EMBRYONIC STEM CELLS/INDUCED PLURIPOTENT STEM CELLS

Transcription

1 EMBRYONIC STEM CELLS/INDUCED PLURIPOTENT STEM CELLS Biphasic Induction of Pdx1 in Mouse and Human Embryonic Stem Cells Can Mimic Development of Pancreatic -Cells ANDREIA S. BERNARDO, a CANDY H.-H. CHO, b SHARON MASON, a HILARY M. DOCHERTY, a ROGER A. PEDERSEN, b LUDOVIC VALLIER, b KEVIN DOCHERTY a a School of Medical Sciences, Institute of Medical Sciences, University of Aberdeen, Aberdeen, United Kingdom; b Laboratory for Regenerative Medicine, Department of Surgery, University of Cambridge, Cambridge, United Kingdom Key Words. Diabetes mellitus Stem cells Pancreatic differentiation Insulin gene Disclosure of potential conflicts of interest is found at the end of this article. ABSTRACT Embryonic stem (ES) cells represent a possible source of islet tissue for the treatment of diabetes. Achieving this goal will require a detailed understanding of how the transcription factor cascade initiated by the homeodomain transcription factor Pdx1 culminates in pancreatic -cell development. Here we describe a genetic approach that enables fine control of Pdx1 transcriptional activity during endoderm differentiation of mouse and human ES cell. By activating an exogenous Pdx1VP16 protein in populations of cells enriched in definitive endoderm we show a distinct lineagedependent requirement for this transcription factor s activity. Mimicking the natural biphasic pattern of Pdx1 expression was necessary to induce an endocrine pancreaslike cell phenotype, in which 30% of the cells were -celllike. Cell markers consistent with the different -cell differentiation stages appeared in a sequential order following the natural pattern of pancreatic development. Furthermore, in mouse ES-derived cultures the differentiated -like cells secreted C-peptide (insulin) in response to KCl and 3-isobutyl-1-methylxanthine, suggesting that following a natural path of development in vitro represents the best approach to generate functional pancreatic cells. Together these results reveal for the first time a significant effect of the timed expression of Pdx1 on the non- -cells in the developing endocrine pancreas. Collectively, we show that this method of in vitro differentiation provides a template for inducing and studying ES cell differentiation into insulin-secreting cells. STEM CELLS 2009;27: INTRODUCTION A great deal of our knowledge of pancreatic development comes from studies in the mouse [1, 2]. Gastrulation occurs at embryonic day 6.5 (E6.5) in the mouse embryo, thereby inducing pluripotent cells of the epiblast to differentiate into the three primary germ layers, namely mesoderm, ectoderm, and endoderm. During this process cells migrate through the primitive streak before forming mesoderm and endoderm, seemingly by way of an intermediate bipotential population (mesendoderm) [3]. The definitive endoderm subsequently gives rise to the lungs, alimentary tract, and the visceral organs including the liver and pancreas. At E8.5, signals from the adjacent ectodermal and mesodermal tissue induce patterning of the forward region of the gut tube, ultimately resulting in the formation of the dorsal and ventral pancreatic buds. The buds expand to form a branching network of epithelial cells, and at around E15.5 the two branches of the pancreas fuse [4]. During this period of expansion, the major cell types of the pancreas form, that is, the exocrine cells, the cells of the pancreatic ducts, and the cells of the endocrine pancreas, which include (insulin), (glucagon), (glucagon), and pancreatic polypeptide (PP) cells. Toward birth, the endocrine cells delaminate from the epithelial network and aggregate to form islets of Langerhans, which then acquire the ability to sense changes in glucose (and other nutrient) levels and to secrete insulin in a regulated pulsatile manner. The development of the various cell lineages of the pancreas is regulated by the temporal and spatial expression of transcription factors that are in turn regulated by cell signaling molecules emanating from adjacent epithelial cells or from the surrounding mesenchyme [5]. Of particular importance is the homeodomain protein Pdx1. Inactivation of the Pdx1 gene leads to agenesis of the pancreas in mice [6, 7] and humans [8]. Pdx1 is initially expressed at around embryonic E8.5 in a narrow band of foregut endoderm that later develops to form the pancreatic buds. It is also present in the dorsal and ventral buds as they form at around E9.5, and it is then expressed throughout the expanding ductal tree up to E13.5. At this point, the basic helix-loop-helix factor Ngn3, under the influence of Delta Notch signaling, determines which cells are to adopt an endocrine fate [9, 10]. The levels of Pdx1 decrease at this stage before reappearing in a second wave of expression in endocrine cells that are destined to become -cells. In the adult pancreas Pdx1 is expressed in -cells, where it regulates expression of a number of -cell genes, including insulin, islet amyloid polypeptide (IAPP), Author contributions: A.S.B.: collection and assembly of data, manuscript writing; C.H.-H.C., S.M., and H.M.D.: collection and assembly of data; R.A.P., L.V., and K.D.: conception and design, manuscript writing. Correspondence: Kevin Docherty, Ph.D., School of Medical Sciences, University of Aberdeen, Institute of Medical Sciences, Foresterhill, Aberdeen, AB25 2ZD, United Kingdom. Telephone: ; Fax: ; k.docherty@abdn.ac.uk Received March 26, 2008; accepted for publication November 11, 2008; first published online in STEM CELLS EXPRESS December 4, AlphaMed Press /2009/$30.00/0 doi: /stemcells STEM CELLS 2009;27:

2 342 Modeling Pancreatic Development in ES Cells NKx6.1, glucokinase, and the glucose transporter (GLUT2). It is also expressed in a subpopulation of islet -cells and in rare endocrine cells of the small intestine [11]. Thus, although Pdx1 is known to play a major role in -cell development in vivo, the importance of each phase of its expression for in vitro -cell development remains unclear. The aim of the present study was to assess the extent to which the development of the endocrine pancreas could be recapitulated in vitro, using mouse and human embryonic stem (ES) cells, by controlling the timing of Pdx1 activity. The strategy involved inducing formation of definitive endoderm and then activating an exogenous Pdx1VP16 transgene in a temporally phased manner. Activation of Pdx1VP16 in an endoderm-enriched background induced a pancreatic phenotype. Varying the timing and duration of Pdx1VP16 led to different pancreatic outcomes, indicating a context-dependent effect of Pdx1 activity. Collectively, the data show that the differentiating cells go through various states of competence, each of which has its own transcriptional fingerprint that is more or less permissive for the formation of endocrine or exocrine pancreas by Pdx1 activation. We further show that biphasic induction of Pdx1VP16, which mimicked the expression of Pdx1 in the developing mouse pancreas, was the most effective in inducing a -cell phenotype. MATERIALS AND METHODS Reagents Leukemia inhibitory factor (LIF) was purified from Escherichia coli cells transformed with the plasmid pgex 2-TLIF. Activin A, basic fibroblast growth factor, and BMP-4 were purchased from R&D Systems Inc. (Minneapolis, LY was from Promega (Madison, WI, and growth factor-reduced Matrigel was from Invitrogen (Carlsbad, CA, Cell Culture The mouse CGR8 ES cell line was cultured on gelatin-coated tissue culture plates in ES medium that comprised KO-Dulbecco s modified Eagle s medium (DMEM) supplemented with 15% (vol/vol) KnockOut Serum Replacement, 1 nonessential amino acids, 1 L-glutamine, 1% -mercaptoethanol (all from Invitrogen), and LIF. Differentiation of mouse embryonic stem cells was carried out in chemically defined medium (CDM) [12] in suspension cultures for 6 days to promote embryonic body (EB) formation. Cells were further differentiated for 21 days as outgrowth cultures by plating the EBs on tissue culture plates coated with growth factor reduced Matrigel in ES medium (described above) in the absence of LIF. For genetic modification the cells (approximately 10 7 ) were stably transfected with ptp6pdx1vp16er T2 (40 g) by electroporation using a Gene Pulser Transfection Apparatus (Bio-Rad, Hercules, CA, at 230 V and 500 F in ES medium lacking serum. The transfected cells were selected in 1.25 g/ml puromycin (Sigma-Aldrich, St. Louis, com). Eight individual Pdx1VP16ER T2 expressing clone lines were randomly chosen and expanded for further analysis. One clonal line, CGR8C10, was used in the experiments reported here. The human embryonic stem cell (hesc) H9 (WiCell Research Institute, Madison, WI, was grown in CDM supplemented with activin and fibroblast growth factor (FGF) as previously described [13]. For endoderm differentiation, hescs were grown for 3 days in CDM supplemented with activin A (100 ng/ml), FGF2 (20 ng/ml), BMP-4 (10 ng/ml), and the phosphoinositide 3 (PI3)-kinase inhibitor LY (10 M). The cells were stably transfected using Lipofectamine 2000 (Invitrogen) [14, 15]. Ten individual Pdx1VP16ER T2 expressing and three green fluorescent protein (GFP)-expressing clone lines were randomly chosen and then expanded for further analysis. One Pdx1VP16ER T2 clone line, H9C7, and one GFP clone line, H9GFP, were used in the experiments reported here. HeLa cells were cultured in DMEM and transfected using Lipofectamine Reagent and Plus Reagent (Invitrogen) [16]. Plasmid Construction The plasmid ptp6pdx1vp16er T2 was generated by joining the Pdx1VP16 sequences from puc18pdx1vp16 [16] with the ER T2 sequence [17] from the plasmid the ptp6creer T2 [18]. Luciferase Assays HeLa cells were cotransfected with the insulin promoter plasmid phins280-luc, ptp6pdx1vp16er T2, and the control plasmid phrl-tk (Promega), and assays were performed as previously described [19]. Western Blot Western blots were performed as previously described [19] with rabbit anti-pdx1 antibody (1:5,000) (provided by C. V. Wright, Vanderbilt University, Nashville, TN) and horseradish peroxidaseconjugated anti-rabbit antibody (1:5,000). Immunocytochemistry Cells were washed twice in phosphate-buffered saline (PBS) and then fixed in 4% (wt/vol) paraformaldehyde, permeabilized with 0.1% (vol/vol) Triton X-100, and blocked with 5% fetal calf serum and 5% serum from the species in which the secondary antibody was raised. The antibodies used in the study included rabbit anti- Pdx1 at 1:400 (from C.V. Wright), rabbit anti-glucagon at 1:300 (Chemicon, Temecula, CA, goat anti- C-peptide at 1:100 (Linco Research, St. Charles, MO, lincoresearch.com), rabbit anti-ngn3 at 1:100 (Santa Cruz Biotechnology Inc., Santa Cruz, CA, goat anti-ptf1a at 1:100 (Everest Biotech, Upper Keyford, Oxfordshire, United Kingdom, rabbit anti-pax6 at 1:200 (Covance, Princeton, NJ, and goat anti-sox17 at 1:100 (R&D Systems). Immunohistochemistry EBs were washed twice in PBS and then fixed in 4% (wt/vol) paraformaldehyde, infiltrated in 50% (wt/vol) Tissue-Tek Compound (Sakura Finetek, Torrance, CA, and snap-frozen in a bath of dry ice and isopropanol. Cryosections were obtained and staining done using rabbit anti-brachyury at 1:150 (Santa Cruz Biotechnology) and goat anti-sox17 at 1:50 (R&D Systems). RNA Extraction and Polymerase Chain Reaction For experiments involving mouse ESCs, RNA was extracted using TRIzol reagent (Invitrogen). One microgram of total RNA was then used to synthesize cdna following a 15-minute digestion with DNase (Invitrogen). Quantitative polymerase chain reactions (QPCR) were done using the TaqMan (Applied Biosystems, Foster City, CA, polymerase chain reaction (PCR) probe sets (supporting information Table 1). For experiments involving hescs and differentiated progenitors, total RNAs were extracted using the RNeasy Mini Kit (Qiagen, Hilden, Germany, Each sample was treated with RNase-free DNase (Qiagen), and 0.6 g of total RNA was reversetranscribed using Superscript II Reverse Transcriptase (Invitrogen). Real-time PCR mixtures were prepared as described (SensiMiX protocol; Quantace, London, United Kingdom, quantace.com) and then denatured at 94 C for 5 minutes and cycled at 94 C for 30 seconds, 60 C for 30 seconds, and 72 C for 30 seconds followed by final extension at 72 C for 10 minutes after completion of 40 cycles. Real-time PCRs were performed using a Stratagene (La Jolla, CA, Mx3005P in triplicate and normalized to PBGD in the same run.

3 Bernardo, Cho, Mason et al. 343 Figure 1. Optimization of the inducible expression system in HeLa cells. (A): HeLa cells were transiently transfected with the ptp6pdx1vp16er T2 plasmid. Forty-eight hours later the cells were treated with 1 M 4OHT for the indicated times and analyzed by standard fluorescence microscopy using an anti-pdx1 antibody (green). Nuclei were visualized using 4,6-diamidino-2-phenylindole staining (blue) (B): Alternatively, the cells were harvested, fractionated into cytosolic and nuclear fractions, and analyzed by Western blotting using an anti-pdx1 antibody. (C): HeLa cells were cotransfected with the human insulin promoter plasmid and ptp6pdx1vp16 or ptp6pdx1vp16er T2 and treated with 4OHT as indicated. Data are expressed as the firefly luciferase activity relative to that in the cells transfected with the empty expression plasmid (control) and represent the average SD (n 3). Abbreviation: 4OHT, 4-hydroxytamoxifen. RESULTS To model endocrine pancreas development during ES cell differentiation we developed an inducible and constitutively active form of the Pdx1 transcription factor by fusing the coding sequence of Pdx1 to the transactivator domain of the VP16 protein from herpes simplex, and to the mutated estrogen receptor ligand-binding domain (ER T2 ) (supporting information Fig. 1). The resulting protein was transcriptional active only after addition in the culture media of the artificial molecule 4-hydroxytamoxifen (4OHT). To confirm that the Pdx1VP16ER T2 was functional, the ptp6pdx1vp16er T2 plasmid was transiently transfected into HeLa cells in the presence or absence of 4OHT. Immunocytochemistry and Western blot analyses showed that the Pdx1VP16ER T2 protein was localized in the cytoplasm (Fig. 1A, 1B) in the absence of 4HOT. When 4OHT was added to the media the Pdx1VP16ER T2 protein could be detected in the nucleus within 30 minutes, and translocation was almost complete by 4 hours. To test whether the chimeric protein was transcriptionally active, the ptp6pdx1vp16er T2 plasmid was transfected into HeLa cells along with a human insulin promoter construct that is responsive to Pdx1VP16 [16]. The Pdx1VP16 construct stimulated promoter activity by 3.2- fold. In the absence of 4OHT the Pdx1VP16ER T2 construct had no effect on the human insulin promoter, but in its presence, the stimulatory effect was similar to that of a constitutively active Pdx1VP16 construct (i.e., lacking ER T2 ) (Fig. 1C). These experiments confirmed that the Pdx1VP16ER T2 protein was responsive to 4OHT and that the addition of the ER T2 sequences did not affect its functional activity. The ptp6pdx1vp16ert 2 was then stably transfected into the CGR8 mouse and H9 human ES cell lines and several puromycin-resistant clones were selected and characterized by immunocytochemistry. Treatment of the clonal lines with 4OHT resulted in the translocation of Pdx1VP16 from the cytoplasm to the nucleus, thus demonstrating that this inducible system works efficiently, as illustrated for the clonal lines CGR8C10 and H9C7 (supporting information Fig. 2). Our objective was to examine the fate of endoderm-enriched ES cell populations that were subjected to continuous or biphasic Pdx1 activation (supporting information Fig. 1A). Therefore, initially we optimized culture conditions to drive ESCs into definitive endoderm cells. The differentiation was done in CDM and involved growing the mouse CGR8C10 cells as EBs for 1 day and then treating the suspension cultures with activin A (100 ng/ml) and/or BMP-4 (10 ng/ml) for up to 7 days (Fig. 2A). The expectation was that the cells would first differentiate into Brachyury/Foxa2-positive (Bry Foxa2 ) mesendodermal progenitors, which would subsequently become Sox17 Foxa2 definitive endoderm [20, 21]. The results (Fig. 2B) show that treatment for 3 days with BMP-4 induced the formation of mesoderm (Flk1 ), a finding consistent with other studies [22]. Both definitive (Sox17 and Foxa2 ) and extraembryonic endoderm (Sox7 ) were also induced. On the other hand, the primitive streak marker Bry was absent from these cultures, as was the neuroectoderm marker Sox1 (data not shown), indicating that these cultures represented a mixed population of mesoderm and both extraembryonic and definitive endoderm lineages. Treatment with high activin A for the same period of time led to the upregulation of the primitive streak marker Bry, as well as the anterior streak marker Foxa2. High activin A also induced formation of definitive endoderm (Sox17 ) and extraembryonic endoderm (Sox7 ) cells. The activin A-treated cultures did not, however, express the mesoderm or neuroectoderm markers Flk1 and Sox1 (data not shown). These results are in keeping with previous results, which showed that activin A at high levels induces an anterior streak fate [23] and promotes endoderm differentiation [24]. The combination of both activin A and BMP-4 treatment during the same period of culture yielded a result similar to that of activin A, except that these cells had less tendency to form definitive (Sox17) and extraembryonic endoderm (Sox7). Collectively, the results indicated that a combination of activin A and BMP-4 in CDM was important to push the cells toward an anterior streak fate. Cells grown in the presence of activin A and BMP-4 or activin A alone were further treated with high activin A for an additional 3-day period and analyzed for the same markers at day 7 of culture. By day 7 the expression of the mesendoderm

4 344 Modeling Pancreatic Development in ES Cells Figure 2. Effects of activin A and bone morphogenetic protein 4 (BMP-4) on the differentiation of CGR8C10 cells to definitive endoderm. (A): Diagram of the different culture conditions. Undifferentiated cells (0) were differentiated as EBs in chemically defined media for 4 (D4) or 7 (D7) days, at which time they were harvested for quantitative polymerase chain reaction (QPCR) analysis. The first day of culture was in the absence of growth factors, and from D1 to D4, cultures were treated with either B, A, or a combination of both (B/A). Cells that were cultured for a further 3 days (until D7) were treated with activin A alone. (B): On D4 or D7 of culture, total RNA was extracted and analyzed for gene expression by QPCR. The data are expressed relative to glyceraldehyde-3-phosphate dehydrogenase and represent the average ( SD) of triplicate cultures of a representative experiment. (C): Cells treated only with activin A (D7 A) or sequentially with BMP and activin for 3 days and then with activin A for a further 3 days (D7 B/A) were stained for Brachyury (green) or Sox17 (red) or with DAPI for nuclei (blue). Abbreviations: A, activin A at 100 ng/ml; B, bone morphogenetic protein-4 at 10 ng/ml; D, day; DAPI, 4,6-diamidino-2-phenylindole. marker Bry was markedly reduced, whereas expression of Foxa2 was retained under both conditions. The mesoderm marker Flk1 and the neuroectoderm marker Sox1 (data not shown) were also expressed at very low levels in both conditions, as was the extraembryonic endoderm marker Sox7. These two culture conditions were clearly distinguished, however, by the differential expression levels of the definitive endoderm marker Sox17. Sox17 was significantly upregulated in EBs derived from cultures previously treated with combined activin A and BMP-4 (Fig. 2C). A combination of BMP-4 and activin A (D7 B/A; Fig. 4) was therefore used in the subsequent experiments to induce definitive endoderm differentiation by the CGR8C10 clonal line. To evaluate the role of Pdx1 in the differentiation of mouse ES cell-derived definitive endoderm-enriched cultures, EBs were plated on tissue culture dishes in the absence or continuous presence of medium containing 4OHT for various lengths of time (Fig. 3A). Cells grew out from the EBs forming outgrowth of cells. These outgrowth cultures were harvested after 21 days and analyzed by QPCR (Fig. 3B; supporting information Fig. 3). In the absence of 4OHT there was no detectable expression of the -cell markers insulin 1, insulin 2 (the two nonallelic mouse insulin genes), IAPP, or the transcription factor Nkx6.1, whereas GLUT2, which can also mark liver cells, was upregulated in this condition, as was the liver marker albumin (Fig. 3B). Expression of these -cell markers (insulin 1 and 2, IAPP,

5 Bernardo, Cho, Mason et al. 345 Figure 3. Effect of continuous Pdx1VP16 activation on the differentiation of definitive endoderm-enriched CGR8C10 cells. (A): Protocol for single-phase 4OHT treatment. CGR8C10 EBs were grown for 3 days in chemically defined medium supplemented with bone morphogenetic protein 4 and activin A plus an additional 3 days in activin A alone (D7B/A). They were then transferred to Matrigel-coated wells and allowed to form outgrowth cultures. Triplicate wells of a six-well plate were treated with 4OHT for the indicated periods. Day 0 represents the day on which the EBs were transferred into wells for outgrowth formation, and all cultures were harvested 21 days later. CGR8C10 outgrowth cultures were grown in the absence of 4OHT (no T) or treated with 4OHT in single phases, according to the protocol described above. (B): Cells were harvested at day 21 and analyzed for expression of -cell markers by quantitative polymerase chain reaction (QPCR). (C): Cells were harvested at day 21 and analyzed for expression of other endocrine and exocrine markers by QPCR. The data are expressed relative to glyceraldehyde-3-phosphate dehydrogenase and represent the average ( SD) of triplicate cultures. This experiment is representative of similar experiments performed on at least two separate occasions. Abbreviations: 4OHT, 4-hydroxytamoxifen; IAPP, islet amyloid polypeptide; PP, pancreatic polypeptide; T, 4-hydroxytamoxifen. and Nkx6.1) was increased when 4OHT was present in the medium during the first 5 or 10 days of culture but decreased markedly when the period of exposure to 4OHT was prolonged beyond day 10, indicating the importance of this early stage of Pdx1 activation for eventual -cell differentiation. When 4OHT treatment was initiated later than day 0 (i.e., at days 5, 10, or 15) (supporting information Fig. 3A), high expression of insulin 2, IAPP, and Nkx6.1 was also observed (supporting information Fig. 3B), thereby revealing that Pdx1 activity can redirect spontaneously differentiating cells, which tend to form liver-like cells, into a -cell phenotype. It is also worth noting that in all the experiments the insulin 1 gene was expressed at levels well below that of the insulin 2 gene, as in the adult mouse islet [25]. In the absence of 4OHT, the -cell marker glucagon, the pancreatic peptide (PP) cell marker PP, and the exocrine cell marker amylase were not detected (Fig. 3C). There was, however, expression of the -cell marker somatostatin. Expression of all these markers was increased in cultures treated with 4OHT for 5 or 10 days. Prolonging the 4OHT treatment for an additional 5 or 10 days significantly decreased glucagon, PP, and somatostatin expression levels but not that of amylase, which was downregulated only when 4OHT treatment was initiated at day 5 or beyond (supporting information Fig. 3B). These results showed that the differentiation of definitive endoderm toward the pancreatic cell lineage was dependent on the activation of Pdx1 and that the timing and duration of activation had a marked effect upon the different pancreatic lineages generated. Thus, the results highlight the importance of regulating Pdx1 expression during differentiation, rather than simply overexpressing it continuously, as even known Pdx1 target genes are not consistently induced by its constitutive expression (Fig. 3C). Accordingly, we hypothesized that activating Pdx1VP16 during differentiation in a manner similar to that which occurs during normal development would provide the best conditions for -cell differentiation. In the mouse fetus, Pdx1 is expressed in two waves: at an early stage in the formation of the pancreas and later in the differentiating -cell. To establish what effect the biphasic expression of Pdx1VP16 would have on the fate of the definitive endoderm-enriched outgrowths, a series of experiments was performed whereby two pulses of 4OHT were given to the cells. After 21 days of outgrowth culture, the cells were harvested and analyzed by QPCR (Fig. 4A). High expression of insulin 1, insulin 2, IAPP, Nkx6.1, Pdx1, and Isl-1 was observed when the Pdx1VP16 construct was activated in two pulses between 0 10 and days (Fig. 4B; supporting information Fig. 4), that is, a pattern of activation that mimicked the expression of Pdx1 in the developing mouse pancreas. Other endocrine cell markers, such as glucagon, somatostatin, and PP, were also strongly activated under these conditions (Fig. 4C). These results suggest that in addition to a role in formation of the -cell, the second phase of Pdx1 expression in the developing pancreas may play a role in the differentiation of -, -, and PP cells. Interestingly, in keeping with the findings of the previous set of experiments, expression of the exocrine cell marker amylase was also dependent on an early pulse of Pdx1VP16 expression. A second pulse of Pdx1VP16 activity was not inhibitory of

6 346 Modeling Pancreatic Development in ES Cells Figure 4. Effect of biphasic Pdx1VP16 activation on the differentiation of definitive endoderm-enriched CGR8C10 cells. (A): Protocol for biphasic 4OHT treatment. CGR8C10 EBs were grown for 3 days in chemically defined medium supplemented with bone morphogenetic protein 4 and activin A plus an additional 3 days in activin A alone (D7B/A). They were then transferred to Matrigel-coated wells and allowed to form outgrowth cultures. Triplicate wells of a six-well plate were treated with 4OHT for the indicated periods. Time of 0-5/10-21 indicates that the cells were treated with 4OHT in two separate pulses over the periods from day 0 to day 5 and from day 10 to day 15. Day 0 represents the day on which the EBs were transferred into wells for outgrowth formation, and all cultures were harvested 21 days later. CGR8C10 outgrowth cultures were grown in the absence of 4OHT (no T) or treated with biphasic pulses of 4OHT, according to the protocol described above. (B): Cells were harvested at day 21 and analyzed for expression of -cell markers by quantitative polymerase chain reaction (QPCR). (C): Cells were harvested at day 21 and analyzed for expression of other endocrine and exocrine markers by QPCR. The data are expressed relative to glyceraldehyde-3-phosphate dehydrogenase and represent the average ( SD) of triplicate cultures. This experiment is representative of similar experiments performed on at least two separate occasions. Abbreviations: 4OHT, 4-hydroxytamoxifen; IAPP, islet amyloid polypeptide; PP, pancreatic polypeptide; T, 4-hydroxytamoxifen. exocrine cell development when commenced at day 10 but clearly resulted in an inhibitory effect when started on day 15. This is in keeping with the separate lineages of exocrine and endocrine cells and the requirement for Pdx1 in the initiation of acinar cell differentiation, possibly via the activation of Ptf1a [26]. These results emphasize once more the requirement for fine regulation of transcription factor expression for appropriate induction of a specific differentiation program and reveal novel subtle effects of Pdx1 on endocrine and exocrine lineages. The outgrowth cultures differentiated in the presence of Pdx1VP16 during the first 5 days of culture were multipotent for pancreatic lineages as they were able to generate both endocrine and exocrine populations. If the onset of Pdx1VP16 activation was delayed until day 5, it was still possible to rescue the endocrine lineage but not the exocrine lineage. A second late Pdx1 pulse (days 15 21) was decisive in patterning the cells toward the endocrine lineage but was inhibitory for exocrine development. In cells differentiated through the biphasic action of Pdx1VP16, C-peptide expression was detected in 32% of the cells as determined by flow cytometry (Fig. 5A). Immunocytochemistry showed that C-peptide was present in populations of cells that grew either as a monolayers or threedimensional clusters of cells (Fig. 5B). Within the monolayer populations, the percentage of C-peptide positive cells was as high as 73%. The cells expressed C-peptide (16.2 pmol/mg protein), as measured by radioimmunoassay of the total culture. Furthermore, the cells exhibited significant insulin secretory response to 3-isobutyl-1-methylxanthine (IBMX) and KCl; however, they were not responsive to glucose, even though a moderate increase of insulin levels was observed upon high glucose stimulation (Fig. 5C). This confirmed that although the biphasic activation of Pdx1VP16 resulted in an enriched pool of -like cells, these cells were not fully differentiated (i.e., they lacked a secretory response to glucose), suggesting that Pdx1 activation alone was not sufficient to generate fully differentiated and mature -cells. To characterize intermediate, and potentially progenitor, cell populations during the 21-day differentiation protocol, outgrowth cultures were harvested at various time intervals following activation of the Pdx1VP16ER T2 protein (Fig. 6A). The subsequent QPCR and immunocytochemistry (Fig. 6B, 6C; supporting information Fig. 5) showed that transcription factors that are expressed early in the developing pancreas, such as Ptf1a and endogenous Pdx1 [27], were present at high levels in the day 5 and day 10 cultures but at lower levels in the day 15 and day 21 cultures. Markers of intermediate pancreas development, such as the transcription factors Ngn3, NeuroD, IA1, MafB, Nkx2.2, and Nkx6.1, were also expressed at higher levels between days 5 and 10 of culture. There was a temporal relationship between Ngn3 and IA1, PAX4 and PAX6, Nkx2.2 and Nkx6.1, and MafA and MafB, which partially mimicked their pattern of expression in the developing mouse pancreas [28 31]. The -cell markers insulin1, insulin 2, IAPP, MafA, and GLUT2 were enriched in the day 21 cultures. This is in keeping with a role for Pdx1 in the late differentiation of -cells [32]. These markers were also expressed in day 10 cultures, which is also consistent

7 Bernardo, Cho, Mason et al. 347 Figure 5. Differentiated CGR8C10 cells express and secrete C-peptide. (A): Flow cytometry graphs showing the number of C-peptide cells expressed in 21-day outgrowth cultures grown in the absence of 4OHT (no T) or in the presence of 4OHT added in a biphasic manner ( T 0-10/15-21). (B): Standard fluorescence micrographs showing the pattern of C-peptide expression in two subpopulations of cells (monolayer and cluster populations) found in 4OHT biphasic ( 4OHT 0-10/15-21) treated cultures. (C): C-peptide content of 21-day outgrowth cultures grown in the absence of 4OHT (no T) or in the presence of 4OHT added continuously from day 5 until day 21 (5-21) or in a biphasic manner (0-10/15-21). C-peptide levels were determined by radioimmunoassay, expressed relative to the protein content of the cells and represent the mean SD of triplicate cultures. (D): C- peptide secreted to the medium during 1 hour by 21-day outgrowth cultures that were treated with 4OHT in a biphasic manner (0 10/15 21). On day 21 the medium was removed, and the cells were washed in phosphate-buffered saline and then incubated for 1 hour with media containing either low Glu (3 mm) or high Glu (25 mm), 50 mm KCl, or 0.5 mm IBMX. C-peptide and insulin levels in the media were determined by radioimmunoassay, expressed relative to the protein content of the cells and represent the mean SD of triplicate cultures, where p.05 ( ) or 0.01 ( ). Abbreviations: 4OHT, 4-hydroxytamoxifen; FL1-fitc, emitted fluorescence detected in an FL1 sensor; Glu, glucose; IBMX, 3-isobutyl-1-methylxanthine; T, 4-hydroxytamoxifen. with the early appearance of -cell markers in the developing pancreas [33]. The presence of somatostatin and PP in the day 5 and 10 cultures is consistent with the expression of somatostatin in the primitive gut and the appearance of PP early in the developing pancreas [34]. It also correlates with the more specific role of Pdx1VP16 activation during days 5 and 10 in inducing an endocrine pancreas fate, while demonstrating that the second pulse of Pdx1 activation does not induce either a -cell or a PP cell phenotype but is instead permissive of their coexistence in the cultures at the later stages. Glucagon, which has been detected early in the developing mouse pancreas [33], was detected in the cultures from the day 5 onward (Fig. 6C). With regard to the exocrine marker amylase, which was observed from day 10 (albeit at much lower levels compared with those from Figs. 3C, 4C), the data suggest that the initial pulse of PdxVP16 was sufficient to induce low levels of exocrine pancreas differentiation, which were sustained even upon the second pulse of Pdx1VP16 activation. This second Pdx1VP16 pulse, however, was not permissive of exocrine expansion, as demonstrated by the low levels of amylase transcripts in comparison with those obtained when Pdx1VP16 activation was done in a single phase started on day 0 or, alternatively, when the biphasic pulses did not include days 5 and 10 of culture, which have been shown to be deleterious for exocrine pancreas differentiation (Figs. 3C, 4C). Together, these findings demonstrate that the events occurring during the 21 days of differentiation recapitulate well the initial and intermediate stages of pancreas development, emphasizing the crucial role of Pdx1 in pancreatic fate decision. The later stages of -cell development were also recapitulated; however, the data em- phasized the importance of the second phase of Pdx1 expression in maintaining other endocrine cell phenotypes. The effect of Pdx1 overexpression on the differentiation of human endoderm progenitors was then analyzed. Pdx1VP 16ER T2 expressing hescs were generated as described above, and the resulting line (H9C7) was differentiated into definitive endoderm enriched cultures, in CDM supplemented with high activin A, FGF2, BMP-4, and the PI3- kinase inhibitor LY (Fig. 7A). The resulting cells homogeneously expressed the definitive endoderm markers Sox17, GSC, CXCR4, Mixl, and FoxA2 (data not shown; Fig. 7B, 7C). The definitive endoderm-enriched cultures were further differentiated in medium containing fetal bovine serum (10%) for an additional 13 days (Fig. 7A). In the absence of 4OHT, the definitive endoderm-enriched cultures followed a hepatocyte lineage, as evidenced by the expression of albumin and -fetoprotein and the absence of pancreatic markers (Fig. 7E). When the definitive endoderm-enriched cultures were treated biphasically with 4OHT and then allowed to mature for a further 3 days, there was a significant increase in endocrine markers (insulin, glucagon, and somatostatin), as well as pancreatic transcription factors, including Ptf1a, Sox9, Ngn3, MafB, and Nkx6.1, suggesting the presence of differentiated and partially differentiated endocrine cells. Immunocytochemistry confirmed the presence of insulin in a substantial fraction of the 4OHT treated cells (Fig. 7F). Therefore, we show that biphasic induction of Pdx1 also drives differentiation of human definitive endoderm-enriched cultures into pancreatic -cells (as it does in mouse), suggesting that evolutionary conserved mechanisms control pancreatic development. Our data further suggest that condi-

8 348 Modeling Pancreatic Development in ES Cells Figure 6. Effect of Pdx1VP16 activation on the course of differentiation of definitive endoderm-enriched cell cultures. (A): Protocol for sampling cells during the 21D of outgrowth culture period of cells treated with 4-hydroxytamoxifen (4OHT) in a biphasic manner. Thus, definitive endoderm-enriched outgrowth cultures were plated in Matrigel-coated wells and harvested at D5 and D10 after incubation with 4OHT. Cells were also harvested at D15 after a 10-day pulse with 4OHT followed by 5 days of no 4OHT and at D21 after two pulses (D0 D10 and D15 D21) of 4OHT. D0 represents the day on which the EBs were transferred into wells for outgrowth formation and arrows point to the days on which cells were collected. (B): CGR8C10 outgrowth cultures treated as per the above protocol were harvested at different time intervals, and RNA was extracted and analyzed for expression pancreatic markers by quantitative polymerase chain reaction. The data are expressed relative to glyceraldehyde-3-phosphate dehydrogenase and represent the average ( SD) of triplicate cultures of two combined experiments. (C E): Standard fluorescence micrographs showing the pattern of C-peptide (red), and glucagon (green), Ngn3 (green), and Ptfa1 (red) expression on cells harvested on the indicated days (D5 D21). The nuclei were stained with DAPI (blue). Abbreviations: D, day; DAPI, 4,6-diamidino-2-phenylindole; IAPP, islet amyloid polypeptide; PP, pancreatic polypeptide. tional overexpression of Pdx1 could be used as an in vitro system to study mechanisms controlling pancreatic organogenesis in mouse and in human. DISCUSSION The aim of this study was to develop an in vitro model to study the development of the endocrine pancreas. This involved inducing expression of an exogenous Pdx1VP16 gene in a definitive endoderm-enriched population generated from mouse or human ES cells. The major findings were that the expression of pancreatic markers was dependent on activation of the Pdx1VP16 transgene, that the timing and duration of activation of the Pdx1VP16 transgene was important in determining the pancreatic lineage generated, that biphasic activation of the Pdx1 transgene in an early period followed by a later period of activation was most effective in inducing a -like cell phenotype, and that in both mouse and human ES cells, biphasic Pdx1VP16 induction led to a multiendocrine pancreas phenotype. The results provide proof of principle that in vitro manipulation of Pdx1 activity in a temporally phased manner can be an effective approach for

9 Bernardo, Cho, Mason et al. 349 Figure 7. Effect of Pdx1VP16 on the differentiation of H9C7 cells. (A): Protocol used for the generation and subsequent differentiation of the clonal human ES cell line H9C7, which stably expressed the ptp6pdx1vp16er T2 plasmid. Cells were differentiated into a definitive endoderm-enriched population following treatment for 3 days with A, B, F, and Ly in chemically defined medium. The cells were cultured for a further 14 days in the presence or absence of 4OHT as indicated. (B): RNA was prepared from cells harvested on D0, D3, and D17 and analyzed by quantitative polymerase chain reaction (QPCR) for the endoderm markers indicated. The data are expressed relative to PBGD expression levels, normalized to a calibrator (D0), and they represent the average ( SD) of triplicate cultures of a representative experiment. (C): Standard fluorescence micrographs showing the pattern of Sox17 (red) expression in D3 cultures. The nuclei were stained with DAPI and are visualized in blue. (D): RNA was prepared from cells harvested on D0, D3, and D17 and analyzed by QPCR for the pancreatic markers indicated. The data are expressed relative to PBGD expression levels, normalized to a calibrator (D0) and represent the average ( SD) of triplicate cultures of a representative experiment. (E): RNA was prepared from cells harvested on D0, D3, and D17 and analyzed by QPCR for the liver markers indicated. The data are expressed relative to PBGD expression levels, normalized to a calibrator (D0) and represent the average ( SD) of triplicate cultures of a representative experiment. (F): Standard fluorescence micrographs showing the pattern of Ins (green) expression in D17 cultures. The nuclei were stained with DAPI (blue). Abbreviations: 4OHT, 4-hydroxytamoxifen; A, high activin A (100 ng/ml); AFP, -fetoprotein; B, bone morphogenetic protein 4 (10 ng/ml); D, day; DAPI, 4,6-diamidino- 2-phenylindole; F, basic fibroblast growth factor (20 ng/ml); IAPP, islet amyloid polypeptide; Ins, insulin; Ly, phosphoinositide 3-kinase inhibitor LY (10 M); PP, pancreatic polypeptide. inducing pancreatic -cell differentiation from pluripotent ES cells. The outcome of our study also emphasizes the importance of native patterns of transcription factor expression during normal

10 350 Modeling Pancreatic Development in ES Cells development as a template for differentiation of ES cells toward a -cell-like phenotype. The beneficial effect of biphasic Pdx1VP16 expression shown here contrasted with the poorer outcome seen when Pdx1 was constitutively overexpressed in mouse ES cells [35]. Similarly, in human ES cells, although constitutive overexpression of Pdx1 enhanced the differentiation toward endocrine and exocrine cell types, it did not drive formation of insulin-expressing cells [36]. On the other hand, regulatable expression of Pdx1 in 4 5-day-old mouse EBs using a Tet-off system resulted in upregulation of a range of -cell markers and high production of insulin in many of the differentiated cells [37]. Our present study goes beyond this latter study to show that by using a definitive endoderm-enriched population and mimicking the natural pattern of expression of Pdx1, an improved -like cell differentiation can be achieved. This in vitro approach can potentially contribute to a better understanding of pancreas development. For example, the transcript levels of the two mouse insulin genes at day 10 of outgrowth culture (Fig. 7B) could reflect the appearance of an insulin positive cell at an early stage in pancreatic development [33]. It is currently not known whether such early embryonic expression is by insulin 1 or insulin 2 genes. Here, in almost all the experiments, the expression of insulin 1 seemed to be more associated with an early pulse of Pdx1 (Fig. 3B). Our data would indicate, therefore, that it is the insulin 1 gene that is predominantly expressed in these early insulin-positive cells. In all experiments the expression of insulin 2 vastly exceeded that of insulin 1. It is, therefore, of interest that insulin 2 was expressed at significantly higher levels than insulin 1 in exocrine cells that were reprogrammed to -cells using three transcription factors, including Pdx1 [38]. The differentiated cells generated after biphasic induction exhibited a secretory response to KCl and IBMX but only a modest response to high glucose. This is in keeping with other studies [39] and may reflect the immature nature of the differentiated cells, which is possibly due to a requirement for aggregate formation, similarly to adult islets. An immature nature of the cells is supported by the presence of MafB in the 21-day outgrowth cultures (supporting information Fig. 4). It is likely that Pdx1VP16, by being permissive of the coexistence of different pancreatic endocrine cells, is not sufficient to allow maturation of the -like cells. Additional factors, such as ngn3 and MafA [38], could help further refine the cultures toward a more specific -like phenotype. CONCLUSION Our in vitro approach substantially recapitulates pancreas development, and by examining the process of differentiation at defined points we were able to identify intermediate cell populations that expressed the predicted early, intermediate, and late genes of pancreatic development. Studying these different intermediate populations will undoubtedly provide novel insights that will further inform strategies toward developing protocols that will allow reprogramming of ES [40] and adult cells [41] toward -cells for transplantation in the treatment of diabetes mellitus. The results also reveal subtle distinctive effects of the timing and duration of Pdx1 expression on the exocrine and endocrine cell lineages. ACKNOWLEDGMENTS This work was supported by grants from the Wellcome Trust and the Juvenile Diabetes Research Foundation. A.S.B. was funded by a research studentship from the Portuguese Foundation for Science and Technology. DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST The authors indicate no potential conflicts of interest. REFERENCES 1 Servitja JM, Ferrer J. Transcriptional networks controlling pancreatic development and beta cell function. Diabetologia 2004;47: Collombat P, Hecksher-Sorensen J, Serup P et al. Specifying pancreatic endocrine cell fates. Mech Dev 2006;123: Lawson KA, Meneses JJ, Pedersen RA. Clonal analysis of epiblast fate during germ layer formation in the mouse embryo. Development 1991; 113: Pictet RL, Clark WR, Williams RH et al. An ultrastructural analysis of the developing embryonic pancreas. Dev Biol 1972;29: Kim SK, Hebrok M. Intercellular signals regulating pancreas development and function 1. Genes Dev 2001;15: Jonsson J, Carlsson L, Edlund T et al. Insulin-promoter-factor 1 is required for pancreas development in mice. Nature 1994;371: Offield MF, Jetton TL, Labosky PA et al. PDX-1 is required for pancreatic outgrowth and differentiation of the rostral duodenum. Development 1996;122: Stoffers DA, Zinkin NT, Stanojevic V et al. Pancreatic agenesis attributable to a single nucleotide deletion in the human IPF1 gene coding sequence. Nat Gen 1997;15: Gradwohl G, Dierich A, LeMeur M et al. neurogenin3 is required for the development of the four endocrine cell lineages of the pancreas. Proc Natl Acad Sci U S A 2000;97: Schwitzgebel VM, Scheel DW, Conners JR et al. Expression of neurogenin3 reveals an islet cell precursor population in the pancreas. Development 2000;127: McKinnon CM, Docherty K. Pancreatic duodenal homeobox-1, PDX-1, a major regulator of beta cell identity and function. Diabetologia 2001; 44: Wiles MV, Johansson BM. Embryonic stem cell development in a chemically defined medium. Exp Cell Res 1999;247: Brons IG, Smithers LE, Trotter MW et al. Derivation of pluripotent epiblast stem cells from mammalian embryos. Nature 2007;448: Vallier L, Reynolds D, Pedersen RA. Nodal inhibits differentiation of human embryonic stem cells along the neuroectodermal default pathway. Dev Biol 2004;275: Vallier L, Rugg-Gunn PJ, Bouhon IA et al. Enhancing and diminishing gene function in human embryonic stem cells. STEM CELLS 2004;22: Docherty HM, Hay CW, Ferguson LA et al. Relative contribution of PDX-1, Mafa and E47/beta2 to the regulation of the human insulin promoter. Biochem J 2005;389: Indra AK, Warot X, Brocard J et al. Temporally-controlled site-specific mutagenesis in the basal layer of the epidermis: Comparison of the recombinase activity of the tamoxifen-inducible Cre-ER(T) and Cre- ER(T2) recombinases. Nucleic Acids Res 1999;27: Vallier L, Alexander M, Pedersen R. Conditional gene expression in human embryonic stem cells. STEM CELLS 2007;25: Hay CW, Sinclair EM, Bermano G et al. Glucagon-like peptide-1 stimulates human insulin promoter activity in part through camp-responsive elements that lie upstream and downstream of the transcription start site. J Endocrinol 2005;186: Kubo A, Shinozaki K, Shannon JM et al. Development of definitive endoderm from embryonic stem cells in culture. Development 2004;131: Yasunaga M, Tada S, Torikai-Nishikawa S et al. Induction and monitoring of definitive and visceral endoderm differentiation of mouse ES cells. Nat Biotechnol 2005;23: Park C, Afrikanova I, Chung YS et al. A hierarchical order of factors in the generation of FLK1- and SCL-expressing hematopoietic and endothelial progenitors from embryonic stem cells. Development 2004;131: Gadue P, Huber TL, Paddison PJ et al. Wnt and TGF-beta signaling are required for the induction of an in vitro model of primitive streak

11 Bernardo, Cho, Mason et al. 351 formation using embryonic stem cells. Proc Natl Acad Sci USA 2006; 103: Tada S, Era T, Furusawa C et al. Characterization of mesendoderm: A diverging point of the definitive endoderm and mesoderm in embryonic stem cell differentiation culture. Development 2005;132: Bengtsson M, Stahlberg A, Rorsman P et al. Gene expression profiling in single cells from the pancreatic islets of Langerhans reveals lognormal distribution of mrna levels. Genome Res 2005;15: Hale MA, Kagami H, Shi L et al. The homeodomain protein PDX1 is required at mid-pancreatic development for the formation of the exocrine pancreas. Dev Biol 2005;286: Burlison JS, Long Q, Fujitani Y et al. Pdx-1 and Ptf1a concurrently determine fate specification of pancreatic multipotent progenitor cells. Dev Biol 2008;316: Mellitzer G, Bonne S, Luco RF et al. IA1 is NGN3-dependent and essential for differentiation of the endocrine pancreas. EMBO J 2006; 25: Sander M, Sussel L, Conners J et al. Homeobox gene nkx6.1 lies downstream of nkx2.2 in the major pathway of alpha and beta-cell formation in the pancreas. Development 2000;127: Collombat P, Hecksher-Sorensen J, Broccoli V et al. The simultaneous loss of Arx and Pax4 genes promotes a somatostatin-producing cell fate specification at the expense of the alpha- and beta-cell lineages in the mouse endocrine pancreas. Development 2005;132: Nishimura W, Kondo T, Salameh T et al. A switch from MafB to MafA expression accompanies differentiation to pancreatic beta-cells. Dev Biol 2006;293: Gannon M, Ables ET, Crawford L et al. Pdx-1 Function is specifically required in embryonic beta cells to generate appropriate numbers of endocrine cell types and maintain glucose homeostasis. Dev Biol 2008; 314: Teitelman G, Alpert S, Polak JM et al. Precursor cells of mouse endocrine pancreas coexpress insulin, glucagon and the neuronal proteins tyrosine hydroxylase and neuropeptide Y, but not pancreatic polypeptide. Development 1993;118: Herrera PL, Huarte J, Sanvito F et al. Embryogenesis of the murine endocrine pancreas; early expression of pancreatic polypeptide gene. Development 1991;113: Blyszczuk P, Czyz J, Kania G et al. Expression of Pax4 in embryonic stem cells promotes differentiation of nestin-positive progenitor and insulin-producing cells. Proc Natl Acad Sci USA 2003;100: Lavon N, Yanuka O, Benvenisty N. The effect of overexpression of Pdx1 and Foxa2 on the differentiation of human embryonic stem cells into pancreatic cells. STEM CELLS 2006;24: Miyazaki S, Yamato E, Miyazaki J. Regulated expression of pdx-1 promotes in vitro differentiation of insulin-producing cells from embryonic stem cells. Diabetes 2004;53: Zhou Q, Brown J, Kanarek A et al. In vivo reprogramming of adult pancreatic exocrine cells to beta-cells. Nature 2008;455: D Amour KA, Bang AG, Eliazer S et al. Production of pancreatic hormone-expressing endocrine cells from human embryonic stem cells. Nat Biotechnol 2006;24: Docherty K, Bernardo AS, Vallier L. Embryonic stem cell therapy for diabetes mellitus. Semin Cell Dev Biol 2007;18: Takahashi K, Yamanaka S. Induction of pluripotent stem cells from mouse embryonic and adult fibroblast cultures by defined factors. Cell 2006;126: See for supporting information available online.