SEARCH

SEARCH BY CITATION

Keywords:

  • Embryonic stem cells;
  • Endoderm;
  • Pancreas;
  • Development;
  • Genetic manipulation

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References

Human embryonic stem cells (HESCs) are pluripotent cells that may serve as a source of cells for transplantation medicine and as a tool to study human embryogenesis. Using genetic manipulation methodologies, we have investigated the potential of HESCs to differentiate into the various pancreatic cell types. We initially created various HESCs carrying the enhanced green fluorescent protein (eGFP) reporter gene under the control of either the insulin promoter or the pancreatic and duodenal homeobox factor-1 (Pdx1) promoter. Our analysis revealed that during the differentiation of HESCs into embryoid bodies (EBs), we could detect green fluorescent cells when eGFP is regulated by Pdx1 promoter but not by insulin promoter. To examine whether we can induce differentiation into pancreatic cells, we have established human embryonic stem cell lines that constitutively express either Pdx1 or the endodermal transcription factor Foxa2. Following differentiation into EBs, the constitutive expression of Pdx1 enhanced the differentiation of HESCs toward pancreatic endocrine and exocrine cell types. Thus, we have demonstrated expression of several transcription factors that are downstream of Pdx1 and various molecular markers for the different pancreatic cell types. However, the expression of the insulin gene could be demonstrated only when the cells differentiated in vivo into teratomas. We conclude that although overexpression of Pdx1 enhanced expression of pancreatic enriched genes, induction of insulin expression may require additional signals that are only present in vivo.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References

Maintaining blood glucose at normal physiological levels is essential for normal body function. Controlled insulin secretion by the β cells of the pancreas is the mechanism that enables glucose homeostasis. The pancreas arises as dorsal and ventral buds that emanate from the embryonic endoderm. In the mouse, at embryonic day 9.5 (E9.5), signaling from the notochord induces the patterning of these endodermal cells that are destined to form the pancreatic buds (reviewed in [1]). Later, there is rapid cellular differentiation and proliferation into exocrine and endocrine cells. The endocrine cells are α, β, δ, and pancreatic polypeptide cells, secreting glucagon, insulin, somatostatin, and pancreatic polypeptide (PPY), respectively. In the mature pancreas, these cells form clusters that comprise the islets of Langerhans. Many transcription factors are involved in the development of the pancreatic cell lineages (reviewed in [2]). In humans at 8 weeks of embryonic development, insulin-containing cells appear, and they coexpress glucagon and somatostatin [3]. The majority of the insulin-containing cells stain negative for the other pancreatic hormones after 9 weeks of embryonic development, which is regarded as a sign of maturation [3].

Human embryonic stem cells (HESCs) are pluripotent cells capable of differentiating into the three embryonic germ layers, that is, ectoderm, mesoderm, and endoderm. The pluripotency of HESCs has been proven both in vivo and in vitro [4, [5]6]. HESCs injected in vivo into immune-deficient mice generate teratomas that are composed of derivatives of all three embryonic germ layers [4, 6]. In vitro, aggregation of HESCs brings about the formation of embryoid bodies (EBs) expressing molecular markers specific to the three embryonic germ layers [5]. Since the differentiation of HESCs is spontaneous, it needs to be directed in order to enrich a specific subset of cells. The ability to induce specific differentiation was demonstrated upon the addition of various growth factors to differentiating HESCs [7]. The isolation of a desired population of differentiated cells was lately demonstrated by genetic labeling of hepatic-like cells derived from HESCs. The hepatic-like cells were labeled by a reporter gene, controlled by hepatic promoter, and sorted from the heterogeneous population of cells using fluorescence-activated cell sorter (FACS) [8].

Differentiation of HESCs into insulin-producing cells was demonstrated by spontaneous differentiation in adherent or suspension culture conditions and by a stepwise protocol of various culture conditions [9, 10]. In mouse embryonic stem cells, there are also several reports of insulin-producing cells by spontaneous and directed differentiation (reviewed in [11]). The conclusions obtained from these experiments are controversial because it was shown that immunostaining for the insulin protein in culture is misleading, since the insulin from the growth media adheres to the cells, and insulin may not be produced by the cells themselves [12, 13].

In this study, by using genetic manipulation methodologies, we have analyzed the potential of HESCs to differentiate into the various cell lineages comprising the pancreas. By introducing a reporter gene under the control of the pancreatic and duodenal homeobox factor-1 (Pdx1) promoter, we could trace putative pancreatic precursor cells during the differentiation of HESCs into EBs. Next, we constitutively expressed either the endodermal transcription factor forkhead box A2 (Foxa2) or Pdx1 in HESCs and examined their pancreatic differentiation. In culture, HESCs overexpressing Pdx1 differentiate into cells expressing pancreatic markers earlier than the control cells. Moreover, we show that further differentiation can be achieved in teratomas derived from the Pdx1- or Foxa2-expressing cells.

Materials and Methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References

Cell Culture

HESCs and their differentiated derivatives were cultured as previously described [5, 7]. EBs were generated by aggregation of the HESCs for different time points: early EBs (2–4 days), mid-EBs (10–14 days), and late EBs (21–30 days). For teratoma formation, 5 × 106 HESCs were injected into the kidney capsule of 4-week-old mice with severe combined immunodeficiency. One month later, the mice were sacrificed, the teratomas were removed, and RNA was extracted from the tissue. All animal experiments were performed according to NIH guidelines.

Reverse Transcription-Polymerase Chain Reaction Analysis

Total RNA was extracted as described [14], and 1 μg of RNA was reverse-transcribed by random hexamer priming using EZ-First Strand cDNA Synthesis Kit (Biological Industries, Kibbutz Beit Haemek, Israel, http://www.bioind.com). cDNA samples were subjected to polymerase chain reaction (PCR) amplification with human-specific primers from different exons. All reverse transcription-polymerase chain reaction (RT-PCR) experiments were performed under nonsaturating conditions. PCR conditions include a first step of 3 minutes at 94°C; a second step of 20–35 cycles of 30 seconds at 94°C, 30 seconds (annealing step) at 55°C–62°C, and 45 seconds at 72°C; and a final step of 5 minutes at 72°C. Primer sequences and the size of the final products are described in Table 1.

Table Table 1.. Primers used for polymerase chain reaction and their expected products
Thumbnail image of

PCR products were assessed by gel electrophoresis on 2% agarose ethidium bromide-stained gels and verified by direct sequencing. We quantified the RT-PCR results using NIH Image 1.63 software. The images of the electrophoresis gels were inverted, and the intensity of each band was measured. The intensity of the background was also measured and subtracted from the intensity of the band.

Plasmid Construction

INS-eGFP expression vector was constructed by deleting the cytomegalovirus promoter sequence from peGFP-NI (basic-eGFP) (Clontech, Franklin Lakes, NJ, http://www.clontech.com) [15] and inserting a 410-base pair rat insulin 1 (NM_019129) minimal promoter sequence [16] into the BamHI and HindIII restriction sites. The construct contained an SV40-driven neomycin selectable marker, which confers resistance to G418 antibiotics. PDX1-eGFP expression vector was established by inserting 4.5 kilobases (kb) of the mouse Pdx1 promoter (NM_008814) [17] into the Sal1 restriction site of basic-eGFP. Actin-PDX1 was constructed by insertion of mouse Pdx1 coding region, 0.86 kb long (NM_008814) [18], into pCAGGS and adding the SV40-driven neomycin selectable marker. PGK-FOXA2 plasmid was based on the plasmid pCA1037 containing the complete rat cDNA sequence of Foxa2 (NM_012743), 2.2 kb long, under the control of phosphoglycerate kinase 1 (Pgk1) [19] and adding the SV40-driven neomycin selectable marker. Transfection and establishment of cell lines were performed as previously described [15].

Immunostaining

HESCs were washed three times with saline and fixed onto the plate with 4% paraformaldehyde. Either mouse anti-chick Foxa2, 1:50 (Developmental Studies Hybridoma Bank, Iowa City, IA, http://www.uiowa.edu/∼dshbwww), or rabbit anti-human Pdx1, 1:200 (Sigma-Aldrich, St. Louis, http://www.sigmaaldrich.com) were used as primary antibodies. As secondary antibodies, we used Cy-3-conjugated goat anti-mouse IgG (H+L), 1:200 (Jackson Immunoresearch Laboratories, West Grove, PA, http://www.jacksonimmuno.com), or Cy-3-conjugated donkey anti-rabbit IgG (H+L), 1:200 (Jackson Immunoresearch Laboratories). The first antibody was incubated with the cells overnight at 4°C. The next morning, the cells were washed twice with saline for 5 minutes. Then, the cells were incubated with the second antibody for 1 hour at room temperature. The cells were washed as described above, and nucleus staining was performed by incubation with 1 μg/ml bis-benzimide, Hoechst 33258 (Sigma-Aldrich) for 10 minutes.

FACS Analysis

To analyze the cells by FACS, the HESCs and their differentiated derivatives were trypsinized, transferred to phosphate-buffered saline, and passed through 70-μm nylon mesh (Becton, Dickinson and Company, Franklin Lakes, NJ, http://www.bd. com). The negative control was nontransfected HESCs that were grown in the same conditions as the PDX/INS/ALB-eGFP cells. Each sample was analyzed twice to verify the results. Analysis was performed on FACSCalibur system (Becton, Dickinson and Company) according to green fluorescent emission, for detection of enhanced green fluorescent protein (eGFP)-positive cells. Analysis was performed by CellQuest software (Becton, Dickinson and Company). Forward and side scatter plots were used to exclude dead cells and debris from the histogram analysis plots.

Statistical Analysis

The significance of the differences in the expression of pancreatic markers between the wild-type (WT) cells and the cells overexpressing Pdx1 was tested using t test analysis. Differences with p < .05 are labeled with an asterisk (*).

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References

Genetic Labeling of HESCs with eGFP Under the Control of Pdx1 or insulin Promoters

Genetic labeling of cells is a valuable tool that enables us to trace a specific cell type within heterogeneous population. As we previously have shown, labeling HESCs with the reporter gene eGFP under the control of albumin promoter allowed us to detect hepatic-like cells among the population of differentiating HESCs and further isolate them [8] (Fig. 1A, I and II). Labeling the β cells is crucial since the percentage of β cells within the developing embryo is very low and might correlate with low abundance within the human EBs.

thumbnail image

Figure Figure 1.. Human embryonic stem cells (HESCs) expressing eGFP under the control of Pdx1 or insulin promoters. HESCs were transfected with eGFP under the control of albumin, Pdx1, or insulin promoters. The HESC clones were subjected to differentiation through EBs and tested for the presence of eGFP-positive cells. Each vector was tested for its eGFP expression using a cell line as a positive control. (A): Various cell lines expressing eGFP under different promoters. I: Hepatoma cells positive for albumin expression, serving as a positive control for the ALB-eGFP vector. II: An embryoid body of an ALB-eGFP HESC clone showing few fluorescent cells. III: Pancreatic cells expressing Pdx1, serving as a positive control for PDX1-eGFP expression. IV: An embryoid body of a PDX1-eGFP HESC clone showing few fluorescent cells. V: Insulinoma cells expressing eGFP under the insulin promoter, serving as a positive control for the INS-eGFP vector. VI: An embryoid body of an INS-eGFP HESC clone without any fluorescent cells. (B): Fluorescence-activated cell sorter analysis for the expression of eGFP under the Pdx1 promoter. Left: Cells from embryoid bodies derived from nontransfected HESCs. Right: Cells from embryoid bodies derived from a PDX1-eGFP clone. The R1 gate defines the PDX1-eGFP-positive cells within the heterogeneous population of cells. (C): Reverse transcription-polymerase chain reaction results for eGFP expression in various clones expressing eGFP and the corresponding GAPDH results for each cDNA sample. RNA samples were taken from cells expressing eGFP under the control of PGK [15], Alb [8], or Pdx1. Abbreviations: Alb, Alb-eGFP cells; eGFP, enhanced green fluorescent protein; GAPDH, glyceraldehyde-3-phosphate dehydrogenase; NT, no template control; Pdx1, Pdx1-eGFP cells; PGK, PGK1 promoter.

Download figure to PowerPoint

A similar methodology was used here to label HESCs by eGFP under the control of either the Pdx1 or the insulin promoters to identify β-like cells within the different cells comprising the EBs. Prior to transfection into HESCs, we examined the effectiveness of the regulatory sequence of Pdx1 by introducing the PDX1-eGFP construct into an established cell line of human pancreatic cells expressing endogenous PDX1 and detected eGFP-positive cells (Fig. 1A, III). The PDX1-eGFP construct was then introduced into HESCs. The cells were subjected to differentiation through EBs and tested along the differentiation for eGFP-positive cells. Using the fluorescent inverted microscope, we have analyzed a few hundred EBs. Fluorescent cells were detected in 2-week-old EBs (Fig. 1A, IV). We also quantified the eGFP-positive cells by FACS. In each FACS analysis, we analyzed 30 to 100 EBs (Fig. 1B). We have previously shown that in the Alb-eGFP clones, 6% ± 2% of fluorescent cells could be observed in the differentiated cells [8]. In the Pdx1-eGFP clones, we now document 5% ± 2% of fluorescent cells in the differentiated EBs.

Pdx1 is expressed early in the endocrine differentiation in pancreas progenitor cells and not only in differentiated β cells. To discriminate between the pancreas progenitors and the β cells, we labeled HESCs by eGFP under the control of the insulin promoter (INS-eGFP). Several clones of HESCs carrying the INS-eGFP plasmid were subjected to differentiation conditions through EBs and were examined for eGFP expression at different time points during the differentiation. We searched for eGFP-positive cells within the EBs using both fluorescent inverted and confocal microscopes. eGFP-positive cells could not be detected during the spontaneous differentiation of HESCs, even in late, 30-day-old EBs (Fig. 1A, VI). Insulinoma cells transformed with the INS-eGFP plasmid expressed the eGFP and served as a positive control (Fig. 1A, V). RNA extracted from the HESCs tested by RT-PCR showed low expression of PDX1 and no expression of insulin (data not shown).

Overexpression of Endoderm Transcription Factors in HESCs

The expression of insulin could not be detected at significant levels following spontaneous differentiation of HESCs through EBs, even by using the reporter gene. Hence, we hypothesized that we should look for specific conditions inducing the differentiation toward β cells. We thus tested whether the overexpression of major transcription factors involved in pancreatic development might enhance the differentiation and enable us to obtain, in culture, insulin-expressing cells. We constitutively expressed two different transcription factors, Foxa2 and Pdx1, that are involved in different phases of early endoderm and pancreatic differentiation (Fig. 2A).

thumbnail image

Figure Figure 2.. Constitutive expression of Foxa2 and Pdx1 transcription factors in human embryonic stem cells (HESCs). (A): A schematic representation of the major transcription factors involved in differentiation of pancreatic cells. The two most upstream regulators, namely Foxa2 and Pdx1, were stably transfected under the control of constitutively expressed promoters into HESCs (based on [2]). The cells were tested for the expression of the exogenous transcription factor by reverse transcription-polymerase chain reaction (RT-PCR) and by immunostaining. (B): Top panel: RT-PCR analysis for the expression of the exogenous Foxa2 in two clones (1 and 2) and in nontransfected cells and the corresponding GAPDH results for each cDNA sample. Bottom panel: Immunostaining with anti-Foxa2 (left), the same field of cells with nuclei staining by Hoechst (right), and overlay of Hoechst and Foxa2 staining (middle). (C): Top panel: RT-PCR analysis for the expression of the exogenous Pdx1 in four clones (1, 2, 3, and 4) and in nontransfected cells and the corresponding GAPDH results for each cDNA sample. Bottom panel: Immunostaining with anti-Pdx1 (left), the same field of cells with nuclei staining by Hoechst (right), and overlay of Hoechst and Pdx1 staining (middle). Abbreviations: Con, nontransfected cells; EX, exogenous; GAPDH, glyceraldehyde-3-phosphate dehydrogenase; GCG, glucagon; PPY, pancreatic polypeptide; SST, somatostatin.

Download figure to PowerPoint

Foxa2 is found in the early endoderm layer from which the pancreas later arises [20]. Targeted disruption of Foxa2 affects the expression of other β cell transcription factors, placing Foxa2 at a very early stage in pancreas development [2]. Overexpression of Foxa2 in mouse embryonic stem cells showed enhanced differentiation toward the endoderm lineage [19]. In this study, HESCs were transfected with Foxa2 regulated by the constitutive PGK promoter. To verify that the cells constitutively express Foxa2, we first examined them for Foxa2 expression by RT-PCR analysis (Fig. 2B, top panel). In addition, expression at the protein level was assessed by immunostaining (Fig. 2B, bottom panel). The protein was found to localize within the nucleus, as expected for this transcription factor.

Pdx1 is a major transcription factor involved in pancreatic development [21]. It is expressed downstream of Foxa2, and its expression in the endoderm lineage is enriched in pancreatic cells [1]. Pdx1 binds the insulin promoter together with other transcription factors, enabling the expression of insulin in the β cells [21, 22]. The precise combination of many transcription factors (e.g., Pdx1, Pax4, and Nkx2.2) is unique to the β cell, and the interaction among them activates transcription in a cooperative manner [22]. Pdx1 regulated by the β actin promoter was introduced into HESCs, and several clones were isolated. These HESCs expressed the exogenous Pdx1, as we show by RT-PCR analysis and by immunostaining (Fig. 2C).

The Effect of Endodermal Transcription Factors on the Pancreatic Differentiation

HESCs expressing Foxa2 or Pdx1 were induced to differentiate via EBs and analyzed for the expression pattern of the pancreatic genes. RNA samples were taken from nondifferentiated HESCs, early EBs, mid Ebs, and late EBs. The expression profile of transcription factors involved in pancreatic development and the proteins produced by the various pancreatic cell types were compared between WT cells and the Foxa2 (Fig. 3B) or the Pdx1-overexpressing cells (Fig. 3A, 3B). This analysis revealed that although we had a high expression of Foxa2 in the cells, it did not significantly change the expression pattern of its downstream gene targets during the differentiation of HESCs. However, overall we saw a small induction in few genes; for example, PAX6 is induced in late EBs, and SST is induced in early EBs (Fig. 3B). Foxa2-expressing EBs express low levels of Pdx1. This is in contrast to the cells transfected with the Pdx1 expression vector that transcribe high levels of the Pdx1 gene. Comparison of the expression pattern of pancreatic genes between WT and Pdx1-overexpressing HESCs revealed induction toward pancreatic differentiation (Fig. 3B). The onset of expression of the transcription factor ISL1 in Pdx1 clones was prematurely induced compared to the WT cells. In Pdx1 clones, the ISL1 showed significantly higher levels of expression during the differentiation of HESCs into EBs. The transcription factors PAX4 and NGN3 and the marker for the endocrine polypeptide cells PPY showed higher expression in Pdx1 clones and were mainly induced in the early EBs. The transcription factor PAX6 showed a more prominent expression in the Pdx1 clones in mid- and late EBs. SST and AMY2B of δ endocrine cells and exocrine cells, respectively, showed a higher level of expression in early, mid-, and late EBs of Pdx1 clones. This induction was significant in several time points. Finally, the transcription factor NKX2.2 was expressed at low or viable levels but only in the Pdx1 clones and not in the WT cells. The overall trend is that the constitutive expression of Pdx1 in the HESCs promoted the expression of most transcription factors involved in the pancreas development and their downstream targets (Fig. 3A, 3B). However, Pdx1-overexpressing cells did not express insulin, and the expression of glucagon was turned off (Fig. 3B).

thumbnail image

Figure Figure 3.. The effect of constitutive expression of Pdx1 and FoxA2 on pancreatic differentiation of human embryonic stem cells (HESCs). Presented is the expression profile of various pancreatic transcription factors and pancreatic cell markers in WT HESCs, in HESCs constitutively expressing Foxa2, and in HESCs constitutively expressing Pdx1. Each cell type was tested for the expression level of the various markers as nondifferentiated HESCs and at three time points during the differentiation as EBs. (A): Representative reverse transcription-polymerase chain reaction (RT-PCR) results of the various markers in ES and their differentiated derivatives early, mid-, and late EBs. (B): Graphs presenting the quantified RT-PCR results for eight pancreatic genes. In each graph, we present the expression levels of the gene at the four time points during HESC differentiation for WT (open bars), Pdx1 (black bars), and Foxa2 (gray bars) cells. The expression levels were normalized by dividing the value of expression for each marker by the value of its glyceraldehyde-3-phosphate dehydrogenase expression. Each bar presents the average normalized RT-PCR results of four replications and the corresponding standard error. In the Foxa2-overexpressing HESCs, we only saw induction of PAX6 in late EBs and induction of SST in early EBs. However, in Pdx1-overexpressing HESCs, the expression pattern of the pancreatic genes revealed induction toward pancreatic differentiation. Most transcription factors involved in the pancreas development and their downstream targets were upregulated. Expression of pancreatic genes in Pdx1-overexpressing cells or Foxa2-overexpressing cells was compared to the expression in WT cells, and significant differences, p < .05, are labeled with an asterisk (*). Abbreviations: E, early embryoid body; EB, embryoid body; ES, WT and Pdx1 human embryonic stem cells; GAPDH, glyceraldehyde-3-phosphate dehydrogenase; L, late embryoid body; M, mid embryoid body; PPY, pancreatic polypeptide; WT, wild-type.

Download figure to PowerPoint

In Vivo Differentiation of HESCs into Insulin-Expressing Cells

The results above show that overexpression of Pdx1 indeed affected the expression pattern of its downstream pancreatic genes and enhanced the expression of markers for pancreatic polypeptide, δ, and exocrine cells. However, it was not sufficient to obtain insulin-expressing cells. Further enhancement of the differentiation is problematic since there are various transcription factors involved in the β cell development, and they can all be candidates for simultaneous over expression. Hence, we decided to investigate whether the in vivo differentiation within teratomas can further differentiate the Pdx1-overexpressing HESCs to become insulin-expressing cells. We injected HESCs from two different Pdx1-overexpressing clones and two Foxa2-overexpressing clones into the kidney capsule of immunocompromised mice and tested the expression of several pancreatic markers in the teratomas. Most of the teratomas, WT and overexpressing Pdx1 or Foxa2 teratomas, expressed the exocrinic marker AMY2B, the pancreatic polypeptide cells marker PPY, and the δ cell marker somatostatin (Fig. 4). Three Pdx1 teratomas and one Foxa2 teratoma expressed the insulin gene. The transcription factor NKX2.2 was not detected in WT cells, but in three insulin-positive teratomas, we detected the expression of NKX2.2. PDX1 was detected in the teratomas overexpressing Pdx1 and not in WT (Fig. 4). We statistically analyzed the data and found that although the genes NKX2.2 and Insulin had higher expression level in Foxa2 teratomas than in WT teratomas, the results were not statistically significant (.05 < p < .1). Likewise, although the genes PDX1, NKX2.2, Insulin, and PPY had higher expression level in Pdx1 teratomas compared to WT teratomas, the results were not statistically significant (.05 < p < .15).

thumbnail image

Figure Figure 4.. In vivo differentiation into pancreatic cells of human embryonic stem cells (HESCs) expressing Foxa2 or Pdx1. WT HESCs and HESCs constitutively expressing Foxa2 or Pdx1 were injected into immunocompromised mice and formed teratomas which are differentiated derivatives of the cells. The teratomas were tested for the expression of various pancreatic markers by reverse transcription-polymerase chain reaction (RT-PCR). The RT-PCR results were quantified, and the average value with standard error for each cell type is presented. The expression levels are normalized by dividing the value of expression for each marker by the value of glyceraldehyde-3-phosphate dehydrogenase expression for the same RNA sample. Abbreviations: PPY, pancreatic polypeptide; WT, wild-type.

Download figure to PowerPoint

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References

Our aim was to examine the differentiation of HESCs into pancreatic cells using genetic manipulation methodologies. First, we aimed to label pancreatic cells using the eGFP gene under the regulation of either the insulin or the Pdx1 promoters. Using this strategy, we could detect fluorescent cells only when PDX1-eGFP was introduced into the cells. These results led us to conclude that during spontaneous differentiation of HESCs into EBs, we could not obtain significant differentiation into pancreatic β cells.

Thus, we tried to constitutively express transcription factors that play a major role in directing differentiation toward pancreatic cells. For this purpose, we introduced into HESCs either Foxa2 or Pdx1 and assessed their effect by examining the expression level of a panel of pancreatic markers during EB formation. In the cells constitutively expressing Foxa2, we almost could not detect better induction of the pancreatic genes compared to the WT cells. Pdx1, on the other hand, enhanced the differentiation of the HESCs toward the pancreatic cell types. The transcription factors promoting the pancreatic differentiation, which are expressed downstream of Pdx1, showed either premature or more prominent expression. The differentiation was mainly toward exocrine cells, although induced expression of markers for pancreatic polypeptide cells and δ endocrine cells was also noted. Still, expression of insulin, a marker for β cells was absent. To test whether HESCs are able to differentiate into β cells in vivo, we examined the expression of pancreatic markers during differentiation of HESCs into teratomas. We tested teratomas derived from WT cells and cells overexpressing Foxa2 or Pdx1. Only in teratomas were we able to detect mRNA transcript encoding for insulin. The differentiation of pancreatic cells relies on growth factors and also on interaction with adjacent cells [23]. Both of these regulations may be more pronounced in the teratomas than in the EBs. The supply of growth factors by the blood vessels may be critical for the β cell differentiation in the teratoma. Candidates for such factors are members of the fibroblast growth factor and epidermal growth factor families. Moreover, the mere presence of blood vessels in the teratoma may be important for the pancreatic differentiation, as it was demonstrated that the vascular endothelium is critical for β cell formation [24]. Future experiments should aim to characterize and isolate the insulin-producing cells either after their differentiation in vivo or in vitro. We conclude that the current culture conditions for differentiating HESCs in vitro into pancreatic cells are still lacking a factor or a combination of factors that exists, at least partially, in vivo. Further differentiation toward a fully matured phenotype might be possible by overexpressing a combination of various transcription factors, together with the identification of factor(s) supplied by the intact animal.

The endodermal progenitors of the various cell types of the pancreas express the transcription factor Pdx1. These cells are the source for the exocrinic and endocrinic cells. Later on, more transcription factors are temporally activated. Of the Pdx1-expressing cells, Ngn3 is expressed only in the cells that are oriented to the endocrinic pathway. In the endocrinic pathway, the expression of Pax6 is detected only in cells that finally become α cells, whereas Pax4 is expressed in the progenitors of β, δ, and pancreatic polypeptide cells. Nkx2.2 is then activated only in pancreatic β cells and is followed by insulin expression [2]. During the spontaneous differentiation of HESCs, we detected the expression of the exocrine cells by AMY2B expression. We could also show the expression of glucagon, somatostatin, and PPY, which are expressed by α, δ, and pancreatic polypeptide cells, respectively. Yet we could not detect the expression of NKX2.2 and its downstream target insulin, representing the β pancreatic cells (Fig. 3). The constitutive expression of Pdx1 caused the differentiating HESCs to express NKX2.2, but it was not enough to cause for expression of insulin. On the other hand, the expression of glucagon was turned off during the differentiation of Pdx1-overexpressing cells in vitro. Since Pdx1 is required for maintaining the hormone-producing phenotype of β cells by positively regulating insulin and by repressing glucagon in β cells [21], we speculate that the elimination of glucagon expression is due to the high expression of Pdx1.

During embryonic development, Foxa2 is expressed early during the differentiation of the endodermal lineage [2]. We speculated that the expression of Foxa2 would promote the cascade of expression of transcription factors and effectively induce pancreatic differentiation. Our results show that in vitro rat Foxa2 had little effect on pancreatic differentiation of HESCs. The rat Foxa2 that we constitutively expressed in the HESCs is a close homologue to the human FOXA2 protein (showing amino acid conservation of 96.4% and almost full identify in the winged helix DNA-binding domain) [25]. We thus speculate that although Foxa2 is actively expressed in our cells, it is not sufficient to induce pancreatic differentiation. Hepatic differentiation, which occurs earlier than that of pancreatic differentiation during embryonic development, was previously shown to be induced by overexpression of Foxa2 in the murine ES cells [19]. Our conclusion is that pancreatic differentiation requires pancreatic-specific transcription factors, and induction of early endodermal factors is not sufficient.

We should also bear in mind that most of the markers used to define pancreatic differentiation are also expressed in other cell types. Since the differentiation is heterogeneous, we cannot rule out the possibility of expression due to other cell types. The transcription factors Isl1, Pax6, Ngn3, and Nkx2.2 are also involved in the differentiation of other cell types. Pax4 is the only pancreas-specific transcription factor. Hence, its expression during the spontaneous differentiation of HESCs reflects differentiation toward pancreatic cells.

The differentiation of HESCs into pancreatic β cells is of major importance to understand the differentiation of the human pancreas and as a source of cells for therapy in diabetes. However, although the differentiation of HESCs was documented for multiple cell types, such as nerve cells and cardiomyocytes [5], the differentiation into β cells was not trivial, and different groups reported different results or interpretations of the results (reviewed in [11]). We aimed at directing the differentiation using two transcription factors. Although differentiation into pancreatic cells was induced, it was not enough to achieve profound differentiation into β cells in vitro. Differentiation into endoderm-derived tissues seems more difficult from differentiation into ectoderm-derived cells, either in mouse or human embryonic stem cells. But even differentiation into endodermal hepatic-like cells may be easier than that of pancreatic β cells. During normal human embryonic development, pancreatic β cells appear later than hepatic cells, and in our body we have more than 100 times more hepatocytes than β cells. If differentiation of EBs in a way mimics normal development, it is conceivable that it will not be trivial to obtain these cell types in culture.

Acknowledgements

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References

We thank our laboratory members for critically reading the manuscript and for their assistance with the generation of teratomas. This research was partially supported by funds to N.B. from the Herbert Cohn Chair (Hebrew University) and by funds from the Ministry of Sciences (Israel) and the Berrie Foundation.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References