High Oxygen Condition Facilitates the Differentiation of Mouse and Human Pluripotent Stem Cells into Pancreatic Progenitors and Insulin-producing Cells*

Background: Oxygen plays a key role in organ development, including pancreatic β-cells. Results: High oxygen conditions increase Ngn3-positive and insulin-positive cells from both mouse and human pluripotent stem cells. Conclusion: Culturing under high oxygen conditions has a facilitative effect on pancreatic differentiation. Significance: This new technique provides an efficient method to utilize patient-specific iPS cells for the treatment of diabetes. Pluripotent stem cells have potential applications in regenerative medicine for diabetes. Differentiation of stem cells into insulin-producing cells has been achieved using various protocols. However, both the efficiency of the method and potency of differentiated cells are insufficient. Oxygen tension, the partial pressure of oxygen, has been shown to regulate the embryonic development of several organs, including pancreatic β-cells. In this study, we tried to establish an effective method for the differentiation of induced pluripotent stem cells (iPSCs) into insulin-producing cells by culturing under high oxygen (O2) conditions. Treatment with a high O2 condition in the early stage of differentiation increased insulin-positive cells at the terminus of differentiation. We found that a high O2 condition repressed Notch-dependent gene Hes1 expression and increased Ngn3 expression at the stage of pancreatic progenitors. This effect was caused by inhibition of hypoxia-inducible factor-1α protein level. Moreover, a high O2 condition activated Wnt signaling. Optimal stage-specific treatment with a high O2 condition resulted in a significant increase in insulin production in both mouse embryonic stem cells and human iPSCs and yielded populations containing up to 10% C-peptide-positive cells in human iPSCs. These results suggest that culturing in a high O2 condition at a specific stage is useful for the efficient generation of insulin-producing cells.


Pluripotent stem cells have potential applications in regenerative medicine for diabetes. Differentiation of stem cells into insulin-producing cells has been achieved using various protocols. However, both the efficiency of the method and potency of differentiated cells are insufficient. Oxygen tension, the partial pressure of oxygen, has been shown to regulate the embryonic development of several organs, including pancreatic ␤-cells. In this study, we tried to establish an effective method for the differentiation of induced pluripotent stem cells (iPSCs) into insulin-producing cells by culturing under high oxygen (O 2 ) conditions. Treatment with a high O 2 condition in the early stage of differentiation increased insulin-positive cells at the terminus of differentiation. We found that a high O 2 condition repressed
Notch-dependent gene Hes1 expression and increased Ngn3 expression at the stage of pancreatic progenitors. This effect was caused by inhibition of hypoxia-inducible factor-1␣ protein level. Moreover, a high O 2 condition activated Wnt signaling. Optimal stage-specific treatment with a high O 2 condition resulted in a significant increase in insulin production in both mouse embryonic stem cells and human iPSCs and yielded populations containing up to 10% C-peptide-positive cells in human iPSCs. These results suggest that culturing in a high O 2 condition at a specific stage is useful for the efficient generation of insulin-producing cells.
Cell replacement therapy has become possible by utilizing artificially generated cells or organs from embryonic stem cells (ESCs), 3 induced pluripotent stem cells (iPSCs), and adult stem cells. These stem cells have marked potential to develop into many different cell types in the body during early life and growth. Over the last decade, with the advent of new techniques and technologies in modern molecular biology, understanding of the underlying mechanism responsible for organ differentiation has developed rapidly. This knowledge has given rise to various new methods of manipulating stem cells in order to generate deficient organs in various diseases. To date, various differentiation methods have been developed for each cell type, including neurons, cardiomyocytes, and pancreatic endocrine cells. Many of these methods are based on mimicking the in vivo development. The development of efficient and safe methods is desired for clinical applications and studying the cause of disease.
Pluripotent stem cells are capable of spontaneous differentiation into insulin-producing cells. This is mainly carried out by preferential differentiation of stem cells into insulin-producing cells by changing the composition of the culture medium and causing the expression of dominant transcription factor genes, which are mainly involved in pancreatic development. Several groups have reported methods of generating pancreatic cell lineages from ESCs and iPSCs (1)(2)(3)(4)(5)(6)(7)(8). These methods induce definitive endoderm differentiation in the first stage and then pancreatic specialization and maturation in the following stages, using combinations of growth factors, small molecules, and extracellular matrix. Lumelsky et al. (6) first demonstrated the successful differentiation of mouse ESCs (mESCs) to insulin-secreting structures, which was concluded to be similar to that of pancreatic islets. However, the limiting factor of this method is that the abundance of differentiated cells is relatively low. Moreover, several reports had the same issue that the differentiated cells are immature and/or not fully functional in culture. Some reports succeeded in generating functional insulin-secreting cells utilizing differentiation under implantation or coculture with organ-matched mesenchyme (7,8). However, such methods have a risk of teratoma or teratomatous tissue element formation in their grafts. Fifteen percent of grafts showed teratoma or a teratomatous tissue element (7). To improve this issue, establishment of safer and more efficient methods is desired.
Oxygen (O 2 ) plays a crucial role in cellular homeostasis (9,10). In normal tissues, the lack of oxygen contributes to cell death, whereas in stem cells, lack of O 2 controls stem cell selfrenewal and pluripotency by activating specific signaling pathways, such as Notch, and the expression of transcriptional factors, such as Oct4 (11,12). Hypoxia is accompanied by the stabilization of hypoxia-inducible factors (HIFs), O 2 -regulated transcriptional factors that regulate an ever increasing number of genes involved in glycolytic metabolism, angiogenesis, erythropoiesis, and metastasis and mediate the adaptation of cells to decreased O 2 availability (13,14). O 2 tension, the partial pressure of O 2 , has been shown to regulate the embryonic development of several organs, including the trachea, heart, lung, limb bud, and bone (15)(16)(17)(18)(19). It is also reported that O 2 tension plays a key role in pancreatic development (20 -23). The embryonic pancreas early in development is poorly vascularized and has a paucity of blood flow, and, at later stages, blood flow increases, and endocrine differentiation occurs at the same time (21). It has also been shown that HIF-1␣ protein is highly expressed in the embryonic pancreas early in development and that increasing concentrations of O 2 in vitro represses HIF-1␣ expression and fosters the development of endocrine progenitors (22,23). Suitable O 2 concentrations should be tested for the differentiation efficiency of ESC and iPSC into pancreatic lineages. However, until now, there has been no report of such an effect on ESC and iPSC differentiation in vitro.
Here we studied the effect of increasing concentrations of O 2 on the differentiation efficiency of mESC and human iPSC (hiPSC) into pancreatic lineages. A high O 2 condition (60% O 2 ) in early stages of differentiation increased the percentage of Ngn3-expressing endocrine progenitors and insulin-positive cells in both mESC and hiPSC. This effect was mediated via the inhibition of HIF-1␣ expression and increase of Ngn3 gene expression. Moreover, a high O 2 condition was found to induce the activation of Wnt signaling. In this study, we demonstrated that culturing ESC and iPSC in a high O 2 condition improved differentiation efficiency into endocrine progenitors and insulin-producing cells compared with normoxic conditions.
Images were taken with an Olympus IX81 fluorescence microscope (Olympus Optical Co. Ltd., Tokyo, Japan). Quantification was carried out using MetaMorph software (Molecular Devices). After images of marker fluorescence and DAPI fluorescence were taken in defined areas of wells in the cell culture plate, each image was thresholded to exclude background noise. The area was quantified, and the percentage of markerpositive cells was calculated by dividing the DAPI-positive area (total cell number) into the marker-positive area (supplemental Fig. S1).
Quantitative Real-time PCR-Total RNA was isolated from cells using a TRIzol reagent (Invitrogen) according to the manufacturer's instructions. cDNA was synthesized using a Superscript III first strand synthesis system (Invitrogen) according to the manufacturer's instructions. Quantitative real-time PCR (qPCR) analysis was performed on an ABI Prism 7300 (Applied Biosystems, Foster City, CA) using a SYBR Premix ExTaq GC (Takara, Shiga, Japan). The primer sequences for each primer set are shown in Table 1. mRNA expression data were normalized against actin expression in a corresponding sample. The data were analyzed using the relative quantification study in Sequence Detection software version 1.2 (Applied Biosystems).
Microarray-Microarray analysis was performed on total RNA samples using a TORAY 3D-gene oligo chip (TORAY, Tokyo, Japan), according to the manufacturer's instructions. The genes induced by a high O 2 environment were determined by global normalization after excluding genes of Ͻ100 intensity in the high O 2 condition-treated group. Genes increased over 8-fold in the ratio of high O 2 condition to normoxia are listed in Fig. 6A. Pathway analysis was performed on up-regulated genes in the high O 2 condition-treated or echinomycin-treated group using GenMapp/MAPP Finder software.
Statistics-Data are shown as the mean Ϯ S.E. Student's t test was used to identify significant differences between two conditions, and one-way analysis of variance or two-way analysis of variance followed by Tukey-Kramer's post hoc analysis was used to compare multiple conditions. p Ͻ 0.05 was considered to be significant.

High Oxygen Condition Facilitates the Differentiation of mESC into
Insulin-producing Cells-We used a modified protocol from a previous report of three-stage stepwise differentiation into insulin-producing cells (25,26) (Fig. 1A). First, mESC ING112 cells were treated with activin A and bFGF to direct the differentiation into definitive endoderm from day 1 to day 7. By day 7, there was a steep reduction in the expression of Oct4 relating to the fact that the cells have transitioned from pluripotency to an endodermal progenitor (Fig. 1B). This is evident from the increased expression of Sox17 and Foxa2 on day 7, both of which are markers of a definitive endoderm. With the change in the composition of the medium containing B27, FGF10, KAAD-cyclopamine, and retinoic acid from day 7 onward, there was a gradual decrease in the expression of Sox17 and Foxa2. Subsequently, there was a marked increase in the expression of Pdx1 and Ngn3 on day 11, indicating the prevalence of pancreatic progenitors and endocrine progenitors in the population of the culture. With the change in the medium composition containing nicotinamide and GLP-1 on day 11, Pdx1 and Ngn3 expressions decreased, whereas the maximum level of Ins1 expression was reached.
Next we examined the effects of a high O 2 concentration condition on the differentiation efficiency of mESC ING112 cells. We cultured cells in a high O 2 condition (60% O 2 ) under stepwise differentiation during stage 1 (days 3-7), stage 2 (days 7-11), or stage 3 (days 11-17) ( Fig. 2A). It was observed that a high O 2 condition during stage 1, the early phase of differentiation, had the greatest effect on differentiation efficiency with an almost 8-fold increase (p Ͻ 0.05) in the percentage of Ins1-GFP-positive cells (Fig. 2, B and C). However, a high O 2 condition during stage 2 or stage 3 had no effect on differentiation efficiency. At the beginning of these experiments, we cultured under high O 2 conditions from day 1; however, earlier treatment had a deleterious effect on the percentage of surviving cells (Fig. 2F). Therefore, we used this protocol as the high O 2 condition from day 3. To determine whether 60% high O 2 is the best condition, we tested different levels of O 2 condition. As a result, 40% O 2 during stage 1 also increased the percentage of Ins1-GFP-positive cells by 4-fold (p Ͻ 0.05; Fig. 2G), but this effect was less than that of 60% O 2 . Instead of high O 2 , we used a hypoxic condition during differentiation, but there was no change compared with the normoxic condition (Fig. 2D). The high O 2 condition also increased the percentage of C-peptidepositive cells in a different mESC line, SK7 (27,28) (Fig. 2E).
High Oxygen Condition Facilitates Differentiation into Endocrine Progenitors-Based on the results in Fig. 2, B and C, we compared gene expression levels between the normoxia and high O 2 condition groups on day 6 of stage 1 (Fig. 3A). We observed a significant decrease (p Ͻ 0.0005) in the expression of Oct4, which is indicative of the fact that the cells lost their pluripotency (Fig. 3B). Whereas there was no difference in the expression of Sox17, a marker gene of definitive endoderm, there was an almost 6-fold (p Ͻ 0.005) and 7-fold increase (p Ͻ 0.0005) in the expressions of Pdx1 and Ngn3, respectively (Fig.  3B). To determine the proportion of cells expressing each marker, immunofluorescence analysis was performed. There was no marked difference in the number of Sox17-positive cells on day 6 ( Fig. 3C), and Pdx1-positive cells on day 11 (Fig. 3D), although induction of its gene expression was observed (Fig.  3B). It was confirmed that the number of Ngn3-positive cells was significantly increased on day 11 (Fig. 3E). Quantification of the percentage of Ngn3-positive cells showed an almost 3-fold (p Ͻ 0.0005) increase (Fig. 3F). These results show that the high O 2 condition reduced the pluripotency of the cells and directed them markedly toward endocrine progenitors.
In our differentiation protocol, stage 1 contained supplements, such as activin A and bFGF, in the medium to direct toward a definitive endoderm. Next, to clarify which high O 2 condition affected undifferentiated mESC or differentiating cells, we examined the effect of a high O 2 condition on Ngn3 expression in undifferentiated cells. Treatment with a high O 2 condition for 3 days did not affect Ngn3 expression in the undif- ferentiated state maintained in mESC medium compared with treatment in differentiation medium, showing that the high O 2 condition affected differentiating cells (Fig. 3G).
High Oxygen Condition Represses HIF-1␣ Protein Level and Hes1 Gene Expression-Even under normoxia, HIF-1␣ is reported to be expressed at a detectable level and participate in the expression of hypoxia-inducible genes in mESCs (29). Therefore, we compared HIF-1␣ protein level and its target gene expression between normoxic and high O 2 condition groups. High O 2 condition during days 3-6 of differentiation repressed HIF-1␣ protein level (Fig. 4, A and B). Under normoxic conditions, expression of Vegfa, a HIF-1␣ targeting gene, increased from day 4 with a peak level on day 5 of differentiation, showing that activation of HIF-1␣ occurs during differentiation, whereas the high O 2 condition significantly decreased Vegfa expression on days 5 and 6 (p Ͻ 0.05, respectively; Fig. 4C). It is reported that HIF-1␣ activates Notch signaling in stem cells and embryonic pancreatic cells (11,22). Hypoxia and subsequent HIF-1␣ expression induced expression of Hes1 (hairy and enhancer of split 1), a Notch downstream gene, and repressed Ngn3 expression, leading to the inhibition of ␤-cell development (22). Therefore, we compared the kinetics of Hes1 and Ngn3 expression in normoxia with those in the high O 2 condition. Under differentiation, Hes1 expression was slightly increased at day 4 and gradually decreased from day 5, whereas high O 2 condition significantly repressed its expression on days 5 and 6 (p Ͻ 0.01 and p Ͻ 0.05, respectively; Fig. 4D). In contrast, Ngn3 expression significantly increased on both day 5 and 6 in the high O 2 condition (p Ͻ 0.05, respectively; Fig. 4E).
Next we examined the effect of HIF-1␣ inhibition on Ngn3 expression under differentiation. Cells were treated with 1 nM echinomycin, an inhibitor of HIF-1␣, from day 3 to day 6. On day 6, Ngn3 expression was significantly increased by echinomycin treatment, whereas Vegfa and Hes1 expressions were decreased (p Ͻ 0.0005, p Ͻ 0.05, and p Ͻ 0.05, respectively; Fig.  4F). These expression profiles were similar to those in the high O 2 condition. Hence, it was shown that HIF-1␣ inhibition and subsequent repression of Notch signaling play a role in facilitated differentiation in the high O 2 condition.
High Oxygen Condition Activates Wnt Signaling Pathway-To further clarify the effect of the high O 2 condition on differentiating cells, we performed microarray analysis on normoxia or high O 2 -treated cells. As a result, many genes were up-regu-lated by the high O 2 condition, and genes showing over 8-fold expression in the high O 2 group compared with the normoxia group are listed in Fig. 5A. Pathway analysis using up-regulated genes in the high O 2 group indicated that several genes were involved in the Wnt signaling pathway, and this pathway was ranked first (p Ͻ 0.01; Fig. 5B). Wnt3, Wnt6 (over 8-fold), Wnt5a, Wnt10a (over 4-fold; data not shown), Wnt4, Wnt7b, Wnt10b, Fzd1, Myc, and Ccnd2 (over 2-fold; data not shown) were increased in the high O 2 group and detected as the Wnt signaling pathway. We also performed microarray analysis on the echinomycin-treated group. Similarly, the Wnt signaling pathway was ranked first when analyzed using up-regulated genes (p Ͻ 0.001; Fig. 5C), suggesting that HIF-1␣ inhibition led to the activation of Wnt signaling.
The Wnt/␤-catenin pathway plays an important role in the regulation of pluripotency and pancreatic development and dif- High Oxygen Condition Facilitates Differentiation of hiPSC into Insulin-producing Cells-We performed pancreatic differentiation from hiPSC clone 23 (25) by our stepwise protocol (Fig. 6A) and analyzed marker expression by immunofluorescence analysis. It was confirmed that the expression of SOX17, a definitive endoderm marker, was not detected in undifferentiated cells (ud) but was markedly expressed on day 7 (stage 1) during differentiation. That signal continued to appear on day 11 (stage 2) to 17 (stage 3) (Fig. 6B). Another definitive endoderm marker, FOXA2, also began to be expressed at stage 1 and also appeared in later periods (Fig. 6C). Expression of PDX1, a marker of pancreatic progenitors, was not detected in undifferentiated and stage 1 cells (data not shown), whereas some signals were detected at stage 2 with a peak signal at stage 3 (Fig.  6D). NGN3, a marker of endocrine progenitors, was also not detected in undifferentiated and stage 1 cells (data not shown), whereas robust signals were detected at stage 2 and continued to be expressed at stage 3 (Fig. 6E). At stage 3, the termination of this differentiation protocol, several insulin-and C-peptidepositive cells were detected (Fig. 6E). Next, we examined the expression dynamics of marker genes by qPCR analysis and concurrently compared our expression dynamics with that of a previously reported protocol (32) (Fig. 7A). It was revealed that the expression dynamics of analyzed genes during differentiation by our protocol (three-step protocol) was similar to that of pancreatic ␤-cell development (33, 34). The D'Amour protocol consists of stage 1 to stage 5. Stage 1 guides pluripotent cells to definitive endoderm, stage 2 and 3 to pancreatic progenitors, stage 4 to endocrine progenitors, and stage 5 to hormone-expressing endocrine cells. By the D'Amour protocol, SOX17 and FOXA2 were expressed higher than in our protocol at stage 1 (Fig. 7A). In contrast, PDX1 gene expression was very high at stage 2 of our protocol and was higher than in the D'Amour protocol. Moreover, robust increases of NGN3, NEUROD1, MAFA, and INS expression were detected at the termination of our protocol (Fig. 7A). The percentages of PDX1-, NGN3-, and C-peptide-positive cells in the population of differentiated cells by our protocol were higher than those by the D'Amour protocol (Fig. 7B). Different protocols are usually used for mESC and hESC/iPSC, especially different lengths of time. Therefore, we examined the effect of altered culture times on pancreatic differentiation of hiPSC. A shorter time frame decreased the percentage of C-peptide-positive cells (p Ͻ 0.05), whereas a longer time had no effect (Fig. 7C).
Next we investigated whether the high O 2 condition has an effect on the pancreatic differentiation of hiPSC (Fig. 8A). The effect of the high O 2 condition during differentiation showed a similar result to that of mESC. Immunofluorescence analysis revealed a significant increase of the percent- age of C-peptide-positive cells by the high O 2 condition during stage 1 (p Ͻ 0.05; Fig. 8B). It was confirmed that there was a significant increase in INS gene expression (p Ͻ 0.05; Fig. 8C). These effects were also observed in another hiPSC line, 201B7 (35) (p Ͻ 0.05; Fig. 8D). We performed immunostaining for glucagon to determine whether differentiated cells are monohormonal or polyhormonal by our protocol. A few glucagon-positive cells were observed in the differentiated cells treated with the high O 2 condition (Fig. 8E). Moreover, co-expression with C-peptide appeared in a few cells, suggesting that some cells were polyhormonal for insulin and glucagon.

High Oxygen Condition Facilitates Differentiation of hiPSC into Endocrine
Progenitors-Corresponding with our results on mESC, we observed that NGN3 expression was significantly increased on day 7 by the high O 2 condition during stage 1, whereas VEGFA and HES1 expression were significantly decreased (p Ͻ 0.0000005 and p Ͻ 0.000005, respectively; Fig. 9, A  and B). Immunofluorescence analysis confirmed that the percentage of NGN3-positive cells was increased by this treatment (Fig. 9C). To further determine whether HIF-1␣ inhibition and Wnt signaling activation are involved in the case of hiPSC, we performed Western blot analysis of HIF-1␣ and microarray analysis of high O 2 -treated hiPSCs. The high O 2 condition repressed the HIF-1␣ protein level (Fig. 9D). By microarray analysis, many genes were found to be up-regulated by the high O 2 condition, as listed in supplemental Fig. S2A. Some of these genes were determined as the Wnt receptor signaling pathway in the GO biological process (supplemental Fig. S2B), suggesting that Wnt signaling was also activated in high O 2 -treated hiPSCs. In addition, Dkk-1 treatment weakened high O 2 -induced NGN3 expression (Fig. 9E). These results in hiPSC are similar to those observed in mouse ESC, thereby indicating that both human and mouse pluripotent cells follow a similar pathway in a high O 2 condition.
We tested whether the high O 2 condition had an effect even in the D'Amour protocol and Nostro protocol (1) (Fig. 10A). Using this protocol, we also saw a significant increase in the percentage of C-peptide-positive cells by the high O 2 condition from day 1 to day 4 (Stage 1; p Ͻ 0.05) and also in the INS expression by high O 2 condition from day 1 to day 4 (Stage 1; p Ͻ 0.05) and from day 4 to day 7 (Stage 2; p Ͻ 0.05) (Fig. 10, B  and C). However, in the Nostro protocol, we did not observe any facilitative effect of the high O 2 condition on the percentage of C-peptide-positive cells (Fig. 10, D and E).

DISCUSSION
Insulin-secreting pancreatic ␤-cells are essential regulators of the mammalian metabolism. The absence of functional ␤-cells leads to hyperglycemia and diabetes, making patients dependent on exogenously supplied insulin. Recent insights into ␤-cell development, combined with the discovery of pluripotent stem cells, have led to an unprecedented opportunity to generate new ␤-cells for transplantation therapy and drug screening (36,37). It is important to mimic the in vivo developmental stages of pancreatic organogenesis in which cells are transitioned through mesendoderm, definitive endoderm, foregut endoderm, pancreatic progenitor, and the endocrine progenitor stage, until mature ␤-cells are obtained from pluripotent stem cells (38). Oxygen tension, the partial pressure of oxygen, has been shown to regulate the stem cell function and embryonic development of several organs, including the pancreas (9,10,(15)(16)(17)(18)(19)(20)(21)(22)(23). In the present study, we demonstrated that a high O 2 condition during the in vitro differentiation of ESC and iPSC has a facilitative effect on generating pancreatic progenitors and insulin-producing cells.
In our stepwise differentiation protocol, the cells transitioned through definitive endoderm, pancreatic progenitor, endocrine progenitor, and insulin-producing cells, as revealed by qPCR analysis. Induction of Pdx1 and Ngn3 gene expressions appeared on day 7 (stage 1), showing that a slight transition to pancreatic progenitors and endocrine progenitors had already started during stage 1 (Fig. 1B). With this protocol, treatment with a high O 2 condition during stage 1 (toward definitive endoderm) increased differentiation efficiency into endocrine progenitors and subsequent insulin-producing cells. This was demonstrated by a significant increase of Ngn3-positive cells and Ngn3 gene expression (Fig. 3, B, E, and F). Ngn3 is a basic helix-loop-helix transcription factor expressed in cell progenitors that is necessary to initiate the endocrine differentiation program in pancreatic development (39,40), and its gene expression is inversely regulated by HIF-1␣ (22). Down-regulation of Notch signaling will yield cells that express Ngn3 (41). Ngn3 gene expression and pancreatic endocrine development are tightly regulated by Hes1, which is an inhibitory bHLH fac- tor activated by Notch signaling and binds to the proximal promoter of Ngn3 and specifically blocks promoter activity (39,42). It has been shown that HIF-1␣ activates Notch-responsive promoters and increases the expression of Notch direct downstream genes, including Hes1 (11). During differentiation, HIF-1␣ signaling is moderately activated even in normoxic conditions, revealed by HIF-1␣ protein expression and an increase of its target gene Vegfa, whereas a high O 2 condition markedly repressed both expressions (Fig. 4, B and C). Supporting our results, it is reported that HIF-1␣ signaling of cultured stem cells is activated during spontaneous differentiation even in normoxic conditions, showing a time-dependent increase of Vegfa (43). A high O 2 condition might increase cellular O 2 concentration and lead to inhibition of HIF-1␣ signaling. The high O 2 condition significantly repressed Hes1 gene expression on days 5 and 6 ( Fig. 4D). Consistent with previous reports, inhibition of HIF-1␣ signaling might lead to repression of Hes1 expression and subsequent induction of Ngn3 expression in a high O 2 condition. Furthermore, the HIF-1␣ inhibitor echinomycin had an effect similar to that of the high O 2 condition (Fig.  4F). These data indicate that inhibition of HIF-1␣ signaling is involved in the facilitative effect of the high O 2 condition on pancreatic differentiation. The high O 2 condition had no effect on Sox17 gene expression and the number of immunoreactive cells (Fig. 3, B and C), suggesting that its treatment might affect the transition from definitive endoderm to pancreatic progenitor or endocrine progenitor. This is supported by the finding that its treatment did not increase Ngn3 expression in the undifferentiated state (Fig. 3G). The high O 2 condition significantly decreased Oct4 gene expression (Fig. 3B). Oct4 gene expression is directly regulated by HIF-2␣, also a hypoxia-dependent factor (12); therefore, it is considered that the high O 2 condition might repress Oct4 gene expression via HIF-2␣ inhibition.
By microarray analysis, we found that the Wnt signaling pathway is activated in high O 2 condition-treated cells (Fig. 5B). Wnt inhibitor Dkk-1 partially repressed high O 2 condition-induced Ngn3 expression (Fig. 5D). It is reported that hypoxia inhibits Wnt signaling via HIF-1␣ competing with T-cell factor-4 (TCF-4) for direct binding to ␤-catenin (44). The high O 2 condition might inhibit HIF-1␣ signaling, and compensatory Wnt signaling was activated. This is supported by the finding that genes induced by the HIF-1␣ inhibitor echinomycin are also involved in the Wnt signaling pathway (Fig. 5C). The canonical Wnt cascade has emerged as a critical regulator of self-renewal and pluripotency in stem cells (30,(45)(46)(47)(48)(49). In contrast, it is also reported that Wnt/␤-catenin signaling promotes the differentiation, not self-renewal, of embryonic stem cells (1,32,50,51). Nostro et al. (1) showed that Wnt signaling induces a posterior endoderm fate, the primed stage from definitive endoderm, and at optimal concentrations enhances the development of pancreatic lineage cells. In this report, Wnt signaling did not affect the levels of PDX1 but did increase INS expression in hiPSC with Wnt3a treatment at the stage of definitive endoderm to pancreatic endoderm. This report is consistent with our findings that a high O 2 condition activates Wnt signaling and facilitates differentiation from definitive endoderm into pancreatic fate. In the developing embryo, a key step in the generation of endoderm-derived cell types is patterning the appropriate region of the gut tube along the anterior-posterior axis. Studies using different model systems have shown that in gastrulation, Wnt signaling is restricted to the posterior region of the embryo and, together with FGF signaling, is responsible for the induction of a posterior phenotype (52,53). Wnt signaling activated by a high O 2 condition might function to promote the development of a posterior phenotype in mESC and hiPSC cultures.
The high O 2 condition also facilitates pancreatic differentiation of hiPSC. Our stepwise differentiation protocol generated insulin-producing cells larger than the D'Amour protocol (32). However, in the case of SOX17 and FOXA2, a robust increase was observed in the D'Amour protocol (Fig. 7A), indicating that the induction of definitive endoderm was more efficient than our three-step protocol. This might have been due to the difference of the activin A concentration (D'Amour protocol, 100 ng/ml; our protocol, 10 ng/ml), because induction of definitive endoderm by activin A is reported to increase in a dose-dependent fashion (54 -56). However, at stage 2 of our protocol, there was a marked increase in the expression of PDX1, greater than at stage 4 of the D'Amour protocol. Furthermore, NGN3, NEUROD1, MAFA, and INS gene expressions were higher in our protocol, indicating that the induction of pancreatic fate from definitive endoderm in our protocol was more efficient than with the D'Amour protocol. Using this protocol, hiPSCs more efficiently differentiated into endocrine progenitors and insulin-producing cells in the high O 2 condition during stage 1, similar to mESCs. The effect of the high O 2 condition was also observed in the D'Amour protocol but not in the Nostro protocol (1) (Fig. 10, A-E). Nostro et al. included VEGF during stage 1, probably to support endothelial development for pancreatic differentiation of hESC (1,57). The high O 2 condition repressed VEGF expression in our study (Fig. 9B), and a similar effect is expected to occur in the Nostro protocol. This effect may compete with the addition of VEGF. Therefore, the high O 2 condition seems to have no facilitative effect in the Nostro protocol.
Insulin-producing cells obtained in our study did not secrete insulin by high glucose stimulation (data not shown), and some cells were polyhormonal because co-expression of insulin and glucagon occurred. During normal human embryogenesis, ␤-cells are not generated until ϳ10 weeks after endoderm specification (58), whereas in hiPSC differentiation cultures, this typically occurs in 2-3 weeks (32). It is possible that pancreatic differentiation in human ESC/iPSC culture may be accelerated by rapid changing of the transcriptional network and/or epigenetic modifications by changing supplements (growth factor and inhibitors, etc). For proper ␤-cells, it may be necessary to change the extracellular environment and signal more precisely to mimic normal human embryogenesis.
In a previous study, Shah et al. (21) stated that the early mammalian embryo is located within the uterus, with a non-existent or immature cardiovascular system and blood supply, but, despite this hypoxic environment, the embryo is still able to undergo rapid growth and organogenesis. Furthermore, they showed that the number of Ngn3-positive cells was not altered by hypoxia treatment in pancreatic explants, whereas the number of insulin-positive cells was decreased by hypoxia, implying that high oxygen may only be required at later stages during pancreatic differentiation, namely endocrine progenitor to ␤-cell. However, in our study, the number of Ngn3-positive cells differentiated from mESCs and hiPSCs was increased by the high O 2 condition (Figs. 3 (E and F) and 9C). There are some differences between the culture environments of dissociated mESCs and pancreatic explants. HIF-1␣ levels seem to be different between mESC culture and pancreatic explants because of spatial and temporal patterns of cell-cell interactions. Moreover, the external signals were different because our study used chemically defined medium, whereas pancreatic explants were maintained in serum-containing medium. These differences seemed to have caused the discrepancy.
A previous study modulated the O 2 environment for pancreatic differentiation (59). Cheng et al. used a 5% O 2 environment for maintaining and differentiating human endodermal progenitor cells into ␤-cells. However, they did not mention the reason for using a hypoxic environment and did not compare the effect on differentiation cultured under hypoxia with normoxia. In our study, a hypoxic condition (5% O 2 ) during differentiation had no facilitative effect on the number of Ins1-GFPpositive cells of mESCs. This discrepancy seems to have been particularly caused by the cell density during differentiation. In their studies, endodermal progenitor cells were plated in 12-well dishes at 3-4 ϫ 10 5 cells/well as dissociated at the start of differentiation, whereas in our study, hiPSCs were grown for 7 days as a colony before the start of differentiation. In the colony state, cells appeared to promote a hypoxic phenotype because HIF-1␣ was expressed at a detectable level even in normoxia (Fig. 9D). Therefore, in our case, the high O 2 condition rather than the hypoxic condition facilitated pancreatic differentiation.
In conclusion, the present study showed that a high O 2 condition during differentiation has facilitative effects on generating insulin-producing cells from mESCs and hiPSCs. This effect was due to the inhibition of Notch signaling and activation of Wnt signaling during definitive endoderm to pancreatic fate. We also found that HIF-1␣ inhibition during differentiation accelerated the generation of pancreatic lineages. These observations would provide an efficient method of utilizing patientspecific iPS cells for the treatment of diabetes.