Human and mouse embryonic and fetal pancreatic tissues

Human embryonic and fetal pancreatic tissues (PCW5–11) were obtained through the Inserm cross-cutting scientific programme Human Developmental Cell Atlas (HuDeCA) from surgical abortions in accordance with the French bioethics legislation and Inserm guidelines [14, 15]. Donors’ written consent was obtained, along with approval from Agence de Biomédecine, the competent authority in France. Fetal ages are displayed as PCW, staged based on morphometric correction [16]. All specimens were initially selected based on macroscopic morphological criteria, excluding samples with obvious malformations. In cases where visual determination of sex was not possible, PCR-based chromosome screening using DNA from biopsies was performed [13]. For labelling, a minimum of three samples per age were employed (electronic supplementary material [ESM] Table 1 and ESM Table 2), while for in vitro experiments we employed a minimum of five pancreases per condition.

Mouse fetal pancreases were obtained from pregnant C57BL/6J mice purchased from the Janvier Breeding Center (LeGenet, St Isle, France). Mice were killed by CO2 asphyxiation according to French Animal Care Committee guidelines and pancreases from embryos at embryonic day (E)10, E12 and E16 (at least six embryos per age) were dissected and fixed [17].

Conventional 2D immunohistochemistry

Human embryonic and fetal pancreases were fixed in formalin, embedded in paraffin and sectioned (5 μm thick) as previously described [14]. The primary and secondary antibodies used are listed in ESM Table 3. EdU revelation was performed using the Click EdU Alexa 555 imaging kit (Thermo Fisher Scientific, Waltham, MA, USA). Nuclei were stained with Hoechst 33342 (0.3 mg/ml, Invitrogen, Waltham, MA, USA).

We manually counted the number of INS+ cells per cluster in all the 5 μm sections of three pancreases at PCW6 (180–240 sections/specimen). The percentage of PDX1+/KI67+ (at PCW7, 9 and 11) and INS+/KI67+ cells (at PCW9 and 11) were quantified on six sections per pancreas (3 pancreases/age). The spatial distribution of PDX1+/KI67+ progenitors was measured by computing the Ripley K function [K(r)] [18]. To do that, the PDX1+/KI67+ progenitors were mapped with x,y coordinates in 2D images at PCW7, 9 and 11 (minimum of n=20 images/developmental stage), using FIJI (version 2.14.0, Eliceirei/LOCI group, USA) [19]. Then, K(r) was computed for every image by implementing the R (version 4.1.2, R Foundation for Statistical Computing, Austria) library spatstat as previously described [20].

3D immunostaining

Embryonic and fetal samples were fixed by immersion in formalin at 4°C for 1 to 5 days depending on size. For tissue bleaching [10], fixed pancreases were dehydrated for 1 h at room temperature (RT) in increasing concentrations of methanol (50%, 80%, 100%) in 1X PBS. Then, the samples were incubated overnight at 4°C in a 6% hydrogen peroxide solution in 100% methanol, re-hydrated for 1hr at RT in decreasing concentrations of methanol (100%, 100%, 80%, 50%) and finally washed for 1 h in 1X PBS. Permeabilisation and blocking were performed under rotation at 70 rev/min at RT in PBSGT (1X PBS, 0.2% gelatin [Prolabo, Nemours, France], and 0.5% Triton X-100 [Sigma-Aldrich, Saint-Louis, MO, USA]) [13].

For whole-mount immunostaining, samples were incubated with the primary antibodies in a solution of PBSGT containing 0.1% saponin during 14 days at 37°C under rotation at 70 rev/min. The primary and secondary antibodies employed are listed in ESM Table 3. Then, tissues were incubated with the secondary antibodies in a solution of PBSGT (0.1% saponin [10 mg/ml]), during 4 days at 37°C under rotation at 70 rev/min. Next, samples were washed for 30 min in PBSGT at RT, embedded in 1.5% agarose (Roth, Goshen, IN, USA) prepared in TAE 1X (Invitrogen) [10, 11] and stored at 4°C until tissue clearing.

In order to achieve tissue transparency, the samples were cleared using a modification from the iDISCO+ protocol [21]. Fetal pancreases were dehydrated in growing concentrations of methanol (20%, 40%, 60%, 80%, 100%, 100%, 1 h each) at RT under rotation (14 rev/min), next incubated overnight in a solution of 2/3 dichloromethane (DCM) and 1/3 methanol, then for 30 min in 100% DCM, and finally transferred to dibenzyl ether (DBE) for further storage and imaging.

3D imaging and processing

Acquisition was performed with Miltenyi Biotec (Bergisch Gladbach, Germany) Ultramicroscope Blaze (sCMOS camera 5.5 MP) and Imspector Pro (version 7.3.2, LaVision Biotec, Germany) acquisition software. Laser light sheets are generated at excitation wavelengths of 488, 561, 640 and 785 nm. Lenses with 4× magnification (MI Plan 4× NA0.35) and 12× magnification (MI Plan NA0.53) objective were used. The z-step size is 3 µm. Images were generated as tiff files, converted to (.ims) by Imaris File Converter (version 9.8, Bitplane, UK), and the 3D images were reconstructed using the volume rendering function. To isolate a specific region of the tissue, the surface tool was used to manually segment. The 3D images, virtual slices and movies were generated using the snapshot and animation tools. To locate the pancreas in full intact PCW7 human embryo, the SOX9+ pancreatic area was delimited in consecutive slices.

Human and mouse pancreatic lengths and diameters were measured with the distance tool in virtual slices. The measurements of the SOX9+ and INS+ cluster volume and density were performed following the automatic segmentation tool. The distribution of human INS+ clusters was measured in three virtual slices from three different pancreases. First, the pancreatic epithelium was manually outlined and the centroid was determined for each slice. Then, the distance from every INS+ cluster was measured to the centroid [19].

Tissue culture

Whole human embryonic and fetal pancreases at PCW6–10 were cultured for a maximum of 7 days at an air-complete medium interface in Petri dishes on 0.45 μm filters (Millipore) [22]. Explants were cultured in G10 medium consisting of RPMI 1640 + glutaMax, 10% FBS, 1% penicillin/streptomycin, 1% HEPES and 1% non-essential amino-acids (61870036, ThermoFisher Scientific). Medium was replaced every 2 days. For EdU incorporation assays, pancreases were treated with 5 µmol/l EdU for 4 h, the reaction was stopped by washing out the tissue with G10 medium and immediate fixation. For experiments with the platelet-derived growth factor (PDGF) AA isoform (PDGFAA), explants were cultured in DMEM/F12 medium (31331, ThermoFisher Scientific) containing ALK5 inhibitor (ALK5i; 10 mmol/l, 221234, Chem Cruz, Dallas, TX, USA), noggin (500 mg/ml, 120–10C, Peprotech, Waltham, MA, USA) and nicotinamide (2.5 mol/l, 481907, Sigma). Explants were treated with PDGFAA (200 ng/ml, PHG0035, Sigma) for 7 days, replacing the medium every 2 days.