Cell and Developmental Biology Program, Fox Chase Cancer Center, Philadelphia, Pennsylvania 19111, USA. zaret@fccc.edu
Understanding how transcription factors control early pancreas development can yield insight into digestive diseases and guide protocols for therapeutic stem-cell differentiation. A new study, using a sensitive lineage-marking approach, shows that the transcription factor Ptf1a is critical to the specification of pancreatic endocrine, exocrine and duct cells in mice. This indicates an earlier and more pervasive influence of Ptf1a on organogenesis than was previously revealed by standard gene inactivation.
Deciphering the function of a gene-regulatory factor in development can be tricky. Gene expression studies and mutant phenotypes generated by gene inactivation (knockout) can reveal where and when a factor must function, but may miss key roles in controlling cell-fate decisions. For example, loss of a regulatory factor may prevent a cell from adopting a particular fate. This can result in a cell differentiating along an alternative route that is normally adopted by other cells in the original progenitor population. The factor-deficient cells adopting this alternative fate would be difficult to distinguish from cells that normally adopt this fate.
The definitive way to assess this is to perform a simultaneous knockout and lineage analysis, where the fate of cells lacking the factor is followed during development. In the accompanying paper1, Yoshiya Kawaguchi and colleagues present such an analysis for the exocrine pancreatic transcription factor subunit Ptf1a. The unexpected finding is that Ptf1a has an earlier role in pancreatic fate decisions than was previously thought, which nicely illustrates how lineage analysis can define the functions of transcription factors in development.
A pancreas primer The pancreas contains exocrine cells that surround ducts and endocrine cells adjacent to blood vessels. The exocrine, endocrine and duct cells seem to arise from common progenitor cells in the dorsal and ventral regions of the embryonic endoderm2, but much remains to be learned about how the progenitor cells are specified. Exocrine cells, ultimately the most common pancreatic cell type, form dense epithelial structures called acini and secrete digestive enzymes into ducts that drain into the gut. Endocrine cells reside as islets within the acinar framework; each of four different endocrine cell types secretes a unique polypeptide hormone into the circulatory system to regulate metabolism.
The protein Ptf1a, along with p75 and p64, comprises the three subunits of the Ptf1 transcription factor; Ptf1 is expressed in the exocrine cells of the adult pancreas and is a potent activator of exocrine-specific genes3,
4,
5. Although expression of the Ptf1 complex was initially detected by mouse embryonic day 15 of gestation (E15), Ptf1a subunit expression was detected by PCR as early as E9 (ref. 6). However, it was unclear whether Ptf1a is exclusive to exocrine progenitors or is initially activated in exocrine and endocrine progenitors and then silenced in endocrine cells.
Ptf1a and pancreatic origins To address this issue, Kawaguchi et al.1 replaced the mouse Ptf1a coding region with a nuclear Cre recombinase (Ptf1a−cre) and crossed this 'knock-in' allele to the R26R reporter background7, which requires Cre-mediated recombination in a cell to express -galactosidase. Thus, if the Ptf1a promoter expresses Cre only transiently in a lineage8, all descendant cells will inherit the mark of -galactosidase recombination and expression.
Embryonic endodermal lineages. Organs derived from endoderm that is proximal to the prospective pancreatic domain are indicated. Hepatic cells are induced by endodermal interactions with cardiogenic mesoderm. New lineage-marking studies of Kawaguchi et al.1 and Gu et al.9 have demonstrated that the Ptf1a and Pdx1 transcription factors are expressed in early pancreatic endoderm cells that are the progenitors of exocrine, endocrine, and duct cells; the transcription factor Ngn3 is expressed later, solely in endocrine progenitors. Although Ptf1a was previously thought to control exocrine progenitors, the new studies show that it controls an earlier step: the specification of pancreatic endoderm. Kawaguchi et al.1 show that in the absence of Ptf1a, the prospective pancreatic endoderm domain differentiates into intestinal lineages instead.
Gu et al.9 have used a variant of this approach10 to definitively identify pancreatic endocrine progenitors. They created mouse transgenes where the promoter for the gene encoding the transcription factor Ngn3 drives the expression of a bifunctional protein containing Cre fused to an inhibitory segment of the estrogen receptor (Cre-ER). The authors showed that Ngn3-cre-ER−positive cells at E8.5−E10.5 give rise to all four endocrine cell types, which is consistent with the previous demonstration that Ngn3 is necessary to generate all endocrine cells11 (see figure). Although they also provide evidence that the Cre-marked progenitor cells do not arise from ducts, it is not clear whether their Ngn3-cre-ER transgenes are expressed as early as endogenous Ngn3, which seems to be expressed initially in structures that may be duct progenitors12. Stepping back further in development, Gu et al.9 showed that cre-ER transgenes driven by the promoter for Pdx1, which encodes a homeobox transcription factor that helps determine the pancreatic endoderm2, marks endocrine, exocrine, and duct cells in the pancreas (see figure). In adult tissues, Pdx1 expression is specific to insulin-secreting cells13.
If an adult endocrine factor such as Pdx1 can be critical for pancreatic determination, what about an adult exocrine factor, Ptf1a? Kawaguchi et al.1 now find that early embryonic expression of their heterozygous Ptf1a−cre knock-in marks all acinar cells and most of the islet and duct cells, thus establishing that Ptf1a is expressed in progenitors of all pancreatic cell types (see figure). Apparently, Ptf1a is silenced in endocrine and duct cells during or shortly after the partitioning of the progenitors into exocrine cells6.
Furthermore, Kawaguchi et al.1 find that in homozygous Ptf1a−cre/Ptf1a−cre mice, acinar cells are completely absent, whereas a small number of endocrine and duct cells differentiate apparently normally, as reported previously by Krapp et al.6. However, the lineage-marking approach provides insights beyond those of the original Ptf1a knockout. For example, in the Ptf1a−cre/Ptf1a−cre embryos, cells expressing -galactosidase descending from the prospective ventral domain become incorporated into the gut epithelium, where they follow intestinal fates instead of those for pancreas. Specifically, the Ptf1a-deficient cells that express -galactosidase contribute to intestinal crypts and villi, where they differentiate into enteroendocrine cells, goblet cells, and other gut cell types. In heterozygous Ptf1a−cre embryos, cells expressing −galactosidase do not label the gut.
Taken together, the combination of lineage-marking and gene targeting shows that in the absence of Ptf1a, cells that normally would become ventral pancreas will assume an alternative endodermal fate for intestinal epithelium, and thus that Ptf1a is necessary to control the initial cell-type decision to make the pancreas (see figure). The authors go on to show that Pdx1 expressed from the Ptf1a promoter can complement an otherwise Pdx1-null background, further illustrating that Ptf1a is expressed in definitive pancreatic progenitors.
Fateful decisions In the prospective dorsal pancreatic domain, there are two different outcomes for the Ptf1a-deficient cells. First, rare cells expressing -galactosidase become incorporated into the gut, as for the ventral cells. Second, pancreatic duct-like structures expressing -galactosidase extend initially from the duodenal epithelium to the spleen, and they contain rare endocrine cells, of which 50% express -galactosidase. As Krapp et al.6 observed in the original Ptf1a knockout, endocrine cells of all four types are eventually found in the spleen, but Kawaguchi et al.1 find that these cells do not express -galactosidase. Thus, a small number of duct and endocrine cells arise either from cells that never express Ptf1a or from cells in which Cre recombinase does not accumulate sufficiently.
Regardless of these latter possibilities, the Ptf1a null phenotypes and the lineage studies together show that Ptf1a is clearly expressed in and required for the majority of progenitors of all major pancreatic cell types. They further suggest that the development of a fraction of duct and endocrine progenitors is apparently not dependent on Ptf1a. Alas, why can't nature be simple? Well, it does keep us in business.
The differences in dorsal and ventral pancreatic phenotypes seen in the Ptf1a-null embryos may relate to other differences in how these domains are specified2 and their eventual contribution to different domains of the mature gland. With regard to the ventral domain, studies of mice14 and zebrafish15 indicate that the endoderm is at least bipotential. Endoderm proximal to the cardiogenic mesoderm develops into liver and actively excludes the pancreatic fate (see figure), whereas endoderm distal to cardiac develops into the pancreas14. Based on the studies of Kawaguchi et al.1, we can conclude that there is a third fate available to such endoderm, as in the following scenario: Pdx1 and Ptf1a are first activated in ventral foregut endoderm that escapes hepatic induction, and the expression of Ptf1a is necessary for those cells to execute all aspects of the pancreatic program. In the absence of Ptf1a and in the absence of signaling that would induce liver, the endoderm cells assume a third available fate—intestinal differentiation. Regardless, it is clear that knockout phenotypes alone can't divulge the beginning of the story—but lineage analysis can.
-galactosidase. Thus, if the Ptf1a promoter expresses Cre only transiently in a lineage
