Diabetes results from the autoimmune destruction of pancreatic β-cells. The last years have witnessed major advances in engineering renewable supplies of β-cells through a combination of extrinsic cues and intrinsic reprogramming of hepatic stem cells. During embryogenesis, liver and pancreas arise from adjacent regions of the anterior foregut endoderm [1] and mutual plasticity between the liver and pancreas in the adult occurs in various pathological and experimental settings [2]. Lineage tracing studies have defined pancreatic master genes determining specification and maturation of the pancreatic endocrine cell lineage, namely pancreatic and duodenal homeobox-1 protein (Pdx1), bHLH proteins pancreatic transcription factor 1a (Ptf1a/P48) and neurogenin 3 (Ngn3) [3, 4]. Conversely, loss of function and misexpression experiments demonstrate that the Notch pathway plays a pivotal role in pancreatogenesis as it maintains an undifferentiated progenitor population. Hairy enhancer and split 1 (Hes1) acts as a key downstream effector of the Notch pathway by repressing pancreatic endocrine cell specification via direct Ngn3 down-regulation [5]. Hes1 repression promotes conversion of biliary cells to pancreatic cells in developing tissues [6]. Although earlier findings demonstrate the feasibility of in vitro and in vivo liver-to-pancreas conversion [7], appropriate allo- or autologuous cell sources need to be characterised. Bipotential mouse embryonic liver cells have been reproducibly isolated from mouse embryos of different strains. BMEL cells have the potential to differentiate into hepatocytes and cholangiocytes in specific 3-D culture systems in vitro or after engraftment in damaged livers in Alb-uPA/SCID mice in vivo [8]. In the present study, we seek for the expression of pancreatic progenitor markers and Hes1 expression in BMEL stem cells in different culture modes.

Liver-to-pancreas conversion offers new possibilities for β-cell engineering for cell therapy of diabetes [14]. Here we focus on untransformed BMEL cells, a new model of bipotential liver progenitor cells that have been reproducibly isolated from different inbred mice strains at 14 dpc. In adherent culture, these cells propagate with a high proliferation index in an undifferentiated state [8]. Strikingly, adherent BMEL cells in our hands also display functional similarities with multipotent pancreatic precursor cells, namely Pdx1 and Ngn3 expression, which does not occur in embryonic liver development. Indeed, Pdx1 is thought to act as a master regulatory gene, sufficient per se to distinguish liver and pancreas in normal development. No spontaneous Pdx1 expression in the hepatic lineage has been described yet during embryogenesis or in the adult liver in the absence of high glucose concentrations [15]. Western blot analysis reveals a 46 kDa endogenous Pdx1 protein, which is in accordance with the apparent weight of the active phosphorylated form of Pdx1 in earlier findings [16]. Ngn3, a true marker of endocrine commitment [4, 17] is detected at the protein, but not at the transcript level in adherent cells. This apparent discrepancy may stem either from differences in the sensitivity of the techniques used here or from a greater stability of the protein. Following aggregate formation, BMEL cells further express Hnf6, and Ngn3 becomes detectable at the transcript as well as the protein level. These results fit with the idea that BMEL cells do not closely phenocopy bipotential hepatoblasts in vivo, and may present an enlarged differentiation potential. Immunocytochemistry results presented in Fig. 1 show that most BMEL cells evenly express Pdx1 and Ngn3 ruling out the idea that pancreatic progenitor-like cells form a sparse subpopulation. It is noteworthy that Ngn3-positive cells have a limited mitotic potential [18], opening the possibility for rapid overgrowth by less differentiated cells in culture. Apart from Pdx1, Hnf6, and Ngn3, BMEL cells do not express any of the downstream effectors of more advanced pancreatic commitment. Moreover, expression of pancreatic markers in aggregates remains unchanged following culture in high glucose medium, which is known to stimulate maturation of pancreatic progenitors (25 mM glucose, data not shown) [19]. Hes1, a direct repressor of Ngn3, is expressed in BMEL cells regardless of culture conditions, albeit quantitative real-time PCR shows a decrease with time in aggregate culture. Although Hes1 is expressed in intrahepatic bile duct cells at the neonatal stage, it is usually absent in hepatoblasts [20]. On the other hand, Hes1 is present in pancreatic progenitor cells where it acts as a dominant factor promoting self-renewal at the expense of β-cell differentiation [21]. Previous microarray data suggest that Notch signalling pathway is operative in maintaining BMEL cells in an undifferentiated state [22]. However, notch signalling disruption experiments failed to initiate activation of late-stage pancreatic transcription factors in BMEL aggregates (Additional_data_1). These results support the idea that factors beyond Hes1 expression, presumably the absence of Ptf1a, impede the extent of expression of Ngn3 and other factors of more advanced β-cell differentiation [23–25]. Indeed, lineage-tracing experiments have shown that endocrine i.e. Ngn3-positive cells derive from progenitors that express Ptf1a and, henceforth, Ngn3 and Ptf1a are usually co-expressed in pancreatic multipotent progenitors [3, 23]. To summarize, here we report that BMEL cells display two stage-specific features of multipotent pancreatic progenitor cells, namely Pdx1 and Ngn3 expression. In the field of β-cell replacement research, BMEL cells are of potential interest for the study of inductive signals critical for β-cell maturation including positive selection of cells, trophic factors, or ectopic expression of pancreatic transcription factors like Ptf1a in complement to endogenous Pdx1 and Ngn3 expression.