To study the effects of FGF10 on the self-renewal of the pancreatic progenitors we have used explant cultures of isolated E10.5 mouse dorsal pancreatic epithelia. The explants were grown into reduced growth factors Matrigel with or without FGF10 and their development analyzed after three or seven days of culture. Our results showed that after three days of culture the untreated pancreatic epithelia did not increase in size while those cultured with FGF10 presented a 4-fold increase in size. The analysis of BrdU incorporation indicated very low proliferation in the untreated rudiments and active proliferation in those receiving FGF10. The untreated pancreatic epithelia were essentially composed by endocrine cells (90% of the total cell surface in the rudiment) and only a few cells remained undifferentiated (10%). In the FGF10 treated epithelia the reverse was observed: the endocrine cells represented only 15% of the rudiment surface area while the remaining 85% was composed of undifferentiated cells. Our results indicate that in the untreated pancreatic epithelia the progenitors cells fail to proliferate and differentiate rapidly. In contrast, in the FGF10-treated epithelia the pancreatic progenitors proliferate actively and remain in an undifferentiated state characterized by the co-expression of the transcription factors Pdx1, Nkx6.1, Ptf1б and Hes1. After seven days of culture the untreated pancreatic epithelia did not increase in size and were composed essentially by endocrine cells (85%) and a few acinar cells (15%). On the contrary during this period the explants treated with FGF10 underwent a 10-fold increase in size. In these rudiments the endocrine cells represented only 10% of the total cell surface, most cells being acinar cells (40%) or undifferentiated cells (50%). Interestingly BrdU incorporation in these explants concerned always the acinar cells (15%) and the undifferentiated cells (20%). Taken together these results indicate that FGF10 has different effects on the proliferation of the pancreatic epithelia. It stimulates the proliferation of pancreatic progenitors and induces the proliferation of differentiated acinar cells. The high BrdU labelling index of the acinar cells in the FGF10 treated rudiments suggests that most of these cells arise by proliferation of the few acinar cells which differentiate from the isolated pancreatic epithelia. It is well established that acinar differentiation is dependent on mesenchymal signals (Rutter et al., 1978; Gittes et al., 1996; Miralles et al., 1998). However, in our culture conditions we observed that a few acinar cells form in isolated pancreatic epithelium. The absence of vimentin staining indicated that there was no contamination by mesechyme cells in these cultures. Therefore these cells probably arise from pancreatic progenitors, which were already engaged in acinar differentiation prior to mesenchyme removal. In this respect it is noteworthy that although not detectable by immunohistochemistry, some specific acinar products like carboxypeptidase A can be detected by RT-PCR as early as E9.5 in the pancreatic rudiments of mice (Gittes and Rutter, 1992), (and our own results). An alternative explanation would be that the role of the mesenchyme in pancreatic development is to furnish permissive rather than instructive signals. The FGF10 (as well as other growth factors) secreted by the mesenchyme would allow the expansion of progenitor cells, that subsequently differentiate into exocrine cells.

The trophic effects of FGF10 on the mouse pancreatic epithelia were not unexpected since we had previously shown that FGF10 stimulates the proliferation of the rat embryonic pancreas (Miralles et al., 1999). Moreover FGF10-null mice presented an hypoplastic pancreas (Bhushan et al., 2001), while transgenic mice over expressing FGF10 in the pancreas displayed pancreatic hyperplasia (Hart et al., 2003;Norgaard et al., 2003). It should be mentioned that transgenic expression of a dominant negative FGFR2b under the control of the Pdx1 promoter did not lead to a pancreatic phenotype (Hart et al., 2000) and only a week pancreatic hypoplasia was observed in the FGFR2b-null mice (Pulkkinen et al., 2003). These studies seem to contradict the hypothesis that FGF10 could play a major role in the control of the proliferation of the undifferentiated pancreatic epithelia. However it must be noted that FGF10 might signal trough other receptors (Powers et al., 2000). Moreover, other factors like FGF2, FGF7, EGF and HGF are also able to stimulate the proliferation of the embryonic pancreatic epithelia and could compensate the loss of FGF10 signalling (Kim and MacDonald, 2002;Edlund, 2002).

Another important effect of FGF10 is its capacity to maintain a considerable number of cells in an undifferentiated state. As indicated above, after seven days of culture in the presence of FGF10 a considerable number of cells in the isolated pancreatic epithelia did not stain positively for endocrine or exocrine markers of pancreatic differentiation. These cells co-expressed the transcription factors Pdx1, Nkx6.1 and p48/Ptf1б, which is a characteristic of pancreatic progenitors. Moreover, most of the undifferentiated cells in the FGF10-treated epithelia express the transcription factor Hes1. This is another characteristic of pancreatic progenitors and indicates that the Notch pathway is active in these cells. These results are similar to what has been reported in the pPdx1-FGF10 transgenic mice, which showed pancreatic hyperplasia, maintenance of pancreatic progenitors in an undifferentiated state and persistent Notch activation (Hart et al., 2003;Norgaard et al., 2003). In our in vitro cultures, acinar differentiation was not so efficiently blocked as it was in the transgenic mice. This could be explained by the fact that in the transgenic mice FGF10 is expressed at the onset of Pdx1 expression, that is E8.5. Thus, in these animals, FGF10 is acting on the early pancreatic progenitors. In our study we have used E10.5 pancreatic epithelium. At this stage most cells in the pancreatic rudiment are progenitor cells, but a few cells have differentiated into endocrine cells and others are probably, as mentioned above, already engaged into acinar differentiation. The mitotic effect of FGF10 on the few acinar cells which differentiate spontaneously probably enhances the relative proportion of acinar cells in our model.

The proportion of undifferentiated cells in the mouse explants is increased comparatively to what was previously observed using isolated E11.5 rat dorsal pancreatic epithelium (Miralles et al., 1999). The E11.5 rat and the E10.5 mouse dorsal pancreatic rudiments are very similar in terms of transcription factors expression and cell differentiation status. Thus, the differences are essentially due to different culture conditions. The cultures were done in serum free conditions in previous studies whereas we added 1% fetal calf serum (FCS) in the present study. We noted that despite the presence of FGF10, a minimal amount of other growth factors was necessary to allow the survival and growth of the pancreatic precursors when cultured in the absence of mesenchyme.

Another unexpected observation is the absence of ductal cells in our cultures. Gittes and co-workers have shown that isolated mouse pancreatic epithelium grown in Matrigel developed into cystic structures formed by differentiated ductal cells (Gittes et al., 1996). In our study we have used reduced growth factor Matrigel, which contains the same basement membrane components than Matrigel but has been greatly depleted in growth factors. Apparently our culture conditions are less favourable to ductal differentiation. However, a few ductal cells were detected in the explants treated simultaneously with FGF10 and compound 1 suggesting that somehow, the blockage of Notch signalling could be required to allow ductal differentiation. It must be noted also that, as in the FGF10-treated explants, differentiated ductal cells were not detected in the pPdx1-FGF10 transgenic mice.

The role of the Notch pathway in the control of pancreatic differentiation is now well established. Loss of function of various Notch pathway genes (Hes1, Delta1, RBPjk) leads to premature and massive differentiation of the pancreatic progenitors into endocrine cells (Apelqvist et al., 1999;Gradwohl et al., 2000;Jensen et al., 2000). A similar phenotype was observed in transgenic mice expressing the bHLHL trascription factor Ngn3 (a gene usually repressed by Notch activation) under the control of the Pdx1 promoter ( pPdx1-Ngn3 ), (Apelqvist et al., 1999). Moreover, when a Notch-IC transgene is activated in the developing mouse pancreas using the Pdx1 promoter, both endocrine and exocrine differentiation are repressed, suggesting that Notch has an inhibitory role in the control of the differentiation of both lineages (Murtaugh et al., 2003). It has been suggested that maintenance of the pancreatic progenitors in an undifferentiated state in the pPdx1-FGF10 mice could result of an eventual effect of FGF10 in inducing persistent activation of the Notch pathway. Our study corroborates this hypothesis. Hes1, a target gene of the Notch pathway, was rapidly downregulated in the isolated E10.5 pancreatic epithelia, but its expression persisted in the explants treated with FGF10. Moreover, the г-secretase inhibitor, compound 1, downregulated Hes1 expression and considerably reduced the growth and the number of undifferentiated cells in the FGF10-treated pancreatic epithelia. Thus, the inhibition of the Notch pathway prevents the effect of FGF10 on the proliferation and the maintenance of the pancreatic progenitors in an undifferentiated state. Therefore, the Notch pathway is required as a downstream mediator of the FGF10 signalling in pancreatic precursors. We do not know how FGF10 maintains the Notch activation. It has been suggested, based on the expression of the Notch ligand genes Jagged 1 and Jagged 2 in the undifferentiated pancreatic epithelium of the pPdx1-FGF10 mice, that FGF10 could induce the expression of these ligands (Norgaard et al., 2003). FGF10 could also down-regulate repressors of Notch activity like Sel1 (Hart et al., 2003) or up regulate Notch expression. In this regard, it has been shown that the FGF1 and FGF2 induce the proliferation and inhibit the differentiation of neuroepithelial precursors trough Notch signalling. Both FGFs efficiently up-regulate the expression of Notch 1 in these neuronal precursors (Faux et al., 2001). However, our RT-PCR analysis did not show any major differences in the levels of expression of these genes between the FGF10-treated and untreated pancreatic epithelia.

It is noteworthy that untreated isolated pancreatic epithelia showed in vitro an outcome similar to that of the pancreas of mice deficient for different genes of the Notch pathway, or the pPdx1- Ngn3 mice. That is, arrested growth, accelerated and almost total differentiation of the pancreatic epithelium into endocrine cells and also, as we observed in our study, rapid down-regulation of Hes1. This suggests that the mesenchyme not only provides signals necessary for the growth of the pancreatic epithelium but it can also regulate the maintenance of the pancreatic progenitors in an undifferentiated state via the Notch pathway. In the developing pancreas FGF10 is essentially produced by the pancreatic mesenchyme, while its receptor FGFR2b is expressed only in the epithelial cells (Miralles et al., 1999). Thus, the pattern of expression of FGF10 is consistent with the hypothesis that FGF10 could be the mesenchymal factor responsible of the maintenance of the Notch signalling. Studies on the development of other organs have also implicated the FGFs in the self-renewal of progenitor cells via the Notch pathway. Of particular interest, in this context, are several studies showing that FGF10 is capable of preventing the differentiation of the odontoblasts by inducing Notch signalling (Mitsiadis et al., 1997;Mustonen et al., 2002). These studies showed that Hes1 expression in the dental precursors is dependent on mesenchymal signals and that FGF10 induces Hes1 expression in explants of isolated dental epithelium. In these explants FGF10 also induces the expression of Lunatic fringe, an enhancer of Notch activity. Interestingly, we have found that Lunatic fringe is expressed throughout pancreatic development and that its maximal expression occurs between E12 and E16, a period corresponding to the expansion of the population of pancreatic precursors. This period coincides with the maximal expression of FGF10 by the pancreatic mesenchyme. Moreover, FGF10 induces Lunatic fringe expression in the E10.5 isolated dorsal pancreatic epithelium. Thus, in vivo FGF10 could maintain Notch activity in the pancreatic precursors by inducing Lunatic fringe.

The present and previous studies indicate that FGF10 is able to maintain active Notch signalling. However, other signalling pathways are also probably implicated in Notch control. In this respect, a recent study has shown that TGFб can induce Notch activation in explant cultures of pancreatic acinar cells. Upon treatment with TGFб these cells expressed high levels of Pdx1 and Hes1 and underwent acinar to ductal metaplasia (Miyamoto et al., 2003). The study of the interactions between the Notch pathway and other signalling cascades implicated in pancreas development will be crucial to further unravel the mechanisms controlling the self-renewal of the pancreatic precursors.