Whereas continuous exposure to all three

Whereas continuous exposure to all three factors (4+) resulted in higher levels of DE genes, there was no significant difference in the proportion of Foxa2+Sox17+ cysteine protease when compared to transient Wnt activation (2+2−). Surprisingly, T expression in a fraction of differentiated cells was detected in both conditions. We assume that this represents residual T protein expressed at the mesendoderm stage (Bakre et al., 2007; Tada et al., 2005). The efficient DE differentiation that we observed is concordant with the findings in EBs treated similarly (80% Cxcr4+ cells), and is further strengthened by the co-labeling with E-Cadherin that excludes Cxcr4-expressing mesoderm cells or visceral endoderm (Li et al., 2011; Morrison et al., 2008; Yasunaga et al., 2005). Furthermore, this method allows for a robust and FACS-detectable expression of the surface marker Cxcr4 at the DE stage, and not 4days later as recently suggested (Naujok and Lenzen, 2012). The profile of Cxcr4 also suggests that it marks the established DE cells and is not required early during their differentiation. It will therefore be possible in the future to purify Cxcr4+ cells exactly at the DE stage and use them for further analysis or differentiation, which will thereby exclude extraembryonic endoderm cells that are actually not expected when the Wnt pathway is so highly activated (Chazaud and Rossant, 2006). We examined the pancreatic differentiation potential of mESC-derived DE cells, following a strategy that we previously applied to hESC cultures (D\'Amour et al., 2006; Mfopou et al., 2010a; Nostro et al., 2011; Sui et al., 2012). This resulted in the differentiation of Pdx1+ and Pdx1+Nkx6.1+ cells that can be considered as posterior foregut and pancreatic progenitors respectively. Expression of Nkx6.1 protein followed that of Pdx1, indicating a progressive differentiation as occurs in vivo (reviewed in (Gittes, 2009; Oliver-Krasinski and Stoffers, 2008)). Our findings further support the view that the main pathways regulating pancreas development are conserved between mouse and human, and therefore validate this model as a screening tool. To this end, the easy manipulation of mESCs in culture represents an asset for the optimization of differentiation protocols that aim at generating functional cells in vitro. Of course, implementation of such protocols back to hESCs might still need some adjustments. Failure to demonstrate beta cell differentiation (Insulin+ or C-peptide+ cells) from our cultures is in line with the current status of ESC research, and suggests that Glucagon+ and Synaptophysin+ cells that we detected are endocrine cells of the primary transition as described during pancreas development and during hESC differentiation (D\'Amour et al., 2006; Oliver-Krasinski and Stoffers, 2008; Van Hoof et al., 2009). This situation has prompted the evaluation of Pdx1+Nkx6.1+ progenitors in diabetes cell therapy, assuming that the in vivo microenvironment will guide the generation of functional beta-cells after transplantation (Eshpeter et al., 2008; Kroon et al., 2008; Rezania et al., 2012; Schulz et al., 2012; Sui et al., 2013; Tuch et al., 2011). In the experiment that we performed with crude cell cultures containing about 25% Pdx1+Nkx6.1+ cells, all transplanted animals developed teratoma in which tissues of endodermal origin were more frequent. This proportion is far higher than the 13 to 46% tumor rates reported with implanted hESC-derived pancreas progenitor cell preparations (Kelly et al., 2011; Kroon et al., 2008; Sui et al., 2013). Therefore pancreatic progenitors should be sorted out before their implantation in vivo. Considering that CD24 does not specifically label the Pdx1+ progenitor cells derived from mouse and human ESC as initially suggested, more attention should be paid to CD142 and to the discovery of new surface markers (Jiang et al., 2011; Kelly et al., 2011; Naujok and Lenzen, 2012).