein AXIN2), efficiently blocked the formation of cystic (GHCre/iDTR) organoids, while dense (–/iDTR) organoids still formed (Supplementary Fig. 7B). WNT3A could not rescue the formation of organoids in the IWP2-treated cultures (Fig. 7A and Supplementary Fig. 7A).

View Full Size

The WNT/LGR pathway in pituitary organoid formation and expansion. (A) Number of organoids formed (P0) from –/iDTR or GHCre/iDTR AP cells in PitOM with or without WNT3A, RSPO1 and/or IWP2 as indicated. Bars represent mean ± s.e.m. (n = 4). *P < 0.05. NS, not significant. (B) Expansion of –/iDTR and GHCre/iDTR pituitary organoids in the presence or absence of the components as indicated. Organoids were first grown till P2 in PitOM, then replated (= P′0) in the presence or absence of WNT3A, RSPO1 and/or IWP2 as indicated, and then again passaged (= P′1–2) in the same medium. Representative light-microscopic pictures are shown at P′0 (after 10 days of culture) and P′1–2 (after 8–10 days of culture). Scale bar: 200 µm. (C) Heatmap from RNA-seq data showing the differential expression of Wnt ligands (from high expression in yellow to low expression in blue) in dense (–/iDTR) and cystic (GHCre/iDTR) pituitary organoids. (D) RT-qPCR expression analysis of Lgr5 in dense (–/iDTR) and cystic (GHCre/iDTR) pituitary organoids (relative to expression in dense organoids set as 1). Bars represent mean ± s.e.m. (n = 3). **P < 0.01. (E) RT-qPCR expression analysis of Lgr5 and Rspo1 in the AP from GHCre/iDTR and control –/iDTR mice (relative to expression in –/iDTR AP set as 1). Bars represent mean ± s.e.m. (n = 3). *P < 0.05. (F) Working model of pituitary organoid formation and expansion. CK8/CK18/SOX2⁺ pituitary stem cells (blue), when activated as occurring after damage, give rise to (large) cystic organoids in a WNT-dependent manner. The organoids expand but stop after a limited number of passages. Dense organoids, consisting of CK5/CK14/SOX2⁺ cells and forming in WNT-independent conditions, may originate from very rare CK5/CK14/SOX2⁺ squamous (stem) cells (or SOX2⁺ cells expressing CK5/CK14 at the mRNA level) present in the pituitary (brown), from (specific) CK8/CK18/SOX2⁺ stem cells switching epithelial phenotype and/or from (specific) CK8/CK18/SOX2⁺ stem cells reverting to a primitive identity in organoid culture. The dense organoids are long-term expandable in a WNT-independent manner.

Citation: Journal of Endocrinology 240, 2; 10.1530/JOE-18-0462

• Download Figure
• Download figure as PowerPoint slide

Also for further expansion and passaging of the organoids, WNT3A was found not essential (in both dense and cystic organoid cultures), whereas RSPO1 was needed to maintain the growth of the cystic organoids over different passages (Fig. 7B). IWP2 impaired cystic organoid expansion, which is only slightly rescued by WNT3A (Fig. 7B). XAV939 also inhibited the expansion of cystic organoids (Supplementary Fig. 7C). Intriguingly, the dense organoids kept expanding regardless of the presence of RSPO1 (and WNT3A), and expansion was only minimally affected by IWP2 or XAV939 (Fig. 7B and Supplementary Fig. 7C).

The dense and cystic organoids thus showed distinct WNT/RSPO growth responsiveness. To investigate whether these functional characteristics correlated with molecular features, we analyzed the endogenous expression of WNT ligands in both organoid types. RNA-seq analysis pointed to an overall higher expression of WNT ligands in the dense organoids (Fig. 7C; validated for a selection by RT-qPCR in Supplementary Fig. 7D) which may be in line with their independency of exogenous WNT3A.

The WNT-amplifier RSPO1 is known as the typical ligand for LGR5. We found high expression of Lgr5 in the cystic organoids (i.e. obtained from the damaged GHCre/iDTR pituitary) while expression is virtually absent in the dense organoids (from control pituitary) (Fig. 7D), in line with the need for RSPO1 in cystic organoid culturing and its redundancy in dense organoid growth. In analogy with Lgr5 gene expression in the cystic organoids, expression was also found upregulated in the GHCre/iDTR pituitary in situ after DT-induced damage (as analyzed by RT-qPCR and RNAscope in situ hybridization) (Fig. 7E and Supplementary Fig. 7E). Emergence of LGR5 expression after damage has also been reported in other tissues with low turnover rate (such as the liver and pancreas; Huch et al. 2013a ,b ). Using the triple transgenic GHCre/iDTR/Lgr5lacZ mouse model, we observed some lacZ⁺ cells (weak signal) arising in the pituitary after damage, some of them in the proposed marginal zone stem cell niche (Supplementary Fig. 7F). Moreover, when AP cells from this mouse model were cultured in organoid-forming culture conditions, a number of lacZ⁺ organoids were detected (as analyzed at P2; Supplementary Fig. 7G). Not all organoids were lacZ⁺ suggesting that (i) not all organoids are derived from the Lgr5^lacZ+ cells emerging after damage; (ii) that the Lgr5^lacZ+ signal is lost upon organoid culture and expansion or (iii) the efficiency of the Lgr5lacZ reporter system is low in the pituitary and the organoids (as supported by the only weak signals appearing in the pituitary in situ). Expression of the LGR5 ligand Rspo1 was also found upregulated in the injured pituitary (Fig. 7E) but was not detected in the organoid cultures (data not shown), explaining why addition of exogenous RSPO1 is needed for the (cystic) organoids to develop and grow. Taken together, our new pituitary organoid models may lead to a valuable tool to study the regulation of putative pituitary stem cells in both normal and activated condition, as exemplified here by revealing particular WNT/LGR pathway aspects.

Finally, since the dense organoid cultures were found to thrive without addition of WNT3A and RSPO1, we also tested whether other (growth) factors were indeed needed for the long-term expansion, and removed FGF2/8/10, SHH, Noggin and cholera toxin (Table 1; the depleted medium further termed ‘EASI’ medium referring to the remaining core factors, viz. EGF, A83-01, SB202190 and IGF1). Surprisingly, the dense organoids remained as expandable in EASI medium as in the original PitOM (Supplementary Fig. 8A). EGF proved to be an essential, but not sufficient growth factor for this dense organoid expansion (Supplementary Fig. 8A), indicating that one or more of the other factors still present in the EASI medium are necessary for optimal organoid growth and expansion. A83-01, the small-molecule inhibitor of TGFB signaling, was found to be required for optimal expansion, while adding TGFB1, in the absence of its inhibitor, further compromised (i.e. arrested) organoid growth (Supplementary Fig. 8B). Finally, we found that EASI medium was even sufficient for the initial formation (P0) of dense (but not cystic) organoids (Supplementary Fig. 8C and data not shown). For the formation of cystic organoids, several PitOM components were found to be crucial, including EGF (Supplementary Fig. 3D), the p38 inhibitor SB202190 and cholera toxin (Supplementary Fig. 8D). Together, these findings may provide hints toward regulatory pathways of the putative pituitary stem cells.

In conclusion, our new pituitary organoid system presents itself as a promising tool toward studying and deciphering pituitary stem cell biology, including identity, behavior and regulation.