Generation of lentiviral vectors carrying transcription factors

We selected 16 candidate TFs: ESC pluripotency factors (OCT4, SOX2, KLF4, CMYC), TFs expressed in early definitive endoderm (MIXL1, GATA4, SOX17, FOXA1, FOXA2, FOXD3, FOXF1), TFs expressed in late endoderm (HNF4α, HNF6, HNF1α, HEX, CEBPα). Primers were designed with including specific restriction enzyme sites in the flanking regions to amplify the cDNAs (S1 Table). The coding sequence for FOXA2, HNF1α, HNF4α, CEBPα, HNF6, and FOXA1, FOXF1, FOXD3 and for HEX, MIXL1, GATA4, SOX17 were PCR amplified from hESC differentiated to hepatocytes for 12 days, 6 days and 4 days, respectively and cloned into the PLVX-IRES-HYGRO constitutive CMV promoter based lentiviral vector plasmid purchased from (Cat No. 632185, Clontech, Cambridge, USA). Human OCT4, SOX2, KLF4 and CMYC cDNA were excised from pMXs-OCT3/4 (Cat.No. 17217), pMXs-SOX2 (Cat No. 17218), pMXs-KLF4 (Cat No. 17219), pMXs-cMYC (Cat No. 17220) respectively. The cDNAs were cloned into the PLVX lentiviral vector (Addgene, Cambridge, USA). Each TF construct was verified by colony PCR, restriction digestion pattern and cDNA sequence analysis. Each TF containing lentiviral vector was co-transfected with 2nd generation lentiviral packaging (gag pol tat rev)/envelope (VSV-G) plasmids into Lenti-X™ 293T Cell line purchased (Cat No. 632180, Clontech, CA, USA) using Fugene transfection reagent (Cat No. E2311, Promega, Madison, WI, USA). Transfer vector- 6μg, Packaging plasmid (psPAX2) (Cat No. 12260, Addgene, Cambridge, USA)- 3.5μg, Envelope plasmid (pMD2. G) (Cat No. 12259 Addgene, Cambridge, USA)– 1.5μg. Supernatants containing the lentiviral particles were collected after 48hr, filtered through a 0.45μm filter and stored at -80°C for future use.

Culture conditions for human MAPCs

Human Multipotent Adult Progenitors (hMAPCs) were derived from human bone marrow as described in Roobrouck et al.[34] MAPCs were cultured as previously described[34]. This cells were cultured in medium containing of 60% DMEM low-glucose (Gibco, Invitrogen, Carlsbad, CA, USA), 40% MCDB-201 (Sigma-Aldrich’s Louis, MO, USA), supplemented with 50mM dexamethasone, 10⁻⁴ M L-Ascorbic acid, 1x Insulin-Transferrin-Selenium(ITS), 0.5x linoleic acid-bovine serum albumin (LA-BSA)(all from Sigma Aldrich), 1% Penicillin/Streptomycin (Gibco, Invitrogen), along with 2% Serum Supreme (Lonza Biowhittaker, Basel, Switzerland) and human PDGF-BB (R&D Systems, Minneapolis, MN,USA) and human EGF (Sigma-aldrich) (both 10ng/ml). Filter medium through 0.22μm filter (Millipore). Human MAPC-cultures were maintained under hypoxic conditions 37°C and 5% O_2, 5% CO₂ humidified incubator at a density of 400 cells/cm² and were passaged every 2–3 days.

Generation of induced endodermal progenitors (iENDO) from human MAPCs

On day 0, 45,000 hMAPCs cells were plated in 10 cm² petri dishes (Cat No. 150350, Thermo Scientific™ Nunclon™ Delta treated, VWR International, Belgium) in triplicates. On day 1, cells were transduced with a cocktail of 14 or 16 un-concentrated viral vector supernatants (MOI of 3) with infection efficiency (96.57% at 300μl) (S15A and S15B Fig). On day 4, transduced cells were harvested and replated on 1.8% differentiation Matrigel (Cat No. 356231, BD biosciences, Bedford, MA, USA) coated six well plates (Cat No. 3516, Corning-Costar^®, MA, USA). A part of the cells was used to evaluate transgene expression. From day 5 onwards, cells were maintained in endoderm induction medium[14] containing 60% DMEM low glucose (Cat No. 31885023, Gibco, Grand Island, NY, USA); 40% MCDB-201 (Cat.No. M6770, Sigma, Saint Louis, MO, USA); 1x-Penicillin-Streptomycin(10,000 U/mL) (Cat No. 15140122, Gibco, Carlsbad, CA, USA); 0.25x LA-BSA (100x) (Cat No. L9530, Sigma, Saint Louis, MO, USA); 0.25x ITS-A (100x) (Cat No. 51300044, Gibco, Grand Island, NY, USA); 100nM,L-ascorbic Acid (Cat No. A4403, Sigma-Aldrich, Saint Louis, MO, USA); 1μM dexamethasone (Cat.No. D2915, Sigma-Aldrich, Saint Louis, MO, USA), 50 μM, 2-mercaptoethanol (50mM) (Cat No. 31350010, Gibco, Grand Island, NY, USA) supplemented with 100ng/ml Activin-A (Cat No. 120-14E), 50ng/ml Wnt3a (Cat No. 315–20) and 5ng/ml BMP4 (Cat No. 120-05ET). All growth factors were purchased from Peprotech, USA. From day 9 to 15 morphological changes from mesenchymal to epithelial cells were assessed by bright field microscopy. Untransduced and PLVX-eGFP empty vector tranduced hMAPCs cultured under similar conditions were used as negative control. The iENDO cells were maintained in a 37°C, 21% O₂, 5% CO₂ incubator. Between days 20–25 transcripts for endogenous mesendoderm/definitive endoderm and late endoderm marker genes were measured by qRT-PCR. Cells were fixed with 4% paraformaldehyde (PFA) (Cat No. P6148, Sigma-Aldrich, Saint Louis, MO, USA) overnight at 4°C to perform immunostaining.

Expansion potential of iENDO cells

14TF iENDO cells were seeded at one million cells/100 cm² petri dish (Cat No. 150350, Thermo Scientific™ Nunclon™ Delta treated, VWR International, Belgium) coated with 0.1% gelatin. Afterwards every 4–5 days, cells were harvested with 0.25% Trypsin EDTA (Cat No. 25200056, Gibco, Grand island, NY, USA) and enumerated using a NUCLEOCOUNTER® NC-100™. Population doublings (PDs) were enumerated as the number of cells initially seeded (C₀) to the number of cells harvested (C₁) at each passage using the following equation: PDnew = PD initial + [log (C₁/C₀]/log2.

Maintenance and expansion of hESCs

H9 hESCs (purchased from WiCell, Madison, WI, USA) were expanded on a 6 well plate (Cat No. 3516, Corning-Costar^®, VWR International, Belgium), in feeder-free conditions on hESC-qualified BD Matrigel^TM, (Cat No. 354277, BD Biosciences, Bedford, MA, USA) using E8 medium Essential 8^TM Basal Medium Essential 8^TM Supplement (Cat No. A1517001, Gibco, Grand Island, NY, USA).

Transduction efficiency in hMAPC

To determine the transduction efficiency of hMAPC, we serially diluted the PLVX-IRES-eGFP lentiviral vector (0–1000μl) and transduced an equal number of hMAPC (50,000 cells/well of 12 well plate). After 3 days, we quantified eGFP expression by FACScanto flow cytometer (BD Biosciences). Data were analyzed with FlowJo Sofware (Tree Star, Ashland, OR, USA).

iENDO cell transplantation under kidney capsule of immunodeficient BALBC Rag2^-/- γc^-/- mice

Immunodeficient BALBC Rag2^-/- γc^-/- mice were maintained in accordance with protocols approved by the ethics committee of KU Leuven. 14 TF iENDO cells were collected from culture dishes by 0.25% trypsin, and injected under the kidney capsule of 10–12-week-old Rag2^-/- γc^-/- mice (N = 16). After 3 weeks (N = 3) and 3 months (N = 12), mice were sacrificed, the kidneys harvested and washed with 1X PBS 2–5 times. A small part of the graft was collected in lysis buffer for RNA extraction and qRT-PCR analysis, while the remainder was fixed with 4% PFA overnight at 4°C for immunohistochemistry and Hematoxyline and Eosin (H&E) staining. Serum from transplanted and untransplanted mice was collected and stored at -80°C for human albumin and C-peptide analysis.

Hepatocyte organoid differentiation from 14TF-iENDO cells

14TF iENDO cells were allowed to form 3D organoids by plating 25,000-iENDO cells/well in low attachment 96 well U bottom cell repellent plates (Cat. No.650970, Greiner-bio one, Belgium) by spinning at 300g for 5 min in Liver Differentiation medium (LDM) consisting of 60% DMEM low glucose; 40% MCDB-201; 1x-Penicillin-Streptomycin; (10,000 U/mL)); 0.25x LA-BSA (100x); 0.25x ITS-A (100x); 100nM, L-ascorbic Acid; 1μM dexamethasone; 50μM, 2-mercaptoethanol (50mM) supplemented with in addition as follows, for protocol-A, day 0 to 4: 50 ng/ml aFGF (Cat.No. 100-17A, Peprotech, NJ, USA) with or without 0.6% DMSO (Cat No.D2650 Sigma-Aldrich, Saint Louis, MO, USA) and, subsequently, from day 4 to 20 with 20 ng/ml HGF (Cat No.100-39H, Peprotech, NJ, USA) with or without 2% DMSO. For protocol B, organoids were cultured in 20 ng/ml HGF and 2% DMSO from day 0 to 16. Compacted organoids were maintained in a 37°C, 21% O₂, 5% CO₂ incubator, with medium change every 2 days until day 20 and 16 (protocol-A and B respectively). On the last day of differentiation organoids were collected for qRT-PCR. For immunostaining, organoids were fixed with 4% PFA overnight at 4°C and subsequently, stored in 70% ethanol at 4°C until embedding. Differentiation medium was collected and stored at -80°C for albumin ELISA.

Hepatocyte organoid differentiation from hESCs

Differentiation of hESCs to definitive endoderm and hepatocyte-like cells was performed as described previously by our group[40] with addition of 0.6% DMSO between day 0 and day 12 and 2% DMSO between day 12 and day 28. On day 20, cells were washed with 1x PBS (Cat No. 10010015, Gibco, Grand Island, NY, USA), incubated with collagenase-1 (200U/ml, 300μl for 20 minutes) and detached by pipetting. Cells were collected, spun for 5 minutes at 300g and re-suspended in LDM containing 10% FBS (Cat No. F6178, Sigma-Aldrich, Saint Louis, MO, USA), Revitacell™ (1:100), (Cat No. A2644501, Gibco, Grand Island, NY, USA), 20ng/ml HGF and 2% DMSO. Cells were plated in low attachment round bottom plates at 7000 cells/well in 20μl. After 24hours, compact organoids formed, after which 100μl LDM containing HGF (20ng/ml) and 2% DMSO was added. Subsequently, medium was changed every alternate day. On day 30, medium was collected for albumin ELISA and organoids were harvested for gene expression analysis.

Pancreatic organoid differentiation from 14TF-iENDO cells

14TF iENDO cells were allowed to form organoids by plating 25,000-iENDO cells/well in low attachment 96-well U-bottom plates by spinning at 300g for 5 min in pancreatic differentiation medium (PDM) consists of 60% DMEM low glucose; 40% MCDB-201; 1x-Penicillin-Streptomycin; 0.25x LA-BSA; 0.5x ITS-A (100x); 100nM L-ascorbic Acid); 50 μM, 2-mercaptoethanol (50mM) without cytokines. Compacted organoids were formed in 2 days. Subsequently organoids were maintained in pancreatic differentiation basal medium supplemented as described below: Stage-1 (day 0–8): PDM with 1% B27® supplement (50x) (Cat No. 17504044, Gibco, Grand Island, NY, USA), 0.1mM Sodium butyrate (Cat No. B5887, Sigma-Aldrich, Saint Louis, MO, USA), 1μM T3 (3,3′,5-triiodo-L-thyronine sodium salt), (Cat No. 64245, Millipore, Calbiochem, San Diego, CA, USA), 3.5mM D-glucose (Cat.No. G7021 Sigma-Aldrich, Saint Louis, MO, USA), 100ng/ml Noggin (Cat No. 120-10C, Peprotech, USA), 5μM ALK5-II Inhibitor (Cat No. ALX-270-445, Enzo life sciences, NY, USA), 5μM SANT-1 (Cat No. 1974, Tocris Biotechnie, Bristol, UK), 10mM nicotinamide (NA), (Cat No. N0636, Sigma-Aldrich, Saint Louis, MO, USA), and 10μM retinoic acid (RA), (Cat No. R2625, Sigma-Aldrich, Saint Louis, MO, USA); and from day 8–12 also 1μM Trichostatin (TSA) ready-made solution (5mM), (Cat No. T1952, Sigma-Aldrich, Saint Louis, MO, USA). Total glucose concentration in the medium is 10mM. Stage-2 (Day 12–30): RPMI-1640 medium (Cat No. 11875093, Gibco, Grand Island, New York) with 0.05% BSA (Cat No. A9418, Sigma-Aldrich, Saint Louis, MO, USA), 1 μM T3 agonist (Cat No. 64245,Calbiochem, Millipore, CA, USA), 5μM ALK5-II Inhibitor (Cat No.ALX-270-445, Enzo life sciences, NY, USA), 10mM NA (Cat No.N0636, Sigma-aldrich, Saint Louis, MO, USA) and 50ng/ml of IGF-II (Cat No.292-G2, R&D systems, MN, USA). Differentiation medium was changed every other day until day 30 and organoids were maintained in a 37°C, 21% O₂, 5% CO₂ incubator. On day 12 and 30 organoids were collected for qRT-PCR. For immunostaining, organoids were fixed with 4% PFA overnight at 4°C and subsequently, stored in 70% ethanol at 4°C until embedding.

Pancreatic organoid differentiation from hESCs

For hESC differentiation to pancreatic β-cells, we followed the protocol described earlier by our group with minor adaptations [18]. On day 30 cells were harvested for qRT-PCR.

RNA extraction, cDNA synthesis and gene expression

Total RNA was purified using the GenElute^™ Mammalian Total RNA Miniprep Kit (Cat No. RTN350, Sigma-Aldrich, Saint Louis, MO, USA) and ZR RNA MicroPrep (Cat No. R1061, Zymo Research, CA, USA). CDNA was generated using 0.5–1μg of RNA with SuperScript® III First-Strand Synthesis SuperMix for qRT-PCR kit (Cat No. 11752050, Invitrogen, CA, USA) and qRT-PCR was performed on a ViiA™ 7 Real-Time PCR System with 384-well plate (Cat No. 4453536, Applied Biosystems, Carlsbad, CA, USA) with a Platinum® SYBR® Green qPCR SuperMix-UDG w/ROX (Cat No. 11744500, Invitrogen, CA, USA) and primers mix at final concentration of 250nM. Gene expression (Cycle threshold) values were normalized based on the PPIG (peptidylprolyl isomerase G) housekeeping gene and the Delta CT calculated. Gene specific primers, purchased from IDT technologies, Leuven, Belgium were designed to distinguish between transgene (CDS-IRES) and endogenous (CDS-3’UTR or Exon-Exon spanning) gene expression. Gene expression graphs or heat maps were represented relative to the housekeeping gene PPIG in log scale. Heat maps were generated using Gene E software (Broad Institute of MIT and Harvard, Cambridge, MA, USA). The efficiency of primers was tested by serial dilution of cDNA and by calculating coefficient of regression (R2). An efficiency of 95‐105% with an R2≥ 97% was considered as good (see S2–S5 Tables for list of all qRT-PCR primers used in this study).

Cell viability staining

The LIVE/DEAD® Viability/Cytotoxicity Kit (Cat No. L3224, Molecular probes, Eugene, USA) was used according to the manufacturer recommendations. Stained organoids were observed under inverted fluorescent microscope (Axiovert, Zeiss). Live and dead cells appear as green and red, respectively.

Albumin secretion assay by ELISA

Albumin in culture supernatants and blood of grafted mice were measured using the Human Albumin ELISA Quantitation Set (Cat No.E80-129, Bethyl Laboratories, TX, USA) and the ELISA Starter Accessory Kit (Cat No. E101, Bethyl Laboratories, TX, USA) as per manufacturer’s recommendations. Absorbance was measured at 450nm using Victor 3 plate reader (Model No. 1420–041, PerkinElmer, MA, USA). The results obtained were normalized with cell number.

Human C-peptide ELISA assay

Secretion of human C-peptide in serum from mice grafted with iENDO cells was analysed using the ultrasensitive human C-peptide ELISA kit (Cat No. 10-1141-01, Mercodia AB, Uppsala, Sweden) as per manufacturer’s recommendations.

Immunostaining of organoids and in vivo grafts

14TF iENDO-hepatocyte organoids and iENDO-endocrine pancreatic organoids were fixed with 4% paraformaldehyde (PFA) (Cat No. P6148, Sigma-Aldrich, Saint Louis, MO, USA) overnight at 4°C and subsequently washed with 1X PBS and stored in 70% ethanol at 4°C. For embedding, fixed organoids were encapsulated in 2% agarose gels and stored in 70% ethanol at 4°C until embedding. Fixed iENDO kidney grafts and 3D hepatic and pancreatic organoids were processed using the Excelsior tissue processor and paraffin embedded using the HistoStar (Thermo Scientific). Some paraffin embedded sections were stained with Hematoxyline & Eosin staining. For other sections, immunostaining was performed after melting the paraffin and rehydrating the sections, which were then washed with PBS + 0.2% Triton-X-100 (PBST) for 5 minutes. Following antigen retrieval (Dako Target Retrieval Solution (1x)) sections were permeabilised with 0.2% PBST for 5 minutes. Undifferentiated 14TF iENDO cells cultured on coverslips in 12 well plate were fixed with 4% PFA and permeabilised with 0.2% PBST), blocked with 0.2% PBST + 5% normal donkey serum, and stained overnight at 4°C with primary antibodies and respective isotype controls in Dako diluent. Slides were washed with 1xPBST three times. Immune complexes were detected by incubation with a AlexaFluor AF488 (Green) and AF555 (Red) (1:500) coupled to secondary antibodies for 30 min at room temperature. The nuclei were visualized using Hoechst or DAPI (1:2,000). After 3 washes, slides were mounted with prolong Gold mounting medium. The signals were detected with a Nikon Eclipse Ti-S and Axioimager.Z1 microscope (Carl Zeiss). Identical exposure times were used for isotype and specific antibodies (See S6–S8 Tables, for list of primary, secondary, isotype antibodies). (Note: For cell surface protein CXCR4, stained with anti-human CXCR4 conjugated with PE antibody (1:100), cell permeabilisation and secondary antibody step was not followed).

hMAPC cell surface maker analysis by flow cytometry

hMAPC cells were detached with 0.05% trypsin-EDTA and washed with 1xPBS containing 3%FBS. After washing, between 1-2x10⁵ cells were incubated for 20min in the dark at room temperature (RT) with conjugated antibodies or isotype controls (1μg/ml). A list of primary monoclonal antibodies can be found in S9 Table). 7-AAD was used as a marker to exclude dead cells from the analysis. All cells were analyzed on a FACScanto flow cytometer (BD Biosciences) with at least 10,000 recorded events. Data were analyzed with FlowJo Sofware (Tree Star, Ashland, OR, USA).

Statistical analysis

Comparisons between two groups were analysed using an unpaired 2-tailed Student’s t-test. P-values < 0.05(*), < 0.01(**), <0.001(***) were considered significant. Data are shown as mean and error bars represent standard error of mean (SEM) of minimum three independent experiments. All results were analysed using GraphPad prism 6 software.

14TFs can reprogram human MAPCs into induced endodermal progenitor (iENDO) cells

Initially, we transduced human MAPCs with 16 selected TFs, including the pluripotency TFs, OCT4, SOX2, KLF4, CMYC (OKSM) and TFs known to be important in mesendoderm specification (S1A Fig). Cells were maintained in endoderm differentiation medium with Activin-A/Wnt3a/BMP4 on Matrigel coated dishes from day 4 onwards (Fig 1B). On day 4, all TFs were highly expressed except for CEBPα (S1B Fig). Between day 9 and 20, we observed morphological changes from a mesenchymal morphology to clusters of cuboidal cells in the transduced cells, but not in untransduced or PLVX-eGFP transduced cells (S1D Fig).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Generation of 14TF iENDO cells from hMAPC.

A) Selected 14TFs for induction of iENDO cells from hMAPC B) Protocol for iENDO generation from hMAPC. C) Morphological changes of 14TF transduced hMAPC from day 0 to day 20 after transduction (scale bar: 100μm). (D-F) Relative gene expression (to PPIG, log scale) in day 20 14TF iENDO cells represented as a heat-map for mesendoderm (D), definitive endoderm and epithelial marker genes (E) and late endoderm marker genes (F) compared with untransduced hMAPCs, and hESC derived endodermal progenitors (d4) or mature endodermal cells (d20). All data represent mean of three independent experiments. *p<0.05, **p<0.01 and ***p<0.001 determined by unpaired 2-tailed Student’s t-test.

https://doi.org/10.1371/journal.pone.0197046.g001

On day 20 cells were harvested and transcripts for endogenous mesendoderm and as well as definitive and late endodermal marker genes assessed by qRT-PCR and compared with untransduced hMAPC cultured under the same conditions and hESCs differentiated to definitive endoderm cells (day 4) and hepatocyte-/endocrine pancreatic-like cells (day 20). For genes that were part of the pool of transduced TFs, primers were used that only detect endogenous transcripts and not the transduced transcripts (denoted by an “e” in front of the name of the gene (e.g. eGATA6). We found a significant induction of several mesendoderm (S1E Fig), definitive endoderm and epithelial cell marker genes (S1F Fig). Of note, the hepatic endoderm marker genes ALB and AFP (S1G Fig) were also induced, which could be consistent with the known role of FOXA2 and HNF1α, as key regulators of hepatocyte differentiation [23, 41, 42].

Therefore, we repeated the studies without using the FOXA2 and HNF1α viral vectors. Human MAPCs were transduced with 14 TFs (Fig 1A) and maintained as described above (Fig 1B). Between day 9 and 20, we again observed morphological changes from a mesenchymal morphology to clusters of cuboidal cells (Fig 1C). On day 4, all TFs were expressed (S1C Fig). When MAPCs were transduced with only OSKM in endoderm induction medium, clusters of cuboidal cells were observed, which were not expandable and did not express endodermal transcripts (S2A Fig). When OSKM was omitted from the 14TFs no cuboidal cells were observed (S2B Fig).

On day 20, reprogrammed cells were harvested and transcripts for the markers described above assessed by qRT-PCR. We again found significant induction of many mesendoderm, definitive endoderm and epithelial marker genes (Fig 1D and 1E) compared to untransduced MAPC cultured in the same medium. Levels of mesendoderm genes were signifcantly lower than in day 4 ESC endodermal progeny and levels of definitive endoderm genes approached levels of hESC-endoderm committed cells, and as hoped for, expression of more mature hepatic and pancreatic endoderm genes remained well below levels found in ESC progeny allowed to further mature to hepatocytes and pancreatic endoderm. The endogenous pluripotency genes eOCT4 and NANOG were not activated (data not shown). iENDO cells also stained positive for endodermal markers (CXCR4: 84% ± 9%, SOX17: 90% ± 16%; CK18: 85% ± 15%; (Fig 2B–2D) and respective isotype controls (S2C–S2E Fig). Most importantly, the late hepatic endodermal markers, ALB and AFP were expressed but significantly less higher level when compared with 16TF-iENDO cells (Fig 1F), which was confirmed by immunostaining (S2F–S2H Fig), Also, iENDO cells express trangene OCT4 at protein level (S2I Fig). We assessed the purity of hMAPCs by cell surface marker expression using flow cytometry (S3A and S3B Fig). We found hMAPC to be negative for the hematopoietic markers, CD45, CD34, CXCR4, and CKIT; the endothelial marker CD31; as well as the CD140a, CD140b, ALP, HLA-class-II, CD271 markers (S3A Fig); and positive for CD44, CD90, CD105, HLA-class 1, CD73, CD146 (S3B Fig). This expression profile is similar to our previous publication [34]. In contrast, in 14TFs iENDO only minimal expression levels of the hMAPC cell surface markers CD 73 and CD105 were detected (S4A and S4B Fig).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. 14TF iENDO cells are expandable long-term in vitro.

A) Expansion curve of 14TF iENDO cells shown as days in culture (horizontal axis) vs. the number of population doublings (PDs) (vertical axis). B) Immunostaining analysis of CXCR4 (C) SOX17 and CK18 (D) and protein expression of 14TF iENDO cells, at passage 4 (magnification 20x, scale bar: 50μm and 40x, scale bar: 100μm), representative sample of n = 3. (E-G) Relative gene expression (to PPIG, log scale) in 14TF iENDO cells at 6–8 and 15–20 passages represented as a heat-map for mesendoderm (E), definitive endoderm and epithelial marker genes (F) and late endoderm marker genes (G) compared with day 20 iENDO cells. H) Relative gene expression (to PPIG, log scale) represented as a heat-map of transgene expression in 14TF iENDO cells on day 20, P6-P8, and P15-P20. All data represent mean of three independent experiments. *p<0.05, **p<0.01 and ***p<0.001 determined by unpaired 2-tailed Student’s t-test.

https://doi.org/10.1371/journal.pone.0197046.g002

14TF iENDO cells can be expanded for at least 20 passages without significant loss of their endodermal progenitor phenotype

The rationale for making iENDO cells, rather than direct reprogramming of somatic cells to mature endodermal cells, was to create an expandable population of cells. Therefore 14TF iENDO cells were passaged in endoderm induction medium with 50ng/ml Activin-A and 25 ng/ml Wnt3a for 6 to 8 passages on Matrigel-coated dishes. In addition we expanded the iENDO cells from passage 6/8 to 20 in a less complex and less expensive medium, namely hMAPC medium on 0.1% gelatin-coated dishes (Fig 2A). We reevaluated expression of the mesendoderm, definitive endoderm, epithelial and late endoderm markers after 6–8 and 15–20 passages. Both after 6–8 and 20 passages, most endogenous transcripts remained unchanged compared to day 20 iENDO cells, even if the definitive endoderm transcripts, SOX17, eFOXA2, CXCR4, CKIT, CXCR4, and eFOXA1, as well as the hepatoblast transcript HNF6 were increased, while eHNF4a decreased (Fig 2E–2G). Importantly, more mature pancreatic and hepatocyte transcripts remained very low. In addition, transcript levels of all transgenes were significantly decreased in 6–8 and 20 passage iENDO cells (Fig 2H). These studies demonstrated that it is possible to create a population of endodermal progenitor cells from hMAPCs that can be expanded at least 20 passages, using a relatively non-expensive medium with a stable transcriptional profile. Genomic stability of early (P4) and late passage (P20) 14TFs iENDO shown that there is no significant genome wide aberrations observed (S5A and S5B Fig).

Grafting of 14TF iENDO cells in vivo

To test their in vivo differentiation capacity, we transplanted one million 14TF ENDO cells (passage 6) under the kidney capsule of BALBC Rag2^-/- γc^-/- mice in 30μl of PBS (Fig 3A). After grafting, the weight and glucose levels of the mice were regularly assessed. Three mice were sacrificed after 3 weeks (N = 3) and the other (N = 12) were maintained untill 3 months after grafting. Unfortunately, of the last group, one mouse died in the 12^th week. In all mice, we found that a tumor was formed surrounding the kidney. The tumor was larger at 3 months after transplantation (Fig 3B).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. In vivo transplantation of 14TF iENDO cells under the kidney capsule forms endodermal tumors but that differentiate into hepatoblasts and pancreatic endocrine like cells.

A) Schematic representation of transplantation of undifferentiated iENDO cells under the kidney capsule of RAG2^-/- γc^-/- mice. B) Morphology of control kidney (half portion shown) and iENDO cell grafted kidney after three weeks (N = 3) and after three months (N = 12). C) H&E stained sections of iENDO graft (10x and 20x magnification) 3 weeks after transplantation (scale bar 50μm) (N = 3). D) H&E stained sections of iENDO graft (20x magnification) 3 months after transplantation (mouse 1 and mouse 2)(scale bar 100μm) Representative for (N = 12) mice. E) Human albumin quantification by ELISA from iENDO cell grafted mice 3 weeks (N = 3) and 3 months (N = 12) after transplantation, untransplanted mouse serum and human blood serum. F) Human C-peptide by ELISA (untransplanted mouse serum, adult human blood serum and 14TF iENDO grafted mouse 3 months after transplantation (N = 12). *p<0.05, **p<0.01 and ***p<0.001 determined by unpaired 2-tailed Student’s t-test. ND- Not determined, NS- Not significant.

https://doi.org/10.1371/journal.pone.0197046.g003

Graft sections were stained with Hematoxylin and Eosin to assess the graft morphology. None of the grafts had invaded into the kidney. In general, at 3 weeks (N = 3) and at 3 months (N = 12), 14TF iENDO cells-derived tumor cells were homogenous containing endodermal cells, akin to a hepatocellular carcinoma (Fig 3C and 3D). However, the graft of the 12 mice sacrificed at 3 months, showed a more heterogeneous morphology, even if most cells still appeared endodermal (Fig 3D).

We immunostained the grafts for the hepatocyte markers AFP, ALB, AAT, HNF4α and the transgene OCT4. Most cells stained positive for OCT4 after 3 weeks (S6A–S6E Fig) (N = 3) and after 3 months (N = 7), cells in the grafts stained positive for the liver markers AFP, ALB and HNF4α, but fewer for AAT (S7A–S7D Fig). In addition, many cells stained postive for OCT4 (S7E Fig). Cells in the graft also stained positive for the intestinal brush-border microvilli marker, VILLIN-C, and the Paneth cell marker, LYSOZYME (S8A Fig), while fewer cells stained positive for the pancreatic endoderm marker PDX1 (S8B Fig). One of the 3-month grafts was morphologically more heterogeneous, and contained fewer AFP, ALB and HNF4α positive cells (S9A–S9D Fig). However, clusters of cells stained positive for the pancreatic endoderm markers PDX1 (S9E Fig and S10A Fig) and NGN3 (S10A Fig), and very few also for SST (S10B Fig). Insulin- or glucagon- positive cells were not detected. In addition, cells in the graft stained positive for the neuroendocrine marker CHGA (S10C Fig), the paneth cell marker, LYSOZYME (S10D Fig) and the intestinal brush-border microvilli marker, VILLIN-C (S10E Fig). Also, many of the cells still stained positive for OCT4 (S10F Fig). In addition, mesodermal lineage markers, as an example the cardiac specification marker NKX2.5 and ectodermal lineage markers, as an example the neuronal marker β-TUBULIN-III were not observed by immunostaining (S11A and S11B Fig) or by qRT-PCR (S11C and S11D Fig). Control, non-grafted kidney sections did not stain positive for any of these markers.

We also performed qRT-PCR on grafts harvested at 3 weeks (N = 3) and at 3 months (N = 9). At 3 weeks transcript levels for ALB, AFP, AAT and TTR were significantly higher in the graft compared with the initial iENDO cell population, even if mature hepatocyte markers G6PC, NTCP, MRP2 and CYP3A4 were not significantly increased (S12A and S12B Fig). Likewise, after 3 months, transcript levels for ALB, AFP and TTR, but not AAT, G6PC, NTCP, MRP2 and CYP3A4 were significantly increased in the graft compared with the initial cell population (S12A and S12B Fig). Consistent with the immunostaining results, we also detected increased levels of the pancreatic endoderm transcripts PDX1, NGN3, NKX6.1, NEUROD1, PAX4, SST, but not INS or GCG; as well as the gut markers, HNF1β and KRT19, CDX2 in mouse 2 (S12C–S12E Fig). Transcripts for the transgenes were similar before and after grafting, both at 3 weeks and at 3 months (S12F Fig).

We analyzed the mouse serum for presence of human albumin. After 3 weeks, human albumin levels were 1,189 ± 615.44 ng/ml (N = 4), and after 3 months 5,777 ± 7,928 ng/ml (N = 12). In comparison, human blood serum albumine levels are 265,499 ± 12,5354 ng/ml (Fig 3E), while human albumin levels in untransplanted mice were 377± 3.28ng/ml. The levels of human albumin depicted are the levels measured minus the levels found in untranplanted mouse serum. Negligible levels of human C-peptide (2.735 ± 0.06 pmol/L) were detected in the blood of (mouse 1&2), 3 months after grafting (Fig 3F), while no c-peptide was detected in the blood of mice 3 months after grafting. c-Peptide levels in human serum were 518.63 ± 584.62 pmol/L.

These studies demonstrate thus that 14TF iENDO cells can differentiate in vivo to multiple endodermal lineages, including hepatoblast, pancreatic endoderm, as well as gut endoderm like cells.

14TF iENDO cells can differentiate into hepatocyte-like cells

As spontaneous differentiation occurred from iENDO cells to cells with hepatoblast features in vivo, we assessed if iENDO cells could be directly differentiated in vitro to cells with mature hepatocyte features, using methods developed in the lab for hPSCs[40]. As iENDO cells have an endodermal phenotype, we started differentiation from day 8 (hepatic endoderm specification) or day 12 (hepatoblast maturation) of the hPSC protocol onwards (protocol A vs. B, respectively (Fig 4A and S13A Fig). To determine if there are morphological differences between iENDO derived hepatocyte organoids cultured following protocol-A (day 20) or following protocol-B (day 16) and cultured with or without DMSO, pictures were taken at the end of each protocol. Not many differences were observed between organoids of the two protocols. However, organoids cultured with DMSO appeared smaller than the spheroids cultured without DMSO (Fig 4B and S13B Fig). No DMSO condition cultured organoid core seems to have dead cells compare to the DMSO by H&E Staining (Fig 4B and S13B Fig). An additional variable was addition of DMSO, as others[11, 15, 43, 44] and we have demonstrated that this improved the overall differentiation of PSCs to hepatocyte-like cells (HLCs)[40]. Initial studies demonstrated that the highest transcript levels for ALB, AFP, AAT, MRP2, CYP3A4, TTR, G6PC, eHNF4α (Fig 4C) were detected when cells were differentiated using protocol A with addition of DMSO compare to the protocol-B (S13C Fig). Therefore, we chose protocol-A + DMSO for further evaluation and comparison with ESCs differentiated as 3D organoids.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. 14TF iENDO cells differentiate into hepatocyte like cells.

A) Protocol time-line for hepatocyte differentiation in 3D from iENDO cells protocol A with or without DMSO. B) Morphology of iENDO differentiated organoids at day 20 with and without DMSO by bright field and H&E staining (N = 3). C) Relative gene expression (to PPIG, log scale) in day 20 iENDO-HLCs ± DMSO represented as a heat-map for hepatocyte markers compared with hESC-HLCs organoids (d30) and PHHs (N = 3). D-F) Immunostaining for AFP, ALB and AAT on day 20 iENDO organoids ± DMSO with respective isotype controls. Scale bar 25 μm (representative example of N = 3). G) Albumin secretion in supernatants of day 20 iENDO organoids ± DMSO, hESC day 30 organoids and primary hepatocytes (PHH) (N = 3). Error bars represents standard deviation of three independent experiments. *p<0.05, **p<0.01 and ***p<0.001 determined by unpaired 2-tailed Student’s t-test. NS- Not significant.

https://doi.org/10.1371/journal.pone.0197046.g004

Expression levels of ALB, AFP, AAT, MRP2 and TTR transcripts were ± 1 log lower than d20 differentiated iENDO compared with day 30 3D ESC-HLCs (differentiated as described[45], until day 20 and as organoids until day 30)(Fig 4C). Transcripts for eHNF4α, and eHNF6, CYP3A4 and G6PC in 3D iENDO differentiated organoids were comparable with 3D-ESC-HLCs (Fig 4C). Addition of DMSO to either iENDO or hESC differentiation improved differentiation (Fig 4C). Pancreatic endocrine gene transcript levels were not induced under hepatic differentiation conditions (S14A and S14B Fig). When compared to primary human hepatocytes (PHHs), 3D iENDO and 3D hESC hepatocyte-like cells continued to express AFP, while transcripts for mature hepatocytes were significantly lower compared with PHHs (Fig 4C). Transgenic tOCT4 transcripts remained detectable in iENDO-HLCs, albeit at decreased levels compared to undifferentiated iENDO cells (Fig 4C). As negative control, iENDO cells were cultured as organoids without growth factors and small molecules (Fig 4C and S13C Fig), which did not induce increased expression of hepatocyte transcripts.

Immunostaining studies revealed that 3D-iENDO-HLC organoids, differentiated with DMSO, contained 67.86% ± 8.66 AFP⁺, 65.32% ± 4.52 ALB⁺, and 13.94% ± 0.644 AAT⁺ cells. The percentage AFP⁺ and ALB⁺ cells was significantly higher than in 3D organoids differentiated without DMSO and levels were comparable with hESC-HLC 3D organoids (Fig 4D–4F).

To study the function of iENDO-derived HLCs, we measured albumin secretion by ELISA: supernatants of 3D iENDO-HLCs contained 78.737 ± 8.70 ng/ml ALB (n = 3) in the presence of DMSO and 104.5 ± 61.68 ng/ml without DMSO (n = 3), while supernatants of 3D ESC HLCs cultured with DMSO contained 149.17 ± 9.93 ng/ml ALB (n = 4). Thus levels were similar for 3D iENDO and 3D hESC-HLCs, but significantly lower than in PHHs(Fig 4G).

Thus, consistent with the in vivo data, iENDO cells can be guided towards HLCs in vitro even though the differentiated cells are less mature than PHHs.

14TF iENDO cells can differentiate into pancreatic endocrine cells

We also assessed if 14TF iENDO cells could be differentiated towards endocrine pancreatic cells. 14TF-iENDO cells (passage 6–12) were cultured as organoids with combinations of growth factors and small molecule epigenetic modifiers known to induce endocrine pancreas differentiation as described in (Fig 5A)[18, 21, 22]. We observed compact spheroids as early as 2 days after plating the cells. The morphology of spheroids on day 12 and day 30 of the protocol is shown (Fig 5B). There was not much difference observed between spheroids on day 12 and day 30, except the size of the organoid which was smaller at day 30 than day 12. On day 12, Live/Dead and H&E staining demonstrated that cells in the organoids were viable (Fig 5C and 5D). In parallel, hESCs differentiated into pancreatic endocrine cells by using the protocol described in our group previously[18].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. Generation of pancreatic endoderm and endocrine progenitor-like cells from iENDO.

A) Protocol time-line for pancreatic endocrine differentiation from iENDO cells. B) Morphology of 3D aggregate at day 12 and day 30 of differentiation, 10x magnification (scale bar 50μm). C) Live/Dead staining of 3D aggregates at day 12 of differentiation (Live-green, Red-dead) (scale bar 50μm). D) H&E staining of day 12 iENDO pancreatic organoids. E)Relative gene expression (to PPIG, log scale) analysis represented in a heat-map of pancreatic endocrine marker 14TFs iENDO day 0, day 12 iENDO pancreatic organoids, day 30 3D iENDO pancreatic organoids and day 30 hESC pancreatic organoids compared with primary human Islets. All data represent mean of three independent experiments. F) Immunostaining for PDX1 (20X) in 293x transduced PDX1 (top) and day 12 iENDO pancreatic organoids (bottom) (scale bar 100μm, representative of N = 3) with respective isotype controls. F) Immunostaining for NGN3 (20X) in day 12 iENDO pancreatic organoids with respective isotype controls (scale bar 50μm, representative of N = 3). *p<0.05, **p<0.01 and ***p<0.001 determined by unpaired 2-tailed Student’s t-test.

https://doi.org/10.1371/journal.pone.0197046.g005

On day 12 (stage 1) and day 30 (stage 2) the expression of pancreatic marker genes was assessed by qRT-PCR (Fig 5E), and compared with undifferentiated 14TF-iENDO cells (day 0), hPSCs differentiated to pancreatic endocrine cells at day 30, and primary human islets. We detected a significant increase in gene expression of PTF1A, PDX1, NGN3, NKX6.1, NEUROD1, NKX2.2, PAX4, MAFA, MAFB, EHNF6 and SST but not for INS and GCG, nor NKX6.1 (master regulator β-cells) or ARX (master regulator α-cells) (N = 5) (Fig 5E). When compared with hPSC progeny, transcript levels in day 30 iENDO progeny were similar to those of hPSC progeny, except that INS and GCG, ARX, PAX6 expression was significantly higher in hPSC progeny. When compared with primary islets, INS, GCG and SST expression in both iENDO and hESC progeny was significantly lower, whereas endocrine pancreatic genes, including NGN3, were significantly higher expressed in iENDO and ESC progeny (Fig 5E). Also, hepatocyte marker transcripts were not induced in iENDO derived pancreatic progeny (S14C and S14D Fig). Following differentiation of iENDO cells, the OCT4 transgene remained expressed. When iENDO cells were cultured without growth factors and small molecules as organoids, pancreatic transcripts were not induced (Fig 5E).

Day 12 organoids were encapsulated in 2% agarose gel and paraffin embedded sections immunostained for the pancreatic endoderm markers. PDX1 and NGN3 were expressed (Fig 5F and 5G lower image).

Thus, consistent with the in vivo data, iENDO cells can be guided towards endocrine pancreatic cells in vitro, even though not to mature β- or α-cells.