Orthotopic transplantation.

NSG mice were used for xenotransplantation of hESC-derived pancreatic cells into the murine pancreas. Mouse were sacrificed 8 weeks after transplantation (56 days for PE transplantation, 52 days for PP transplantation). While the experimental procedure of the transplantation has been detailed previously [23], cell preparation will be described in the pancreatic differentiation section.

In situ hybridization (ISH).

For riboprobe generation, PCR was performed using GoTaq Flexi Polymerase (Promega) with the following primer pairs: Ppdpf_fwd 5’-gtcttcccaaagccgaccc-3’ and Ppdpf_rev 5’-gggtggctgctgtgagttta-3’, Pdx1_fwd 5’-gtcttcccaaagccgaccc-3’ and Pdx1_rev 5’-caaacagcctaaagacaaggct-3’. Ppdpf and Pdx1 were amplified from cDNA derived from (E11) total mouse RNA (Takara) and adult mouse pancreas of C57BL/6J mouse strain, respectively. Purified PCR products were cloned into pCR 2.1 TOPO plasmid (Thermo) according to manufacturer’s recommendations with insert:vector ratios of 2:1 and 5:1 with 50 ng vector overnight (o/n) at 16°C. Insert sequences were verified by sanger sequencing (Eurofins Genomics). Plasmid DNA was linearized using HindIII (NEB). To generate DIG-labeled riboprobes, in vitro transcription using T7 RNA polymerases (Roche) was performed. Pdx1 antisense [58] and Ppdpf sense probes served as positive and negative controls, respectively. For in situ hybridization, mouse tissue was fixed in 4% PFA, cryoprotected in 20% sucrose, and embedded in OCT compound. 18 µm thin frozen sections through the pancreatic anlage of E14.5 C57BL/6J embryos were collected on SuperFrost Plus slides (Thermo). In situ hybridization was performed according to [59] with minor modifications. Briefly, after proteinase K treatment and acetylation, sections were permeabilized using 1% TritonX100 and incubated o/n at 63°C with riboprobes diluted in hybridization buffer [50% formamide, 5x SSC, 5x Denhardt’s solution, 250 µg/mL yeast RNA, 500 µg/mL herring sperm DNA]. Sections were washed three times in 0.2x SSC at 65°C, blocked with 10% goat serum, and incubated with alkaline phosphatase conjugated anti-DIG antibody (Roche) at 4° o/n. Alkaline phosphatase activity was detected using nitro blue tetrazolium (NBT) and 5-bromo-4-chloro-3-indolyl-phosphate (BCIP) (Roche).

Maintenance culture.

In this study, the human embryonic stem cell line HUES8 (Harvard University; RRID:CVCL_B207) was used. Culture and differentiation towards the pancreatic lineage was performed with permission from the Robert Koch Institute according to the “79. Genehmigung nach dem Stammzellgesetz, AZ 3.04.02/0084”. HUES8 cell authentication was confirmed with a DNA profile using nonaplex PCR of Short Tandem Repeats (STRs) done by the Leibniz-Institute.

Culture conditions for hESCs were 5% CO₂, 5% O₂, and 37°C. Cells were cultured on hESC Matrigel (Corning) coated plates (according to manufacturer’s recommendations) in FTDA medium [60] or mTesR1 medium (STEMCELL Technologies) with daily media change. Splitting in a 1:4 to 1:6 ratio was done twice a week with TrypLE Express (Thermo) to enable feeder-free homogenous monolayer cultures. For this, cells were washed once with PBS and incubated with TrypLE for 3-5 min at 37°C. By adding DMEM-F12+Glutamax (Gibco) the enzymatic reaction was stopped and after centrifugation at 200 x g for 5 min, cells were resuspended in FTDA or mTesR1 supplemented with 10 µM ROCK inhibitor (Y-27632, Abcam) and seeded on a new, coated well.

Pancreatic progenitor differentiation for in vitro experiments.

Differentiation of hESCs into PPs was performed as described in [14]. Briefly, hESCs were seeded on 24-well plates, which were coated with growth factor reduced Matrigel (Corning, diluted 1:18 in DMEM-F12+Glutamax), 24 h prior to the start of differentiation. Cells were cultured in basal media BE1 at d0-d5: MCDB131 (Thermo) with 2 mM L-Glutamine (Gibco), 1.174 g/l sodium bicarbonate (Sigma), 0.8 g/l cell culture tested glucose (Sigma) and 0.1% fatty acid free BSA (Sigma/Proliant). At d6-14 BE3 was used as basal medium: MCDB131 with 2 mM L-Glutamine, 1.754 g/l sodium bicarbonate, 0.44 g/l glucose, 0.5% ITS-X (Gibco), 44 mg/l L-Ascorbic acid (Sigma) and 2% fatty acid free BSA. At a confluency of about 70-95%, differentiation was initialized. For that, cells were washed with PBS (Gibco), and BE1 medium containing 3 μM CHIR99021 (Axon Medchem) and 100 ng/ml ActivinA (Peprotech/R&D) (d0 medium) was added. After initializing differentiation, cells were cultured at 5% CO₂, 21% O₂, and at 37°C. After 24 h, medium was replaced by BE1 medium supplemented with 100 ng/ml ActivinA (d1 - d3 medium). During the next two days d1 – d3 medium was added to allow formation of definitive endoderm (DE), and on day 4 and day 5 BE1 medium containing 50 ng/ml KGF (Peprotech) was added inducing foregut specification. From day 6 until day 9, cells were cultured in BE3 medium supplemented with 0.25 μM SANT1 (Sigma), 2 μM retinoic acid (Sigma), 0.2 μM LDN193189 (Sigma), and 0.5 μM PD0325901 (Calbiochem). From day 10 on cells received 50 ng/ml FGF10 (Peprotech/R&D), 0.33 μM indolactamV (Stemcell Technologies), 10 μM SB431542 (Axon Medchem), and 16 mM glucose (resulting in a final glucose concentration of 24 mM) in BE3 medium for four days.

Pancreatic progenitor differentiation for in vivo experiments.

For orthotopic transplantation, our most recent protocol was implemented as detailed in [34]. This refined protocol allows higher PP efficiencies and better representation of the multipotency marker GP2. For PE transplantations, planar cultures were harvested at d7 2 days prior to PE and transferred to Costar ultra-low attachment plates (Corning) at 37°C 5% CO₂ without orbital shaking. Suspension culture spheres were harvested at the day of transplantation and cells corresponding to 1x10⁶ PE cells were transplanted. For PP transplants, cells were GP2 MACS sorted at d12 as optimized in [34] (based on [61,62]) and cultured in ultra-low attachment plates at 37°C 5% CO₂ without orbital shaking for 1 day. While 0.56x10⁶ sorted PP cells were transplanted in most cases, 10x10⁶ unsorted planar PP cells were transplanted when sorted PP cells were not sufficient for 3 transplantation experiments. In summary, following cells have been transplanted:

1. (1) 1x10⁶ suspension PE: 3x PPDPF^WT/WT, 3x PPDPF^KO/KO cl.1, 3x PPDPF^KO/KO KO cl.2 (all engrafted well).
2. (2A) 0.56x10⁶ GP2-sorted suspension PP: 2x PPDPF^WT/WT (1 not engrafted), 2x PPDPF^KO/KO cl.1, 3x PPDPF^KO/KO KO cl.2
3. (2B) 10x10⁶ unsorted planar PP: 1x PPDPF^WT/WT, 2x PPDPF^KO/KO cl.1 (1 teratom, excluded from analysis).

Based on our experience, we suggest transplanting 1x10⁶ unsorted PE or PP cells that were cultured as spheroids at least 1 day prior to transplantation for future experiments.

In vitro differentiation into pancreatic endocrine organoids.

We performed stem cell-derived islet differentiation as recently detailed in [36].

Construct generation.

A large deletion in the PPDPF gene locus was generated by CRISPR/Cas9 gene editing using a four times nicking strategy, thereby inducing two DSBs. Two nicks at each intended cleavage site were required to lead to one DSB, thereby minimizing off-target cleavage. crRNAs were designed in intronic regions flanking exon 3 of the PPDPF gene locus: guide 1a: GCTGGTTCCGCAGCCTGCAC; guide 1b: GCCTGAAGGCCGGTGGGCTG; guide 2a: GGGGCACTCGGTACTGCTGC; and guide 2b: GAAGCCATTCCCCACCCCCC. CRISPR plasmids were based on pX335-U6-Chimeric_BB-CBh-hSpCAS9n(D10A)-2A-GFP vector (a gift from Boris Greber at MPI Münster [63], itself based on Addgene #42335, a gift from Feng Zhang [64]). Phosphorylation for subsequent ligation of complementary oligonulceotides was performed using the T4 Polynucleotide Kinase (NEB). Oligonucleotides were annealed at 37°C for 30 min and 95°C for 5 min with subsequent cooling to 25°C at 0.1°C/sec. Restriction digestion of the vector was performed using FastDigest BbsI (Thermo, FD1014) for 30 min at 37°C. After digestion, the vector was purified using the JetQuick Genomic DNA Clean up Spin Kit (Genomed) according to the manufacturer’s protocol. Ligation of the annealed oligonucleotides with the BbsI-digested vector was performed with a T7 DNA Ligase (NEB) for 1h at room temperature (RT). To prevent undirected recombination, ligation products were treated with a Plasmid Safe ATP-dependent DNase (Biozym, 161010). Electrocompetent DH5A E. coli cells were transformed with final constructs for plasmid amplification. The final DNA of amplified plasmids was checked by PCR and Sanger sequencing.

Cell line generation.

Before transfection, HUES8 cells were dissociated into single cells using Accutase and 200,000 cells were seeded per well of a 6-well plate. Cells were cultured o/n in FTDA media supplemented with 10 µM ROCK inhibitor. 2 μg DNA (consisting of 0.5 μg of each construct) was transfected via FuGene HD Transfection Reagent (Promega) in OptiMEM medium (Thermo) to one well of 6-well plate for a 4 h incubation. After 1 day, transfected hESCs were selected with 1 µg/mL puromycin for 1 day. Multiple single cell-derived colonies were picked mechanically as soon as small colonies had formed for screening of gene edited clones.

Screening of edited clones: DNA isolation and PCR reaction.

Clonal colonies were mechanically dissociated and one half of the cells was further cultivated and expanded while the other half was used for genotyping. DNA was isolated using the QIAamp DNA Micro Kit (Qiagen), according to manufacturer’s instructions and amplified using a high-fidelity PrimeSTAR GXL DNA Polymerase (Takara). PCR screening was based on a primer pair spanning the target region (fwd: GAGCCCCGCCTCGGTAAATAAC; rev: AGGGTGGACTTCCCGAAAAAGAA). PCR products were sent for Sanger sequencing (GATC Biotech AG Tuebingen) for validation of clonal genotypes (sequencing primer: CTCGGTAAATAACCCAGC). For off-target sequencing, following primer pairs were used: For MGMT, fwd: TCAGGAGCACCCATTTGGTC, and rev: GGCCCCCATTTAAGGGTTCA and for TTC34, fwd: AAACGCACCCACAGACCG, and rev: TGAGGGTGGGGTTTCTGTTC.

Co-Immunoprecipitation experiments.

Construct generation

While expression plasmids of GATA4, FOXA1, FOXA2, PDX1, NKX6-1 with C-terminal Flag- and GFP-tag have been previously generated in a pcDNA3 backbone [13], PPDPF, KCTD12, and MAGED1 expression plasmids with the same tags were cloned in this study. PPDPF constructs were commercially synthesized (Eurofins) into a pEX-A backbone and recloned into our library of pcDNA3, pcDNA3-Flag, and pcDNA3-GFP via EcoRI and NotI restriction enzymes. For MAGED1 and KCTD12, the coding sequence (CDS) was amplified from human cDNA and first TOPO-cloned into pCR XL TOPO and pCR 2.1 TOPO, respectively, using TA cloning kits (Thermo). During PCR amplification, enzyme recognition sites were introduced with following primers and GoTaq Flexi Polymerase (Promega): MAGED1 (NotI fwd: ctgaacGCGGCCGCaaGCTCAGAAAATGGACTGTGGTG; XbaI rev: ctgcacTCTAGAtgcTCACTCAACCCAGAAGAAACCAA); KCTD12 (EcoRI fwd: ctgaacGAATTCGCTCTGGCGGACAGCAC; Xba rev: ctgcacTCTAGATCACTCCCTGCAGAAGACGTAC). For KCTD12 PCR reaction, 3% DMSO was added. Topo cloning was performed according to manufacturer’s recommendation with insert:vector ratios of 2:1 and 5:1 with 50 ng vector and o/n TA reaction at 16°C. For transformation and bacteria amplification, chemical competent DH5a were used for PPDPF and MAGED1, while dam incompetent chemical competent bacteria (JM110) were employed for KCTD12 due to a Dam methylation blocked XbaI site. After topo cloning and validation by sanger sequencing (Eurofins Genomics), CDSs of interest were cloned into target pcDNA3-Flag and pcDNA-GFP vectors via restriction enzyme digestion. Sticky ends were produced for MAGED1 plasmids with NotI-XbaI, for PPDPF and KCTD12 with EcoRI-XbaI and cloned via T4 DNA ligation (NEB) using insert:vector ratios of 3:1 with 100 ng vector input.

Lipofection and protein lysis of HEK293T cells

2.1 x 10⁶ HEK293T cells were seeded in DMEM with 10% FCS and 1x Penicillin/Streptomycin on a 10 cm dish 1 day prior to transfection. For transfection, 10 µg plasmid DNA was transfected with 18 µL Lipofectamine 2000 (Thermo). After 24 h, cells were washed with PBS and harvested with 0.25% Trypsin/EDTA. After one additional PBS washing step, cells were resuspended in 600 µL chaps lysis buffer (0.01 M Chaps + 50 mM Tris-HCL (pH 7.8) + 150 mM NaCl) supplemented with 5 mM NaF + 0.5 mM PMSF + 1x protease inhibitor cocktail (Sigma) and lysed on ice for 1 h.

Immunoprecipitation experiments

After washing agarose-conjugated α-Flag M2 beads (Sigma) 6 times with 0.5% BSA in PBS, beads were centrifuged at 420 x g for 1 min at 4°C and protein extracts were added to 40 µL beads and put on a shaker o/n at 4°C. After incubation, samples were washed 6 times with chaps buffer and resuspended after centrifugation at 20,800 x g for 1 min at 4°C in 1x laemmli buffer. Proteins were detected via Western Blotting as detailed below.

Western Blotting.

For protein extraction of hESCs and differentiated pancreatic cells, cell lysates were generated by incubating cell pellets in RIPA buffer supplemented with 1 mM PMSF, phosphatase inhibitor (1x) and protease inhibitor cocktail (1x) for 30 min on ice and vortexing every 10 min. After 8 min centrifugation at 10,000 x g, supernatant containing the protein fraction was collected. Protein concentration was determined using a Bradford reagent (Bio Rad) and equalized amounts of protein lysates were separated on a 10-12% polyacrylamide gel in SDS-buffer followed by blotting to a methanol-activated Immobilon-P PVDF membrane (Millipore) by using semidry transfer buffer (32 mM glycine (Applichem), 44 mM Tris, and 20% methanol (Sigma)) and the Transblot semidry transfer system (Bio-Rad). Effective protein transfer was confirmed by Ponceau staining (AppliChem) before membrane was blocked with 5% BSA and 0.1% Tween20 (Sigma) in TBS (TBS-T) for at least 1 h at RT. Membranes were incubated with primary antibodies diluted in blocking solution o/n (o/n) at 4°C. After washing three times with TBS-T, incubation with secondary antibody anti-mouse-horseradish peroxidase (HRP) or anti-rabbit-HRP (ECL anti-rabbit or mouse IgG, GE Healthcare) was performed for 1 h at RT. HRP was detected with the SuperSignal West Dura kit (Thermo) on a chemiluminescence imaging fusion SL system (VILBER). The following primary antibodies were used: anti-GFP (Roche, 1181 446 0001, 1:500), anti-Flag M2 (Sigma, F3165, 1:500), anti-PDX1 (R&D, AF2419, 1:500), anti-PPDPF (Atlas Antibodies, HPA040929, 1:150), and anti-Vinculin (Sigma, V9264, 1:1000).

Flow cytometry of surface marker.

At PSC stage, pluripotency marker SSEA4 and TRA1-60 were checked after gene editing. At DE stage, differentiation efficiency was determined by KIT (CD117) and CXCR4 (CD184) surface marker staining. For each measurement, cells from one well of a 24-well plate were harvested with TrypLE, enzymatic reaction was stopped with FC buffer containing 2% FCS in PBS and washed once with FC buffer (200 x g, 5 min). Cells were blocked for at least 20 min on ice with blocking buffer consisting of 10% FCS in PBS and washed again with FC buffer. After resuspension of the cell pellets in 50 µl FC buffer, DE cells were incubated with PE-conjugated CXCR4 antibody (Thermo, MHCXCR404, 1:50) and APC-conjugated KIT antibody (Thermo, CD11705, 1:100) for 45 min on ice. hESCs were incubated with PE-conjugated SSEA4 (BD Bioscience, 560128, 1:10) and FITC-conjugated TRA1-60 (BD Bioscience, 560380, 1:10) for 60 min on ice. Samples were washed and resuspended in FC buffer and filtered using a 50 µm polyamide mesh (Hartenstein). DAPI was added after the last washing step at a concentration of 150 ng/µl to distinguish viable (DAPI^-) and dead (DAPI⁺) cells during analysis.

Flow Cytometry of Annexin V.

For Annexin V staining, cells were harvested as described for surface marker staining and staining was performed according to manufacturer’s instructions of the FITC Annexin V Apoptosis Detection Kit with 7-AAD (Biolegend, 640922). Instead of using the recommended volumes of reagents, applied volumes were halved without changing reaction mix concentrations.

Flow cytometry of intracellular marker.

Differentiation efficiency was analyzed at PE and PP stage by FC-based analysis of intracellular PDX1 and NKX6-1 expression. Endocrine organoids were singularized with TrypLE for around 9-12 min with pipetting up and down in-between and control of proper singularization under the microscope. Planar cells were harvested and singularized as for surface marker staining. Singularized cells were washed with PBS (200 x g, 5 min) and fixed on ice for 25 min in 4% PFA in PBS with 100 mM sucrose (Sigma). Samples were washed twice with PBS and blocked with 5% donkey serum (Jackson ImmunoResearch) in 0.1% Triton-X/PBS for 30 min on ice. After centrifugation (3000 x g, 5 min) and two rounds of washing with 2% donkey serum, 0.1% Triton-X in PBS (wash solution), cells were resuspended in blocking solution with primary antibodies. Incubation was performed o/n at 4°C or 90 min on ice with combination of following antibodies: anti-PDX1 (R&D, AF2419, 1:500), anti-NKX6-1 (DSHB, F55A12, 1:150), anti-Ki67 (Thermo, MA5-14520, 1:1000 or DAKO, M7240, 1:200 or BV421-coupled (BD Horizon, 562899, 1:20)), anti cl-CASP3 (Cell Signaling, 9661, 1:800), GCG (Sigma, G2654, 1:500), C-pep (Cell Signaling, 4593, 1:100), and INS-Alexa647 (Cell Signaling, 3014, 1:80). Next, the samples were washed twice and incubated with Alexa Fluor secondary antibodies (Thermo, 1:500) for 90 min on ice. After one additional washing step, cells were finally resuspended in washing solution and filtered to obtain single cells before measurement. FC measurements were performed on an LSR II (BD) or an AttuneNxT (Thermo) flow cytometer.

Flow cytometry of EdU staining.

EdU staining was performed using the Click-iT EdU Alexa Fluor 647 Assay Kit (Thermo, C10635). EdU at a concentration of 10 μM was added to the cell cultures prior to harvesting. Chosen incubation time was optimized for each cell type: At hPSC stage, EdU was incubated for 45 min, at DE stage for 2 h and at PE and PP stage for 4 h. Cells were harvested, fixated as described in the intracellular flow cytometry section and staining was performed according to manufacturer’s recommendations with minor changes. Instead of using the recommended volumes of reagents, applied volumes were halved without changing reaction mix concentrations. DNA was finally incubated with 1 µg/µL DAPI for around 30min prior to measurements on an AttuneNxT (Thermo) flow cytometer.

Flow cytometry analysis.

Analysis of flow cytometry data including gating was performed using FLowJo version 10.8.1. Briefly, Cells were gated on the area of forward and sideward scatters. Single cells were subsequently defined by the area and width of the forward scatter followed by an additional stringency step using the area and width of the sideward scatter. In case of live cell measurements, DAPI^- or 7AAD^- cells were gated as living cells for subsequent marker analysis (area of specific channels).

Immunocytochemistry staining.

Prior to immunocytochemistry (ICC/IF) in-well staining, stem cells were cultivated and differentiated on Ibidi-precoated glass-bottom IBIDI-24-well plates (IBDI) with additional Matrigel coating as outlined in maintenance and differentiation culture. Cells were washed with PBS, fixed in 4% PFA+100 mM sucrose solution at RT for 20 min, and washed with PBS three times. Quenching was performed with 50 mM NH₄Cl (Sigma) for 10 min and wells were washed three times with PBS before permeabilization with 0.1% Triton-X/PBS was performed for 30 min at 4°C. Cells were blocked with 5% normal goat or donkey serum (Jackson ImmunoResearch) in 0.1% Triton-X/PBS for 45 min and incubated with the primary antibody solution at 4°C o/n. The following antibodies were used: anti-OCT4 (Santa Cruz, sc-5279, 1:100), anti-SOX2 (R&D, MAB2018, 1:300), anti-SOX17 (R&D, AF1924, 1:500), anti-PDX1 (R&D, AF2419, 1:300), and anti-NKX6-1 (DSHB, F55A12 (concentrate), 1:150). On the next day, cells were washed three times with PBS and incubated with secondary Alexa Fluor antibodies (Thermo) 1:500 diluted in blocking solution at RT for 1 h in the dark. After washing with PBS, 500 ng/ml DAPI in PBS was added to the cells for 10 min. Wells were either directly imaged or stored in PBS at 4°C prior to imaging. Cells on IBIDI-plates were imaged on an AxioObserver 7 microscope (Zeiss) equipped with a Axiocam 702 camera (Zeiss) using ZEN blue software (Zeiss).

ICC, IF, and IHC image analysis.

For ICC/IF and IF, brightness and contrast was adjusted in ZEN blue (Zeiss), for IHC in Photoshop (Adobe) according to good scientific practice. Images were then exported as jpeg files and cropped in Adobe Illustrator (Adobe) for final image compilation.

Analysis of the PPDPF interactome.

The SILAC-based analysis of the PPDPF interactome was performed as previously described (PMID: 250s59612). Briefly, HEK293T cells were metabolically labeled with lysine and arginine containing light (K0/R0) (Sigma; control cells) or heavy (K8/R10) carbon, nitrogen or hydrogen isotopes (Silantes) for ten passages to allow complete labeling of proteins. Labeled cells were transfected with N-terminally Strep/FLAG-tagged PPDPF or a vector control by lipophilic transfection via a home-made PEI reagent. To rule out labeling artifacts, labels were also switched between the conditions. The bait protein and associated proteins were purified via StrepTactin resin (IBA) and eluted with Desthiobiotin elution buffer (IBA). The eluates of the heavy and light condition were mixed 1:1 and subjected to mass spectrometric analysis. For this purpose, samples were precipitated by Methanol/Chloroform and cyteines were alcylated by DTT/Iodoacetamide prior to a tryptic proteolysis following standard conditions. After the proteolysis step, samples were desalted by StageTips (Thermo-Fisher) and analysed on a Q-Exactive Orbitrap mass spectrometer (Thermo-Fisher) coupled to a nano-flow high-performance liquid chromatography (HPLC) system (Dionex Ultimate 3000 RSLC, Thermo-Fisher). For separation, 180 min standard gradients were used. To obtain MSMS spectra, the ten most intense precursor peptides were subjected to fragmentation (TOP10 method). Resulting RAW files were analysed by MaxQuant (ver. 1.6.2.10) setting K8/R10 as heavy and K0/R0 as light condition. The search was performed against the human subset of the Swissport/UniprotKB database (release: 2016_09, 23442 entries) using carbamylated cysteine as fixed and methionine oxidation and N-terminal acetylation as variable modifications. The statistical analysis was performed with Perseus (ver. 1.6.7.0).

RNA isolation, reverse transcription and qPCR.

Cells were harvested with TrypLE as outlined above. RNA was extracted from cell pellets with the GeneJET RNA Purification Kit (Thermo) according to manufacturer’s instructions. Reverse transcription of 200-1000 ng of total RNA was performed with the iScript cDNA Synthesis Kit (Bio-Rad) and cDNA was utilized for quantitative real-time PCR (qPCR) with SensiMix SYBR No-ROX Kit (Bioline) on the QuantStudio 3 Real-Time PCR System (Thermo). Following commercial QuantiTect primers (Qiagen) of target genes were employed: Hs_SOX17_1_SG (QT00204099), Hs_CXCR4_2_SG (QT02311841), Hs_GATA6_1_SG (QT00233331), Hs_GATA4_1_SG (QT00031997), Hs_FOXA1_1_SG (QT00212828), Hs_SOX9_1_SG (QT00001498), Hs_PDX1_1_SG (QT00201859), Hs_PROX1_1_SG (QT01006670), Hs_NKX6-1_1_SG (QT00092379), Hs_PTF1A_2_SG (QT01033396), Hs_GP2_1_SG (QT00010535), Hs_CPA1_1_SG (QT00001736), Hs_PPDPF_1_SG (QT00203469). Target genes were normalized against the housekeeping gene Hs_HMBS_1_SG (QT00014462) using 2^-ΔΔCt.

RNA isolation, library preparation and sequencing.

RNA from hPSC, DE, PE and PP stage was isolated as described above. For DE, PE, and PP, RNA was isolated from four independent differentiations (differentiation started on different days). For hPSC control samples, 2 RNA samples were isolated. Quality and concentration were measured on an RNA Pico chip on an Agilent Bioanalyzer. Library preparation and sequencing was performed as detailed in [65]. Briefly, the Sequencing library preparation has been performed with the TrueSeq RNA Sample Prep Kit v2-Set B (RS-122–2002, Illumina Inc, San Diego, CA) producing 275 bp fragments including adapters in average size. Pooled libraries have then been clustered on the cBot Instrument from Illumina using the TruSeq SR Cluster Kit v3—cBot—HS(GD-401–3001, Illumina Inc, San Diego, CA). Sequencing was performed as 50 bp single reads and 7 bases index reads on an Illumina HiSeq2000 instrument using the TruSeq SBS Kit HSv3 (50-cycle) (FC-401–3002, Illumina Inc, San Diego, CA).

Data processing.

RNA-seq reads were aligned to the human genome using STAR Aligner v.2.5.2a [66] with the Ensembl 86 reference genome. Sequenced read quality was checked with FastQC v0.11.2 (http://www.bioinformatics.babraham.ac.uk/projects/fastqc/) and alignment quality metrics were calculated using the RNASeQC v1.18 [67]. Following read alignment, duplication rates of the RNA-seq samples were computed with bamUtil v1.0.11 to mark duplicate reads and the dupRadar v1.4 Bioconductor R package for assessment [68]. Unique read counts were retrieved from the “featureCounts” software package [69]. As quality control measure, Reads Per Kilobase of transcript per Million mapped reads (RPKM) were quantified using Cufflinks software version 2.2.1 [70].

For quality control, total read counts, the rate of ribosomal RNA, the rate of intergenic and intronic reads and distribution of logarithmic RPKM values were assessed for each sample. While all samples except of one were of very good quality with 20-40 million 85 bp single end reads and very low ribosomal, intergenic, and intronic rates, one sample (PPDPF^WT/WT cl. 1 at DE, replicate 3) was excluded from downstream analysis. Following samples were finally analyzed:

1. hESC/hPSC (day 0): 4 genotypes x 2 replicates = 8 samples
2. DE (day 4): 4 genotypes x 4 replicates - 1 sample exclusion = 15 samples
3. PE (day 10): 4 genotypes x 4 replicates = 16 samples
4. PP (day 14): 4 genotypes x 4 replicates = 16 samples

RNA-seq data analysis.

Generation of DEG lists and transformed read counts

Raw read counts were imported to R v4.2.1 and differential expression was analyzed with “DESeq2” [71]. A gene was determined to be differentially expressed if the absolute log2 FC > I1I and the adjusted P-value < 0.01. Read counts were Rlog transformed, before subjected to downstream statistical analysis.

PCA

Prior PCA analysis, Rlog transformed read counts were filtered for the top 10% most variant genes. For the PE and PP subset PCA, Rlog transformed read counts were first filtered for PE and PP samples before filtering for the top 10% most variant genes. Subsequently, PCA analysis was performed using the “prcomp” function and standard settings in R. PCA was visualized using “ggbiplot” function in conform package.

Hierarchical clustering

Sample distances have been calculated from Rlog transformed read counts using the “sampleDists” function with default parameters such as Euclidean distance calculation.

Overrepresentation analysis

Overrepresentation analysis was conducted on pairwise DEG lists using the web browser tool of gProfiler [56].

Venn diagrams and Volcano plots

Venn diagrams were generated using “ggvenn” function and package, while “ggplot” was used for plotting Volcano plots.

Line plots and heatmaps

Line plots and heatmaps illustrating target gene expression were performed on Rlog transformed read counts. The Mean ± SEM was plotted with “ggplot” along the time course of differentiation for line plots. Heatmaps were generated using the packages (“pheatmap”) for plotting and the package (“biomaRt”) for retrieval of gene sets from common databases. The function (“pheatmap”) was used with parameter scale=”row” and clustering_method=”ward.D2” with further parameters set as default.

Gene set enrichment analysis (GSEA)

GSEA was conducted with the desktop version of GSEA version 4.0.3 (Broad Institute) [72]. Run GSEA function (not preranked function) was conducted on the Rlog transformed read count matrix and an additionally prepared data information file. Except of Collapse parameters (“No_collaps”), default settings including 1000 permutations and a weighted statistical analysis were used. For comparison with reference gene sets, gene lists have been either directly compiled from literature or raw data has been reanalyzed. A complete list of applied gene sets can be found in S5 Table.

Generation of reference gene list

Differentiation: For stage-specific gene sets of hPSC-based pancreatic differentiations, two different RNA-seq data sets were utilized. The gene sets for Definitive endoderm_Xie2013, Gute tube_Xie2013, Pancreatic endoderm_Xie2013, Pancreatic progenitor_Xie2013 were directly retrieved from [27]. For Pluripotent stem cell_Breunig2021 upregulated DEGs of the pairwise RNA-seq comparison between hPSC and DE from [23] were extracted, for Definitive endoderm_Breunig2021 the downregulated DEGs of the same comparison and for Pancreatic progenitor_Breunig2021 upregulated DEGs of the comparison PPs versus DE were listed.

Mouse Ptf1a Targets_Thompson2012: Gene sets from an RNA microarray of Ptf1a^KO/KO versus Ptf1a^KO/WT murine pancreatic progenitors could be directly retrieved from [31] (Ptf1a^KO/KO up_Thompson2012 and Ptf1a ^KO/KO down_Thompson2012).

Human Fetal Tissue_Jennings2017: A human fetal gene set was generated from raw FASTQ files (friendly provided by Neil Hanley) [29]. Briefly, FASTQ files have been mapped to the human genome, normalized using batch correction, and significant gene lists have been generated by Limma pairwise comparisons between all investigated organs. All genes found significant in at least on pairwise comparison were subsequently sorted according to their peaked expression to assign organ-specific gene lists (Liver, Extrahepatic biliary duct, Pancreas).

Human Fetal Pancreas_Gonçalves2021: Gene lists were directly retrieved from cluster analysis in supplementary tables of [73] (Unknown1_Gonçalves2021, Erythroblasts_Gonçalves2021, Proliferative_Gonçalves2021, Mesenchyme_Gonçalves2021, Tip_Gonçalves2021, Trunk_Gonçalves2021, Endocrine_Gonçalves2021, Neural_Gonçalves2021).

Human Fetal Tissue_Cao2020: Organ-specific pseudobulk expression profiles were retrieved from S2 Table of [30]. For final gene lists, genes with P-value < 0.01 and log2 FC > I2I were selected (Adrenal_Cao2020, Placenta_Cao2020, Cerebellum_Cao2020, Lungs_Cao2020, Liver_Cao2020Stomach_Cao2020, Intestine_Cao2020, Kidney_Cao2020, Brain_Cao2020, Muscle_Cao2020, Eye_Cao2020, Spleen_Cao2020, Heart_Cao2020, Pancreas_Cao2020, Thymus_Cao2020).

Reanalysis of human fetal and adult pancreas scRNA-seq data.

The processed gene expression matrix and annotation data (including developmental time point, cell type annotation and UMAP coordinates) of the human fetal pancreas dataset [32] were downloaded from OMIX database under accession code OMIX001616 and reanalyzed using seurat v5.1.0 [74]. Counts were normalized using Seurat´s “NormalizeData” function. The expression of selected genes across developmental time points and cell types and within UMAP space was visualized using Seurat´s “VlnPlot” and “FeaturePlot” functions, respectively, and kernel density estimates of gene expression were plotted using “Nebulosa´s plot_density” function.

The adult pancreas scRNA-seq data, downloaded from GSE84133 has been previously reanalyzed as detailed here [23].

ChIP-seq and ATAC-seq analysis.

ChIP-seq and ATAC bigwig files have been downloaded from publicly available data as described in the data availability section. Data has been plotted at the PPDPF locus with the desktop version of Integrative Genomics Viewer (IGV) version 2.6.3 [54].

PheWAS studies.

Association results were extracted from HugeAMP (https://hugeamp.org) [20], cis-eQTLS from (https://eqtlgen.org/cis-eqtls.html) [22], FORGEdb scores from https://forgedb.cancer.gov [55], and SNP frequencies in ethnical cohorts (only listed in S1 Table) from https://www.ncbi.nlm.nih.gov/snp.

Figure preparation.

Adobe Illustrator was used for compilation and arrangement of multipaneled figures and for the generation of schematic views. The in Figs 9A and S6A depicted scheme of an orthotopic transplantation was produced with the help of the generic artificial intelligence function of Adobe Illustrator.

Statistical analysis.

If not stated elsewise, statistical analysis was performed using the GraphPad Prism 8 software and detailed information regarding the different applied tests are indicated in figure legends. In general, data summarize three independent experiments (independently started hPSC differentiations) with each analysis performed in duplicates (two wells per condition), unless otherwise stated. Statistical analyses was performed on the respective means of technical replicates. Statistical significance was defined as follows: * P-value < 0.05, ** P-value < 0.01, *** P-value < 0.001, **** P-value < 0.0001. For differentiation experiments, independently started wells were considered independent experiments, while wells of the same differentiation were treated as technical replicates.