Hepatic Competence

Hepatic competence, or the ability to form liver from the foregut endoderm, is considered the first step in a two-step process for the specification of the liver in vertebrates. Competence is a prerequisite for the endoderm to respond to specific signals, such as Fgfs from the surrounding mesoderm, which then leads to the second step, the induction of liver specific genes such as ALB (albumin). This “competence” is facilitated through the FOXA gene transcription factor family that includes FOXA1 and FOXA2 (forkhead box proteins A1 and A2). FOXA gene expression precedes the induction of the hepatic program by FGF signals in the endoderm, and FOXA2 binding revises α-fetoprotein (AFP) gene transcription in vitro (Crowe et al, 1999). Deletion of FOXA1 and FOXA2 led to no liver bud formation with the loss of AFP expression in the ventral foregut, which would indicate that hepatic specification had failed to initiate (Lee et al, 2005). The forkhead box (FOX) proteins are an extensive family of transcription factors that share homology with the winged helix/forkhead DNA binding domains (Kaestner et al, 2000) that play important roles in regulating expression of genes involved in cellular differentiation, proliferation, transformation, and metabolic homeostasis (Wang et al, 2001; Zaret et al, 1999; Duncan, 2000). In vitro explant cultures demonstrated that Fgf was sufficient to induce ventral foregut endodermal cells to differentiate into hepatoblasts (Guadi et al, 1996). However, when FOXA1/FOXA2-deficient endoderm was cultured with exogenous Fgf2, no liver expression was seen (Lee et al, 2005). This result would indicate the need for FOXA1 and FOXA2 for the onset of hepatogenesis. In vivo DNA-binding studies revealed that the liver-specific albumin gene ALB has an important upstream regulatory binding site for FOXA factors (Guadi et al, 1996; Bossard & Zaret, 1998). Prior to FOXA or GATA4 binding, the ALB gene is transcriptionally silent with a closed chromatin. After binding with FOXA and GATA4, the chromatin domain is thought to become exposed (Cirillo et al, 2002), thus increasing the ability of the gene to be activated. By the E9.5 gestation stage, other transcription factors—CCAAT/enhancer binding protein-β (CEBPB) and NF1, nuclear factor 1—bind to sites adjacent to the FOXA site, and as a result, the albumin gene becomes active (Zaret, 2002). The critical binding of FOXA to chromatin increases the ability of the gene expression to be activated by allowing other transcription factors to bind, hence the hepatic competence.

Hepatic Induction

Fgf from the cardiogenic mesoderm and Bmps from the septum transversum mesenchyme have been implicated in the induction of liver fate in mouse and chick embryonic endoderm. In vitro studies were performed in which the ventral foregut and cardiogenic mesoderm were cocultured in the presence of fibroblast growth factor inhibitors, resulting in the inhibition of liver induction (Jung et al, 1999). Culturing foregut endoderm without the cardiogenic mesoderm in the presence of a low concentration of exogenous FGF2 (2 to 5 ng/mL) rescued hepatic gene expression (Jung et al, 1999) and simultaneously suppressed the expression of the pancreas program (PDX1 expression) (Deutsch et al, 2001), whereas a high concentration of FGF2 (10 to 500 ng/mL) induced NKX2-1, an early marker for respiratory epithelium but not albumin gene expression (Serls et al, 2005). Ventral foregut endoderm cocultured with noggin (Nog), a Bmp inhibitor, inhibited hepatic gene induction, an effect that was rescued when Bmp2 or Bmp4 were added. Despite these results, however, embryos that were homozygous mutant for BMP4 still exhibited normal hepatic gene induction (Rossi et al, 2001; Smith & Harland, 1992). These data indicate that the cardiac mesoderm, a source for FGF signaling, is crucial for the hepatic induction from the ventral foregut endoderm and subsequent morphogenesis, whereas Bmp aids in the process as well (Fig. 1A.3).

FIGURE 1A.3 Schematic for liver ontogeny controlled by factors released from the cardiogenic mesenchyme (1) and septum transversum mesenchyme (STM, 2), which induce hepatic specification. Signaling molecules (e.g., Vegfr2, HGF) and transcription factors (e.g., Prox1, Hlx) from STM and endothelial cells leads to delamination and migration of endodermal cells into the STM to subsequently form the liver bud. BMP, bone morphogenetic protein; HGF, hepatocyte growth factor; c-Met, HGF receptor; Vegfr2, vascular endothelial growth factor receptor 2.

(Image courtesy Mariam Hobeldin.)

Morphogenesis of the Hepatic Bud

Following hepatic specification (E8.5 to E9.0), the liver starts to express liver-specific genes (ALB, AFP, HNF4A) that eventually form the liver bud. Hepatic bud morphogenesis is facilitated by two transcription factors, HHEX (hematopoietically expressed homeobox, discussed above) (Crompton et al, 1992) and PROX1 (prospero-related homeobox 1) (Oliver et al, 1993). HHEX is expressed in the anterior endoderm at E7.0, eventually giving rise to the ventral pancreas and liver. HHEX-null embryos grow without a liver, thyroid and develop forebrain defects at E11.5; however, evidence of endodermal epithelial thickening suggest a possible defect in differentiation as evidence of possible hepatic induction (Martinez-Barbera et al, 2000).

In a separate study on HHEX-null embryos, liver genes ALB and PROX1 were expressed in the ventral foregut endoderm at around E8.5, with the thickening of the hepatic endoderm region at E9.0 being smaller compared with heterozygote embryos. In addition, a significantly lower proliferation rate was seen in the prospective hepatic domain compared with controls, as demonstrated by bromodeoxyuridine (BrdU) staining, with no evidence of apoptosis (Bort et al, 2004). The basal membrane layer, which is rich in laminin, surrounds the hepatic endoderm and degrades around E9.0 to E9.5, so that the hepatocytes start migrating into the septum transversum mesenchyme (STM) to form the liver bud. This degradation is facilitated by the hepatoblasts, which under normal conditions downregulate E-cadherin. However, in PROX1-null mutant embryos, the progenitor liver cells fail to migrate into the STM as a result of excess E-cadherin and basement membrane proteins laminin and collagen 4. The basal lamina fails to degrade, and the cells remain trapped in the hepatic diverticulum with an overall reduction in liver size (Sosa-Pineda et al, 2000). The bulk of the liver lobe lacks hepatocytes, suggesting that the mesenchymal component contributes most of the liver mass in PROX1-null mutant embryos. A similar phenotype is seen with ONECUT1 and ONECUT2 double mutants, which are required for basal lamina degradation (Margagliotti et al, 2007). Furthermore, pharmacologic inhibition of matrix metalloproteinases (MMPs), extracellular matrix remodeling enzymes usually expressed by hepatoblasts and STM cells, inhibits hepatoblast migration in culture (Margagliotti, 2008). To further illustrate the importance of the extracellular matrix in hepatic bud morphogenesis, hepatoblasts deficient in laminin receptor β1-integrin fail to colonize the liver bud (Fassler & Meyer, 1995). These β1-integrins are among those that act as receptors for extracellular matrix proteins, such as laminins and collagens (Hynes, 1992).

In summary, PROX1 and ONECUT factors are important in regulating delamination and in controlling hepatoblast migration through regulating MMPs and hepatoblast interactions with the extracellular matrix. Without an appropriate extracellular matrix, cell migration into the STM will be disrupted.

The role of endothelial cells in liver organogenesis has also been studied, with the liver vasculature providing a vital source for hematopoiesis in early life. Endothelial cells are positioned as a loose chain of cells interceding between the thickening hepatic epithelium and the STM (Matsumoto et al, 2001). Embryos that were null mutant for VEGFR2 (also known as KDR) had failure of delamination and subsequent migration by the hepatoblasts (Matsumoto et al, 2001). These results imply that the endothelial cells interact with nascent hepatic cells and aid in liver bud outgrowth.

Liver Bud Growth

The STM cells are closely related to the ventral endoderm and contribute to hepatic induction and growth. This epithelial–mesenchymal interaction is essential for the formation of the liver bud as well as for further expansion and differentiation. During this phase (E9.5 to E15), the liver bud undergoes a tremendous amount of growth and also becomes an important site for hematopoiesis. Factors and signals regulating this critical stage arise from both the hepatic mesenchyme and the STM. These signals include Fgf and Bmp, which promote liver growth in addition to their role in hepatic specification. Bmp4 is strongly expressed in the STM and continues to be expressed at E9.0, during which the liver bud migrates into the STM (Rossi et al, 2001). BMP4-null mutant embryos were developed to assess for a role in liver organogenesis. These null-mutant mice had a delay in the growth of the liver bud, which would indicate that Bmp constitutes an important growth signal for the liver (Rossi et al, 2001).

Another factor that has been implicated in liver growth is HLX, which is expressed prominently in the visceral mesenchyme (STM), into which the liver will expand (Hentsch et al, 1996). HLX-null embryos exhibited severe liver hypoplasia without affecting liver specification (Hentsch et al, 1996). A similar finding was observed in embryos that were null mutant for hepatocyte growth factor (Hgf) (Schmidt et al, 1995). In contrast to HLX−/− embryos, apoptosis was the underlying cause for the severe reduction in liver size. Hgf is produced by the cells lining the sinusoids, which are of mesenchymal origin, mediating its effects through the c-Met tyrosine kinase receptor produced by hepatocytes (Schmidt, 1995). TGF-β signaling is also involved in liver outgrowth. It mediates its signals through SMAD2 and SMAD3, which after phosphorylation translocate to the nucleus to either upregulate or downregulate gene expression. Embryos that were heterozygous mutants for both SMAD2 and SMAD3 exhibited severe liver hypoplasia as a result of a decrease in β1-integrin expression (Weinstein et al, 2001). Intriguingly, this phenotype was rescued when HGF, a potent hepatotrophic growth factor, was added to the culture medium. Presumably, this was a result of the restoration of β1-integrin expression (Kagami et al, 1996), which plays an important role in hepatocytic adhesion to the extracellular matrix; TGF-β and HGF are known to induce the gene (Kagami et al, 1996; Kawakami-Kimura et al, 1997).

Biliary Epithelial Cell Differentiation and Formation of the Ductal Plate

The exact origin of the biliary epithelial cells has been greatly debated; however, the popular school of thought is that they are derived from bipotential hepatoblasts that can differentiate into either hepatocytes or biliary epithelial cells. This theory is based on the observation that immature hepatoblasts coexpress markers of both hepatocytes (ALB) and biliary epithelial cells (KRT19). The biliary-specific marker KRT19 becomes strongly expressed at a later gestational age, as the cells become ductal cells, whereas other cells transiently express the hepatocyte markers ALB and AFP as they develop into mature hepatocytes (Shiojiri et al, 1991). This theory was further supported when embryonic liver was transplanted into the testes of syngeneic animals, before the emergence of intrahepatic biliary ducts, as it subsequently gave rise to both hepatocytes and typical bile ducts (Shiojiri et al, 1991). However, this finding does not exclude the possibility that ductal cells may be derived from other embryonic cell types. Suzuki and colleagues (Suzuki et al, 2002) used in vitro studies on hepatic “stem” cells from E13 embryonic livers identified as having self-renewing capability and multilineage differentiation potential, to show that these cells could form differentiated hepatocytes, biliary epithelial cells, pancreatic, and intestinal cells.

The first step in triggering the initiation of the transition from a hepatoblast to a biliary epithelial cell is thought to be facilitated through the ONECUT transcription factor hepatocyte nuclear factor 6 (HNF6), which is expressed in the biliary epithelial cells of the developing intrahepatic biliary ducts and in hepatoblasts, gallbladder primordium, and the extrahepatic bile ducts (Landry et al, 1997; Rausa et al, 1997). HNF6−/− embryos displayed severe biliary anomalies that resulted in an absent gallbladder and the extrahepatic bile duct network being replaced with an enlarged structure connecting the liver to the duodenum. In the intrahepatic biliary ducts, however, it caused abnormal differentiation of biliary epithelial cells, resulting in cholestasis. Closer histologic examination revealed an increased number of KRT19-positive cells compared with controls at E13.5, with the development of abnormal large cysts at E15.5 to E16.5 that contained an epithelium of KRT19-expressing cells. These abnormal cysts are similar to those seen in Caroli disease, a disorder with ductal plate malformation and ectasias. Although HNF6 might play a role in the disease, not all of the malformations seen in HNF6−/− mice are manifested in Caroli disease, such as the absence of the gallbladder. The abnormal increase in KRT19-positive cells at E13.5 lacked any proliferative marker, suggesting that they are postmitotic, resulting from an increased number of hepatoblasts that have differentiated toward a biliary lineage prematurely (Clotman et al, 2002). In addition, the excess KRT19-positive biliary epithelial cells formed cordlike extensions within the liver parenchyma, rather than being restricted to the vicinity of the portal vein, as seen in the controls. This observation supports the role of HNF6 in controlling the differentiation of hepatoblasts into biliary epithelial cells and the morphogenesis of the intrahepatic biliary ducts, confining biliary epithelial cells to the periportal area. A similar morphologic defect of intrahepatic biliary ducts was observed in HNF1B−/− embryos. These findings suggest that HNF6 controls intrahepatic biliary duct development via HNF1B (Clotman et al, 2002).

With regard to mesenchymal–epithelial induction of liver primordium and gallbladder, studies have suggested that the mesenchyme contributes to biliary epithelial cell differentiation, where differentiation of hepatoblasts into biliary epithelial cells was stimulated when cocultured with hepatic or lung mesenchyme (Shiojiri & Koike, 1997). It was also noted in the HNF6−/− study that biliary epithelial cell differentiation occurred at the interface between the portal mesenchyme and the liver parenchyma (Clotman et al, 2002). A recent study revealed that the forkhead box f1 (FOXF1) transcription factor may play an important role in mesenchymal–epithelial signaling, an interface required for the development of organs derived from foregut endoderm, such as the pancreas, liver, gallbladder, and lung (Kalinichenko et al, 2002). FOXF1 expression is restricted to the gallbladder mesenchyme and STM. In FOXF1+/−, the gallbladder develops severe structural abnormalities with significant reduction in size and reduced mesenchymal cell numbers, an absent biliary epithelial cell layer, and a deficient external smooth muscle layer. The reduction in mesenchymal cell numbers was attributed to the reduction in the vascular cell-adhesion molecule (VCAM1) and cell-adhesion α5 integrin (ITGA5) genes, both of which are essential for mesodermal formation (Mahlapuu et al, 2001; Yang et al, 1993). Defective smooth muscle layer formation was attributed to diminished levels of platelet-derived growth factor receptor α (PDGFRA), which is essential for smooth muscle cell differentiation (Jacob et al, 1994). FOXF1 is not expressed in intrahepatic biliary duct mesenchyme in wild-type mice, and no defects were seen in the intrahepatic biliary ducts of FOXF1+/− mice; however, FOXF1 mRNA levels in the liver were increased, suggesting that it may be a compensatory mechanism to prevent defects in the liver (Kalinichenko et al, 2002). All these data suggest that FOXF1 is crucial for the development of the extrahepatic bile ducts and gallbladder, with the mesenchyme playing an essential role in biliary epithelial cell differentiation. The interaction between the biliary epithelial cells and the extracellular matrix is also thought to contribute to biliary epithelial cell differentiation.

Integrins are membrane receptors for extracellular matrix proteins that play an important role in mediating the interaction between differentiating cells and the extracellular matrix (Hynes, 1992; Couvelard et al, 1998). Hepatoblasts express integrin heterodimers containing the β1 subunit (α1β1, α5β1, α6β1, and α9β1), and when hepatoblasts differentiate into immature intrahepatic biliary epithelial cells when in contact with the mesenchyme, the morphology of the integrins changes. The primitive intrahepatic biliary epithelial cells upregulate α6β1 expression and lose α1β1, while acquiring several other integrin dimers not previously expressed on hepatoblasts, including α2β1, α3β1, α5β1 and α6β4 (Couvelard et al, 1998). This change is likely due to the different pericellular environment. Intrahepatic biliary epithelial cells are contacted by a basement membrane containing collagen, entactin, and laminin (Desmet, 1985); however, hepatocytes are surrounded by the perisinusoidal matrix, which is devoid of laminin and entactin (Schuppan, 1990). The increase in α6β1 expression—along with the acquisition of biliary-specific expression of α2β1, α3β1, and α6β4, which are integrin receptors for laminin—correlated with the deposition of laminin at the contact points of the portal mesenchyme with the ductal plate (Couvelard et al, 1998).

Remodeling of the Ductal Plate

As mentioned earlier, the double-layered ductal plate around the portal vein begins to form focal dilations incorporated into the portal mesenchyme to form intrahepatic biliary ducts, while the parts of the ductal bilayer plate not involved in the formation of the ducts progressively regresses. This mechanism of regression is thought to be carried out by apoptosis (Terada & Nakanuma, 1995; Sergi et al, 2000). Cell-matrix interactions are also thought to contribute to the remodeling process. Tenascin was found to be expressed in the mesenchyme around the biliary epithelial cells of primitive ducts migrating into the mesenchyme. In contrast, tenascin was absent in the mesenchyme of peripheral ducts (Terada & Nakanuma, 1994).

Tubulogenesis of ductal cells is thought to be controlled by soluble factors secreted from hepatocytes or biliary epithelial cells. When coculturing human biliary epithelial cells with hepatocytes, a marked ductular morphogenic response was induced, and the biliary epithelial cells formed well-organized luminal ducts. This result was reproduced when biliary epithelial cells were grown in a conditioned medium from previous hepatocyte and biliary epithelial cell coculturing (Auth et al, 2001); however, it remains to be seen whether soluble factors do contribute to tubulogenesis in in vivo studies (Fig. 1A.5).

FIGURE 1A.5 Overview of intrahepatic bile duct formation around the portal vein. Hepatoblasts in the portal mesenchyme vicinity begin to acquire the biliary epithelial cell marker (KRT19) and downregulate hepatic genes, first forming a single layer at E14 then a bilayer at E18 with focal dilations that eventually form the intrahepatic bile ducts. The rest of the bilayer regresses.

(Image courtesy Mariam Hobeldin.)

Developmental Relationship Between the Ducts, Vessels, and Mesenchyme of the Portal Tract

A functional relationship appears to exist between the contents of the portal tract (bile duct, hepatic artery, portal vein) based on observations of a number of human diseases termed ductal plate malformations. These diseases include biliary atresia, Caroli disease, and Meckel and Alagille syndromes, in which abnormal biliary ducts are associated with anomalies of the portal mesenchyme and of the portal blood vessels (Lemaigre, 2003). This association was also demonstrated in studies on Notch pathway defects. In Alagille syndrome, an autosomal dominant disease, bile ducts are absent in the portal tract associated with an increased number of arteries and fibrosis. Haploinsufficiency of JAG1, a Notch receptor ligand, is associated with Alagille syndrome, where JAG1 is persistently expressed in the ductal epithelium in humans (Li et al, 1997; Louis et al, 1999). It is also expressed in the endothelial cells of the developing portal vasculature (Crosnier et al, 2000). The animal model for Alagille syndrome was replicated in double-heterozygous mice for mutations in the JAG1 and NOTCH2 genes (J1N2+/−). JAG1 protein was expressed in the hepatic vasculature, and NOTCH2 was expressed in a subset of hepatoblasts surrounding the portal vein, hepatic artery, and bile ducts. Interestingly, although neither JAG1 nor NOTCH2 protein was expressed in the bile duct epithelium of these mice, JAG1 is expressed in bile duct epithelium in humans (Louis et al, 1999; McCright et al, 2002). The differences in JAG1 expression between human and mouse most likely reflects species specificity rather than technical artifacts, as mice that are heterozygous or homozygous for the JAG1 gene did not show the liver symptoms of Alagille syndrome (Xue et al, 1999). Similar biliary abnormalities were observed in NOTCH2-null mutant mice (McCright et al, 2002).

Hepatocyte Differentiation

During later stages of development, hepatocytes undergo a transition period from a hematopoietic support role to a mature adult hepatocyte. This change occurs under the control of the transcription factor CEBPα and HNF4, the latter being a crucial factor in hepatocyte differentiation. Loss of HNF4 function led to disruption of the expression of several genes associated with the mature hepatocyte phenotype. In HNF4−/− mice, hepatocytes failed to express many mature hepatic enzymes and lacked normal morphology, leading to low glycogen stores, disrupted sinusoids, and gap junction disruption. These results demonstrate that HNF4 is essential for liver differentiation.

Other factors that have been shown to promote hepatocyte differentiation include oncostatin M (OSM), an interleukin-6 family cytokine, Hgf, and Wnt (Michalopoulos et al, 2003; Tan et al, 2008). Although as discussed earlier in the chapter, the repression of Wnt signaling in the foregut endoderm is necessary for hepatic specification, its role is reversed at later gestational stages to promoting hepatocyte differentiation. β-Catenin, a central component of the canonical Wnt pathway, is essential for normal development; its aberrant activation in liver was associated with tumors, including hepatic adenomas and hepatocellular cancers (Peifer & Polakis, 2000; de La Coste et al, 1998; Zucman-Rossi et al, 2006). Livers that were deficient in β-catenin displayed decreased numbers of hepatocytes with hepatoblasts that lacked maturation, proliferation, and function (Tan et al, 2008). In addition, CEBPα, a fundamental regulator of hepatocyte differation and maturation (Tan et al, 2008), was decreased. Also noted in these livers was a complete absence of CK-19-positive intrahepatic biliary ducts, suggesting that β-catenin may play a role in biliary differentiation (Tan et al, 2008). In vitro studies showed that OSM, produced by hematopoietic cells in midfetal liver, and Hgf induce hepatic differentiation in the presence of dexamethasone, both working through different pathways (Kamiya et al, 2001). It was demonstrated that embryonic livers at E14.5 expressed G6Pase, tyrosine amino transferase (TAT), and accumulated glycogen, all signs of a mature and differentiated liver when cultured with Hgf or Osm. This finding was further supported when TAT levels, as well as glycogen storage, was significantly reduced in livers derived from mice null mutant for gp130, the common receptor subunit of the IL-6 family cytokines (Kamiya et al, 2001). CEBPα is also critical for the acquisition and maintenance of hepatocyte differention, with knockout mice exhibiting defects in liver growth and architecture as well as an increase in cell proliferation (Flodby et al, 1996). CEBPα is also thought to be a key factor in controlling the switch in the differentiation of bipotential hepatoblasts to become either biliary epithelial cells or hepatocytes. CEBPα starts to be expressed in the endodermal liver primordium at E9.5, and its expression in the nuclei of hepatoblasts and hepatocytes becomes stronger with development. During biliary cell differentiation, CEBPα expression was suppressed in periportal biliary cell progenitors, suggesting that its suppression may be a prerequisite to biliary cell differentiation from hepatoblasts (Shiojiri et al, 2004). Lineage tracing experiments have recently shown that Sox9+ biliary epithelial cells themselves differentiate into mature hepatocytes under normal physiologic conditions (Furuyama et al, 2011). Hepatocyte differentiation from the lineage tagged Sox9+ bile duct cells increased during the hepatic regenerative process, suggesting that the biliary and pancreatic ductal tree (Sox9-expressing domains) contains a previously unappreciated pool of progenitor cells.

Basic Pancreatic Embryology

During early development, the pancreas is formed by the dorsal and ventral bud, each originating from the endodermal lining of the caudal part of the primitive foregut tube after regional specification with transcription factor Pdx1 followed by Ptf1a (Sadler & Langman, 2006; Grapin-Botton, 2005; Moore-Scott et al, 2007). The first morphologic evidence of pancreas development is in the form of mesenchymal condensation overlying the dorsal aspect of the primitive foregut tube at the level of the duodenal anlagen, distal to the stomach. Shortly after, at around E9.5 gestation in mice and day 26 of gestation in humans, the dorsal bud begins to evaginate into the overlying mesenchyme (Kallman & Grobstein, 1964; Picte et al, 1972). Approximately 12 hours later in mice, and 6 days after dorsal bud evagination in humans, the ventral bud begins to arise.

The morphologic development of the ventral pancreatic bud is similar to the dorsal bud; however, it is markedly different at the molecular level. After this initial bud evagination, the pancreatic bud then undergoes elongation of the stalk region, a precursor to the main pancreatic duct and branching morphogenesis of the apical region of the bud. Unlike growth patterns seen in the developing lung, kidney, and salivary glands, in which the branching morphogenesis undergoes a typical 90-degree pattern, the pancreas grows in an acute-angled branching pattern, which can lead to the exclusion, or “squeezing out,” of mesenchyme from between the closely apposed branches of epithelium. This exclusion of mesenchyme may influence epithelial–mesenchymal interactions and lineage selection (Fig. 1A.6). The ventral pancreatic bud, which arises from the base of the hepatic diverticulum, rotates dorsally and eventually fuses with the dorsal pancreatic bud (6 to 7 weeks gestation in humans, E12 to E13 in mice), contributing to the formation of the uncinate process and inferior part of the head of the pancreas, with the rest of the pancreas arising from the dorsal pancreatic bud. The entire ventral pancreatic duct and the distal part of the dorsal pancreatic duct fuse together to form the main pancreatic duct of Wirsung. The remaining proximal part of the dorsal pancreatic duct is either obliterated or persists as a small accessory pancreatic duct, the duct of Santorini (Sadler & Langman, 2006).

FIGURE 1A.6 Pancreatic branching morphogenesis is different than in other organ systems—such as lungs, kidneys, and salivary glands—which branch in a 90-degree pattern to be fully exposed to the mesenchyme. However, the pancreas has an acute branching pattern; thus some areas are excluded from mesenchymal contact. Mesenchyme contains factors that regulate pancreatic growth and differentiation.

(Image courtesy Mariam Hobeldin.)

During development the pancreas undergoes a major amplification of the endocrine cell population through two waves of differentiation: an early primary transition (pre-E13.5 in mice) followed by the secondary transition (E13.5 to E16.5 in mice) (Picte et al, 1972; Jensen, 2004). In a similar gestational window, the exocrine pancreatic precursors undergo rapid branching morphogenesis and acinar cell differentiation (see Fig. 1A.6).

Endodermal Patterning of the Pancreas

As mentioned earlier, the pancreas develops through epithelial–mesenchymal interactions. Prior to this, the dorsal foregut became competent to receive propancreatic signals from the notochord (FGFs) (Wells & Melton, 2000) and from the anterior foregut (retinoic acids and BMPs) (Stafford & Prince, 2002; Tiso et al, 2002). The prepancreatic region of the definitive foregut comes into contact with the notochord, whereas the ventral pancreas contacts the lateral plate mesoderm (Kumar et al, 2003).

Dorsal and Ventral Pancreatic Bud Development

The notochord is key to the development of the dorsal pancreas, where it remains in contact with the dorsal prepancreatic endoderm until the paired dorsal aortae fuse in the midline (∼E8.0). The dorsal pancreas failed to form when the notochord was removed from early chick embryos in vitro (Kim et al, 1997), but no effect was seen on ventral pancreas development, further emphasizing the different molecular controls for dorsal versus ventral pancreas development. The ventral pancreas develops under the control of signals from the overlying cardiogenic mesenchyme, which also produces prohepatic signals (FGFs) to induce liver formation. Lack of prohepatic FGF signaling in regions of the cardiogenic mesenchyme will lead to the endoderm by “default” differentiating into ventral pancreas (Deutsch et al, 2001). Sonic Hedgehog (SHH) in the prospective pancreatic endoderm is suppressed by the notochord, which is necessary for dorsal pancreatic development, and SHH is otherwise expressed along the entire primitive gut tube epithelium. When the notochord was deleted in chick embryo cultures, ectopic SHH was seen in the pancreatic region of the foregut endoderm, and the pancreas failed to develop (Hebrok et al, 1998).

Pancreatic Mesenchyme

The pancreatic epithelium becomes enveloped by mesenchyme, which contains important factors that are pivotal for pancreatic morphogenesis. The mesenchyme has a general proexocrine effect on the pancreas through cell contact and through proendocrine effect, mediated by diffusible factors secreted from the mesenchyme (Li et al, 2004). Mesenchymal contact with the epithelium both enhances notch signaling (HES1), which favors the acinar lineage and also inhibits neurogenin 3 expression (NEUROG3), leading to the suppression of endocrine differentiation (Duvillie et al, 2006). The “default” differentiation of the pancreatic epithelium in the absence of mesenchyme is endocrine (Gittes et al, 1996). Thus the mesenchyme has both an inductive effect on the exocrine tissue development and a repressive effect on the development of endocrine cells.

The interaction between the mesenchyme and the pancreatic epithelium is regulated by several factors, the most important of which appears to be the FGFs, which are expressed in several epithelial–mesenchymal interface regions of the pancreas and are particularly important in regulating branching morphogenesis (Hogan, 1999). FGFs 1, 7, and 10 mediate their effect through FGF receptor 2B (FGFR2B) to induce epithelial proliferation, favoring exocrine differentiation (Elghazi et al, 2002; Celli et al, 1998; Dichmann et al, 2003). Null mutation for FGF10 or FGFR2B leads to blunting of branching pancreatic morphogenesis with a lack of proliferation of endocrine progenitor cells, indicating that FGF10 induces proliferation of epithelial cells and prevents endocrine differentation (Bhushan et al, 2001; Pulkkinen et al, 2003). These roles for FGFs are primarily for ongoing development of the buds, in contrast to the role of FGF from the cardiogenic mesenchyme, which inhibits ventral pancreatic bud formation in favor of liver development (Deutsch et al, 2001).

TGF-β Signaling

The TGF-β superfamily has been implicated in early pancreatic development. TGF-β isoforms (TGF-β1, -β2, and -β3) are localized to the pancreatic embryonic epithelium as early as E12.5 and progressively become focused to the acinar cells (Crisera et al, 1999). The ligands mediate their actions through TGF-β receptor type II (TBR-II), a heterodimer receptor; however, in late gestation it localizes to the pancreatic ducts (Tulachan et al, 2007). TBR-II inhibition has been shown to recruit an inappropriately high number of ductal cells to become endocrine progenitor cells, suggesting that TBR-II signaling is important for regulating the flow of endocrine progenitors out of ducts (Sanvito et al, 1995; Lee et al, 1995).

Activins, part of the TGF-β superfamily, also play a role in pancreatic development. When exogenous activin was added to embryonic pancreas explant cultures, it inhibited branching morphogenesis, a process associated with ductal and acinar differentiation (Ritvos et al, 1995). This observation was further supported when follistatin, an activin inhibitor expressed in the pancreatic mesenchyme by E12.5, was added to the pancreatic epithelium; it mirrored the mesenchymal effect, promoting exocrine and repressing endocrine development (Miralles et al, 1998). These observations would support a role for activin in promoting and regulating endocrine differentiation and suppressing exocrine differentiation in the early pancreas. Along similar lines, transgenic mice that express a dominant-negative activin receptor display reduced levels of differentiated islet cells (Kim et al, 2000; Shiozaki et al, 1999; Yamaoka et al, 1998). In vitro in cell lines, activin A favors β-cell differentiation.

SMADs are intracellular signals for the TGF superfamily and have gained interest among developmental biologists, since SMAD4 was specifically identified to be mutated in 50% of pancreatic cancers (Hahn et al, 1996). However, a key role for SMAD4 has not yet been identified in pancreatic development (Simeone et al, 2006; Bardeesy et al, 2006). SMAD7 on the other hand, which acts as a general TGF signaling inhibitor when expressed under the PDX1 promoter, results in a dramatic decrease in β cells at birth—by around 90%—and replacement with glucagon cells (Smart et al, 2006).

Notch Signaling

A key step in pancreatic development is the lineage decision by pancreatic progenitor cells between the endocrine and exocrine lineage. Notch signaling has been identified as a master regulator of this fate decision switch (Apelqvist et al, 1999). However, the exact mechanism for Notch signaling in pancreatic lineage selection remains elusive, and ambiguity still surrounds the exact role of Notch signaling through pancreatic development. It appears that the downstream factors of Notch targets are variably expressed at certain times in development of the pancreas. The early pattern of downstream molecules is a key mediator of Notch signaling in determining endocrine versus exocrine differentiation of progenitor cells and in maintaining a stem cell state in some progenitor cells. Later in gestation, the downstream molecules for Notch signaling change to foster cell maturation and differentiation beyond this initial lineage decision between endocrine and exocrine differentiation.

Hedgehog Signaling

As mentioned earlier, Shh suppression in the prospective pancreatic endoderm is a prerequisite for pancreas formation, however, Hedgehog signaling in pancreatic development appears to be complex. SHH-null mutant mice do not exhibit an expanded pancreatic field; in contrast, Indian Hedgehog (IHH) null mutant mice that develop with a small pancreas indicate a propancreatic role for IHH (Hebrok et al, 2000). However, when combining SHH-null mutation with heterozygous mutation for IHH, an annular pancreas develops. Mice that are null mutant for the Hedgehog Patched receptor Ptc lack Pdx1 and glucagon expression at E9.5, supporting the need to suppress Shh for early pancreatic development (Hebrok et al, 2000). Similarly, transgenic overexpression of Shh or Ihh under the Pax4 promoter led to a significant decrease in endocrine and exocrine tissue mass (Smith et al, 2000).

WNT

As mentioned earlier in the chapter, β-catenin repression in the anterior endoderm is necessary to initiate liver and pancreas development. Consistent with this, PDX1–WNT1 and PDX1–WNT5A transgenic mice developed pancreatic agenesis and severe hypoplasia, respectively, confirming that early Wnt signaling is detrimental for pancreatic development (Heller et al, 2002). Similar to Notch signaling, the role of Wnt signaling beyond endodermal patterning is complex and depends on the time and place of Wnt signaling. β-catenin, which is mostly expressed in the mesenchyme (Heller et al, 2002), when ablated from Pdx1-positive cells resulted in the loss of acinar tissue at birth, confirming earlier studies that exocrine tissue development is mesenchyme dependent (Murtaugh et al, 2005; Wells et al, 2007). Others found a role for Wnt signaling in promoting postnatal pancreatic growth (Heiser et al, 2006), illustrating the complex and multiple roles that Wnts play in pancreatic development.

Endothelial Cells

Recently, an important role for endothelial cells in pancreatic endocrine development has been demonstrated. As mentioned earlier, the notochord separates from the gut endoderm by the fusion of the dorsal aortas around E8.0. Lammert and colleagues (2001) demonstrated that the removal of the dorsal aorta from Xenopus embryos led to the absence of pancreatic endocrine development. In addition, aortic endothelial cells were able to induce pancreatic bud formation and subsequent insulin cell differentiation in the adjacent endoderm. Importantly, ectopic VEGFA in the intestine and stomach led to ectopic formation of islets in the stomach (Lammert et al, 2001). Interestingly, it appears that ventral pancreas development seems to be not dependent on the endothelium despite its close proximity to the vitelline veins (Yoshitomi & Zaret, 2004).

Glucagon

The first detectable endocrine cells during pancreatic development are the glucagon-containing cells at around E9 in the mouse, and recent studies have shown that glucagon signaling is necessary for early differentiation of insulin-expressing cells, which appear around E13.0 (Picte et al, 1972; Prasadan et al, 2002). Glucagon hormone is generated from proglucagon by the action of prohormone convertase 2 (PC2). When either PC2 or the glucagon receptor is knocked out, mutant mice are generated without glucagon, displaying a delay in islet cell differentiation and maturation but still retaining the large amplification of insulin-positive cells (“secondary wave”) later in gestation (Vincent et al, 2003). Exogenous addition of exendin-4, a glucagon-like peptide-1 (GLP-1) analog, was able to rescue the delay in early insulin differentiation (Prasadan et al, 2002). Exendin-4 was also shown also to convert AR42J cells (Zhou et al, 1999; Yew et al, 2004) and ARIP cells (Hui et al, 2001) into insulin-expressing cells. Together these studies strongly support the role of glucagon signaling, through its receptor, in initiating early insulin differentiation.

Extracellular Matrix

The pancreatic epithelium is contained within a continuous sheath of basement membrane, which constitutes the epithelial-mesenchymal interface and plays an important role in guiding pancreatic development (Hisaoka et al, 1993). Basement membrane has also been shown to play an important role in regulating branching morphogenesis in many other organs. Laminin-1 was found to induce duct formation in isolated E11 mouse pancreatic epithelium (Gittes et al, 1996) and to mediate the proexocrine inductive effect of the mesenchyme and the pro–β-cell role later in gestation (Li et al, 2004; Jiang et al, 2001). Cadherins—calcium-dependent adhesion molecules, critical for cell–cell adhesions—play an essential role in migration and differentiation of pancreatic endocrine progenitor cells. E-cadherin and R-cadherin are initially localized to the ducts, but they become downregulated as cells move out of the ducts to form islets (Dahl et al, 1996; Sjodin et al, 1995). N-CAM, another cell-adhesion molecule expressed in mature α and PP cells (Cirulli et al, 1994), when knocked out led to improper aggregation of endocrine cells within the islet (Esni et al, 1999). Thus, cell-adhesion molecules regulate the adhesive properties of endocrine cells, and these adhesive properties are essential for endocrine cell aggregation into islets.

Transcription Factors

The role of transcription factors has been extensively studied in pancreatic development. The most heavily studied is Pdx1, an early marker for pancreatic progenitors that later restricts to the β cells. It is expressed in the prepancreatic region of the primitive foregut tube at E8.5 then expands to be expressed in distal stomach, common bile duct, and duodenum by E10.5 to 11.5 (Offield et al 1996; Guz et al, 1995). Pdx1 is initially expressed throughout the epithelium; however, its expression becomes suppressed in cells as they commit to the endocrine (Jensen et al, 2000) lineage or ducts (Gu et al, 2002). Its expression reappears as cells differentiate to the insulin-positive β-cell lineage. PDX1-null mutant mice and humans developed pancreatic agenesis (Jonsson et al, 1994; Yee et al, 2001; Stoffers et al, 1997a; Stoffers et al, 1997b). Delayed Pdx1 inhibition using a tetracycline-regulatable transgenic knock-in system demonstrated severe blunting of pancreatic development with complete absence of acini and β cells (Holland et al, 2002), indicating that Pdx1 continues to play a role in pancreas development beyond regional specification during endodermal patterning.

SOX9

SOX9 acts as a marker for pancreatic progenitor cells that can give rise to all pancreatic cell types, and it is necessary for maintaining cells in a progenitor state (Akiyama et al, 2005). SOX9 mutants display pancreatic hypoplasia as a result of depletion of the pancreatic progenitor pool (Seymour et al, 2007). It has also been suggested that SOX9 mediates HNF6 and the control of NEUROG3-cell formation, because conditional deletion of SOX9 led to the formation of cystic structures similar to those seen in HNF6−/− embryos, associated with absent pancreatic endocrine cells (Seymour et al, 2007). Furthermore, it was found that SOX9 is transiently coexpressed with NEUROG3 and frequently colocalizes with HES1, which acts as a precursor to NEUROG3-positive cells. It was suggested that SOX9 helps mediate the transition of endocrine progenitors from the nonendocrine HES1-positive, NEUROG3-negative state to an HES1-negative, NEUROG3-positive endocrine-committed state (Seymour et al, 2007). Thus, SOX9 acts as a key mediator of the commitment of NEUROG3-positive endocrine progenitors (Fig. 1A.7).

FIGURE 1A.7 Overview of the pancreatic endocrine and exocrine cell lineage.