• We are sorry, but NCBI web applications do not support your browser and may not function properly. More information
Logo of nihpaAbout Author manuscriptsSubmit a manuscriptNIH Public Access; Author Manuscript; Accepted for publication in peer reviewed journal;
Mech Dev. Author manuscript; available in PMC Nov 28, 2011.
Published in final edited form as:
Mech Dev. Sep 2011; 128(7-10): 412–427.
Published online Aug 10, 2011. doi:  10.1016/j.mod.2011.08.001
PMCID: PMC3225072

Activin, BMP and FGF pathways cooperate to promote endoderm and pancreatic lineage cell differentiation from human embryonic stem cells


The study of how human embryonic stem cells (hESCs) differentiate into insulin-producing beta cells has two-fold significance: first, it provides an in vitro model system for the study of human pancreatic development, and second, it serves as a platform for the ultimate production of beta cells for transplantation into patients with diabetes. The delineation of growth factor interactions regulating pancreas specification from hESCs in vitro is critical to achieving these goals. In this study, we describe the roles of growth factors bFGF, BMP4 and Activin A in early hESC fate determination. The entire differentiation process is carried out in serum-free chemically defined media (CDM) and results in reliable and robust induction of pancreatic endoderm cells, marked by PDX1, and cell clusters co-expressing markers characteristic of beta cells, including PDX1 and insulin/C-peptide. Varying the combinations of growth factors, we found that treatment of hESCs with bFGF, Activin A and BMP4 (FAB) together for 3 to 4 days resulted in strong induction of primitive-streak and definitive endoderm-associated genes, including MIXL1, GSC, SOX17 and FOXA2. Early proliferative foregut endoderm and pancreatic lineage cells marked by PDX1, FOXA2 and SOX9 expression are specified in EBs made from FAB-treated hESCs, but not from Activin A alone treated cells. Our results suggest that important tissue interactions occur in EB-based suspension culture that contribute to the complete induction of definitive endoderm and pancreas progenitors. Further differentiation occurs after EBs are embedded in Matrigel and cultured in serum-free media containing insulin, transferrin, selenium, FGF7, nicotinamide, islet neogenesis associated peptide (INGAP) and exendin-4, a long acting GLP-1 agonist. 21–28 days after embedding, PDX1 gene expression levels are comparable to those of human islets used for transplantation, and many PDX1+ clusters are formed. Almost all cells in PDX1+ clusters co-express FOXA2, HNF1ß, HNF6 and SOX9 proteins, and many cells also express CPA1, NKX6.1 and PTF1a. If cells are then switched to medium containing B27 and nicotinamide for 7 to 14 days, then the number of insulin+ cells increases markedly. Our study identifies a new chemically defined culture protocol for inducing endoderm- and pancreas-committed cells from hESCs and reveals an interplay between FGF, Activin A and BMP signaling in early hESC fate determination.

Keywords: definitive endoderm, beta cell, bone morphogenetic protein, fibroblast growth factor, human embryonic stem cells, Activin A

1. Introduction

Human embryonic stem cells (hESCs) possess a strong proliferative ability and the potential to differentiate into all somatic cell types of the embryo and adult. As such, they provide an excellent platform for cell-based therapeutics and the study of human embryonic development. Mouse ES studies have shown a correlation between the developmental pathways followed by cells in vitro and in vivo, and also between factors that induce and pattern ESCs and the embryo during differentiation (Tam et al., 2006; Jackson et al., 2010). It is logical to assume that differentiation of hESCs in vitro mimics human embryonic development in vivo and can be used as a tool to study the mechanisms regulating cell fate decisions during development. Early studies defined the effects of single growth factors on hESCs differentiation fates (D'Amour et al., 2005; Schuldiner et al., 2000; Xu et al., 2002; Xu et al., 2005a). However, given the knowledge that a limited number of different morphogens exert a diverse repertoire of effects in various tissues and at different stages of embryonic development in lower organisms, it seems obvious now to study combinatorial effects of different morphogens on tissue differentiation from hESCs. Several important studies in human pluripotent stem cells have begun to explore the interplay of multiple signaling pathways, including TGFbeta/Nodal/Activin, fibroblast growth factor (FGF), and bone morphogenetic protein (BMP) signaling on early fate decisions, and endoderm differentiation in particular (Vallier et al., 2009; Nostro et al., 2011; Rezania et al., 2011; Green et al., 2011). As seemingly minor manipulations of similar protocols lead to divergent outcomes, we sought to further define the combinatorial effects of these signaling pathways on early endodermal cell fate decisions in hESCs and the ability to differentiate into pancreatic lineages.

As a member of the TGFß superfamily, Activin A has been reported to induce endoderm in Xenopus animal caps in vitro (Smith et al., 1990; Gamer and Wright, 1995; Ninomiya et al., 1999); in mouse ES cell differentiation (Tada et al., 2005; Yasunaga et al., 2005); and in human ES cell differentiation cultures (D'Amour et al., 2005). On the other hand, Beattie et al. (Beattie et al., 2005) and Vallier et al. (Vallier et al., 2005) showed that Activin A plays an important role in the maintenance of pluripotency of hESCs. It is intriguing that treatment with the same growth factor can result in such different fates. A confounding factor is probably insulin/insulin-like factor. In medium containing insulin, wherein the phosphatidylinositol 3-kinase (PI3K) signaling pathway is presumably activated, Activin A maintains the pluripotency of hESCs. Inhibition of the Activin signaling pathway with the Activin receptor inhibitor SB431542 results in hESC differentiation. However, when PI3K signaling is suppressed, Activin A induces definitive endoderm (DE) (McLean et al., 2007). The interaction of growth factors is also observed with BMP4 in early hESC differentiation. Whereas BMP4 has been shown to promote trophectoderm differentiation of hESCs and blockade of BMP4 signaling is essential in maintaining the pluripotency of hESCs (Xu et al., 2002; Pera, 2004; Pera et al., 2004; Xu et al., 2005b), we have found when BMP4 is combined with bFGF or Activin A, its induction of trophectoderm is greatly reduced.

Previous studies have found that BMP and FGF signaling modulate Activin-induced mesendoderm differentiation in mouse ES differentiation (Valdimarsdottir and Mummery, 2005; Morrison et al., 2008; Hansson et al., 2009). Vallier et al. (Vallier et al., 2009) presented a three-step protocol demonstrating that the combination of Activin, FGF and BMP signaling promotes hESC differentiation into mesendoderm. An FGF signaling inhibitor SU5402 is used in the second step of this three-step protocol. Results of our separate studies on early differentiation are consistent with Vallier's report. A recent paper by Nostro et al. (Nostro et al., 2011) also uses these growth factors, in combination with Wnt and VEGF, to induce endoderm differentiation from human pluripotent stem cells. Our protocol, which differs from both protocols in timing, growth factors used, and doses of growth factors, also demonstrates that a combination of bFGF, Activin A, and BMP4 (FAB) without insulin, in serum-free chemically-defined medium, results in reliable anterior primitive streak (APS)/DE production and FAB-treated cells can be further differentiated into pancreatic lineage cells.

The basic biology of these processes of further commitment of APS/DE to the pancreatic lineage and final maturation to endocrine cells are still poorly understood despite some knowledge from studies of other species (Vesque et al., 2000; Lammert et al., 2001; Kumar et al., 2003; Gu et al., 2002; Wells and Melton, 1999) (Stafford et al., 2004; Lau et al., 2006). Induction of posterior foregut is marked by expression of FOXA2, HNF1ß and HNF6 (D'Amour et al., 2006; Kroon et al., 2008). Co-expression of SOX9 and PDX1 marks pancreatic commitment (Seymour et al., 2007). Zhou et al. (Zhou et al., 2007) reported that a multipotent progenitor domain exists at the tip of the branching pancreatic tree in murine embryos. These progenitors, which are PDX1+PTF1a+cMychighCPA+, go through a competence window for the temporary expression of NGN3. Lineage tracing experiments show that NGN3+ cells are the progenitors of all endocrine cells (Gradwohl et al., 2000).

Using information gathered from the literature regarding development of the pancreas, we have devised a unique 4 stage differentiation protocol, based on FAB treatment during the first stage and EB formation in the presence of FGF2 during the second stage (Fig. 1). In the in vitro differentiation protocol presented here, late-stage cells are embedded in Matrigel because we hypothesize that extracellular matrix may help cells form three dimensional structures that promote cell-cell contact and create a more islet-like environment. Distinct sphere-structured cell clusters, called PDX1+-spheres, spontaneously form under these conditions. The cells in PDX1+-spheres are almost entirely PDX1+, SOX9+, FOXA2+, HNF1ß+, and HNF6+, and some of them are PTF1a+, CPA1+, NGN3+ and NKX6.1+. Budding/branching structures are observed in some PDX1+-spheres. At the end of the culture period, we demonstrate cells that co-stain with PDX1 and insulin, and are able to detect C-peptide in the media, secreted by the cells.

Figure 1
Diagram of the differentiation protocol.


2.1 Combined treatment with bFGF, Activin A and BMP4 in serum-free, chemically defined media results in mesendoderm and APS/DE induction

Our long-term objective is the differentiation of hESCs into pancreatic beta cells to be used as a transplantable cure for patients with Type 1 diabetes. It is widely believed that differentiation protocols that mimic in vivo developmental processes and begin with APS/DE formation will be the most successful in achieving this goal. Starting with a protocol aimed to initiate APS/DE formation involving treatment of hESCs with BMP4 and bFGF in conditioned medium, we ultimately developed a 4 stage protocol (Fig. 1) in which cells are differentiated in chemically-defined media through stages that mimic pancreatic differentiation in embryos, to an end-point of PDX1+/C-peptide+/Insulin+ cells.

Aiming to direct the differentiation of hESCs to APS/DE, we considered prior reports studying the effects of BMP4 on lineage differentiation from ESCs. In one paper, it was shown that BMP4 blocks neuroectoderm differentiation from mESCs (Ying et al., 2003); other reports suggest that BMP4 in combination with other growth factors, such as bFGF, results in mesendoderm and DE formation (Phillips et al., 2007; Willems and Leyns, 2008; Vallier et al., 2009; Micallef et al., 2007) and reviewed in (Zorn and Wells, 2009). Furthermore, data from non-mammalian systems indicate a potential role for BMP signaling in pancreatic endoderm formation (Kumar et al., 2003)

We initially tested a differentiation protocol in which hESCs grown on Matrigel (MG) in MEF-conditioned medium (CM) were treated for 4 days with 10 ng/ml BMP4 and 100 ng/ml bFGF. Although growing cells for 7 days with a relatively high dose of BMP4 (100 ng/ml) induces differentiation into trophoblast cells expressing human chorionic gonadotrophin (hCG) (Supp. Fig. 1 and (Xu et al., 2002; Vallier et al., 2009)), and treatment with 25 ng/ml BMP4 for 24 hours results in mesoderm induction (Zhang et al., 2008), treatment with an intermediate dose of 50 ng/ml BMP4 in combination with bFGF for 4 days resulted instead in significant increases in APS/DE gene expression, including GSC (p=0.0002), MIXL1 (p<0.0001), Brachyury (p<0.0001), SOX17 (p<0.0001), and FOXA2 (p=0.0005) (Fig. 2), when compared with undifferentiated hESCs. Increases in transcript levels for these five genes were much greater in cells treated with both BMP4 and bFGF, compared to conditions in which either factor was added alone, and hCG expression was lower in BMP4/bFGF treated cultures compared to high dose BMP4 alone treated cells (Supp. Fig. 1). Thus, bFGF can inhibit or partially override the ability of BMP4 to induce trophoblast, and together, bFGF and BMP4 appear to promote an APS/mesendoderm fate, as characterized by expression of Brachyury, GSC, MIXL1 and SOX17. However, expression of the trophectoderm marker CDX2 is still present (Supp. Fig. 2), suggesting that not all cells have adopted an APS fate.

Figure 2
Effect of different combinations of growth factors on endoderm, mesoderm, and primitive streak development. QPCR analysis was performed on cells after 4 days of differentiation with growth factors included as indicated. Data are shown as fold change versus ...

Further differentiation of BMP4/bFGF treated cells through Stage 2 and Stage 3 resulted in cultures containing cells positive for PDX1/Insulin/C-peptide (Supp. Fig. 3), as well as cells expressing somatostatin or glucagon (Supp. Fig. 4). Addition of the BMP4 antagonist noggin during Stage 1 led to significantly reduced levels of expression of SOX17, FOXA2, Brachyury, and PDX1 at the end of stage 2, suggesting that the observed differentiation requires BMP4 signaling (Supp. Fig. 5).

These observations suggested that BMP4 and bFGF were sufficient to induce DE from hESCs; however, the role of Activin A in this process is indisputable (D'Amour et al., 2005; Hansson et al., 2009; Vallier et al., 2009). Activin A activity is present in undifferentiated ESC cultures grown on MEFS, as it has been suggested that bFGF treatment of MEFs may induce MEFs to secrete Activin A (Greber et al., 2007; Greber et al., 2010). Hence, to further characterize the mechanism of mesendoderm promotion by BMP4 and bFGF, we examined the role of Activin A in Stage 1 of our differentiation protocol, by adding the Activin receptor inhibitor SB431542 to cells treated with bFGF+BMP4 in CM. This treatment completely inhibited expression of GSC (p=0.4204), MIXL1 (p=0.89710, T (p=0.2113), and FOXA2 (p=0.1271) (Fig. 2), which was not significantly different from that of undifferentiated hESCs. This finding indicates that the differentiation to a mesendoderm fate occurs through the activation of the Activin pathway. Although no exogenous Activin A was added to our media, both CM and Matrigel contain Activin A/TGFβ activity (Vukicevic et al., 1992; Beattie et al., 2005). Indeed, when unconditioned media (UM) was substituted for CM, the inductive effect of bFGF+BMP4 on SOX17, FOXA2, and Brachyury was lost (data not shown).

Unlike UM, CDM contains no serum or serum-replacement, both of which contain insulin. Recent papers have shown that Activin A promotes DE differentiation from hESCs but that this occurs only when PI3K signaling pathway is inhibited (McLean et al., 2007). When we found that adding Activin A to CM did not improve the differentiation of hESCs to mesendoderm, we posited this was due to the presence of insulin, a PI3K agonist, which is a component of SR, contained in this media. Thus, we studied gene expression in Activin A treated cultures in the presence and absence of insulin. When Activin A is combined with insulin, the induction of SOX17 is significantly decreased (p=0.0154) and induction of Brachyury is significantly increased at the first stage (p=0.0017). Thus, it appears insulin promotes or perhaps maintains mesendoderm (T), as well as further mesoderm differentiation (MEOX1 and TBX6 and KDR) but inhibits DE differentiation (CXCR4; Supp. Fig. 6).

To date, many published protocols that use Activin A treatment to induce DE differentiation include a small amount of FBS (D'Amour et al., 2005), or use primarily low dose Activin A which is known to favor mesoderm fates (Vallier et al., 2009). To determine whether treatment with Activin A alone is sufficient to promote differentiation from APS through pancreatic endoderm to pancreatic endocrine cells in our protocol, we developed a chemically-defined serum free medium (CDM). Cells grown in CDM supplemented with Activin A alone showed significant induction of MIXL1, GSC, T, SOX17 and FOXA2 (p<0.0001 compared to undifferentiated cells in CM; Fig. 2). Despite this promising early differentiation, EBs formed from Activin A-treated hESCs failed to develop. Thus, it was impossible to assess these cells in later stages of differentiation. We next tried treating cells with bFGF+BMP4 in CDM, and found good expression of SOX17 and T, although levels of MIXL1, GSC, and FOXA2 expression were much lower than in cultures treated with Activin A (Fig. 2). However, when the three growth factors were combined (bFGF+Activin A+BMP4, hereafter referred to as FAB), expression levels of all five transcripts were comparable to or greater than those seen with Activin A alone (Fig. 2), a finding that is consistent with the DE induction achieved by different Activin, bFGF, and BMP4 based serum-free protocols (Nostro et al., 2011; Vallier et al., 2009; Rezania et al., 2011; Green et al., 2011). Immunostaining confirmed the expression of FOXA2, SOX17, and T at the protein level (Fig. 3 and Supp. Fig. 7) following this treatment. Whereas the SOX17 and FOXA2 staining was similar among the groups tested, there appears to be a greater proportion of T+Foxa2 cells in cultures grown in FAB, FAB+insulin, and BMP4+Activin A compared with cells treated with Activin A alone (Supp. Fig. 7), which correlates with the ability of the treated cells to form EBs, as described above. In addition, FAB-treated cells show upregulation of T and NKX2.5 transcripts, but not KDR or TBX6 transcripts compared to undifferentiated cells (Supp. Fig. 8). Thus, it is possible that mesoderm produced at Stage 1 and/or during EB formation allows definitive endoderm cells to survive, as has been hypothesized previously in murine (Tada et al., 2005) and human ESC systems (Xu et al., 2006). Activin A also inhibited the ability of BMP4 to induce differentiation of hESCs to trophoblasts. CDX2, an early marker of trophoblast differentiation induced by treatment with bFGF+BMP4, is not expressed when cultures are grown in FAB (Supp. Fig. 2).

Figure 3
Differentiation of hESCs to APS/DE following growth factor treatment. Cells were differentiated for 4 days (Stage 1) and then stained for SOX17 (red) and FOXA2 (green). Topro3 (blue) marks nuclei. Growth factor concentrations: bFGF, 100 ng/ml; Activin ...

We assessed the effect of the PD173074, a specific inhibitor of FGF receptor signaling (Bansal et al., 2003), on FAB-treated cells. Transcript levels of GSG, MIXL1, SOX17, and FOXA2 are reduced dramatically following addition of 100 nM PD173074 (Fig. 2), suggesting that signaling controlled by bFGF is important in this differentiation protocol. To confirm the role of PI3K in endoderm differentiation, we added the PI3K inhibitor LY294002 during FAB treatment, and analyzed gene expression after three days. We found that specific PI3K inhibition with 50 μM LY294002 did not have a dramatic effect on transcript levels of APS/DE genes when compared with FAB-treated cells: SOX17 and GSC expression were increased slightly, while T and FOXA2 expression were reduced somewhat (Fig. 2). This is likely due to the fact that CDM is insulin-free, and is consistent with our result showing that insulin (a PI3K agonist) has an inhibitory effect on FAB-induced DE induction and enhances mesendoderm gene transcription (Fig. 2). These results recapitulate the role of PI3K inhibition in endoderm formation from human pluripotent stem cells as first elucidated by McLean et al. (McLean et al., 2007).

These findings confirm those of Vallier et al. (Vallier et al., 2009) who also identified the mesendoderm induction capabilities of the combination of BMP4, bFGF, and Activin A. Furthermore, EBs can be readily generated from FAB-treated human ES cells, enabling further differentiation to pancreatic endoderm and endocrine cells (see Section 2.2). We therefore used FAB treatment in CDM for Stage 1 differentiation of hESCs to DE to test further differentiation strategies to pancreatic lineages.

2.2 FAB-treated hESCs differentiated as embryoid bodies (EBs) begin to express pancreatic progenitor markers

Our previous experiments showed that differentiation through an EB stage, in which inductive tissue interactions may occur in three dimensions among the early embryonic germ layers, positively influences development of pancreatic lineage cells, compared to differentiation under two-dimensional conditions (Xu et al., 2006). We have tried different combinations of growth factors at different stages to get maximum expression of PDX1 and Insulin. Cells differentiated following our initial protocol of treatment with BMP4+bFGF in CM in Stage 1 followed by growth as EBs in CM+bFGF (bFGF concentrations ranging from 20ng/ml to 100ng/ml) during Stage 2 resulted in a maximum level of PDX1 transcript accumulation at EB14 (14 days of suspension culture). In these cultures, the PDX1 transcript levels were comparable to the level of PDX1 expression in a 50% pure adult human islet preparation. Immunostaining showed a greatly increased number of PDX1+ cells, and some of these cells also expressed Ki67, suggesting PDX1+ cells are proliferative at this stage (Supp. Fig. 9). In addition, many cells in these cultures expressed PTF1a. However, we found that this extremely strong expression of PDX1, even in conjunction with expression of PTF1a, did not lead to robust Insulin expression at later stages (data not shown).

On the other hand, cells treated with FAB in stage 1 and then grown with ITS+bFGF in suspension culture during stage 2 have a relatively later peak of PDX1+ expression and display strong Insulin expression at later times. At the end of stage 2, QPCR showed a significant increase in PDX1 transcripts in EBs made from FAB-treated cells compared to EBs grown from untreated hESCs (fold change of 175, Fig. 4). Transcript levels of FOXA2 were increased by 40-fold in FAB-treated cultures at Stage 2 versus untreated controls. The increased expression of SOX9, HNF6, CPA, and NGN3 in FAB-treated cells (Supp. Fig. 10) suggest that cells have begun to differentiate into foregut progenitor cells, which is supported further by immunostaining. At the end of Stage 2, FAB-treated hESCs express FOXA2, SOX9 and PDX1 (Fig. 5). FOXA2 appeared widely distributed, and some cells also expressed SOX9. PDX1+ cells appeared in small clusters among FOXA2+ cells. Although not all cells at this stage expressed PDX1, PDX1+ cell clusters were found in 78% ±.07% of EBs. Thus, there is a transition from endoderm-committed cells to foregut progenitors from Stage 1 to the end of Stage 2 during suspension culture, but this is not complete until Stage 3.

Figure 4
FAB-treatment of cells in Stage 1 results in the highest transcript levels of FOXA2 and PDX1 at Stages 2 and 3. Fold change values of foxa2 (a) and pdx1 (b) transcripts at Stage 2 and 3 for cells treated at Stage 1 with either CMBF, ITSFAB or FAB compared ...
Figure 5
Expression of FOXA2, PDX1, and SOX9 at the end of Stage 2. Stage 2 EBs were plated overnight and then stained for FOXA2 (red) and PDX1 or SOX9 (green). Nuclei are marked with Topro3 (blue). Small focal areas of cells stain for PDX1 or SOX9; the majority ...

2.3 Formation of cell spheres expressing FOXA2, SOX9, HNF6, HNF1β, PDX1 as well as CPA, Ki67, NKX6.1, NGN3, MAFA and Insulin

To investigate further the differentiation potential of the cells grown through Stages 1 and 2, Stage 2 cells were plated in MG on coverslips in serum-free ITSFINE media containing insulin, transferrin, selenium, FGF7, INGAP, nicotinamide, and exendin 4 for an additional 21–28 days (Stage 3). These reagent additives were chosen rationally based on their defined properties as previously described in the literature. Duct-like structures started to form around day 14 of Stage 3 (not shown). Starting around day 21, sphere-like cell clusters were formed (Fig. 6, top). Note that the cell clusters shown are not EBs; these cell clusters formed after plating. Immunostaining revealed that the cell clusters were highly enriched for FOXA2 (98% ±0.1%), SOX9 (95% ±.03%), HNF6 (97% ±.04%), HNF1β (100%) and PDX1 (97% ±.04%) positive cells (Figs. 6 and and7).7). Within a single well (1.88 cm2) there are an average of 47 spheres (referred to as PDX1+-spheres), of which 89% have expression of these markers. By the end of Stage 3, most cells in the wells that are not PDX1+-spheres die, resulting in a highly purified population. Furthermore, the development of discrete PDX1+-spheres that contain many pancreatic lineage cells enables for a novel type of purification, in which trypsinization can be used to remove individual cells that are not part of a PDX1+-sphere, while cell clusters are retained for Stage 4. The significant increase in the number of PDX1+ cells from stage 2 to stage 3 (e.g. compare Fig. 5 with Figs. 6, ,77 and and8)8) parallels the increase in pdx1 transcripts (Fig. 4) and suggests either that Stage 3 medium is selective for growth of this population or that the subset of Stage 2 DE cells that were not PDX1+ progressively differentiated into PDX1+ cells. Within these large areas of PDX1+ cells, some cells also co-express NKX6.1, PTF1a (Fig. 7), or Insulin (Fig. 8). Branch-like structures sometimes appear within cell-clusters, and are co-stained with PDX1 and CPA, and a substantial number of PDX1+ cells begin to express NGN3 at this stage (Fig. 7). Cells outside of the cell clusters were never found to express these markers. Proliferating Ki67+ cells were found both within and outside of PDX1+-sphere cells (Fig. 8). In many areas of PDX1+ cells, there is co-expression of MafA, which is not seen outside the PDX1+ regions (Fig. 8). Mature beta cells express MafA but not MafB (Nishimura et al., 2006). In our hands, MafB staining has not been successful, so full interpretation of MafA staining is not possible. Insulin-positive cells were found within cell clusters throughout Stage 3, and co-stain with both PDX1 and C-peptide (not shown). At the end of Stage 3, the mean number of positive cells per culture well (in 24-well plates) is 168±7.

Figure 6
Expression of endoderm and pancreatic markers in Stage 3 cells. Top, bright field image of Stage 3 cells; panels are the same field at different magnifications. Cells were stained for FOXA2, PDX1, HNF6, and HNF1beta, as shown. Topro 3 (blue) marks nuclei. ...
Figure 7
Expression of pancreatic markers in Stage 3 cells. Cells were stained for PDX1, SOX9, NKX6.1, PTF1a, and carboxypeptidase a (CPA), and NGN3. Nuclei (blue) were marked by Topro 3. Scale bars 50 μm.
Figure 8
Expression of endocrine and late stage pancreas markers in Stage 3 cells. Many cells co-express PDX1 and insulin, or C-peptide and insulin. Many PDX1+ cells at this stage are proliferative, as shown by co-expression of PDX1 and Ki67. MAFA is present in ...

QPCR data suggest that Stage 1 treatment of hESCs influences their differentiation capacities at subsequent stages. For example, FAB treatment results in higher expression of PDX1 and FOXA2 at Stages 2 and 3 compared to CMBF or ITSFAB treatment (Fig. 4) and increased PDX1 transcript accumulation than observed in cells cultured using the D'Amour protocol (Supp. Fig. 10). The level of FOXA2 transcripts in FAB-treated cultures diminishes from Stage 2 to 3 (41 and 21 fold changes versus untreated spontaneously-differentiated EBs), whereas PDX1 transcripts increase from 175 to 1158 fold change in Stage 2 and 3, respectively, as would be expected if endoderm is further differentiating to the foregut/pancreas lineage. A timecourse of gene expression reveals that SOX9, HNF6, CPA, and NGN3 transcript levels are very low or undetectable in undifferentiated hESCs and Stage 1 cells. At Stage 2 there is a modest increase in SOX9 transcripts, and a 73 fold increase in HNF6 expression, compared with hESCs. By Stage 3, expression levels of each of these transcripts have increased compared to the baseline expression level: SOX9, 27.5 fold; HNF6, 311 fold; CPA, 13 fold; NGN3, 78 fold (Supp. Fig. 10).

2.4 A final differentiation stage encourages terminal differentiation and results in an increase in Insulin+ cells

For Stage 4, ITSFINE media was switched to NB media (nicotinamide and B27) for an additional 5–14 days in order to promote endocrine differentiation. At the end of this stage, the insulin content increased dramatically, both at transcript level (delta CT went from 9 to 5) and at protein level (Fig. 4; Fig. 9) compared to Stage 3. The mean number of Insulin+ cells per culture well (in a 24-well plate) was 1795±71, ten times the number seen at the end of Stage 3. The total cell number between Stage 3 and Stage 4 is relatively constant (data not shown), indicating that the percentage of insulin+ cells increases from Stage 3 to Stage 4. In Stage 3 cultures, Insulin+ cells were more scattered, whereas by the end of the fourth stage, some Insulin+ cells are found clustered together. The clustering and increase in number of cells suggests the possibility that these new Insulin+ cells arose by either proliferation of existing Insulin+ cells or differentiation from some type of progenitor cell, or both. Immunostaining of cells at the end of Stage 3 demonstrated that some Insulin+ cells co-expressed Ki67, suggesting that some Insulin+ cells were dividing at this time. However, the majority of Insulin+ cells do not stain with Ki67, and in the 4th stage, virtually no Insulin+ Ki67+ co-stained cells are observed (not shown). Therefore, it is likely that most Insulin+ cells in these late stage cultures arise by differentiation from Insulin-negative cells.

Figure 9
Expression of endocrine markers in Stage 4 cells. The number of insulin+ cells increases dramatically from Stage 2 (average of 168 insulin+ cells per coverslip) to Stage 4 (average of 1795 insulin+ cells per coverslip). Most insulin+ cells also express ...

We have compared the results of this 4 Stage FAB protocol with differentiation observed when following the protocol described by D'Amour et al. (D'Amour et al., 2006). Supp. Fig. 10 shows the ΔCT values for FOXA2, PDX1, and Insulin during the FAB protocol and the D'Amour protocol. Transcript levels of FOXA2 are higher in Stage 5 of the D'Amour protocol compared with FAB-treated cells at Stage 4, whereas expression of PDX1 and insulin are higher in FAB-treated cells at the same late timepoints. Staining with antibodies to PDX1 and insulin revealed positive cells from both protocols (data not shown).

The Insulin+ cells in these cultures also co-express C-peptide (Fig. 9), demonstrating that the cells are synthesizing Insulin and not taking it up from the media. The cultures also have many somatostatin- and glucagon-expressing cells (Fig. 9). In some cases, these cells co-express Insulin, revealing that they are immature endocrine cells. However, the vast majority of Insulin+ cells at this stage do not co-express other pancreatic hormone proteins. The literature (Cabrera et al., 2006) shows the ratio of Insulin+:Glucagon+:Somatostatin+ cells in human islets as approximately 1 : 0.74 : 0.19. In our cultures at Stage 4, the ratio is 1:0.46:1.09, revealing that our cultures have fewer glucagon+ cells and many more somatostatin+ cells than typically found in adult human islets. The average C-peptide concentration detected in media at the fourth stage is approximately 83 ±16 pM (n=8). The presence of various endocrine cell types indicates the authenticity of pancreatic lineage differentiation. Additionally, we see low levels of expression of markers of other endoderm-derived organs, including TAT, CDX1, and TITF1 (data not shown), suggesting that the differentiation promoted by this protocol is specific to pancreas.


A better understanding of extrinsic signals regulating early fate choices of hESCs will drive the discovery of successful differentiation protocols for producing functional cell types for clinical applications. Here, we examine how bFGF, Activin A/Nodal, BMP4 (FAB) and insulin modulate mesendoderm and DE differentiation of hESCs grown under chemically defined conditions in adherent monolayer culture.

Prior reports exploring the effects of Activin/Nodal, FGF, and BMP signaling in mouse and human ESCs have yielded conflicting results which need to be resolved. Previous differentiation studies in human ESCs showed that the expression of GSC and other anterior primitive streak genes were induced by high dose Activin A but were prevented by treatment with BMP4 and concomitant inhibition of FGF signaling (D'Amour et al., 2005). On the other hand, high dose BMP4 treatment alone readily promotes extra-embryonic endoderm or trophoblast lineage differentiation (Xu et al., 2005b; D'Amour et al., 2005). However, studies in mESCs indicate that either BMP4 alone, Activin A alone or both together induced primitive streak markers, but that BMP4 exerts a dominant effect (Nostro et al., 2008; Jackson et al., 2010). Other studies suggest that a BMP4/Activin-based culture protocol is useful for initiating the production of hESC-derived cardiomyocytes (Laflamme et al., 2007). Similarly, short-term BMP4 treatment alone of hESCs promotes mesoderm differentiation (Zhang et al., 2008), while Activin A alone treatment has been shown to maintain self-renewal properties (Xiao et al., 2006) and Activin A/Nodal and FGF coordinately maintain pluripotency in hESCs (Vallier et al., 2005). On the other hand, Vallier et al. and others also studied the effects of stimulating or inhibiting BMP4, bFGF and Activin signaling pathways in hESCs and found that Activin signaling is permissive but not sufficient for differentiation of hESCs toward mesendoderm fate but the combination of all three factors efficiently induced mesendoderm differentiation (Vallier et al., 2009). Like these earlier studies, our experimental results indicate that early cell fates are highly dependent on the combinatorial effects of these growth factors. We find that the induction of GSC, MIXL1, SOX17 and FOXA2 expression by Activin A is not adversely affected by the addition of BMP4. Moreover, adding both BMP4 and bFGF to Activin A-treated cells in serum/serum replacement-free defined media also maintains the expression of GSC, MIXL1, and FOXA2 induced by Activin A and further enhances the expression of T and SOX17. Importantly, hESCs grown in FAB media survive better than if they are differentiated in Activin A alone, and are able to form EBs, unlike cells treated with Activin A alone. This may be due to the increased presence of mesendoderm and/or mesoderm in these cultures as evidenced by the presence of T+FOXA2 cells, such as shown by Zhang et al. (Zhang et al., 2008) in short-term BMP4 treated cells. On the other hand, our culture conditions produced numerous early Oct4+T+ cells prior to the induction of FOXA2 expression (data not shown) consistent with the early mesendodermal cells produced by Yu et al. (Yu et al., 2011). Thus, the T+FOXA2 cells in our Stage 1 cultures may also represent short-term persistence of these early mesendodermal cells that have not yet transitioned into DE.

We find that insulin also plays a significant inhibitory role in early differentiation. Adding insulin to Activin A does not affect the induction of GSC and MIXL1, but significantly decreases the expression of SOX17 and FOXA2. Similar gene expression changes and reduced DE marker expression occurs when insulin is added to FAB. On the other hand, insulin enhances the expression of T and other mesoderm markers TBX6, MEOX1 and KGR. These gene expression changes can be interpreted as indicating that insulin signaling promotes hESC differentiation into primitive streak and mesendoderm, but not endoderm. Under these conditions, cells may be unable to exit the primitive streak fate, or may be pushed towards the mesoderm lineage. These findings and interpretation are consistent with previously published studies, such as those of McLean et al. who reported that insulin antagonizes DE differentiation of hESCs induced by Activin A (McLean et al., 2007).

Our protocol leads to reliable production of sphere-shaped cell clusters in which essentially all cells express HNF1β, HNF6, FOXA2, SOX9 and PDX1. This pattern of expression is consistent with a posterior foregut fate, and it remains a possibility that some cells within these spheres may adopt a stomach or intestine fate. However, it is clear that many cells within the PDX1+-spheres are committed to a pancreas fate, as judged by their expression of NKX6.1, PTF1a, NGN3, MAFA, and CPA1, which is highly reminiscent of pancreatic progenitor epithelium (Oliver-Krasinski and Stoffers, 2008). The FAB-based culture protocol in chemically-defined media described here has been shown, in preliminary experiments, to yield similar quantities of DE and pancreatic progenitors from three human iPS cell lines, IMR-90, Foreskin and DF 19-9-7T (data not shown), in addition to hESC lines H1 and H9.

Endocrine hormone-producing cells were also generated in this differentiation protocol. Insulin, glucagon, and somatostatin were all found in these cell clusters. Although a small percentage of cells are polyhormonal, most hormone-positive cells express a single hormone and are in close relationship with each other. This is in contrast to a recent paper by Nostro et al., in which a majority of the C-peptide+ cells generated from their protocol also express glucagon or glucagon and somatostatin (Nostro et al., 2011). In our protocol described here, nearly all insulin+ cells also were PDX1+ and C-peptide+, indicative of beta-like cells. Numerous functional studies suggest the importance of the three-dimensional structure of the islet and direct interactions between various islet endocrine cells, including beta cell-beta cell and beta cell-alpha cell interactions (Miller et al., 2009). Compared to a two-dimensional hESC culture system, the PDX1+-sphere is one step advanced in terms of promoting interactions among cells expressing different hormones. Within the PDX1+-sphere, however, some but not all cells are in direct contact with another hormone-producing cell, possibly contributing to suboptimal glucose responsiveness. Notably, all pancreatic lineage-associated marker stained cells were expressed only in cells within PDX1+-spheres, even though there were cells present in the cultures outside of the PDX1+-spheres. This characteristic may facilitate the incorporation of straightforward methods for further purification in the future.

Despite the similarities in the cellular makeup of PDX1+-spheres and islets, there are some important differences. First, PDX1+-spheres are probably not vascularized as are islets, although we have not tested this assumption formally. PDX1+-spheres are three dimensional cell clusters, some of which have branch-like structures and some of which contain different-sized hollow cavities. Without blood vessels feeding these structures, it is hypothesized that cells inside the PDX1+-sphere would obtain nutrients through diffusion. This may explain that when cultured in media containing low glucose concentrations, such as CMRL or low-glucose DMEM (glucose concentration is 5mM), survival and differentiation of PDX1+-spheres were considerably less robust than that seen in high glucose concentration containing media, such as DMEM/F12 (glucose concentration is 17.5 mM). Second, we observed an abnormal ratio of hormonal cells, i.e. a much higher percentage of somatostatin positive cells-to-insulin positive cells in hESC-derived pancreatic cells than in normal human islets. The third difference is that the insulin content of PDX1+-spheres is considerably lower than that of islets. Although the described methods generate numerous PDX1+ cells (transcript level is comparable to that in 50% pure human islets), and an easily detectable number of insulin+ cells, it has proven difficult to produce enriched populations of insulin+ cells despite testing many different growth factor supplements and media formulations as other studies have noted (D'Amour et al., 2006; Nostro et al., 2011; Rezania et al., 2011). In stage 4, culture in RPMI 1640 supplemented with B27 and nicotinamide did result in a 10 fold increase in the number of insulin+ cells. We observed some cells co-stained with insulin and Ki67 suggesting that the increase in the number of insulin+ cells may have been due to the proliferation of pre-existing insulin+ cells present at the end of stage 3 or beginning of stage 4. Most insulin+ cells were not Ki67+ though, in which case they would more likely have recently differentiated directly from progenitors. In spite of a significant increase in the number of insulin+ cells, the insulin content of cells at stage 4 culture is not comparable to that of human islets. The low number of insulin+ cells may be reflective of the relatively low endocrine fate commitment (i.e. low neurogenin 3 protein expression or inappropriate timing of expression or context of expression) and efforts should be focused now on increasing neurogenin 3 expression in PDX1+-spheres. In our current protocol, it takes ~55 days of in vitro culture to obtain about 80 pM of C-peptide secreted into the culture media. Interestingly, in the report published by Kroon et al., they required 12 days of in vitro culture plus 44 days of in vivo maturation to produce 100 pM of C-peptide content in mice serum (Kroon et al., 2008). The time frame required to produce this amount of insulin in these two different protocols is very similar, suggesting there may be a fixed time requirement of human beta cell maturation. Nonetheless, we have not been able to demonstrate glucose-stimulate insulin secretion of FAB-treated cells in vitro. In glucose stimulated insulin release assays, challenge with glucose+IBMX results in a stimulation index of approximately 2. However, Stage 4 cells do not secrete insulin/C-peptide in response to glucose alone.

It is speculated that the extracellular matrix is an important part of a niche for islet maturation. Stage 3 began with EB14 cells embedded in Matrigel. We have observed that cells migrate and form PDX1+-spheres at the end of third stage when plated in Matrigel. Laminin-111 exists in vascular basement membrane in murine fetal islets and is reported to promote insulin gene expression and proliferation of β cells (Jiang et al., 1999; Nikolova et al., 2006). In human islets, the most prominent component of this matrix is laminin-511 (Otonkoski et al., 2008). Another difference is the presence of two basement membranes surrounding blood vessels of human islets instead of a single membrane in murine islets (Otonkoski et al., 2008). Matrigel is rich in mouse laminin-111 and was found to be essential for the induction of islet differentiation from expanded human pancreatic duct cells (Gao et al., 2003; Bonner-Weir et al., 2000). These species differences may help guide further protocol modifications.

In summary, the work presented here sheds light on the early signals regulating formation of primitive streak, mesendoderm and DE from hESCs, and in particular reveals an interplay between FGF, BMP, Activin/Nodal and insulin signaling. Three-dimensional PDX1+-spheres are routinely formed in our culture system, which strongly resemble pancreatic progenitor epithelium. With these refined methods for producing enriched populations of pancreatic progenitors, we believe that efforts should be redoubled to better understand the extrinsic signaling promoting endocrine fate, including the generation of NGN3+ progenitors and conversion of these progenitors into functional endocrine cells.


4.1 Cell culture and differentiation

NIH approved hESC lines, H1 (WA01) and H9 (WA09) were used between passage 18 to 42 and had normal karyotype checked by WiCell cytogenetics services (Madison, WI). Media for undifferentiated ESCs was comprised of 80% DMEM/F12 and 20% Knockout serum replacement supplemented with 1 mM L-glutamine, 1% nonessential amino acids, 0.1 mM 2-mercaptoethanol and 4 ng/ml bFGF (all from Invitrogen). hESCs were cultured in 6-well plates on a feeder layer of irradiated mouse embryonic fibroblasts (MEF) in either ESC media (control group), ESC media plus 50 ng/ml BMP4 (BMP4 group; R&D systems), or ESC media plus 50 ng/ml BMP4 plus 300 ng/ml noggin (noggin group; R&D systems) for 4 days. In experiments with Matrigel, hESCs were grown on growth-factor depleted Matrigel (BD Biosciences) instead of MEF and culture media was MEF-conditioned media (CM, control group), CM plus 50 ng/ml BMP4 or CM plus 50 ng/ml BMP4 plus 100 ng/ml bFGF. In FAB experiments, hESCs plated on MG were cultured in chemically defined medium (CDM) containing DMEM/F12, 1 mM L-glutamine, 1% nonessential amino acids, 0.1 mM 2-mercaptoethanol and 2% BSA, supplemented with 100 ng/ml Activin A, 50 ng/ml BMP4 and 100 ng/ml bFGF for 3–4 days (stage 1), which were equally effective.

Colonies were transferred by incubating with 2 mg/ml dispase (Invitrogen) and put into 100-mm non-treated suspension culture dishes with CM or medium containing DMEM/F12, ITS supplement (BD, 5 μg/ml insulin + 5 μg/ml transferrin + 5 ng/ml selenous acid), and 20–100 ng/ml bFGF on a shaker for 14 days to form embryoid bodies (EBs) (stage 2). EBs formed for 14 days were found to produce better pancreatic differentiation in later stages than 7 day EBs. EBs were then suspended in diluted Matrigel (0.66 mg/ml in cold DMEM/F12) and replated in ITSFINE medium, a serum-free medium comprised of DMEM/F-12 (17.5 mM glucose) with ITS, 10 ng/ml FGF7 (R&D), 10 mM nicotinamide (Sigma), 10 nM exendin-4 (Sigma), 4 μg/ml insulin, 200 nM INGAP (PSN-4765, sequence: IGLHDPSHGTLPNGS) and 2 g/L BSA (Sigma) for 14, 21 or 28 days (stage 3). Longer periods than 28 days under these conditions resulted in excessive cell detachment. At stage 4, media was switched to RPMI 1640 supplemented with B27 and nicotinamide for 4 –14 days. A fraction of each culture from stage 3 and 4 was used for RT-PCR and Q-PCR and the remaining cells were either embedded in OCT (EB14; Tissue-Tek) or fixed on coverslips for immunostaining. Media were collected at the end of stage 3 and 4 for the measurement of C-peptide levels.

4.2 Quantitative PCR, RT-PCR, and C-peptide measurement

Total cellular RNA was extracted with Trizol (Invitrogen). cDNA was synthesized from 1μg total RNA using a SuperScript First-Strand Synthesis kit (Invitrogen). Quantitative real time RT-PCR (QPCR) was performed using Assays-on-demand agents (Applied Biosystems) on an ABI PRISM 7700 Sequence Detection System (Applied Biosystems) See Supplemental Table 1 for a list of the assay numbers. QPCR was performed according to the equipment manufacturer's instructions. Relative quantification was carried out using the comparative cycle threshold (CT) method recommended by the supplier. Fold change was calculated as: 2−ΔΔCT. Mean ΔΔCT values from QPCR analyses were compared using the unpaired, two-tailed Student's t-test. P values < 0.05 were considered significant. hCG was measured on the Dade Behring Dimension Clinical Chemistry System, according to the instructions of the manufacturer.

For non-quantitative RT-PCR, oligonucleotide primer pairs were generated against human transcripts using Genbank sequences. See Supplemental Table 1 for primer sequences, amplicon sizes, and cycle numbers. Primers were selected from two exons that spanned at least one intronic sequence. PCR was performed using HotStarTaq DNA polymerase (Qiagen) and reaction conditions were as follows: initial denaturation at 95°C for 15 min, then cycles of 94°C for 30 sec, 30 sec at annealing temperature, 1 min at 72°C, and a final 10 min extension at 72°C. Primers were annealed at 53°C except for pdx1 (56°C), sox17 (55°C) and foxa2 (50°C; with Qiagen's Q-solution). A control sample without reverse transcriptase (-RT) was amplified with GAPDH primers in all cases, and human adult pancreas RNA was used as a positive control. C-peptide levels in media from stage 3 or 4 cultures were measured using the ultrasensitive C-peptide ELISA (Mercodia).

4.3 Immunofluorescence staining and cell counts

Immunofluorescence staining of coverslips was carried out as previously described (Kahan et al., 2003). The following primary antibodies were used at the listed dilutions: PDX1 rabbit anti-mouse serum 1:4000 (gift of C. Wright); insulin guinea pig anti-human 1:200 (Linco); glucagon mouse monoclonal 1:2000 (Sigma); somatostatin mouse monoclonal 1:2000 (Novonordisk); amylase rabbit 1:2000 (Accurate); Ki-67 mouse monoclonal 1:25 (BD Pharmingen); C-peptide rat monoclonal 1:3000 (BCBC 1921); Brachyury goat anti-human 1:20 (R&D); OCT4 mouse anti-human 1:100 (Santa Cruz); Sox17 goat anti-human 1:40 (R&D); Sox17 rat anti-human 1:400 (Gift of K. D'Amour); FOXA2 rabbit anti-rat1:4000 (Gift of R. Costa); HNF6 rabbit anti human 1:100 (Santa Cruz); HNF1ß goat 1:100 (Santa Cruz); CPA rabbit anti bovine 1:200 (AbD); ngn3 rabbit 1:2000 (Gift of M. German); ptf1a rabbit 1:800 (Gift of C. Wright); NKX6.1 mouse anti rat 1:10 (Developmental Studies Hybridoma Bank); Sox9 rabbit 1:500 (Chemicon). Secondary antibodies (Goat anti-mouse lgG Alexa Fluor 488, 1:2000; Goat anti-rabbit Alexa Fluor 568, 1:4000; Goat anti-rat Alexa Fluor 488, 1:2000; Goat anti-rabbit, Alexa Fluor 647, 1:4000; Goat anti-mouse 568, 1:2000; Donkey anti-goat Alexa Fluor 568, 1:2000; Donkey anti-mouse Alexa Fluor 488, 1:2000) were obtained from Invitrogen/Molecular probes (Eugene, OR).

To determine the percentage of cells within PDX1+-spheres that expressed specific proteins, cells were co-stained with the marker of interest and the nuclear marker Topro3. The number of positive cells for each antibody was determined by dividing the number of nuclei that stained with a given marker by the number of nuclei within each cell cluster, as marked by Topro3.

Supplementary Material

Supp. Fig 1

supp fig 9

supp legends

supp table 1

supp fig 10

supp fig 2

supp fig 3

supp fig 4

supp fig 5

supp fig 6

supp fig 7

supp fig 8


The authors would like to thank C. Kibbe for excellent technical assistance, G. Leverson for assistance with statistics, and K. D'Amour, R. Costa, C. Wright, M. German, and the Beta Cell Biology Consortium for kindly providing antibodies. This study was generously funded by grants from the JDRF (2007-75 and 2009-496), NIH ARRA (DK-78889-1A2) and ADA (7-05-RA-103).


Disclosure Statement: Patents applications involving the work described in this manuscript have been filed by the Wisconsin Alumni Research Foundation (WARF): US 11/799,659 and P110310US01 (application #61495817). XX and JSO have assigned licensing rights to WARF.

Reference List

  • Bansal R, Magge S, Winkler S. Specific inhibitor of FGF receptor signaling: FGF-2-mediated effects on proliferation, differentiation, and MAPK activation are inhibited by PD173074 in oligodendrocyte-lineage cells. J.Neurosci.Res. 2003;74:486–493. [PubMed]
  • Beattie GM, Lopez AD, Bucay N, Hinton A, Firpo MT, King CC, Hayek A. Activin A maintains pluripotency of human embryonic stem cells in the absence of feeder layers. Stem Cells. 2005;23:489–495. [PubMed]
  • Bonner-Weir S, Taneja M, Weir GC, Tatarkiewicz K, Song KH, Sharma A, O'Neil JO. In vitro cultivation of human islets from expanded ductal tissue. Proc.Natl.Acad.Sci.U.S.A. 2000;97:7999–8004. [PMC free article] [PubMed]
  • Cabrera O, Berman DM, Kenyon NS, Ricordi C, Berggren PO, Caicedo A. The unique cytoarchitecture of human pancreatic islets has implications for islet cell function. Proc.Natl.Acad.Sci.U.S.A. 2006;103:2334–2339. [PMC free article] [PubMed]
  • D'Amour KA, Agulnick AD, Eliazer S, Kelly OG, Kroon E, Baetge EE. Efficient differentiation of human embryonic stem cells to definitive endoderm. Nat.Biotechnol. 2005;23:1534–1541. [PubMed]
  • D'Amour KA, Bang AG, Eliazer S, Kelly OG, Agulnick AD, Smart NG, Moorman MA, Kroon E, Carpenter MK, Baetge EE. Production of pancreatic hormone-expressing endocrine cells from human embryonic stem cells. Nat.Biotechnol. 2006;24:1392–1401. [PubMed]
  • Gamer LW, Wright CV. Autonomous endodermal determination in Xenopus: regulation of expression of the pancreatic gene XlHbox 8. Dev.Biol. 1995;171:240–251. [PubMed]
  • Gao R, Ustinov J, Pulkkinen MA, Lundin K, Korsgren O, Otonkoski T. Characterization of endocrine progenitor cells and critical factors for their differentiation in human adult pancreatic cell culture. Diabetes. 2003;52:2007–2015. [PubMed]
  • Gradwohl G, Dierich A, LeMaur M, Guillemot F. neurogenin 3 is required for the development of the four endocrine cell lineages of the pancreas. Proc.Natl.Acad.Sci.U.S.A. 2000;97:1607–1611. [PMC free article] [PubMed]
  • Greber B, Lehrach H, Adjaye J. Fibroblast growth factor 2 modulates transforming growth factor beta signaling in mouse embryonic fibroblasts and human ESCs (hESCs) to support hESC self-renewal. Stem Cells. 2007;25:455–464. [PubMed]
  • Greber B, Wu G, Bernemann C, Joo JY, Han DW, Ko K, Tapia N, Sabour D, Sterneckert J, Tesar P, Scholer HR. Conserved and divergent roles of FGF signaling in mouse epiblast stem cells and human embryonic stem cells. Cell Stem Cell. 2010;6:215–226. [PubMed]
  • Green MD, Chen A, Nostro MC, d'Souza SL, Schaniel C, Lemischka IR, Gouon-Evans V, Keller G, Snoeck HW. Generation of anterior foregut endoderm from human embryonic and induced pluripotent stem cells. Nat.Biotechnol. 2011;29:267–272. [PubMed]
  • Gu G, Dubauskaite J, Melton DA. Direct evidence for the pancreatic lineage: NGN3+ cells are islet progenitors and are distinct from duct progenitors. Development. 2002;129:2447–2457. [PubMed]
  • Hansson M, Olesen DR, Peterslund JM, Engberg N, Kahn M, Winzi M, Klein T, Maddox-Hyttel P, Serup P. A late requirement for Wnt and FGF signaling during activin-induced formation of foregut endoderm from mouse embryonic stem cells. Dev.Biol. 2009;330:286–304. [PMC free article] [PubMed]
  • Jackson SA, Schiesser J, Stanley EG, Elefanty AG. Differentiating embryonic stem cells pass through 'temporal windows' that mark responsiveness to exogenous and paracrine mesendoderm inducing signals. PLoS.ONE. 2010;5:e10706. [PMC free article] [PubMed]
  • Jiang FX, Cram DS, DeAizpurua HJ, Harrison LC. Laminin-1 promotes differentiation of fetal mouse pancreatic beta-cells. Diabetes. 1999;48:722–730. [PubMed]
  • Kahan BW, Jacobson LM, Hullett DA, Oberley TD, Odorico JS. Pancreatic precursors and differentiated islet cell types from murine embryonic stem cells: an in vitro model to study islet differentiation. Diabetes. 2003;52:2016–2024. [PubMed]
  • Kroon E, Martinson LA, Kadoya K, Bang AG, Kelly OG, Eliazer S, Young H, Richardson M, Smart NG, Cunningham J, Agulnick AD, D'Amour KA, Carpenter MK, Baetge EE. Pancreatic endoderm derived from human embryonic stem cells generates glucose-responsive insulin-secreting cells in vivo. Nat.Biotechnol. 2008;26:443–452. [PubMed]
  • Kumar M, Jordan N, Melton D, Grapin-Botton A. Signals from lateral plate mesoderm instruct endoderm toward a pancreatic fate. Dev.Biol. 2003;259:109–122. [PubMed]
  • Laflamme MA, Chen KY, Naumova AV, Muskheli V, Fugate JA, Dupras SK, Reinecke H, Xu C, Hassanipour M, Police S, O'sullivan C, Collins L, Chen Y, Minami E, Gill EA, Ueno S, Yuan C, Gold J, Murry CE. Cardiomyocytes derived from human embryonic stem cells in pro-survival factors enhance function of infarcted rat hearts. Nat.Biotechnol. 2007;25:1015–1024. [PubMed]
  • Lammert E, Cleaver O, Melton D. Induction of pancreatic differentiation by signals from blood vessels. Science. 2001;294:564–567. [PubMed]
  • Lau J, Kawahira H, Hebrok M. Hedgehog signaling in pancreas development and disease. Cell Mol.Life Sci. 2006;63:642–652. [PubMed]
  • McLean AB, D'Amour KA, Jones KL, Krishnamoorthy M, Kulik MJ, Reynolds DM, Sheppard AM, Liu H, Xu Y, Baetge EE, Dalton S. Activin a efficiently specifies definitive endoderm from human embryonic stem cells only when phosphatidylinositol 3-kinase signaling is suppressed. Stem Cells. 2007;25:29–38. [PubMed]
  • Micallef SJ, Li X, Janes ME, Jackson SA, Sutherland RM, Lew AM, Harrison LC, Elefanty AG, Stanley EG. Endocrine cells develop within pancreatic bud-like structures derived from mouse ES cells differentiated in response to BMP4 and retinoic acid. Stem Cell Res. 2007;1:25–36. [PubMed]
  • Miller K, Kim A, Kilimnik G, Jo J, Moka U, Periwal V, Hara M. Islet formation during the neonatal development in mice. PLoS.ONE. 2009;4:e7739. [PMC free article] [PubMed]
  • Morrison GM, Oikonomopoulou I, Migueles RP, Soneji S, Livigni A, Enver T, Brickman JM. Anterior definitive endoderm from ESCs reveals a role for FGF signaling. Cell Stem Cell. 2008;3:402–415. [PubMed]
  • Nikolova G, Jabs N, Konstantinova I, Domogatskaya A, Tryggvason K, Sorokin L, Fassler R, Gu G, Gerber HP, Ferrara N, Melton DA, Lammert E. The vascular basement membrane: a niche for insulin gene expression and Beta cell proliferation. Dev.Cell. 2006;10:397–405. [PubMed]
  • Ninomiya H, Takahashi S, Tanegashima K, Yokota C, Asashima M. Endoderm differentiation and inductive effect of activin-treated ectoderm in Xenopus. Dev.Growth Differ. 1999;41:391–400. [PubMed]
  • Nishimura W, Kondo T, Salameh T, El K, I, Dodge R, Bonner-Weir S, Sharma A. A switch from MafB to MafA expression accompanies differentiation to pancreatic beta-cells. Dev.Biol. 2006;293:526–539. [PMC free article] [PubMed]
  • Nostro MC, Cheng X, Keller GM, Gadue P. Wnt, activin, and BMP signaling regulate distinct stages in the developmental pathway from embryonic stem cells to blood. Cell Stem Cell. 2008;2:60–71. [PMC free article] [PubMed]
  • Nostro MC, Sarangi F, Ogawa S, Holtzinger A, Corneo B, Li X, Micallef SJ, Park IH, Basford C, Wheeler MB, Daley GQ, Elefanty AG, Stanley EG, Keller G. Stage-specific signaling through TGFbeta family members and WNT regulates patterning and pancreatic specification of human pluripotent stem cells. Development. 2011;138:861–871. [PMC free article] [PubMed]
  • Oliver-Krasinski JM, Stoffers DA. On the origin of the beta cell. Genes Dev. 2008;22:1998–2021. [PMC free article] [PubMed]
  • Otonkoski T, Banerjee M, Korsgren O, Thornell LE, Virtanen I. Unique basement membrane structure of human pancreatic islets: implications for beta-cell growth and differentiation. Diabetes Obes.Metab. 2008;10(Suppl 4):119–127. [PubMed]
  • Pera MF. Unnatural selection of cultured human ES cells? Nature Biotechnology. 2004;22:42–43. [PubMed]
  • Pera MF, Andrade J, Houssami S, Reubinoff B, Trounson A, Stanley EG, Ward-van Oostwaard D, Mummery C. Regulation of human embryonic stem cell differentiation by BMP-2 and its antagonist noggin. J.Cell Sci. 2004;117:1269–1280. [PubMed]
  • Phillips BW, Hentze H, Rust WL, Chen QP, Chipperfield H, Tan EK, Abraham S, Sadasivam A, Soong PL, Wang ST, Lim R, Sun W, Colman A, Dunn NR. Directed differentiation of human embryonic stem cells into the pancreatic endocrine lineage. Stem Cells Dev. 2007;16:561–578. [PubMed]
  • Rezania A, Riedel MJ, Wideman RD, Karanu F, Ao Z, Warnock GL, Kieffer TJ. Production of functional glucagon-secreting alpha-cells from human embryonic stem cells. Diabetes. 2011;60:239–247. [PMC free article] [PubMed]
  • Schuldiner M, Yanuka O, Itskovitz-Eldor J, Melton DA, Benvenisty N. Effects of eight growth factors on the differentiation of cells derived from human embryonic stem cells. Proc Natl Acad Sci U S A. 2000;97:11307–11312. [PMC free article] [PubMed]
  • Seymour PA, Freude KK, Tran MN, Mayes EE, Jensen J, Kist R, Scherer G, Sander M. SOX9 is required for maintenance of the pancreatic progenitor cell pool. Proc.Natl.Acad.Sci.U.S.A. 2007;104:1865–1870. [PMC free article] [PubMed]
  • Smith JC, Price BM, Van NK, Huylebroeck D. Identification of a potent Xenopus mesoderm-inducing factor as a homologue of activin A. Nature. 1990;345:729–731. [PubMed]
  • Stafford D, Hornbruch A, Mueller PR, Prince VE. A conserved role for retinoid signaling in vertebrate pancreas development. Dev.Genes Evol. 2004;214:432–441. [PubMed]
  • Tada S, Era T, Furusawa C, Sakurai H, Nishikawa S, Kinoshita M, Nakao K, Chiba T, Nishikawa S. Characterization of mesendoderm: a diverging point of the definitive endoderm and mesoderm in embryonic stem cell differentiation culture. Development. 2005;132:4363–4374. [PubMed]
  • Tam PP, Loebel DA, Tanaka SS. Building the mouse gastrula: signals, asymmetry and lineages. Curr.Opin.Genet.Dev. 2006;16:419–425. [PubMed]
  • Valdimarsdottir G, Mummery C. Functions of the TGFbeta superfamily in human embryonic stem cells. APMIS. 2005;113:773–789. [PubMed]
  • Vallier L, Alexander M, Pedersen RA. Activin/Nodal and FGF pathways cooperate to maintain pluripotency of human embryonic stem cells. J.Cell Sci. 2005;118:4495–4509. [PubMed]
  • Vallier L, Touboul T, Chng Z, Brimpari M, Hannan N, Millan E, Smithers LE, Trotter M, Rugg-Gunn P, Weber A, Pedersen RA. Early cell fate decisions of human embryonic stem cells and mouse epiblast stem cells are controlled by the same signalling pathways. PLoS.ONE. 2009;4:e6082. [PMC free article] [PubMed]
  • Vesque C, Ellis S, Lee A, Szabo M, Thomas P, Beddington R, Placzek M. Development of chick axial mesoderm: specification of prechordal mesoderm by anterior endoderm-derived TGFbeta family signalling. Development. 2000;127:2795–2809. [PubMed]
  • Vukicevic S, Kleinman HK, Luyten FP, Roberts AB, Roche NS, Reddi AH. Identification of multiple active growth factors in basement membrane Matrigel suggests caution in interpretation of cellular activity related to extracellular matrix components. Exp.Cell Res. 1992;202:1–8. [PubMed]
  • Wells JM, Melton DA. Vertebrate endoderm development. Annual Review of Cell Developmental Biology. 1999;15:393–410. [PubMed]
  • Willems E, Leyns L. Patterning of mouse embryonic stem cell-derived pan-mesoderm by Activin A/Nodal and Bmp4 signaling requires Fibroblast Growth Factor activity. Differentiation. 2008;76:745–759. [PubMed]
  • Xiao L, Yuan X, Sharkis SJ. Activin A maintains self-renewal and regulates fibroblast growth factor, Wnt, and bone morphogenic protein pathways in human embryonic stem cells. Stem Cells. 2006;24:1476–1486. [PubMed]
  • Xu C, Rosler E, Jiang J, Lebkowski JS, Gold JD, O'sullivan C, Delavan-Boorsma K, Mok M, Bronstein A, Carpenter MK. Basic fibroblast growth factor supports undifferentiated human embryonic stem cell growth without conditioned medium. Stem Cells. 2005a;23:315–323. [PubMed]
  • Xu RH, Chen X, Li DS, Li R, Addicks GC, Glennon C, Zwaka TP, Thomson JA. BMP4 initiates human embryonic stem cell differentiation to trophoblast. Nature Biotechnology. 2002;20:1261–1264. [PubMed]
  • Xu RH, Peck RM, Li DS, Feng X, Ludwig T, Thomson JA. Basic FGF and suppression of BMP signaling sustain undifferentiated proliferation of human ES cells. Nat.Methods. 2005b;2:185–190. [PubMed]
  • Xu X, Kahan B, Forgianni A, Jing P, Jacobson L, Browning V, Treff N, Odorico J. Endoderm and pancreatic islet lineage differentiation from human embryonic stem cells. Cloning Stem Cells. 2006;8:96–107. [PubMed]
  • Yasunaga M, Tada S, Torikai-Nishikawa S, Nakano Y, Okada M, Jakt LM, Nishikawa S, Chiba T, Era T, Nishikawa S. Induction and monitoring of definitive and visceral endoderm differentiation of mouse ES cells. Nat.Biotechnol. 2005;23:1542–1550. [PubMed]
  • Ying QL, Nichols J, Chambers I, Smith A. BMP induction of Id proteins suppresses differentiation and sustains embryonic stem cell self-renewal in collaboration with STAT3. Cell. 2003;115:281–292. [PubMed]
  • Yu P, Pan G, Yu J, Thomson JA. FGF2 sustains NANOG and switches the outcome of BMP4-induced human embryonic stem cell differentiation. Cell Stem Cell. 2011;8:326–334. [PMC free article] [PubMed]
  • Zhang P, Li J, Tan Z, Wang C, Liu T, Chen L, Yong J, Jiang W, Sun X, Du L, Ding M, Deng H. Short-term BMP-4 treatment initiates mesoderm induction in human embryonic stem cells. Blood. 2008;111:1933–1941. [PubMed]
  • Zhou Q, Law AC, Rajagopal J, Anderson WJ, Gray PA, Melton DA. A multipotent progenitor domain guides pancreatic organogenesis. Dev.Cell. 2007;13:103–114. [PubMed]
  • Zorn AM, Wells JM. Vertebrate endoderm development and organ formation. Annu.Rev.Cell Dev.Biol. 2009;25:221–251. [PMC free article] [PubMed]
PubReader format: click here to try


Related citations in PubMed

See reviews...See all...

Cited by other articles in PMC

See all...


Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...