Through a conditional Fgf10
gain-of-function model, we present evidence of a competence window of differentiation during pancreatic development. Endocrine, ductal, and exocrine cell terminal differentiation normally occur during the secondary transition between E13.5 and E15.5 in the mouse. Our study reveals that prolonged exposure to Fgf10 differentially affects these terminal differentiation events. We found that the competence to form the endocrine and ductal cell lineages depended upon the gestational timing, not the duration, of Fgf10 exposure. We propose a model of this observation (). Based on this model, we predict that sustained Fgf10 expression throughout the secondary transition causes an irreversible loss of endocrine competence through inhibition of Ngn3 activation. Sustained Fgf10 expression during this period also negatively regulates the competence towards the ductal cells fate, as evidenced by the lack of several ductal-specific markers (Supp. Fig. 4
). We conclude that the competence window for ductal and endocrine cell formation is irreversibly lost as a result of ectopic Fgf10 expression during the secondary transition. We also attribute the shift towards the exocrine cell fate in our conditional Fgf10
model as a result of the maintenance of Ptf1a in the Fgf10-arrested progenitors and the rapid restoration of bhlh8
/MIST1 upon Fgf10 withdrawal.
An endocrine competence window in pancreatic development
What causes the loss of endocrine and ductal cell competences when Fgf10 is overexpressed? We observed that Ngn3 expression is negatively influenced by the continued presence of Fgf10 (). This inhibition of Ngn3 expression by Fgf10 may explain the general loss of endocrine cell formation; Ngn3 activity is required for endocrine cell fate determination (Gradwohl et al., 2000
). We also found that when the arrested organ is released from Fgf10 suppression, the endocrine cell types normally forming late (somatostatin and ghrelin) were preferentially formed (, ), possibly reflecting delayed Ngn3 activation. In support of this observation, Johansson et al. (2007)
reported that the timing of Ngn3 activation significantly affected which endocrine cell sub-types were capable of forming; competence towards the γ cell subtype occurs after E14.5 (Johansson et al., 2007
). A possible explanation for suppression of Ngn3 expression is active Notch signaling as evidenced by the increased levels of Hes1 (). Hes1 maintains the progenitor state and represses Ngn3 target genes, such as NeuroD
(Jensen et al., 2000
). Increased levels of Sox9 expression in the Fgf10-arrested epithelium () may also contribute to the high levels of Hes1 as Sox9 has been shown to regulate Hes1 (Seymour et al., 2007
). Furthermore, Fgf10 signaling has also been shown to stimulate the Notch pathway in the pancreatic epithelium (Miralles et al., 2006
). However, Fgf10-mediated Notch signaling cannot explain why Ngn3 expression was not restored upon later attenuation of ectopic Fgf10 expression (). This implies that other factors necessary for Ngn3 expression are only present prior to and during the secondary transition. We did observe several Ngn3+
cells after the secondary transition in WT and DTG embryos (, ), although far fewer of these cells were observed at E16.5 than during the secondary transition. This suggests that endocrine progenitors persist beyond the secondary transition. In support of this observation, others have reported the existence of Ngn3+
cells in the adult pancreas, although the presence of these cells is extremely rare (Gu et al., 2002
; Xu et al., 2008
). The concomitant loss of ductal cell competence implies that the instructional cues for this particular cell fate are active in a common ductal/endocrine precursor. However, as the necessary factors for duct cell fate determination are currently unknown, it is difficult to speculate on why ductal competence is lost; we cannot determine whether the extended duration of Fgf10 expression affected a common ductal/endocrine precursor, the individual ductal cell lineage, or both. Identification of the factors that underlie ductal cell competence will be necessary to distinguish between these possibilities.
Why is exocrine competence unaffected by Fgf10 overexpression throughout the secondary transition? Widespread Fgf10 expression may commit the pancreatic progenitors towards this cell fate. This notion was suggested by pancreatic explant studies that showed increased exocrine differentiation, at the expense of endocrine cell development, occurred in the presence of Fgf10 (Miralles et al., 1999
). However, it is clear that Fgf10 does not control exocrine fate determination because exocrine differentiation proceeds in Fgf10
null mice (Bhushan et al., 2001
). Widespread Hes1 staining in the Fgf10-arrested epithelium () suggests that Notch signaling may be involved in regulating exocrine competence. Yet, Notch gain-of-function models showed that activation of the Notch pathway potently suppresses terminal differentiation of both endocrine and exocrine lineages (Murtaugh et al., 2003
; Hald et al., 2003
). Also, loss-of-function of Hes1
demonstrated that exocrine differentiation proceeds in that model (Jensen et al., 2000
). Consequently, active Notch signaling cannot solely be used to explain the persistent exocrine competence in the Fgf10-arrested progenitors. Competence for the exocrine lineage throughout the secondary transition may be attributable to sustained Ptf1a expression in the Fgf10-arrested epithelium (, ). Expression of Ptf1a is necessary for development of the exocrine compartment (Dong et al., 2008
). Ptf1a is unlikely a direct transcriptional target of Fgf10 because Fgf10 would otherwise have an instructive role in exocrine development; we demonstrated that exocrine differentiation is also suppressed in the presence of Fgf10 (, , ). Furthermore, we would have expected a strong induction of Ptf1a in the presence of Fgf10, which was not observed (). In zebrafish, Ptf1a expression remains in the Fgf10
mutant pancreas, indicating its expression is independent of Fgf10 (Dong et al., 2007
). Therefore, other downstream effectors of Fgf10 signaling potentially influence Ptf1a activity in establishing the exocrine compartment. Notably, we have identified two targets of Fgf10 signaling, Etv4
, with similar expression patterns to Ptf1a ( and (Kobberup et al., 2007
)). Further studies using gene ablation of Etv4
specifically in pancreas will address their role in regulating the activation of the pro-exocrine factors, such as Ptf1a.
Segregation of the epithelial progenitor population into distal and central domains potentially patterns the developing pancreas. Zhou et al.
(2007) described the distal domain using a “tip and trunk” analogy of the epithelial progenitor pool prior to differentiation (Zhou et al., 2008
). These authors demonstrated that the distal “tips” of the branching pancreatic epithelium contain the pancreatic progenitors that become committed to the exocrine cell fate, while cells in the “trunk” are committed to endocrine and ductal cell lineages (Zhou et al., 2008
). In support of this hypothesis, Ptf1a expression is clearly distalized and significantly down-regulated in the centralized region as the mouse organ approaches the secondary transition (Hald et al., 2008
). In the Fgf10 overexpressing pancreas, the distal segregation of Ptf1a expression is not observed (). Ptf1a remains expressed throughout the Fgf10-arrested cells. MIST1, a marker of fully differentiated exocrine cells (Johnson et al., 2004
), is only expressed at the distal tips of the wild-type pancreas (). However, in the Fgf10-arrested organ, MIST1 appears widely distributed at low levels (). The loss of Ngn3, which is normally only observed in centralized progenitors, in combination with the widened expression pattern of Ptf1a and MIST1, suggest a loss of the central domain in response to prolonged Fgf10 expression and a possible replacement by a more distalized one. Such a result is in agreement with the observed loss of the endocrine and ductal cells and would argue that the time of Fgf10 exposure gradually shifts competency zones within the organ.
The absence of instructional cues for particular fates to emerge in the presence of elevated Fgf10 expression likely underlies the loss of endocrine and ductal competence. Our results indicate that spatial segregation of certain cell intrinsic components is required for establishing the patterning domains. For example, Hnf6 is maintained in presence of Fgf10 (). Although Hnf6 was reported to be an upstream activator of Ngn3 gene expression (Jacquemin et al., 2000
), it fails to activate Ngn3 in the Fgf10-arrested pancreas upon attenuation of Fgf10 expression. Moreover, the increased levels of Hnf6 in the Fgf10-arrested progenitors () could not specify the ductal cell fate (). Thus, we conclude that Hnf6 expression alone is insufficient in generating a pro-endocrine/pro-ductal field. It is currently unknown to what extent loss of Ngn3 may be causal to loss of central patterning. Given that distal/central patterning of the developing pancreas is likely to involve a large battery of cell signaling components, some of these are likely to have been impacted by ectopic Fgf10 expression. Rescue experiments using Ngn3 gain-of-function should be useful to address to what extent endocrine and possibly ductal competence relies on this factor.
Although our observations pertain to the effects of Fgf10 protein on patterning during later pancreatic development, it is of interest to compare our results to the role of Fgf10 during early endoderm formation. The initial study of Fgf10
deficient mice concluded that Fgf10 is required for pancreatic growth (Bhushan et al., 2001
), although studies on endodermal patterning were not performed. Analyses of Fgf10 function in zebrafish by Dong et al. (2007)
have provided a more detailed picture of the role of Fgf10 in establishing cell fates in the hepatopancreatic ductal system (HPD) (Dong et al., 2007
). At time of pancreatic budding, Fgf10 is expressed in Isl1-positive mesenchymal cells adjacent to the core structure of the HPD, and its absence leads to a generally failed formation and thinning of these ducts. These regions become populated by ectopically forming endocrine and liver cells. In mice, Hes1
mutants display ectopic formation of endocrine cells in developing pancreas (Jensen et al., 2000
) and in the bile duct epithelium (Sumazaki et al., 2004
). As the bile duct is part of the HPD system, this argues that Fgf10 controls Notch signaling in the early endoderm. The observation that Fgf10 is involved in suppressing differentiation in the proximal pancreas and liver adjacent to the HPD in zebrafish and our observation that overexpression of Fgf10 suppresses the centralized pancreatic cell fates in mice (i.e.
endocrine and ductal cell types) later in pancreatic development suggest of the timing and location of Fgf10 expression is crucial for organ patterning.
Our observations have clinical implications as Fgf10 is currently used in cell-based therapeutic approaches for diabetes. Fgf10 has already been implemented in the successful conversion of human embryonic stem cells towards the beta-cell fate through a directed differentiation strategy (D'Amour et al., 2006
). The D’Amour study convincingly suggests that hES cells are highly suitable for a cellular replacement therapy (Madsen and Serup, 2006
). However, the current methodology fails to efficiently generate clean fractions of insulin-producing cells. The competence window for endocrine cell formation described herein should not be overlooked. Our findings illustrate that comprehension of cellular competence during pancreatic development may provide an additional basis of understanding endocrine differentiation. This highlights a current need to characterize the local morphogen environment preceding and during the secondary transition and how this is controlled by Fgf10. Knowledge of such an environment should be highly beneficial to optimize the current directed differentiation protocols.