Collection of human fetal gonads and ethical approval

Human fetal gonads were collected from elective terminations of first trimester pregnancies. The elective terminations were carried out at the Departments of Gynaecology at Copenhagen University Hospital (Rigshospitalet), Hvidovre Hospital and Herlev Hospital, Denmark. The study was approved by the regional ethics committee (# H-1-2012-007). All women gave their consent after being given oral and written information about the project. None of the terminations were for pathological reasons of the pregnancy or fetal abnormality. A total of 117 gonads was included in this study with age ranging from 6 to 10 post conceptional weeks (PCW) from 41 female foetuses and 30 male foetuses. Fetal age was calculated by scanning crown-rump length and by foot-length evaluation [31]. The nomenclature used to describe fetal age of samples was PCW followed by days (e.g., 8 + 5 PCW) and indicate the age at the time of ex vivo culture setup. The sex of the foetus was determined by quantitative RT-PCR using QuantStudio3 (Applied biosystems, Thermo Fischer Scientific, Denmark). DNA was isolated from surplus fetal tissue using NucleoSpin Tissue kit (MACHEREY NAGEL, 740952.250) as described by the manufacturer. Quantitative RT-PCR analysis of SRY gene expression (forward sequence: GAATATTCCCGCTCTCCGGA, reverse sequence: GCTGGTGCTCCATTCTTGAG) was conducted in duplicates with SYBR Green master mix (Agilent, 600828-51) and ZFYX gene expression (forward sequence: ACCRCTGTACTGACTGTGATTACAC, reverse sequence: GCACYTCTTTGGTATCYGAGAAAGT) was included as positive control. Additionally, control samples were included for both XY tissue and XX tissue in each analysis.

Ex vivo gonad tissue culture

Human fetal gonads were cultured using the ex vivo culture model as previously described [28,29,30]. In brief, human fetal gonads were dissected in ice-cold phosphate buffered saline (PBS) and cut into 1 mm3 tissue fragments. The tissue fragments were subsequently cultured for 24 h awaiting sex determination in 37°C and 5% CO2 in a hanging drop of 40 µl media consisting of MEMα medium (Gibco) supplemented with 1 × MEM non-essential amino acids, 2 mM sodium pyruvate, 2 mM L-Glutamine, 1 × Insulin-Transferrin-Selenium (ITS) supplement (Sigma-Aldrich), 1 × penicillium/streptomycin and 10% Fetal bovine serum. All supplements were from Gibco (Nærum, Denmark), except ITS (Sigma-Aldrich, Brøndby, Denmark).

To manipulate with WNT/β-catenin signalling in human fetal gonads, two small molecule inhibitors were used; IWR-1 (which stabilizes the β-catenin destruction complex and thus ensures degradation of β-catenin) in fetal ovaries and CHIR99021 (which inhibits GSK-3 that normally ensures degradation of β-catenin in the absence of a WNT signal) in fetal testes. Following sex determination of the foetus, gonadal tissue fragments from female foetuses were thus transferred to a new hanging drop of media supplemented with either: (1) 1 µM IWR-1, abbreviated IWR (2) 1 µM IWR-1 + 50 ng/ml FGF9, abbreviated IWR+FGF9 or (3) 1 µM IWR-1 + 50 ng/ml FGF9 + 25 ng/ml Activin A + 25 ng/ml Activin B + 25 ng/ml TGFβ1, abbreviated IWR+FAT (See Table 1 for details). Gonadal tissue fragments from male foetuses were transferred to media supplemented with either: (1) 3 µM CHIR99021, abbreviated CHIR (2) 3 µM CHIR99021 + 100 ng/ml RSPO1, abbreviated CHIR+RSPO1 or (3) 3 µM CHIR99021 + 100 ng/ml RSPO1 + 100 ng/ml WNT4, abbreviated CHIR+RW. For all treatment setup at least one tissue fragment from each foetus was cultured as vehicle control (either with 0.1% DMSO, 0.1% BSA in PBS or 0.1% DMSO, 0.1% BSA, 4mM HCl in PBS). Importantly, due to a considerable biological variation between the biological replicates (foetuses), mainly due to differences in the time from the surgical termination of pregnancy until the tissue fragments can be setup in ex vivo cultures, all treatment-induced changes are shown as a ratio relative to vehicle control treated samples from the same foetus. The doses selected and used were based similar types of ex vivo culture studies in mice [19, 32] and on pilot experiments in which human fetal testis (n = 3) and ovaries (n = 3) were cultured ex vivo with the treatment doses and combinations of pharmaceutical inhibitors and recombinant proteins listed above. Media from these pilot experiments were initially analysed for effects on the secretion of AMH and Inhibin B (tendency to treatment-induced increase in fetal ovaries and decrease in fetal testes). Gonadal tissue was cultured for 14 days with a complete media change every second day. At the end of ex vivo culture, tissue fragments were either fixed in formalin and subsequently paraffin embedded for immunohistochemistry/immunofluorescence analysis or fixed in 4% PFA for 3D imaging analysis. The media were collected and pooled for each individual tissue fragment throughout the culture period and was stored at -20 °C until further analysis.

Table 1 Media supplementsImmunohistochemistry

Immunohistochemical analysis of gonadal tissue was conducted as previously described [33], with few modifications. In brief, paraffin sections (4 μm) of formalin fixed tissue were deparaffinized and rehydrated. Antigen retrieval was accomplished in a pressure cooker (medical decloaking chamber, Biocare, Concord, CA, USA) at 110°C for 30 min in retrieval buffer or by microwaving sections for 15 min in retrieval buffer. Blocking of endogenous peroxidase was performed with 1% (v/v) H2O2 in methanol for 30 min. Sections were incubated with 0.5% (w/v) milk powder in Tris buffered saline (TBS). After incubation overnight at 4°C with primary antibody (details listed in Supplementary Table 1), the sections were incubated with the appropriate secondary antibody (ImmPRESS, MP-7401, MP-7402, MP-7405, Vector Laboratories, CA, USA) for half an hour at RT and signal development was performed with AEC (ImmPACT, SK-4205, Vector Laboratories, CA, USA) visualised as red staining. The sections were washed with TBS between the incubation steps except between blockade for cross-reactivity and the primary antibodies. In all experiments positive controls were included as well as negative controls in which the primary antibody was replaced by the dilution buffer alone. None of the negative control sections showed any staining. Counterstaining was conducted with Mayer’s haematoxylin.

Immunofluorescence

Immunofluorescence was performed on ex vivo cultured tissue as previously described [28]. In brief, immunofluorescence was performed with TBS washes (3 × 5 min) between each step and all incubations were carried out in in a humidity box (Fisher Scientific, UK). Sections (4 μm) of formalin fixed tissue were dewaxed and rehydrated using standard procedures, followed by heat-induced antigen retrieval (pressure cooker) in 0.01 M citrate buffer (pH 6) and peroxidase block in 3% (v/v) H2O2 in methanol for 30 min. Next, the sections were blocked in normal chicken serum (NCS; Biosera, Ringmer, UK) diluted 1:5 in TBS containing 5% (w/v) BSA (NCS/TBS/BSA), followed by incubation with COUPTF-II antibody diluted in NCS/TBS/BSA overnight at 4°C. The next day, sections were incubated with peroxidase-conjugated chicken anti-mouse secondary antibody (Santa Cruz), diluted 1:200 in NCS/TBS/BSA for 30 min at room temperature (RT), and followed by incubation with Tyr-Cy3 (Perkin Elmer-TSAPlus Cyanine3 System; Perkin Elmer Life Sciences, Boston, MA, USA) according to manufacturer’s instructions. Before the next primary antibody dilution was added, the sections were again subjected to antigen retrieval by blocking in NCS/TBS/BSA and overnight incubation at 4°C with AMH or OCT4 antibody diluted in NCS/TBS/BSA. On the third day, slides were incubated with peroxidase-conjugated appropriate secondary antibody (Santa Cruz) diluted 1:200 in NCS/TBS/BSA for 30 min at RT, followed by incubation with Tyr-Fluorescein (Perkin Elmer-TSA-Plus Fluorescein System; Perkin Elmer Life Sciences) according to manufacturer’s instructions. Sections were again subjected to antigen retrieval followed by blocking in NCS/TBS/BSA and incubation with SOX9 or FOXL2 antibodies diluted in NCS/TBS/BSA overnight at 4°C. Sections were then incubated with peroxidase-conjugated chicken anti-rabbit secondary antibody (Santa Cruz), diluted 1:200 in NCS/TBS/BSA for 30 min at RT, followed by incubation with Tyr-Cy5 (Perkin Elmer-TSA-Plus Cyanine5 System; Perkin Elmer Life Sciences, Boston, MA, USA) according to manufacturer’s instructions. Sections were counterstained with DAPI (Sigma-Aldrich) diluted 1:500 in TBS for 10 min. Finally, slides were mounted with Permafluor (Thermo Scientific, UK) and fluorescent images captured using an Olympus BX61 microscope (Olympus). Antibody dilutions are listed in Supplementary Table 2.

BrdU incorporation

Before end of ex vivo culture period, tissue fragments were cultured with BrdU labeling agent (Life Technologies, Nærum, Denmark) diluted 1:10 in media for 6 h to allow for the detection of proliferating cells in the tissue. After 6 h, tissue fragments were formalin fixed and paraffin embedded as described above. Proliferating cells were visualized by immunohistochemical analysis using a BrdU antibody (Supplementary Table 1) as described in the Immunohistochemistry section.

Quantification of cells

The number of OCT4+ (gonocyte and oogonia marker), BrdU+ (proliferation marker) and cPARP+ (apoptosis marker) cells were quantified per area of tissue using one entire tissue section. The area of fetal gonadal tissue was calculated following scanning of sections on a NanoZoomer 2.0 HT (Hamamatsu Photonics, Germany) and images captured using the software NDPview version 2.6.13 (Hamamatsu Photonics). Images from the NanoZoomer were subsequently analysed using the counting tool in Adobe Photoshop CC version 20.0.1. For all quantifications, tissue samples from at least five embryos/foetuses were included.

3D imaging and quantitative analysis

For 3D imaging analysis, ex vivo cultured tissue was washed in ice-cold PBS and fixed in 4% PFA for 2–4 h before it was transferred to PBS containing 0.01% Sodium Azide (Sigma-Aldrich) and kept at 4°C until iDISCO + whole-mount immunostaining analysis which was carried out as previously described [34] and detailed below.

Sample pre-treatment with methanol

As previously described [35, 36] samples were embedded in 2% agarose (Roth) prepared in 1× PBS (Invitrogen) prior to clearing and processing. Samples were then dehydrated for 1 h at RT in ascending concentrations of methanol in water (20%, 40%, 60%, 80% and 100% two times). Next, samples were incubated in dichloromethane 66% (Sigma-Aldrich #270,997) with 33% methanol overnight at 4°C. After 2 washes of 1 h with 100% methanol, samples were incubated at 4°C with the bleaching solution (5% H2O2 in methanol) to remove the pigmentation. Samples were then rehydrated in descending concentrations of methanol (100% × 2, 90%, 80%, 70%, 60% and 50%) and stored in 1× PBS at 4°C.

Whole-mount immunostaining

Samples were incubated at room temperature (RT) on an adjustable rotator in a permeabilized blocking solution (PBSGT) of 1 × PBS containing 0.2% gelatin (Sigma), 1% Triton X-100 (Sigma-Aldrich) and 0.01% Sodium Azide for two nights. For immunostaining, samples were transferred in PBSGT together with the primary antibodies SOX9 (rabbit polyclonal antibody), AMH (mouse monoclonal antibody), and Cytochrome P450 17A1/CYP17A1 (goat polyclonal antibody) and incubated at 37°C in agitation for one week. This was followed by 6 washes of 1 h in PBSGT at RT. Samples were then incubated in secondary antibodies (anti-rabbit AlexaFluor 488, anti-mouse AlexaFluor 568 and anti-Goat AlexaFluor 647, Thermofisher) diluted at 1:500 in 1×PBS containing 0.2% gelatin (Prolabo), and 0.5% Triton X-100 (Sigma-Aldrich) (PBSGT) at RT for three days. After six washes of 30 min in PBSGT at RT, samples were stored in the dark at 4°C until tissue clearing. Antibody details and dilutions are listed in Supplementary Table 3.

Tissue clearing

All incubation steps were performed at RT in a fume hood, on a tube rotator at 14 rpm covered with aluminium foil to avoid contact with light. Samples were dehydrated in a graded series (20%, 40%, 60%, 80% and 100% × 2) of methanol diluted in PBS × 1. Next, samples were incubated in dichloromethane 66% with 33% ethanol overnight at 4 °C. This was followed by a delipidation step of 45 min in 100% dichloromethane (DCM; Sigma-Aldrich). Samples were cleared in dibenzyl ether (DBE; Sigma-Aldrich) for 2 h at RT on constant agitation and in the dark. The next day, samples were stored in individual light-absorbing glass vials (Rotilabo, Roth) at RT. In these conditions, samples could be stored and imaged for up to 9 months without any significant fluorescence loss.

Imaging

3D imaging was performed as previously described [37]. An ultramicroscope (LaVision BioTec) and an Andor Neo 5.5 sCMOS camera using InspectorPro software (LaVision BioTec) were used to perform imaging. The light sheet was generated by a laser (wavelength 568–647 nm, Coherent Sapphire Laser, LaVision BioTec) and two cylindrical lenses with a 1.1×/0.1NA and 4×/0.3NA objectives were used. Samples were placed in an imaging reservoir made of 100% quartz (LaVision BioTec) filled with DBE and illuminated from the side by the laser light. The step size between each image was fixed at 4 μm. Analysis of 3D imaging was performed using the Imaris software (version 9.9, Bitplane, Oxford Instruments).

3D analysis

Images, 3D volume, and movies were generated using Imaris ×64 software (version 9.9.0, Bitplane). Stack images were first converted to imaris file (.ims) using ImarisFileConverter and 3D reconstruction was performed using “volume rendering”. Optical slices of samples were obtained using the “orthoslicer” tools. The “surface” tool was used and precisely the machine learning plugin Labkit (pixel classification) from Fiji (National Institute of Health, Bethesda) to segment SOX9 positive cells and remove the background noise. After several machine-learning based training sessions for pixel classification, the classifier of segmentation was applied to all 3D acquisitions. This was followed by the use of the « spots » tool to automatically count the SOX9 positive cells of the new channel after image segmentation. The number of cells was obtained by the tab « statistics ». The staining of AMH was segmented using Labkit Fiji plugin. In this case, the quantification was done on the volume of AMH inside the mask previously created. This information was extracted by the tab « statistics » of « surface » tool in µm3.

RSPO1 ELISA assay

RSPO1 measurements were performed on media from ex vivo tissue cultures using Human Rspondin-1 ELISA Kit (R&D Systems, DY4546-05) according to the manufacturers instructions. RRID: AB_2936293. Media samples were diluted 25% (75 µl media sample + 25 µl culture media). The detection limit of RSPO1 assay was 31.1 pg/ml and the intra-assay variation was ≤ 10% for all controls except one control in each of the two batches, for those the intra-assay variance was below ≤ 15%.

AMH hormone measurements

Quantification of AMH concentration was measured in ex vivo culture media by ELISA using the Beckman Coulter ACCESS AMH assay Reagents Kit (Ref. B13127) and Calibrator kit (Ref. B13128) as previously described [28]. RRID: AB_2892998. Collected media samples were diluted 1:10 in culture media prior to analysis, with additional sample dilution (1:25) and (1:50) necessary for a few of the samples. The detection limit of the AMH assay was 0.14 pmol/L and the intra-assay variation was ≤ 9%.

Inhibin B hormone measurements

Inhibin B concentration in ex vivo culture media was measured by ELISA using the Beckman Coulter INHIBIN B GEN II ELISA assay Reagents Kit (Ref. A81303) and Calibrator kit (Ref. A81304) as previously described [28]. RRID: AB_2827405. Collected media samples were diluted 1:50 for samples originating from male ex vivo tissue cultures and 1:20 for samples from female ex vivo tissue cultures. The detection limit of the inhibin B assay was 3 pg/ml, and the intra-assay variation was ≤ 10%.

Androgen measurements

Quantification of androgens; testosterone, androstenedione, and dehydroepiandrosterone sulphate (DHEAS) in media from ex vivo cultures were performed using a highly sensitive LC-MS/MS method as previously described [38], and adjusted for measurement in culture media [28]. In brief, the modifications were the following: calibration curves were prepared in culture media, control samples were prepared by spiking with high and low concentration of steroids, and all collected media samples were diluted (1:4) in culture media prior to analysis. Each batch included standards for calibration curves, approximately 20 unknown samples, one blank and three pooled controls spiked with steroid standards at low and high levels. Limit of quantification was 0.012 nM for testosterone, 0.042 nM for androstenedione and 19 nM for DHEAS. The inter-day variation expressed as the relative standard deviation (RSD) was ≤ 14% for controls spiked in low levels and ≤ 3% for controls spiked in high levels.

INSL3 measurements by LC-MS/MS

Quantification of INSL3 in media from ex vivo cultures was performed using LC-MS/MS as previously described [39], with the exception that INSL3 calibrators were diluted in culture media instead of serum. Limit of detection was 0.03 µg/l and limit of quantification was 0.15 µg/l. Intra-assay variation was ≤ 10%.

Statistics

Due to the substantial heterogeneity between the biological replicates (foetuses), a paired analysis approach was used in which treated samples were always compared to vehicle control treated samples from the same foetus. For vehicle controls, the mean of all tissue fragments from each fetus was set to 1 and the individual ratio for each was calculated. Statistical analysis was performed using the ratio paired t-test in GraphPad Prism software version 8, which utilizes the logarithm of the ratio to test the difference between a given control and treatment group. The ratio paired t-test was performed individually for each of the treatments compared with vehicle controls. This was due to the small amount of tissue available for the youngest foetuses which only allowed the setup of one treatment group. The data are shown as ratios compared with the internal controls and illustrated for each treatment group as individual data points and mean ± SEM, with the indicated “n” corresponding to the number of foetuses. Asterisk indicates statistical significance with * P < 0.05, ** P < 0.01, *** P < 0.001 and **** P < 0.0001.