PMCCPMCCPMCC

Search tips
Search criteria 

Advanced

 
Logo of nihpaAbout Author manuscriptsSubmit a manuscriptHHS Public Access; Author Manuscript; Accepted for publication in peer reviewed journal;
 
Dev Cell. Author manuscript; available in PMC 2010 July 1.
Published in final edited form as:
PMCID: PMC2734336
NIHMSID: NIHMS119897

Sox17 regulates organ lineage segregation of ventral foregut progenitor cells

SUMMARY

The ventral pancreas, biliary system and liver arise from the posterior ventral foregut, but the cell-intrinsic pathway by which these organ lineages are separated is not known. Here we show that the extrahepatobiliary system shares a common origin with the ventral pancreas and not the liver, as previously thought. These pancreatobiliary progenitor cells coexpress the transcription factors Pdx1 and Sox17 at e8.5 and their segregation into a Pdx1+ ventral pancreas and a Sox17+ biliary primordium is Sox17-dependant. Deletion of Sox17 at e8.5 results in the loss of biliary structures and ectopic pancreatic tissue in the liver bud and common duct, while Sox17 overexpression suppresses pancreas development and promotes ectopic biliary-like tissue throughout the Pdx1+ domain. Restricting Sox17+ biliary progenitor cells to the ventral region of the gut requires the notch effector Hes1. Our results highlight the role of Sox17 and Hes1 in patterning and morphogenetic segregation of ventral foregut lineages.

Keywords: endoderm, Pdx1, pancreas, biliary, extrahepatic, liver, Hes1, stem cell

INTRODUCTION

The ventral foregut endoderm of the mammalian embryo is a simple epithelium of several hundred cells, yet it gives rise to a spectrum of organs and structures including the lungs, thyroid, liver and hepatobiliary system (intrahepatic bile ducts (IHBD), extrahepatic bile ducts (EHBD), common duct, gall bladder, cystic duct) and ventral portions of the pancreas, stomach and duodenum. Studies indicate that the most anterior part of the ventral foregut gives rise to thyroid and lung whereas the posterior region give rise to the liver, biliary system and the ventral pancreas (Fukuda and Mizuno, 1978; Le Douarin, 1968; Le Douarin, 1988; Rosenquist, 1971; Tremblay and Zaret, 2005). Several studies have demonstrated how signaling cascades and transcription factors influence foregut organ specification (see below). However, the exact lineage relationship between the liver, extraheptobiliary system and the ventral pancreas is not clear nor have the cell intrinsic factors been identified that mediate separation of these lineages from a common group of foregut progenitor cells.

Studies using embryonic explants, lineage tracing, gene expression analyses and mouse genetics indicate that several different signaling pathways and transcription factors are involved in the specification and development of foregut organs. Foregut explant cultures and genetics have identified that mesoderm-derived signals, including FGF and BMP ligands, act to pattern the ventral foregut into liver, pancreas and lung (Deutsch et al., 2001; Dong et al., 2007; Gualdi et al., 1996; Jacquemin et al., 2006; Serls et al., 2005; Shin et al., 2007) and in the separation of the liver and pancreatic lineages (Chung et al., 2008). Other pathways implicated in development of hepatic, pancreatic and biliary systems include the FGF10 and Hedgehog signaling pathways (Bhushan et al., 2001; Dong et al., 2007; Manfroid et al., 2007) (Hebrok et al., 2000; Kim and Melton, 1998; Litingtung et al., 1998). A large body of work has detailed many transcription factors that are required for specification of individual organs, including the liver (Foxa2, Foxa1 and Hhex) and pancreas (Ptf1a/p48, Pdx1, Hhex) (Bort et al., 2004; Bort et al., 2006; Kawaguchi et al., 2002; Krapp et al., 1998; Lee et al., 2005; Offield et al., 1996). Despite these significant advances, the cell-intrinsic mechanisms by which this common pool of foregut progenitor cells gives rise to distinct organ lineages, and how organ boundaries are subsequently established and maintained is less clear.

The biliary system is derived from the same region of the ventral foregut as the liver and pancreas, yet relatively little is known about early biliary development. Based on histological evidence, it has been suggested that the hepatic diverticulum gives rise to the liver, as well as the IHBD, EHBD, and collecting duct and gall bladder primordium. Knockout experiments in the mouse have identified several transcription factors, Hhex, HNF6 and HNF1β that are required for biliary development (Clotman et al., 2002; Coffinier et al., 2002; Hunter et al., 2007). Several of these also affect liver development, again suggesting the liver and biliary system share a common origin. The ventral pancreas also arises from the same region of the ventral foregut as the hepatic and biliary system and loss of the Notch effector Hes1 results in gall bladder agenesis and ectopic pancreatic tissue in the common duct. This finding suggests a developmental link between the biliary system and the pancreas (Burke et al., 2004; Fukuda et al., 2006; Sumazaki et al., 2004). This possibility is further supported by reports of congenital defects in humans in which linked anomalies affect both the pancreas and biliary system (Ashraf et al., 2005; Chappell et al., 2008; Galan-Gomez et al., 2007; Heij and Niessen, 1987; Iguchi et al., 2005; Martinoli et al., 1980; Mehes et al., 1976; Mitchell et al., 2004). Despite all of this correlative evidence, the lineage relationship between the biliary system and either the liver or the pancreas has not been experimentally demonstrated.

Sox17 is necessary for definitive endoderm formation in many vertebrate species including Xenopus, zebrafish and mouse. However, early embryonic lethality in Sox17-null mice has made studying its role during organogenesis impossible (Kanai-Azuma et al., 2002). In Xenopus, Sox17 interacts with co-factors to activate transcription of target genes, such as HNF1β, that are known to be required for later development of endodermal organs (Sinner et al., 2006; Sinner et al., 2004; Zorn et al., 1999). Moreover, our data indicated that Sox17 expression is maintained in the posterior foregut, suggesting to us that Sox17 might have a function during ventral foregut organogenesis.

We have used a combination of gene expression analysis, lineage tracing and mouse genetics to investigate the lineage relationship between the biliary system, pancreas and liver and to identify cell-intrinsic pathways that separate these organ lineages from a common progenitor. We have identified that: 1) The biliary system and ventral pancreas arise from a common progenitor that is distinct from the liver at e8.5. 2) Sox17 is a key regulator in the early segregation these organ lineages. 3) Loss of Sox17 results in Pdx1 expressing cells in the liver bud and ectopic pancreatic tissue in the common duct. 4) Sox17 is necessary and sufficient for specifying a ductal fate. 5) Hes1 is required for separating the biliary and pancreatic lineages and may act in a feedback loop with Sox17.

Together with published findings, our data supports a model where Sox17 is broadly expressed in foregut progenitor cells and is progressively downregulated in cells as they become lineage restricted. For example, down-regulation of Sox17 in Pdx1-expressing cells is required for pancreas development. This model illustrates how one transcription factor, Sox17, might act in combination with others factors like Hex1 and Pdx1 to specify different organ lineages from a common pool of progenitor cells.

RESULTS

Pdx1 and Sox17 coexpression defines a common ventral foregut progenitor

Sox17 is required for endoderm formation but its role in later endoderm development is not known. We evaluated the expression pattern of Sox17 mRNA and protein at e8.5 and found both in ventral foregut domain that largely overlaps with Pdx1 (Fig. 1A, Fig. S1A). This Sox17+/Pdx1+ domain is caudal to the thickening pseudostratified epithelium of the hepatic diverticulum, which is Hhex positive at this stage (Fig. 1A). Very few cells at the boundary of the two domains were Hhex+/Sox17+/Pdx1+ (Fig. 1A). Over the next 24 hours of development (e9.5) the Sox17 and Pdx1 co-expression domain progressively separates into a Sox17+ biliary primordium (also called primordial gall bladder) and a Pdx1+ pancreatic primordia (Fig. 1, Fig. S1). Analysis of Sox17 mRNA by in situ hybridization completely coincided with Sox17 protein expression (Fig. S1B,C). These expression patterns indicate a role for Sox17 development of the biliary system and pancreas.

Figure 1
Sox17 and Pdx1 are coexpressed in ventral foregut progenitor cells

The biliary primordium, extrahepatobiliary system and ventral pancreas develop from Pdx1+ cells

Our expression data led us to hypothesize that the biliary system does is not derived from the liver primordium as previously suggested, but rather shares a common origin with the ventral pancreas and comes from the Pdx1+/Sox17+ progenitor cells in the ventral foregut. To determine if that the biliary system is derived from a Pdx1+ progenitor we performed a lineage tracing experiment with a Pdx1-Cre transgenic line (Wells et al., 2007) and a Rosa reporter mouse. Pdx1-expressing cells were lineage labeled starting at e8.5 and analysis at e9.5 and e10.5 embryos demonstrated that the biliary primordium was LacZ-positive and thus were derived from Pdx1-expressing progenitor cells (Fig. 1C, E). Since the e10.5 biliary primordium is negative for Pdx1 protein, these cells must be a descendent of an earlier Pdx1-cre expressing cell. Analysis of e16.5 embryos indicated that the epithelia of the duodenum, dorsal and ventral pancreas, and the extrahepatobiliary system (comprised of the gall bladder, cystic duct, common duct and extrahepatic ducts) were either derived from earlier Pdx1-expressing cells (Fig. 1F) or from de novo Pdx1 expression (Offield et al., 1996). Although the extrahepatobiliary system is derived from Pdx1-expressing cells, our analysis of Pdx1 null embryos (Pdx1tTA/tTA) demonstrates that Pdx1 is not necessary for development of Sox17-expressing biliary primordium or subsequent development of the gall bladder, collecting duct, and common duct (Fig. 1G), which is consistent with previous reports (Jonsson et al., 1994; Offield et al., 1996). These data indicate that the biliary primordium and ventral pancreas are both derived from a Pdx1+ ventral foregut progenitor.

Sox17 is necessary for development of the biliary primordium

To investigate the function of Sox17 in the pancreatobiliary development, we deleted Sox17 in the ventral foregut starting at e8.5 using two different Cre lines (Pdx1-cre and FoxA3cre) (Lee et al., 2005a; Wells et al., 2007) and two different Sox17 alleles (a conditional Sox17Flox allele and a null Sox17GFP allele) (Kim et al., 2007) (Fig. S2). Deletion of Sox17 was efficient in e.9.5 embryos (FoxA3cre;Sox17Fl/Fl - LOF) and resulted in an expansion of Pdx1 expression throughout the ventral gut, which is the presumptive gall bladder region where Pdx1 is normally down regulated (Fig. 2A,B). At E10.5, control embryos had a distinct Pdx1+ ventral pancreas and Sox17+ biliary primordium, whereas the biliary primordium was absent in FoxA3cre;Sox17Fl/Fl embryos (Fig. 2C,D). Similar phenotypes were observed in E10 Pdx1-cre;Sox17Fl/GFP embryos (Fig. 2E and F). The loss of the biliary primordium in e9.5 Sox17-LOF embryos was not due to any quantitative change of cell proliferation (phosphohistone H3 staining) or cell death (caspase-3 staining) (n=4 different animals) (Fig. S4). These results demonstrate that Sox17 is required for development of the biliary primordium and suggest that biliary cells adopt a Pdx1+ pancreatic fate. In these experiments, Pdx1-cre;Sox17Fl/Fl embryos had a less severe phenotype than Pdx1-cre;Sox17GFP/Fl because the latter embryos already lack one functional allele of Sox17 (Sox17GFP) (Fig. S3).

Figure 2
Sox17 is required for development of the biliary primordium and for establishing and/or maintaining organ boundaries between the pancreas and liver

Sox17 is required for establishing distinct organ domains between the liver and pancreatobiliary system

As described above, Sox17-LOF embryos had Pdx1 expressed throughout the ventral gut endoderm (Fig. 2B, F), which is not due to loss of presumptive biliary cells through cell death or reduced proliferation (Figure S4). This suggests that Sox17 is required to restrict Pdx1 expression to the proper domain along the dorsal-ventral axis. We next investigated if Sox17 is also required for establishing or maintaining boundaries between the pancreatobiliary and liver domains. In control embryos between e9.0-9.5, there are discrete boundaries between the liver bud (Prox1+) and the pancreatobiliary domain (Pdx1+ and Sox17+). Pdx1 expression is almost never seen in the liver bud of control embryos. In contrast we observed ectopic Pdx1 expressing cells throughout the liver bud of E9.5 FoxA3cre;Sox17Fl/Fl embryos (Fig. 2G,H,G′,H′,I), and in Pdx1-cre;Sox17Fl/GFP and FoxA3cre;Sox17Fl/GFP embryos (not shown). Quantitation of the number of Pdx1+ cells in the Prox1+ liver bud showed that Sox17-LOF embryos had a 10-fold higher number of ectopic Pdx1+ cells within the liver bud (7.8 +/− 3 cells per liver bud section, p<0.05, n=5) relative to control embryos (0.7 +/− 0.7, n=3). This was confirmed by analysis of sagittal sections of Pdx1-cre;Sox17Fl/GFP E9.0-E9.5 that clearly showed ectopic Pdx1+ cells in the liver bud (Fig. S5A, B). When we analyzed embryos that were totally null for Sox17 (Sox17GFP/GFP), the Pdx1 domain was almost completely contained within the Prox1 domain (Fig. S5C, D). These data suggest that Sox17 is required for establishing and/or maintaining distinct domains between the liver and the developing pancreatobiliary system along the anterior posterior axis (Fig. 2J).

Embryos lacking Sox17 have biliary agenesis and ectopic pancreatic tissue in the ducts

E10.5 Sox17-LOF embryos had agenesis of the biliary promordium, and this manifested as a loss of gall bladder and cystic duct at E16.5 (Fig. 3A,B). Moreover 50% of the E16.5 FoxA3cre; Sox17Fl/Fl embryos had visible ectopic pancreatic tissue in the common duct, consistent with the presence of ectopic Pdx1 cells at e10 (Fig. 3A′,B′) (n=4). Hematoxylin and Eosine (H&E) and antibody staining of sections through the common duct showed pancreatic tissue budding from the common duct in Sox17-LOF embryos (Fig. 3C, D). We confirmed that this was not due to a plane of section by dissecting the common duct away from the pancreas and liver. Immunohistochemical analysis of the dissected common duct of E16.5 controls showed a single layered epithelium that was positive for the duct specific lectin, DBA and for Pdx1, which are normally expressed in the duct at this stage (Fig. 3C-C[triple prime]). In contrast, the marker expression profile of Sox17-LOF ducts was remarkably similar to that of the pancreas; branching amylase positive tissue and small islet-like clusters of insulin/glucagon positive cells budding off of the DBA positive duct (Fig. 3D-D[triple prime]) and reduced levels of Pdx1. Control ducts had rare insulin and glucagon positive cells as previously reported (Dutton et al., 2007) and never express amylase. The ectopic endocrine clusters in Sox17-LOF ducts were often negative for the duct markers HNF6 and DBA suggesting that they had lost a ductal phenotype (Fig. 3 and data not shown). This suggests that removing Sox17 in the common duct is sufficient to induce ectopic pancreas development. It is therefore not surprising that deleting Sox17 has no obvious effect on normal pancreas development (data not shown).

Figure 3
Loss of Sox17 leads to gall bladder agenesis and ectopic pancreas development at e16.5

Sox17 misexpression suppresses pancreas development

Our data suggest a model where Sox17 represses pancreatic fate in the bilary primoridium, and that Sox17 must be down regulated in the ventral pancreatic cells as they separate from the biliary lineage. We tested this possibility by maintaining Sox17 in the pancreatic lineage using a Pdx1-driven tetracycline transactivator mouse (Pdx1tTa/+) and a tetracycline regulated Sox17 transgene (Pdx1tTa/+;tetO-Sox17 = Sox17 gain-of function (GOF) (Holland et al., 2002; Park et al., 2006). In this system the Sox17 transgene is ON in the absence of doxycyline, where as administering Dox to turns the Sox17 transgene OFF. In Sox17-GOF embryos that were never fed doxycycline (ON), Sox17 was ectopically expressed in Pdx1-expressing cells starting at e9.5 (data not shown) and by e10.5 Sox17 was expressed throughout the Pdx1 domain. These embryos had a remnant of a dorsal and ventral pancreatic bud, which were fused to the gut tube (Fig. 4A compared to control in Fig. 1). Formation of the biliary epithelium was relatively normal in Sox17-GOF embryos at e10.5, indicating that ectopic Sox17 can suppress pancreas formation without affecting segregation of the biliary lineage.

Figure 4
Sox17 suppresses pancreas development and ventralizes the gut

Since the dorsal and ventral pancreas do not bud properly in Sox17-GOF embryos, we hypothesized that maintaining Sox17 in the pancreatic domains suppresses pancreas development. Consistent with this, expression of the pancreatic transcription factor Nkx2.2 is dramatically reduced in Sox17-GOF embryos relative to control embryos (Fig. 4B and C). Nkx2.2 is only expressed in areas where Sox17 was not ectopically expressed (Fig. 4C, boxed region and inset). Other pancreatic markers, such as Pdx1 and Ptf1a are not affected by ectopic Sox17 expression (Fig. 4A, H, I). These data suggest that Sox17 inhibits pancreas development downstream of Pdx1 and Ptf1a and upstream of Nkx2.2. It also suggests that Sox17 does not directly down regulate Pdx1 during separation of the pancreatic and biliary lineages.

We examined expression of Hnf6 and Hhex, which are also expressed in this region of the gut. Hhex is normally restricted to the ventral gut structures at E10.5, including the ventral pancreas, biliary and liver primordia (Bort et al., 2004; Hunter et al., 2007) but is not normally expressed in the midgut or dorsal pancreas. However, in Sox17-GOF embryos Hhex expression was dorsally expanded into the gut and even expressed in the presumptive dorsal pancreas (Fig. 4F,G). The normal expression of Hnf6 throughout the biliary primordium, dorsal and ventral pancreas and midgut is not changed by Sox17 misexpression at E10.5 (Fig. 4D,E). As expected, liver bud development was not altered in Sox17-GOF embryos since the Pdx1tTa is not active in the liver (data not shown). Taken together these results suggest that Sox17 suppresses pancreatic fate and that down-regulating Sox17 from the Pdx1 domain is crucial for normal pancreas development.

Sox17 misexpression causes organ dysgenesis throughout the Pdx1 domain

Sox17 misexpression suppressed early pancreatic development and mispatterned the gut, but it was not clear what these cells had become in response to Sox17 expression. Since Sox17 is required for biliary specification, one possibility is that Sox17 misexpression directed some of these cells into a biliary and ductal fate. In order to test this we analyzed the fate of Sox17 expressing cells at e16.5. Compared to control embryos, Sox17-GOF embryos had a large mass at the pyloric junction (Fig. 5A-D) and had a dramatic decrease in the amount of pancreatic tissue (Fig. 5B, Supplemental Fig. 4). In lineage traced embryos (Pdx1-tTa;tetO-cre;tetO-Sox17;R26R) we found that the entire ectopic structure was derived from Sox17-overexpressing cells (Fig. S6).

Figure 5
Sox17 expression is sufficient to induce ectopic ductal tissue, pancreas agenesis and gut malformations at e16.5

Histological analysis of Sox17-GOF embryos at e16.5 revealed many ectopic duct-like structures that expressed the duct marker HNF6 (Fig. 5E, F). The stomach epithelium and mesenchyme in the GOF embryos both appeared hyperplasic (Fig. 5D). While some ectopic ductal tissue was negative for the gut epithelial marker HNF4α, other regions were HNF6/HNF4α double positive, suggesting that these cells had the properties of both gut and ductal epithelium (Fig. 5F). The ectopic tissue mass contained regions of duodenal-like villi that were interspersed with a hyperplasic ductal-like epithelium. The gut epithelial marker villin (Braunstein et al., 2002) (Fig. 5G) was absent from large regions of the protuberance in Sox17-GOF embryos (Fig. 5H). These data suggest that persistent Sox17 expression suppresses pancreatic development and is sufficient to induce ectopic ductal tissue in the stomach and duodenum. Since there is no marker that can distinguish one type of ductal epithelia from another, it is impossible to determine if the ectopic ductal tissue present in the Sox17-GOF embryos is biliary.

The role of Sox17 in directing a biliary versus pancreatic cell fate choice is temporally restricted to e8.5-e10.5

Since the tetO-Sox17 transgene was constitutively misexpressed between e8.5 and e16.5 of development, it was not clear when during development Sox17 misexpression was causing these phenotypes. To investigate this, we gave a pulse of Sox17 expression for different lengths of time between e8.5 to e12.5 and then analyzed the embryos at e16.5. We found that misexpression of Sox17 between e8.5-10.5 gave the same phenotypes as maintained expression from e8.5-e16.5: loss of pancreatic tissue and the formation of a tissue mass containing ectopic ductal tissues and disorganized stomach and duodenal epithelium (Fig. 6A, D). In the converse experiment where the Pdx1tTa/+; tetO-Sox17 transgene was induced after e12.5, pancreatic development was grossly normal and no ectopic mass was detected at the pyloric-duodenal junction (data not shown). We based the timing of tetO-Sox17 on previous experiments that demonstrated that the Pdx1 tet-transactivator initiates transgene expression at e8.5 and is down regulated within one day of doxycycline treatment (Hale et al., 2005).

Figure 6
Use of the Tet-regulatible system shows that Sox17 has distinct temporal activities

Sox17 misexpression is sufficient to induce a ductal cell fate in the pancreas at the expense of other lineages

Given the appearance of ectopic ductal tissue in the duodenum the Sox17-GOF embryos we hypothesized that Sox17 might have the general property of promoting a ductal fate. To test this we misexpressed Sox17 in the pancreas of animals starting at E12.5, where it is normally not expressed (Sox17 ON in Pdx1-cells at E12.5) and examined the pancreas at E16.5 and at 6 weeks (Fig. 6G-J). At E16.5, the Sox17-GOF pancreas was the same size as in control animals, however histological analysis showed a dramatic increase in DBA positive ductal cells relative to control animals (Fig. 6G, H). This increase in DBA positive ductal tissue appeared to come at the expense of endocrine cells, which were dramatically reduced in these animals (data not shown). Furthermore, H&E staining showed that the pancreas of 6-week old Sox17-GOF animals had a massive expansion of ductal tissue, whereas acinar tissue was reduced and mature endocrine cells were nearly absent (Fig. 6I, J and data not shown). These results suggest that Sox17 misexpression in the pancreas is sufficient to promote a pancreatic ductal cell fate at the expense of the other lineages.

Together with the loss-of-function data, we conclude that Sox17 is required between e8.5 and e10.5 to properly segregate the foregut progenitors into the biliary and pancreatic lineages, and that sustained Sox17 expression in the Pdx1+ cells in this period is sufficient to suppress pancreas and induce ectopic biliary fate.

Hes1 regulates the segregation of Sox17/Pdx1 lineages from common foregut progenitors

Previous studies have implicated the Notch, FGF10, and hedgehog signaling pathways in the development of the ventral pancreas and biliary tree. We used existing mouse mutant lines to determine if any of these pathways were involved in establishing the Sox17/Pdx1 pancreatic-biliary progenitor domain. Embryos lacking either FGF10 (FGF10 null) or the hedgehog co-receptor smoothened (Pdx1-cre;smoothenedFl/Fl) formed a normal Sox17+/Pdx1+ progenitor domain that resolved into biliary and pancreatic primordial that were indistinguishable from control embryos at E10.5 (Fig. S7). As previously reported, FGF10 null embryos at later stages had pancreatic and biliary defects that occur after specification and segregation of the ventral pancreas and gall bladder primordium (data not shown).

In contrast, our analysis of embryos lacking Hairy Enhancer of Split-1 (Hes1−/−), a transcriptional effector of notch signaling, demonstrated that Hes1 is required for formation of the ventral pancreatic and biliary bud. In Hes1−/− embryos at e9.5, the Sox17 expression domain was expanded dorsally in a scattered fashion, and we even observe Sox17+ cells in the dorsal pancreas (Fig. 7A,B and inset. NOTE: Fig. 7 is presented with separated channel colors in Fig. S8). This suggests that Hes1 acts to restrict Sox17+ cells to the ventral gut. At e10.5 Sox17 and Pdx1 expression was nearly absent in ventral endoderm of Hes1−/− embryos, and the ventral pancreas and biliary primordium did not develop into distinct structures (Fig. 7C, D). Failure to separate these domains is catastrophic and results in biliary agenesis, hypoplasia of the EHBD and replacement of the common duct with pancreatic tissue as previously reported (Fukuda et al., 2006; Sumazaki et al., 2004). We have not determined if Hes1 is acting as a Notch effector in this context, as it can also have Notch-independent effects.

Figure 7
Hes1 and Sox17 may act coordinately to segregate the pancreatobiliary primordia

In addition, our analysis of Sox17-LOF and Sox17-GOF embryos suggests a functional connection between Sox17 and Hes1 in controlling biliary development. First, the Sox17-LOF phenotype is strikingly similar to Hes1−/− animals, with gall bladder agenesis and ectopic pancreatic tissue in the common duct (Fig. (Fig.22,,3)3) (Fukuda et al., 2006; Sumazaki et al., 2004). Second, misexpression of Sox17 in Pdx1-positive cells in the dorsal and ventral pancreas coincides with an increase in HES1 protein levels at e9.5 (Fig. 7E, F) and e10.5 (Fig. S8). Third, HES1 protein expression is reduced in Sox17-LOF animals (Fig. 7G, H) suggesting that Sox17 appears to positively effect Hes1 expression but is not absolutely required for its expression. Lastly, Sox17+/Pdx1+ progenitor cells express higher levels of HES1 protein than Sox17−/Pdx1+ cells (Fig. (Fig.7E7E and S8E, compare dorsal Sox17−/Pdx1+ vs. ventral Sox17+/Pdx1+). Taken together our results suggest a model where between E8.5 and 9.5, Sox17 and Hes1 may be involved in a regulatory feedback loop where high levels of Sox17 in Pdx1 positive cells act to promote Hes1 expression, which then in turn progressively restricts Sox17-expressing, Pdx1 negative cells to the ventral gut (Fig. 7I). Early cell segregation defects in the Hes1 mutant embryo are associated with a reduction of both Sox17 and Pdx1 expressing cells at E10.5 and disruption of organ development.

While our results clearly show that Hes1 is required for proper segregation of Sox17/Pdx1 lineages from common foregut progenitors, further studies are needed to elucidate the molecular mechanism by which Hes1 and Sox17 work to restrict biliary progenitors to the ventral-most domain of the pancreatic/biliary bud.

DISCUSSION

A Sox17-dependent mechanism for specifying organ cell fate and establishing boundaries is required for normal pancreas, biliary and liver development

The liver, ventral pancreas and biliary system are all derived from several hundred endoderm cells found in the caudal region of the ventral foregut. We presented data to demonstrate a requirement for Sox17 in lineage choice between a pancreatic and biliary cell fate. We propose a model (Fig. S10) where Sox17 is broadly expressed in the presumptive foregut and is then progressively down regulated in the different organ domains as they are specified. For example, at E8.5 Sox17 is down regulated in the presumptive Hhex1+ domain as hepatic cells are specified, then it is down regulated in the Pdx1+ cells as ventral pancreatic cells are specified. Finally, Sox17 expression is maintained in a subset of biliary progenitors that give rise to the gall bladder, cystic duct, common duct and extrahepatic bile ducts. Our data further suggests that the Sox17 is required either to establish or maintain distinct boundaries between the liver, biliary and pancreatic domain since the loss of Sox17 results in ectopic Pdx1 cells in the liver bud and biliary system. Furthermore, we have shown that the Sox17-dependant segregation of pancreatic and biliary domains requires Hes1 (discussed below and Fig. S10). The factors that regulate the dynamic expression of Sox17 during foregut organogenesis still need to be investigated.

Feedback between Sox17 and Hes1 during biliary and pancreas development

The Sox17 loss-of-function phenotype is strikingly similar to Hes1−/− animals, with gall bladder agenesis and ectopic pancreatic tissue in the common duct (Fukuda et al., 2006; Sumazaki et al., 2004), suggesting that these two molecules act in the same pathway. While Sox17 is not required for Hes1 expression our data indicates that Sox17 is able to regulate Hes1 levels in Pdx1-expressing endoderm; Hes1 protein is reduced in Sox17-LOF animals and is elevated in Pdx1+ cells that overexpress Sox17 in the dorsal pancreas. Importantly we observe higher Hes1 protein levels in the normal context of ventral cells that co-express Pdx1 and Sox17, suggesting that progenitors are defined as Pdx1+/Sox17+/high Hes1.

Interestingly, our data suggest a negative feedback loop exists between Hes1 and Sox17 since Sox17 cells expand dorsally in Hes1−/− embryos and Sox17+ cells are even found in the dorsal pancreas. The absolute levels of Sox17 do not change, suggesting that Hes1 does not directly repress Sox17 expression, rather, Hes1 restricts Sox17 cells to the ventral part of the bud in an indirect manner. While more work is needed to work out the exact details of how Sox17 and Hes1 cooperate to regulate lineage segregation, it is interesting to speculate that Pdx1 and Sox17 may be cooperating between e8.5 and e9.5 to drive high Hes1 levels, and that Hes1 feeds back to limit Sox17-expressing cells to the ventral gut (Fig. S10). Our data shows that failure to ventrally restrict Sox17+ cells in Hes1−/− embryos at e9.5 corresponds with loss of the biliary primordium at e10.5. It is possible that some Sox17 cells at e9.5 are converted to a pancreatic fate, consistent with the formation of ectopic pancreatic tissue in Hes1−/− embryos (Fukuda et al., 2006; Sumazaki et al., 2004). Alternatively, early mislocaliztion of Sox17+ cells might trigger their apoposis, resulting in loss of biliary structures by e10.5. This possibility remains to be determined.

Our data explain why mutations in either Sox17 or Hes1 cause similar phenotypes, and suggest that the reported phenotypes in Hes1−/− embryos are due in part to Sox17-dependent defects and visa-versa. Since Sox17 cells are not properly localized Hes1−/− embryos, and Hes1 expression is reduced in Sox17-LOF embryos, mutations in either would feed back and cause failure in pancreas and biliary lineages to properly segregate, loss of biliary tissue, and de-repression of pro-pancreatic factors. Ectopic pancreas formation in Hes1−/− embryos was attributed to suppressing pancreatic transcription factors Ptf1a/p48 and Ngn3 in ductal epithelia that already expressed endogenous Pdx1 (Esni et al., 2004; Fukuda et al., 2006; Ghosh and Leach, 2006; Jensen et al., 2000; Sumazaki et al., 2004). It is not clear if ectopic pancreatic tissue in Sox17-LOF embryos is due to early mis-segregation of biliary and pancreatic lineages or a reduction in Hes1 protein expression in the duct at later stages. Pancreatic markers have been observed at low levels in biliary system (Dutton et al., 2007), suggesting that this epithelium has latent pancreatic potential that could be enhanced by removal of biliary determinants like Sox17 or Hes1.

Sox17 is upstream in a cascade of transcription factors that regulates biliary tree development

The transcription factors Hhex, HNF6 (Onecut One), HNF1® and HNF4left angle bracket play roles in different aspects of biliary tree development (Clotman et al., 2002; Hunter et al., 2007; Matthews et al., 2004). Since our studies demonstrate that Sox17 is also involved in early biliary development, it is possible that Sox17, Hhex and HNF6 form a transcriptional network that regulates development of the biliary system. Consistent with this, misexpression of Sox17 throughout the dorsal Pdx1 domain is sufficient to induce ectopic expression of Hhex in the duodenum and dorsal pancreas at E10.5. Moreover, Sox17 misexpression is sufficient to induce the biliary/ductal molecular program resulting in HNF6 expression in ectopic ductal structures throughout the Pdx1 expression domain. These data suggest that Sox17 might act a master regulator of the biliary/ductal lineage upstream of Hhex and is sufficient for diverting other cell types into the biliary lineage. We have also shown that Sox17 overexpression in the pancreas at E12.5, when it is not normally expressed, is sufficient to promote a ductal fate, apparently at the expense of endocrine cells. This supports the idea that Sox17 acts at the top of a hierarchy and can activate a transcriptional program resulting in a ductal fate.

An alternative model for biliary development

Our lineage tracing and gene expression data indicates that the biliary system shares a common origin with ventral pancreas and not the liver as previously thought (Figure S10) (Clotman et al., 2002; Hunter et al., 2007; Shiojiri, 1997). The fact that Sox17 and Hes1 LOF affected the biliary and pancreatic lineages also supports this conclusion. Lastly, congenital defects have been reported in humans that have linked anomalies in both the pancreas and biliary system. It is likely that a common progenitor for liver, biliary system and pancreas exists at earlier stages of development, when the anterior definitive endoderm is forming the foregut, and that the first of these lineages to separate is the liver. This possibility is supported by studies in Zebrafish, where single lineage traced cells gave rise to committed liver cells and Pdx1-positive cells. Those Pdx1 positive cells, our data would suggest, are the pancreatobiliary progenitors.

These data describe a Sox17-based model for extrahepatic biliary development from a Pdx1 progenitor. It is important to point out that the intrahepatic bile ducts (IHBD) were not lineage labeled using Pdx1-cre and were not affected in Sox17-LOF or Sox17-GOF studies (data not shown). Therefore, development of the entire biliary tree might be dependant on cells derived from different populations of progenitor cells. The EHBDs, common duct, cystic duct and gall bladder are derived from a Pdx1+ progenitor cell, whereas the IHBDs are derived from a hepatic progenitor cell (Fig. S10). This model could provide insight into the causes of human congenital abnormalities affecting both pancreas and biliary tree. Moreover, a better understanding of how endoderm organ lineages are initially segregated will inform efforts at directed differentiation of human ES cells into endoderm derivatives (Spence and Wells, 2007).

EXPERIMENTAL PROCEDURES

Immunohistochemistry

Tissues were either fixed overnight and embedded in paraffin, or fixed for 1 hour, equilibrated in 30% sucrose/1xPBS, and frozen in OCT. Sections were cut 6-10 micrometers thick. Paraffin sections were deparaffinized, dehydrated and we performed antigen retrieval by steaming slides in sodium citrate buffer for 30 minutes. Sections were blocked in the appropriate serum (5% serum in 1x PBS + 0.5% triton-X) for 30 minutes. Primary antibodies were diluted in blocking buffer and incubated on tissue sections overnight at 4 degrees Celsius. Slides were washed and incubated in secondary antibody in blocking buffer for 2 hours at room temperature. For a list of antibodies used and dilutions, see supplemental experimental procedures. Slides were washed and mounted using Fluormount-G.

LacZ staining and histology analysis

Beta-galactosidase activity was detected in fixed whole tissue using the Histomark X-gal substrate system (Kireguard and Perry Laboratories, MD). For hematoxylin and eosin staining, 6 μm paraffin sections were dewaxed in xylene, rehydrated, and stained.

Mice

All mice used in these studies were housed at the Cincinnati Children's Hospital Research Foundation mouse facility, and were maintained according to institutional protocols. Pdx1-cre, Pdx1-tTa, FoxA3cre, Sox17-GFP, FGF10, SmoothenedFl and Hes1 mice have all been previously described (Bellusci et al., 1997; Holland et al., 2002; Ishibashi et al., 1995; Kim et al., 2007; Lee et al., 2005b; Wells et al., 2007; Zhang et al., 2001). The generation of the Sox17Fl mice is described in the Supplemental Materials and Methods.

Supplementary Material

01

ACKNOWLEDGEMENTS

We thank Vladimir Kalinichenko, Chris Wylie, Ben Stanger, Gail Deutsch and Aaron Zorn and the Wells and Zorn labs for input on the work. We also thank Sean Morrison (University of Michigan) and Raymond MacDonald (University of Texas Southwestern) for the Sox17GFP and Pdx1tTA mice. Nadean Brown (Cincinnati Children's Hospital Medical Center) and Clifford Bogue (Yale University) kindly provided the Hes1 and Hhex antibodies, respectively. This work was supported by a career development award from the Juvenile Diabetes Research Foundation (JDRF-2-2003-530) and a grant from the NIH (R01GM072915). JRS was supported by postdoctoral fellowship from the JDRF and an NIH training grant in Developmental and Perinatal Endocrinology (T32 HD07463).

Footnotes

Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

References

  • Ashraf A, Abdullatif H, Hardin W, Moates JM. Unusual case of neonatal diabetes mellitus due to congenital pancreas agenesis. Pediatric Diabetes. 2005;6:239–243. [PubMed]
  • Bhushan A, Itoh N, Kato S, Thiery JP, Czernichow P, Bellusci S, Scharfmann R. Fgf10 is essential for maintaining the proliferative capacity of epithelial progenitor cells during early pancreatic organogenesis. Development. 2001;128:5109–5117. [PubMed]
  • Bort R, Martinez-Barbera JP, Beddington RS, Zaret KS. Hex homeobox gene-dependent tissue positioning is required for organogenesis of the ventral pancreas. Development. 2004;131:797–806. [PubMed]
  • Bort R, Signore M, Tremblay K, Martinez Barbera JP, Zaret KS. Hex homeobox gene controls the transition of the endoderm to a pseudostratified, cell emergent epithelium for liver bud development. Developmental Biology. 2006;290:44–56. [PubMed]
  • Braunstein EM, Qiao XT, Madison B, Pinson K, Dunbar L, Gumucio DL. Villin: A marker for development of the epithelial pyloric border. Dev Dyn. 2002;224:90–102. [PubMed]
  • Burke ZD, Shen CN, Tosh D. Bile ducts as a source of pancreatic beta cells. Bioessays. 2004;26:932–937. [PubMed]
  • Chappell L, Gorman S, Campbell F, Ellard S, Rice G, Dobbie A, Crow Y. A further example of a distinctive autosomal recessive syndrome comprising neonatal diabetes mellitus, intestinal atresias and gall bladder agenesis. Am J Med Genet A. 2008;146A:1713–1717. [PubMed]
  • Chung WS, Shin CH, Stainier DY. Bmp2 signaling regulates the hepatic versus pancreatic fate decision. Dev Cell. 2008;15:738–748. [PMC free article] [PubMed]
  • Clotman F, Lannoy VJ, Reber M, Cereghini S, Cassiman D, Jacquemin P, Roskams T, Rousseau GG, Lemaigre FP. The onecut transcription factor HNF6 is required for normal development of the biliary tract. Development. 2002;129:1819–1828. [PubMed]
  • Coffinier C, Gresh L, Fiette L, Tronche F, Schutz G, Babinet C, Pontoglio M, Yaniv M, Barra J. Bile system morphogenesis defects and liver dysfunction upon targeted deletion of HNF1beta. Development. 2002;129:1829–1838. [PubMed]
  • Deutsch G, Jung J, Zheng M, Lora J, Zaret KS. A bipotential precursor population for pancreas and liver within the embryonic endoderm. Development. 2001;128:871–881. [PubMed]
  • Dong PD, Munson CA, Norton W, Crosnier C, Pan X, Gong Z, Neumann CJ, Stainier DY. Fgf10 regulates hepatopancreatic ductal system patterning and differentiation. Nature Genetics. 2007;39:397–402. [PubMed]
  • Dutton JR, Chillingworth NL, Eberhard D, Brannon CR, Hornsey MA, Tosh D, Slack JM. Beta cells occur naturally in extrahepatic bile ducts of mice. J Cell Sci. 2007;120:239–245. [PubMed]
  • Esni F, Ghosh B, Biankin AV, Lin JW, Albert MA, Yu X, MacDonald RJ, Civin CI, Real FX, Pack MA, et al. Notch inhibits Ptf1 function and acinar cell differentiation in developing mouse and zebrafish pancreas. Development. 2004;131:4213–4224. [PubMed]
  • Fukuda A, Kawaguchi Y, Furuyama K, Kodama S, Horiguchi M, Kuhara T, Koizumi M, Boyer DF, Fujimoto K, Doi R, et al. Ectopic pancreas formation in Hes1 - knockout mice reveals plasticity of endodermal progenitors of the gut, bile duct, and pancreas. J Clin Invest. 2006;116:1484–1493. [PubMed]
  • Fukuda S, Mizuno T. Hepatic parenchyma, biliary ducts and gall bladder forming potency in the hepatic primordium in the quail embryo. Anatomy & Embryology. 1978;155:15–21. [PubMed]
  • Galan-Gomez E, Sanchez EB, Arias-Castro S, Cardesa-Garcia JJ. Intrauterine growth retardation, duodenal and extrahepatic biliary atresia, hypoplastic pancreas and other intestinal anomalies: further evidence of the Martinez-Frias syndrome. Eur J Med Genet. 2007;50:144–148. [PubMed]
  • Ghosh B, Leach SD. Interactions between hairy/enhancer of split-related proteins and the pancreatic transcription factor Ptf1-p48 modulate function of the PTF1 transcriptional complex. Biochem J. 2006;393:679–685. [PubMed]
  • Gualdi R, Bossard P, Zheng M, Hamada Y, Coleman JR, Zaret KS. Hepatic specification of the gut endoderm in vitro: cell signaling and transcriptional control. Genes Dev. 1996;10:1670–1682. [PubMed]
  • Hale MA, Kagami H, Shi L, Holland AM, Elsasser HP, Hammer RE, MacDonald RJ. The homeodomain protein PDX1 is required at mid-pancreatic development for the formation of the exocrine pancreas. Dev Biol. 2005;286:225–237. [PubMed]
  • Hebrok M, Kim SK, Jacques B, McMahon AP, Melton DA. Regulation of pancreas development by hedgehog signaling. Development. 2000;127:4905–4913. [PubMed]
  • Heij HA, Niessen GJ. Annular pancreas associated with congenital absence of the gallbladder. Journal of Pediatric Surgery. 1987;22:1033. [PubMed]
  • Holland AM, Hale MA, Kagami H, Hammer RE, MacDonald RJ. Experimental control of pancreatic development and maintenance. Proceedings of the National Academy of Sciences of the United States of America. 2002;99:12236–12241. [PubMed]
  • Hunter MP, Wilson CM, Jiang X, Cong R, Vasavada H, Kaestner KH, Bogue CW. The homeobox gene Hhex is essential for proper hepatoblast differentiation and bile duct morphogenesis. Dev Biol. 2007;308:355–367. [PMC free article] [PubMed]
  • Iguchi H, Ikeda Y, Okamura M, Tanaka T, Urashima Y, Ohguchi H, Takayasu S, Kojima N, Iwasaki S, Ohashi R, et al. SOX6 attenuates glucose-stimulated insulin secretion by repressing PDX1 transcriptional activity and is down-regulated in hyperinsulinemic obese mice. J Biol Chem. 2005;280:37669–37680. [PubMed]
  • Jacquemin P, Yoshitomi H, Kashima Y, Rousseau GG, Lemaigre FP, Zaret KS. An endothelial-mesenchymal relay pathway regulates early phases of pancreas development. Developmental Biology. 2006;290:189–199. [PubMed]
  • Jensen J, Pedersen EE, Galante P, Hald J, Heller RS, Ishibashi M, Kageyama R, Guillemot F, Serup P, Madsen OD. Control of endodermal endocrine development by Hes-1. Nat Genet. 2000;24:36–44. [PubMed]
  • Jonsson J, Carlsson L, Edlund T, Edlund H. Insulin-promoter-factor 1 is required for pancreas development in mice. Nature. 1994;371:606–609. [PubMed]
  • Kanai-Azuma M, Kanai Y, Gad JM, Tajima Y, Taya C, Kurohmaru M, Sanai Y, Yonekawa H, Yazaki K, Tam PP, et al. Depletion of definitive gut endoderm in Sox17-null mutant mice. Development. 2002;129:2367–2379. [PubMed]
  • Kawaguchi Y, Cooper B, Gannon M, Ray M, MacDonald RJ, Wright CV. The role of the transcriptional regulator Ptf1a in converting intestinal to pancreatic progenitors. Nat Genet. 2002;32:128–134. [PubMed]
  • Kim I, Saunders TL, Morrison SJ. Sox17 dependence distinguishes the transcriptional regulation of fetal from adult hematopoietic stem cells. Cell. 2007;130:470–483. [PMC free article] [PubMed]
  • Kim SK, Melton DA. Pancreas development is promoted by cyclopamine, a hedgehog signaling inhibitor. Proc Natl Acad Sci U S A. 1998;95:13036–13041. [PubMed]
  • Krapp A, Knofler M, Ledermann B, K B.r., Berney C, Zoerkler N, Hagenb chle O, Wellauer PK. The bHLH protein PTF1-p48 is essential for the formation of the exocrine and the correct spatial organization of the endocrine pancreas [In Process Citation] Genes Dev. 1998;12:3752–3763. [PubMed]
  • Le Douarin N. [Synthesis of glycogen in hepatocytes undergoing differentiation: role of homologous and heterologous mesenchyma] Dev Biol. 1968;17:101–114. [PubMed]
  • Le Douarin NM. On the origin of pancreatic endocrine cells. Cell. 1988;53:169–171. [PubMed]
  • Lee CS, Friedman JR, Fulmer JT, Kaestner KH. The initiation of liver development is dependent on Foxa transcription factors. Nature. 2005;445:944–947. [PubMed]
  • Litingtung Y, Lei L, Westphal H, Chiang C. Sonic hedgehog is essential to foregut development. Nat Genet. 1998;20:58–61. [PubMed]
  • Manfroid I, Delporte F, Baudhuin A, Motte P, Neumann CJ, Voz ML, Martial JA, Peers B. Reciprocal endoderm-mesoderm interactions mediated by fgf24 and fgf10 govern pancreas development. Development. 2007;134:4011–4021. [PubMed]
  • Matthews RP, Lorent K, Russo P, Pack M. The zebrafish onecut gene hnf-6 functions in an evolutionarily conserved genetic pathway that regulates vertebrate biliary development. Dev Biol. 2004;274:245–259. [PubMed]
  • Mehes K, Vamos K, Goda M. Agenesis of pancreas and gall-bladder in an infant of incest. Acta Paediatrica Academiae Scientiarum Hungaricae. 1976;17:175–176. [PubMed]
  • Mitchell J, Punthakee Z, Lo B, Bernard C, Chong K, Newman C, Cartier L, Desilets V, Cutz E, Hansen IL, et al. Neonatal diabetes, with hypoplastic pancreas, intestinal atresia and gall bladder hypoplasia: search for the aetiology of a new autosomal recessive syndrome. Diabetologia. 2004;47:2160–2167. [PubMed]
  • Offield MF, Jetton TL, Labosky PA, Ray M, Stein RW, Magnuson MA, Hogan BL, Wright CV. PDX-1 is required for pancreatic outgrowth and differentiation of the rostral duodenum. Development. 1996;122:983–995. [PubMed]
  • Park KS, Wells JM, Zorn AM, Wert SE, Whitsett JA. Sox17 influences the differentiation of respiratory epithelial cells. Developmental Biology. 2006;294:192–202. [PubMed]
  • Rosenquist GC. The location of the pregut endoderm in the chick embryo at the primitive streak stage as determined by radioautographic mapping. Dev Biol. 1971;26:323–335. [PubMed]
  • Serls AE, Doherty S, Parvatiyar P, Wells JM, Deutsch GH. Different thresholds of fibroblast growth factors pattern the ventral foregut into liver and lung. Development. 2005:35–47. [PubMed]
  • Shin D, Shin CH, Tucker J, Ober EA, Rentzsch F, Poss KD, Hammerschmidt M, Mullins MC, Stainier DY. Bmp and Fgf signaling are essential for liver specification in zebrafish. Development. 2007;134:2041–2050. [PubMed]
  • Shiojiri N. Development and differentiation of bile ducts in the mammalian liver. Microsc Res Tech. 1997;39:328–335. [PubMed]
  • Sinner D, Kirilenko P, Rankin S, Wei E, Howard L, Kofron M, Heasman J, Woodland HR, Zorn AM. Global analysis of the transcriptional network controlling Xenopus endoderm formation. Development. 2006;133:1955–1966. [PubMed]
  • Sinner D, Rankin S, Lee M, Zorn A. Sox17 and beta-catenin cooperate to regulate the transcription of endodermal genes. Development. 2004;131:3069–3080. [PubMed]
  • Spence JR, Wells JM. Translational embryology: Using embryonic principles to generate pancreatic endocrine cells from embryonic stem cells. Developmental Dynamics. 2007;236:3218–3227. [PubMed]
  • Sumazaki R, Shiojiri N, Isoyama S, Masu M, Keino-Masu K, Osawa M, Nakauchi H, Kageyama R, Matsui A. Conversion of biliary system to pancreatic tissue in Hes1-deficient mice. Nat Genet. 2004;36:83–87. [PubMed]
  • Tremblay KD, Zaret KS. Distinct populations of endoderm cells converge to generate the embryonic liver bud and ventral foregut tissues. Developmental Biology. 2005;280:87–99. [PubMed]
  • Wells JM, Esni F, Boivin GP, Aronow BJ, Stuart W, Combs C, Sklenka A, Leach SD, Lowy AM. Wnt/beta-catenin signaling is required for development of the exocrine pancreas. BMC Developmental Biology. 2007;7:4. [PMC free article] [PubMed]
  • Zorn AM, Barish GD, Williams BO, Lavender P, Klymkowsky MW, Varmus HE. Regulation of Wnt signaling by Sox proteins: XSox17 alpha/beta and XSox3 physically interact with beta-catenin. Mol Cell. 1999;4:487–498. [PubMed]