Introduction

Endocrine cells of the digestive tract include pancreatic cells clustered in the islets of Langerhans (Yamaoka and Itakura, 1999) and scattered cells distributed throughout the digestive epithelium, from the stomach to the colon (Hocker and Wiedenmann, 1998), known as enteroendocrine cells. Despite the fact that the relative percentage of endocrine cells is low within the gastrointestinal epithelium, quantitatively enteroendocrine cells constitute the major endocrine organ of the organism. The various peptide hormones produced by these endocrine cells control important physiological functions such as glycemia, exocrine pancreatic secretion, growth and repair of the gut epithelium, motility of the gut wall and gastric emptying (Skipper and Lewis, 2000). In mammals, the gastrointestinal epithelial cells continue to proliferate actively throughout life. In the small intestinal mucosa, enteroendocrine cells, as well as the three other principal cell types of the gut (enterocytes, goblet cells and Paneth cells), arise from a multipotent stem cell located near the base of the crypts of Lieberkühn (Gordon et al., 1992). It is thought that progenitor cells, located in the proliferative compartment, derive from these stem cells and, while leaving the crypts, stop dividing and differentiate further into each of the intestinal cell lineages during their migration to the villus tip (Booth and Potten, 2000; Marshman et al., 2002). Paneth cells migrate downwards to the base of the crypts. The intestinal enteroendocrine cells consist of at least 15 different cell types classified essentially on the basis of their hormonal content with a specific geographical distribution (Hocker and Wiedenmann, 1998). The stomach epithelium is organized in invaginations known as gastric glands, and it is thought that presumptive stem cells located in a central position in the isthmus region of the gland give rise to all the gastric epithelial cell types (Karam et al., 1997). A majority of the newly formed enteroendocrine cells migrate to the base of the gastric gland (Karam and Leblond, 1993). Transgenic studies have provided important insights into the potential lineage relationship between the different enteroendocrine cell types (Roth et al., 1990; Roth and Gordon, 1990; Lopez et al., 1995; Rindi et al., 1999). However, the early steps of the specification of the endocrine cell lineages in the digestive tract are poorly understood. In particular, the molecular mechanisms controlling endocrine commitment in multipotent stem/progenitor cells in the embryonic as well as the adult intestine and stomach remain to be clarified.

Proteins of the basic helix−loop−helix (bHLH) family are important transcriptional regulators of cell fate determination and differentiation in a number of cell types in both invertebrate and vertebrate species (Hassan and Bellen, 2000; Massari and Murre, 2000). For example, the products of the proneural genes neurogenin 1 and 2 are bHLH transcription factors required for neural precursor determination in the peripheral and central nervous system (Fode et al., 1998, 2000; Ma et al., 1999). We and others showed recently that the third member of the neurogenin family, neurogenin3 (ngn3), is transiently expressed in endocrine progenitors during pancreas development (Apelqvist et al., 1999; Gradwohl et al., 2000; Jensen et al., 2000a; Schwitzgebel et al., 2000). Mice lacking ngn3 die postnatally from diabetes because they fail to generate any islet cells, and endocrine progenitors are missing (Gradwohl et al., 2000). These results, together with gain-of-function studies (Apelqvist et al., 1999; Grapin-Botton and Melton, 2000; Schwitzgebel et al., 2000), suggest that ngn3 controls endocrine fate decisions in multipotent pancreatic endodermal progenitors. Endocrine cell development in the intestinal epithelium is also regulated by members of the bHLH transcription factor family. For example, in the mouse intestine, the loss of Math1 leads to depletion of the secretory cell lineages, including Paneth, goblet and enteroendocrine cells (Yang et al., 2001). In addition, the bHLH protein BETA2/NeuroD has been shown to control terminal differentiation of the enteroendocrine secretin-producing cells by coordinating transcription of the secretin gene with cell cycle arrest (Naya et al., 1997; Mutoh et al., 1998; Ratineau et al., 2002). In contrast, the bHLH transcription factor HES1 negatively regulates enteroendocrine differentiation (Jensen et al., 2000b).

In the present study, we investigated the expression and function of ngn3 in the developing and adult intestine. We also generated transgenic mice to characterize the activity of the ngn3 promoter, and took advantage of the stability of the reporter protein to trace ngn3 progeny in the gastrointestinal tract. Our results indicate that ngn3 is expressed in enteroendocrine progenitors all along the proximo-distal axis of the developing intestine and in the adult crypts. No endocrine cells are found in the intestinal epithelium of ngn3-deficient mice, and endocrine progenitors are lacking. Based on these results, we propose that, similarly to the developing pancreas, ngn3 specifies an endocrine fate in uncommitted cells of the intestinal mucosa during both embryogenesis and adult life. In the stomach, enteroendocrine differentiation is both ngn3 dependent and independent.

Results

ngn3 is expressed in isolated cells in the embryonic gut endoderm and subsequently in the intestinal crypts

The ngn3 spatio-temporal expression pattern in embryonic and adult mouse intestines has been investigated by in situ hybridization (ISH; Figure 1A−E). ngn3-expressing cells are first found as early as E12.5 days post-coitum (d.p.c.) in scattered cells within the gut endoderm (Figure 1A). ngn3 transcripts precede the expression of NeuroD, which is detected in the intestinal epithelium from E14.5 d.p.c. onwards (not shown). The number of ngn3⁺ cells peaks at E14.5−E15.5, and labeled cells are found from the duodenum to the colon (Figure 1B−D) in a decreasing gradient reflecting the rostro-caudal morphogenesis of the gut. In the adult small intestine, ngn3-expressing cells are located exclusively in the proliferative compartment, the crypts (Figure 1E). Double labeling experiments showed that ngn3 transcripts are found in dividing bromodeoxyuridine-positive (BrdU⁺) cells (Figure 1E; arrowheads) but we also observed a few ngn3⁺/BrdU^- cells in the crypts (data not shown). ngn3⁺ cells were not detected in the villi and do not co-stain with chromogranin A, a general marker for differentiated endocrine cells; in contrast, NeuroD-expressing cells are found mainly in the villi and co-express chromogranin A (data not shown). Immunohistochemistry with an anti-ngn3 antibody indicated that ngn3 protein is also restricted to crypt cells (Figure 1F; arrows).

Intestinal enteroendocrine cells derive from ngn3 progenitors

bHLH genes are transiently expressed in progenitor cells; their expression is turned off as the cells become post-mitotic and differentiate. Therefore, it is difficult to determine the identity of the cell types derived from these progenitor cells. One strategy to identify the progenitor descending cells is to use a reporter gene coding for a protein more stable than the endogenous bHLH protein. This approach has been used to define the interneurons derived from Math1 progenitors (Helms and Johnson, 1998) and validated in a LacZ knock-in of Math1 (Bermingham et al., 1999; Ben Arie et al., 2000). Here, we characterize the activity of the mouse ngn3 promoter in the gastrointestinal tract of transgenic mice and take advantage of the stability of the -galactosidase protein (Smith et al., 1995) to define the cells deriving from ngn3-expressing progenitors.

We generated transgenic mice expressing a nuclear LacZ reporter gene under the control of 6.9 kb of 5' genomic sequences. In recent cell lineage tracing experiments, a very similar ngn3 promoter fragment has been used to drive the phage CRE recombinase and successfully tag the ngn3 progeny in the pancreas (Gu et al., 2002). We obtained three founder mice with an identical expression pattern of the transgene. Similarly to the ngn3 transcript and protein, -galactosidase activity has been detected all along the proximo-distal axis of the developing intestines (Figure 2A, C and E) as well as in the adult crypts (Figure 2D and F) in dividing progenitors (Figure 2F, arrow). These results indicate that the tested ngn3 regulatory sequences faithfully recapitulate endogenous gene expression in the gut as well as in the pancreas (Figure 2A; data not shown). The -galactosidase is only present in a subset of ngn3-expressing cells (Figure 2E, arrows) which is likely to be due to the mosaic expression of transgenes and/or the differences in the maturation and degradation of the two gene products. A similar observation has been reported in the analysis of the human ngn3 promoter (Lee et al., 2001). In addition, we observed -Gal⁺ cells in the villi of the adult intestinal epithelium (Figure 2B, D and G) where they co-stain with the pan-endocrine marker chromogranin A (Figure 2G). In contrast, the LacZ transcript was found exclusively in the crypt compartment (data not shown). Therefore, the presence of the -galactosidase in differentiated endocrine cells is not due to an ectopic expression of the transgene but results from the persistence of the LacZ gene product in the enteroendocrine lineage although ngn3 transcription is extinguished. Our data thus demonstate that intestinal enteroendocrine cells derive from ngn3-expressing progenitors.

Intestinal endocrine cells do not develop in ngn3-deficient mice

The function of ngn3 in the development of the intestine and in particular in the differentiation of the enteroendocrine lineage was analyzed in mice homozygous for the ngn3 deletion that we generated previously. No gross morphological abnormalities have been observed in the digestive tract of ngn3-deficient newborn animals, except that they had a smaller stomach (Figure 3A and A'). We also recorded that milk often stagnates in the intestine of the mutant mice (Figure 3A'). To determine whether endocrine cells are present in the mutant intestinal epithelium, an immunocytochemical analysis was performed with antibodies against chromogranin A and the principal intestinal hormones. Endocrine cells can be detected in the wild-type developing gut starting from E15.5 d.p.c. Cells expressing chromogranin A are found in the wild-type small (Figure 3B) and large (not shown) intestine at birth, whereas they could not be detected all along the proximo-distal axis of the intestine (Figure 3B'; data not shown) in the absence of ngn3. In addition, the principal intestinal hormones, cholecystokinin (CCK), secretin, gastrin, serotonin, peptide YY (PYY), glucagon-like protein (GLP), gastric inhibitory protein (GIP) and somatostatin are not produced in ngn3 homozygous intestine (exemplified in Figure 3C' and D'; data not shown) compared with the controls (Figure 3C and D; data not shown). ngn3 mutant mice die shortly after birth; we could thus have missed delayed entero-endocrine differentiation because intestinal maturation proceeds during the 3 weeks of postnatal life. This possibility was investigated by grafting 12-day fetal intestinal anlagen under the skin of nude mice to rescue the lethal pancreatic phenotype and allow development beyond postnatal day 1. Intestinal grafts were recovered after 4 weeks; they formed vascularized and well developed structures composed of intestinal villi, as well as the mucosal, submucosal and muscular layers (not shown). No endocrine cells could be found in the crypt and villus epithelium of the grafts analyzed, as assessed by the absence of chromogranin A and of hormone (gastrin/CCK, serotonin, GIP, PYY and GLP)-expressing cells, compared with the wild-type intestinal grafts (Figure 3E and E'; data not shown). These data suggest that all the intestinal enteroendocrine cell differentiation, at all time points, is impaired in the absence of ngn3.

Enteroendocrine progenitors are lacking in the absence of ngn3

Previous studies have shown that members of the bHLH transcription factor family regulate intestinal cell differentiation. Math1 controls the specification of a common multipotent progenitor for the intestinal secretory cells (goblet, Paneth and enteroendocrine cells), whereas NeuroD is known to coordinate terminal differentiation of enteroendocrine cells by inducing cell cycle arrest and activating the transcription of hormone genes. To determine at which stage of enteroendocrine cell development ngn3 is required, we studied the expression of Math1 and NeuroD in mutant intestinal epithelium. Math1 transcripts were detected in immature crypt cells but also in differentiated cells in the villi of both wild-type and mutant mice at birth (Figure 4A and A'; arrowheads and arrows, respectively), suggesting that ngn3 lies downstream of Math1 in the endocrine lineage. In contrast, NeuroD expression is turned off in the absence of ngn3 (Figure 4B and B'). ngn3 would thus act upstream of NeuroD and control the development of the enteroendocrine progenitor cells, before they leave the cell cycle and differentiate. The loss of Pax6 expression (Figure 4D), a paired box gene reported to be expressed widely in the enteroendocrine lineage and controlling GIP cell differentiation (Larsson et al., 1998), confirmed the upstream position of ngn3 in the cascade of transcription factors regulating enteroendocrine differentiation.

To define further the function of ngn3, we examined the expression of the LacZ reporter gene of ngn3 promoter::LacZ animals in a ngn3^-/- background (genotype: ngn3-promoter::LacZ; ngn3^-/-). Interestingly, we could not detect any -galactosidase activity in the intestinal epithelium of mutant animals (Figure 4C and C'), suggesting that enteroendocrine progenitors did not develop in ngn3 mutant mice.

The major intestinal epithelial cytotypes develop normally in ngn3^-/- mice

The consequences of the ngn3 mutation on the overall development of the intestinal mucosa as well as on the differentiation of the other intestinal cell types were examined in both newborn animals and grafted embryonic intestines as described above. Through histological analysis, we did not detect any obvious differences in the development of the mucosal layer, and the crypt and villi organization was not altered in the intestine of mice lacking ngn3 (Figure 5A and A'). Similarly, the muscular layers, the subepithelial myofibroblasts and the muscle fibers within the villus conjunctive core developed correctly, as determined by the expression of specific markers (data not shown). However, we consistently observed that the thickness of the mucosal and muscular layers was increased, at least in the proximal intestine of ngn3 homozygous mice at late postnatal stages (data not shown). In contrast to the endocrine cell types, the other epithelial cytotypes differentiate as enterocytes, mucus and Paneth cells. Indeed, differentiation markers of enterocytes, such as lactase activity in newborns (Figure 5B and B') and sucrase activity at later stages (grafts, data not shown), were detected in the intestinal epithelium of control and mutant animals. In addition, the shift between the two digestive enzymes, signified by a decrease in lactase activity and induction of sucrase, occurred in the grafting conditions at stages corresponding to weaning in both knock-out and wild-type intestines, suggesting that the enterocyte lineage differentiated properly in the absence of ngn3. Goblet cells were clearly identified in the small and large intestine by the production of mucus in mice lacking ngn3 (Figure 5A and A'; data not shown) but, interestingly, their number increased from an average of 363 45 to 585 28 goblet cells/mm² of intestinal mucosa in the P1 small intestine (an average of three wild-type and mutant animals were analyzed). The colon did not show a similar increase in goblet cell density. Finally, Paneth cells were seen in grafted intestines at their expected position in the bottom of the crypts as clusters of cells filled with secretory granules both in ngn3 homozygous mice and in controls (Figure 5C and C', arrowheads).

Gastric endocrine development is unevenly affected in ngn3^-/- mice

ISH with an antisense ngn3 riboprobe revealed that ngn3 transcripts are present in the gastric epithelium at postnatal stages and in the adult (data not shown). As described above for the gut, we took advantage of the -galactosidase protein stability to determine the identity of the ngn3-expressing cells and to trace their progeny in the ngn3 promoter::LacZ transgenic mice. Whole-mount analysis revealed that the reporter protein is expressed, in addition to the intestine, in the glandular part (oxyntic and antro-pyloric region) of the stomach at postnatal stages (Figure 6A) and in the adult (data not shown). This region corresponds to the area that normally contains gastric endocrine cells. No staining was observed in the proximal third of the stomach (forestomach) which does not contain any glands. To determine if ngn3 was expressed in the gastric endocrine lineage, the stomach of the transgenic mice was sectioned and double stained for -galactosidase activity and chromogranin A, which is expressed in all the gastric endocrine cell types (Norlen et al., 2001). The LacZ transgene product was expressed mainly in isolated cells in the gastric epithelium and some of the -Gal⁺ cells co-stained for chromogranin A (Figure 6B, arrow). To characterize the gastric ngn3 lineage further, similar double staining experiments have been performed with specific markers for the different gastric endocrine cell types (Solcia et al., 2000). Gastrin (G cells), somatostatin (D cells), serotonin (enterochromaffin EC cells), ghrelin (X/A cells) and histamine [enterochromaffin-like ECL cells expressing the specific marker histidine decarboxylase (Chen et al., 1999)] secreting cell types were all marked by the transgene (Figure 6C−G). These results suggest that ngn3, as is the case in the pancreas and the intestine, is also expressed in gastric endocrine progenitors and that these ngn3 progenitors will give rise to all the principal gastric endocrine cell types.

To determine whether gastric enteroendocrine differentiation was affected in ngn3-deficient mice, an immunohistochemical analysis was performed to detect chromogranin A and specific endocrine cell types. Cells positive for chromogranin A were still present in the ngn3^-/- gastric epithelium (Figure 7A and A'), indicating that endocrine differentiation occurred in ngn3-deficient stomach. However, gastrin- and somatostatin-expressing G and D cell differentiation was impaired in mice lacking ngn3 (Figure 7B' and C') compared with the wild-type gastric mucosa (Figure 7B and C). In contrast, serotonin- and ghrelin-producing EC and X/A endocrine cell types were still present in the mutant gastric epithelium (Figure 7D' and E'). Since ECL cells are not yet present in the mouse oxyntic mucosa at birth (our own observation), to address the role of ngn3 in ECL cell differentiation we grafted control and ngn3^-/- stomachs, as described above, to reach postnatal stages. Histidine decarboxylase is detected in both wild-type and mutant grafted stomachs (Figure 7F and F'), suggesting that the differentiation of ECL cells, which represent 65−75% of the gastric endocrine cell population (Chen et al., 1999), is not controlled by ngn3. The lack of G and D cells observed at birth in ngn3 mutant gastric epithelium (Figure 7) has been confirmed in the grafted ngn3-deficient stomachs (not shown).

Discussion

The intestinal epithelium is replaced every 3−4 days in rodents by the generation of multiple cell lineages from multipotent epithelial stem cells. The different intestinal cell types probably do not derive directly from this stem cell, rather it is thought that they are produced from an intermediate population of committed progenitor cells (Bjerknes et al., 1999). Recent studies support the hypothesis that the secretory cell lineages (goblet, enteroendocrine and Paneth cells) derive from a common progenitor expressing the bHLH transcription factor Math1 (Yang et al., 2001). From these results, it is reasonable to suggest that the secretory and absorptive lineages arise from two types of progenitors specified on the basis of the expression of the bHLH transcription factor Math1 (Figure 8A). However, very little is known about the genetic programs which control the commitment of the Math1-positive multipotent progenitors into the different secretory lineages and in particular the enteroendocrine lineage. Gene targeting experiments demonstrated that the bHLH protein NeuroD/BETA2, which is expressed in all enteroendocrine cells (Naya et al., 1997), is required specifically for the differentiation of the secretin- and CCK-expressing cells (Naya et al., 1997), whereas the remaining enteroendocrine cells develop properly in the absence of NeuroD/BETA2 (Rindi et al., 1999). Other experiments demonstrated that, in the intestine, NeuroD/BETA2 coordinates the transcription of the secretin gene with cell cycle arrest (Mutoh et al., 1998). Thus NeuroD/BETA2 is promoting the terminal differentiation of secretin cells rather than controlling early steps in the development of the enteroendocrine cell types.

In the present study, we show that another member of the bHLH transcription factor family, ngn3, is expressed in the fetal intestinal epithelium as well as in immature cells located in the proliferative compartment of the crypts in the adult small intestine. Differentiated endocrine cells in the villi which express NeuroD (Rindi et al., 1999) do not express ngn3. However, lineage tracing experiments of the ngn3 progeny shown herein demonstrate that crypt ngn3-expressing cells finally give rise to chromogranin A-positive enteroendocrine cells. This conclusion has been drawn from the analysis of transgenic mice that we produced expressing the LacZ gene under the control of ngn3 regulatory sequences. In these mice, the reporter recapitulates endogenous intestinal ngn3 expression, and the long half-life of -galactosidase allowed us to follow the fate of the ngn3-expressing cells. Together, these expression data suggest that, as is the case in the developing pancreas, ngn3 marks an endocrine committed progenitor population, which does not yet express endocrine hormones. Gene targeting experiments showed that all the intestinal endocrine cells fail to develop in ngn3^-/- mice at embryonic stages, at birth and also at later postnatal stages that could be reached by grafting mutant intestines under the skin of nude mice. This latter experiment demonstrates that early lethality cannot explain the loss of enteroendocrine cells. Importantly, the expression of NeuroD is lost in the ngn3^-/- intestine, whereas Math1-expressing cells are still detected in the crypts and the villi. Therefore, we propose that ngn3 has an early function in the development of the intestinal endocrine lineage, before the cells exit the cell cycle. ngn3 probably acts in a regulatory cascade downstream of or parallel to Math1 and upstream of NeuroD, and would thus be required for the endocrine fate commitment of a Math1⁺ multipotent progenitor of the secretory lineages (Figure 8A). This hypothesis is supported further by the lack of -galactosidase protein in the embryonic intestine of ngn3 promoter::LacZ mice in a ngn3^-/- background, which suggests that endocrine progenitors are lacking in the absence of ngn3. Similar observations have been made in the pancreas (M.Jenny and G.Gradwohl, unpublished results). One can therefore hypothesize that the development of the enteroendocrine lineage is blocked at the Math1-expressing progenitor stage in ngn3^-/- intestine. However, we cannot completely rule out that some endocrine progenitors are generated but then fail to proliferate. One other possible explanation for the failure of LacZ transgene expression in an ngn3^-/- background could be that ngn3 gene transcription is auto-regulated. We find that the latter hypothesis is unlikely since -galactosidase⁺ progenitors can be detected in the ngn3-deficient gastric epithelium (data not shown). The common requirement for ngn3 function in the generation of CCK-, secretin-, gastrin-, GIP-, GLP-, PYY-, serotonin- and somatostatin-producing intestinal cell types suggests that these enteroendocrine lineages originate from a common endocrine progenitor. However, we cannot rule out that the different endocrine cells arise from distinct progenitors that each express ngn3. We are inclined to believe in the possibility that ngn3 acts to specify a common intestinal endocrine progenitor, as several studies have noted co-expression of certain combinations of intestinal hormones (Lopez et al., 1995; Upchurch et al., 1996; Rindi et al., 1999).

Although none of the other main intestinal cytotypes displayed altered differentiation characteristics, an interesting observation is that we found an increase in the number of goblet cells in ngn3^-/- small intestine at birth. This suggests, but does not prove the existence of a bipotential progenitor (endocrine/goblet). In the absence of ngn3, endocrine commitment is no longer possible and the cells would then adopt the alternative goblet fate. This hypothesis that enteroendocrine and goblet cells might share part of their developmental program is supported further by the increase of goblet cells observed in Hes1 knock-out mice (Jensen et al., 2000b) and the reported switch of pure enteroendocrine cells to a more goblet-like phenotype in Crohn's disease (Poulsom et al., 1993). One alternative explanation could also be that the increase in goblet cells is due to their filling up of a compartment that would otherwise be filled by an expanding population of endocrine cells. The high number of goblet cells in the colon (major cell type) could explain why such an increase was not observed in the colonic crypts or suggest that goblet cell regulation might be different in the colon and the small intestine.

Thus, the data presented here provide evidence that ngn3 is a determination gene which controls an endocrine fate decision in multipotent intestinal progenitors. In addition, our results demonstrate that ngn3 specifies the intestinal endocrine lineage in a cell-autonomous manner since we proved that enteroendocrine cells derive from ngn3-expressing cells. Other as yet unidentified bHLH genes might specify the fate of the other intestinal cytotypes. Our results are consistent with a previous report suggesting that the Notch−Hes signaling system is involved in endodermal endocrine fate determination in the developing gut (Jensen et al., 2000b). Indeed, Jensen et al. observed an increase of all endocrine populations and an up-regulation of ngn3 and NeuroD in the gut endoderm of mice lacking the bHLH repressor gene Hes1. Together with our findings, one can hypothesize that the positive regulation of ngn3 on enteroendocrine cells is antagonized by Hes-1 in neighboring cells upon the activation of the Notch receptor by Delta ligand and subsequent up-regulation of Hes1.

The intestinal and pancreatic endocrine phenotype of the ngn3 mutation prompted us to analyze endocrine differentiation in the stomach, another endodermally derived tissue containing a number of different endocrine cell types (Solcia et al., 2000). We showed that gastrin- and somatostatin-producing G and D cell differentiation is impaired in the stomach of mice lacking ngn3 (Figure 8B). However, we observed that a significant number of serotonin-expressing enterochromaffin EC cells were still present in the ngn3^-/- gastric epithelium. While this manuscript was in preparation, Lee et al. (2002) reported a similar observation and showed in addition that ngn3 is important for the maintenance of gastric epithelial cell identity. Our studies extend these findings by demonstrating that the differentiation of two additional gastric endocrine cell types is not affected in ngn3 mutant mice: the X/A cells producing ghrelin (Date et al., 2000), a novel growth hormone secretagogue and anorexigenic peptide (Wang et al., 2002), and the ECL cells secreting histamine which play a key role in the regulation of gastrin-stimulated acid secretion (Chen et al., 1999). ECL cells are not detected at birth when ngn3 mutants die; therefore, the differentiation of this particular cell type could only be studied in rescued intestines. However, the persistence of the ngn3 promoter-driven -galactosidase in all the differentiated gastric endocrine cell types studied suggests that although ngn3 does not control endocrine differentiation of serotonin- (EC), grehlin- (X/A) and histamine- (ECL) producing cells, they derive from ngn3-expressing progenitors. Interestingly, serotonin-producing cells are present both in the small intestine and in the antral mucosa of the stomach. Although these cells secrete the same hormone in both tissues and derive from ngn3-expressing progenitors, they are differentially affected in the gut (lost) and the stomach (retained) in the absence of ngn3. Thus the differentiation of similar endocrine cell types is controlled by different genetic programs in the gut and the stomach. Similar observations have been made in the ventral telencephalon where there are some GABAergic neurons which express the bHLH transcription factor Mash1 which are Mash1 independent while others are dependent (Casarosa et al., 1999).

In conclusion, our results indicate that ngn3 is expressed specifically in endocrine progenitors of the developing and adult intestine as well as of the stomach. Our loss-of-function experiment demonstrates that ngn3 is required for the specification of the endocrine fate in multipotent intestinal progenitors (Figure 8A) and that endocrine differentiation is completely impaired in the mutant intestine. In contrast, gastric endocrine differentiation does not rely entirely on ngn3 function. As a consequence, at least two different gastric endocrine lineage pathways exist, one dependent on and the other independent of ngn3, and controlled by an as yet unidentified gene (Figure 8B) Taken together, the work described here and our previous data demonstrate that endocrine cell fate determination is similar in the intestine and the pancreas, but different in the stomach. These findings on the common mechanisms occurring in the gut and the pancreas may initiate the development of novel strategies to derive insulin-secreting cells from intestinal endocrine progenitors.

Materials and methods

Transgene construction and ngn3-deficient mice

To generate the ngn3 promoter::LacZ construct, a 6.86 kb XbaI−XhoI fragment (6696 bp of 5' genomic and untranslated region sequences and 176 bp of ngn3 coding region) of mouse ngn3 genomic DNA (Gradwohl et al., 2000) was cloned upstream of the IRES-NLS-LacZ-pA sequence in pBS-INL vector (Fode et al., 2000), resulting in plasmid pngn3(6.8)-INL. The 10.6 kb ngn3−LacZ insert was released by NotI digestion and microinjected into murine oocyte pronuclei, and three independent transgenic lines were generated and maintained by crossing into a CD1 outbred background. Transgenic progeny were identified by PCR using primer in the LacZ gene, VW233 5'-GCACATCCCCCTTTCGCCAG CTGGCGTAAT-3' and VW234 5'-CGCGTCTGGCCTTCCTGTAGCC AGCTTTCA-3'. Ngn3^+/- animals were generated as described previously (Gradwohl et al., 2000).

Multiplex RT−PCR

Multiplex RT−PCR was performed on dissected P1 duodenum, jejunum and colon as described by Jensen et al. (2000b). TBP (encoding TATA-binding protein) was co-amplified as an internal standard.

-galactosidase and lactase activity detection

Tissues were fixed for 10−20 min at room temperature in 0.2% glutaraldehyde, 5 mM EGTA pH 7.3, 2 mM MgCl₂ in 0.1 M sodium phosphate pH 7.3, washed three times for 10 min in LacZ wash buffer (2 mM MgCl₂ in 0.1 M sodium phosphate pH 7.3, 0.02% NP-40) and whole mount staining was performed in a solution containing 1 mg/ml X-Gal, 5 mM potassium ferrocyanide and 5 mM potassium ferricyanide in LacZ wash buffer at 37°C for 2−4 h. After staining, samples were washed in phosphate-buffered saline (PBS), post-fixed with Bouin and processed for wax sections. For double labeling experiments, X-Gal stainings were realized directly on paraformaldehyde-fixed frozen tissue sections, followed by immunostaining. Enterocyte lactase activity was revealed as detailed in Jost et al. (1998).

RNA in situ hybridization (ISH), immunohistochemistry and immunofluorescence

RNA ISH experiments were performed on frozen paraformaldehyde-fixed tissues sections as described previously (Cau et al., 1997; Gradwohl et al., 2000). In some cases, ISH was followed by immunostaining. The following cRNA probes were used: ngn3 (Gradwohl et al., 2000), math1 (kindly provided by R.Kageyama, Kyoto University, Japan) and NeuroD (Fode et al., 1998). Immunohistochemistry and immunofluoresence were performed on paraffin and cryosections as described previously (Cau et al., 1997). The following antibodies were used: rabbit anti-chromogranin A at 1:300 (Diasorin), guinea pig anti-ngn3 at 1:1000 (kindly provided by M.German, UCSF, San Francisco, CA), rabbit anti--galactosidase 1:500 (ICN), rabbit anti-somatostatin at 1:200 (Dako), rabbit anti-CCK/gastrin at 1:750 (INSERM U.45, 8E), rabbit anti-serotonin at 1:5000 (Incstar) or mouse anti-serotonin at 1:50 (Dako), rabbit anti-secretin at 1:2000 (kindly provided by P.Robberecht, Brussels, Belgium, 13/4), rabbit anti-GIP at 1:500 (INSERM U.45, 0.59A), rabbit anti-PYY at 1:1000 (INSERM U-45, A4D), rabbit anti-ghrelin at 1:2000 (kindly provided by C.Tomasetto, IGBMC, Strasbourg, France), rabbit anti-GLP at 1:1000 (INSERM U.45, 199D), rabbit anti-HDC at 1:800 (Progen) and mouse anti-proliferating cell nuclear antigen (PCNA) at 1:100 (Dako). Secondary antibodies used were: Alexa 488 anti-rabbit at 1:500 (Molecular probes), Cy3 anti-guinea pig at 1:500 (Jackson immunoresearch), and peroxidase-coupled anti-rabbit, anti-guinea-pig and anti-mouse at 1:200 (Vector Laboratories). BrdU incorporation and detection experiments were as described by Parras et al. (2002). Goblet cells were stained for mucin using the PAS reaction.

Grafting experiments

To analyze ngn3 knock-out tissues beyond postnatal day 1, 12 day fetal intestine and stomach dissected out from ngn3^-/- and wild-type mice were grafted under the skin of nude mice. Fetal intestines were subdivided into four parts corresponding to the presumptive duodenum, jejunum, ileum and colon. The intestinal grafts were recovered after 4 weeks, at a stage corresponding to weaning. Grafted stomachs were recovered after 2−3 weeks, because the gastric mucosa suffered from longer development. A mean of three grafts developed from each proximo-distal intestinal segment and from the stomach were analyzed.

Acknowledgements

We thank Dr Andrew Leiter for critical reading of the initial version of the manuscript, Christiane Arnold, Caroline Daurat and Viviane Hauer for excellent technical assistance, Drs Catherine Tomasetto, Rolf Hakanson and Lo Persson for helpful discussions, Drs Mike German and Ryochiro Kageyama for providing us with the ngn3 antibody and the Math1 cDNA, respectively, and Marianne LeMeur, Elisabeth Metzger and all the members of the mouse facilities of the IGBMC for their help in the generation of the ngn3-promoter::LacZ transgenic mice. This work was supported by funds from the Institut National de la Santé et de la Recherche Médicale (INSERM; AVENIR grant to G.G.), the Ministère de la Recherche et de la Technologie (ACI Biologie du développement et Physiologie Intégrative 2000) and grants from the Association pour la Recherche sur le Cancer (ARC), the Juvenile Diabetes Research Fundation (JDRF; 4-2001-434) and National Institutes of Health (NIH; 1U19-DK61244-01) to G.G. M.J. is a recipient of PhD studentship from the Ministère de la Recherche et de la Technologie.

References

Apelqvist A, Li H, Sommer L, Beatus P, Anderson DJ, Honjo T, Hrabe de Angelis M, Lendahl U and Edlund H (1999) Notch signalling controls pancreatic cell differentiation. Nature, 400, 877–881. | Article |

Ben Arie N, Hassan BA, Bermingham NA, Malicki DM, Armstrong D, Matzuk M, Bellen HJ and Zoghbi HY (2000) Functional conservation of atonal and Math1 in the CNS and PNS. Development, 127, 1039–1048. | PubMed | ISI | ChemPort |

Bermingham NA, Hassan BA, Price SD, Vollrath MA, Ben Arie N, Eatock RA, Bellen HJ, Lysakowski A and Zoghbi HY (1999) Math1: an essential gene for the generation of inner ear hair cells. Science, 284, 1837–1841. | Article | PubMed | ISI | ChemPort |

Bjerknes M and Cheng H (1999) Clonal analysis of mouse intestinal epithelial progenitors. Gastroenterology, 116, 7–14. | Article | PubMed | ISI | ChemPort |

Booth C and Potten CS (2000) Gut instincts: thoughts on intestinal epithelial stem cells. J Clin Invest, 105, 1493–1499. | PubMed | ISI | ChemPort |

Casarosa S, Fode C and Guillemot F (1999) Mash1 regulates neurogenesis in the ventral telencephalon. Development, 126, 525–534. | PubMed | ISI | ChemPort |

Cau E, Gradwohl G, Fode C and Guillemot F (1997) Mash1 activates a cascade of bHLH regulators in olfactory neuron progenitors. Development, 124, 1611–1621. | PubMed | ISI | ChemPort |

Chen D, Zhao CM, Lindstrom E and Hakanson R (1999) Rat stomach ECL cells up-date of biology and physiology. Gen Pharmacol, 32, 413–422. | Article | PubMed | ISI | ChemPort |

Date Y, Kojima M, Hosoda H, Sawaguchi A, Mondal MS, Suganuma T, Matsukura S, Kangawa K and Nakazato M (2000) Ghrelin, a novel growth hormone-releasing acylated peptide, is synthesized in a distinct endocrine cell type in the gastrointestinal tracts of rats and humans. Endocrinology, 141, 4255–4261. | Article | PubMed | ISI | ChemPort |

Fode C, Gradwohl G, Morin X, Dierich A, LeMeur M, Goridis C and Guillemot F (1998) The bHLH protein NEUROGENIN 2 is a determination factor for epibranchial placode-derived sensory neurons. Neuron, 20, 483–494. | Article | PubMed | ISI | ChemPort |

Fode C, Ma Q, Casarosa S, Ang SL, Anderson DJ and Guillemot F (2000) A role for neural determination genes in specifying the dorsoventral identity of telencephalic neurons. Genes Dev, 14, 67–80. | PubMed | ISI | ChemPort |

Gordon JI, Schmidt GH and Roth KA (1992) Studies of intestinal stem cells using normal, chimeric and transgenic mice. FASEB J, 6, 3039–3050. | PubMed | ISI | ChemPort |

Gradwohl G, Dierich A, LeMeur M and Guillemot F (2000) neurogenin3 is required for the development of the four endocrine cell lineages of the pancreas. Proc Natl Acad Sci USA, 97, 1607–1611. | Article | PubMed | ChemPort |

Grapin-Botton A and Melton DA (2000) Endoderm development: from patterning to organogenesis. Trends Genet, 16, 124–130. | Article | PubMed | ISI | ChemPort |

Gu G, Dubauskaite J and Melton DA (2002) Direct evidence for the pancreatic lineage: NGN3⁺ cells are islet progenitors and are distinct from duct progenitors. Development, 129, 2447–2457. | PubMed | ISI | ChemPort |

Hassan BA and Bellen HJ (2000) Doing the MATH: is the mouse a good model for fly development. Genes Dev, 14, 1852–1865. | PubMed | ISI | ChemPort |

Helms AW and Johnson JE (1998) Progenitors of dorsal commissural interneurons are defined by MATH1 expression. Development, 125, 919–928. | PubMed | ISI | ChemPort |

Hocker M and Wiedenmann B (1998) Molecular mechanisms of enteroendocrine differentiation. Ann NY Acad Sci, 859, 160–174. | PubMed | ChemPort |

Jensen J, Heller RS, Funder-Nielsen T, Pedersen EE, Lindsell C, Weinmaster G, Madsen OD and Serup P (2000a) Independent development of pancreatic - and -cells from neurogenin3-expressing precursors: a role for the notch pathway in repression of premature differentiation. Diabetes, 49, 163–176. | PubMed | ISI | ChemPort |

Jensen J et al. (2000b) Control of endodermal endocrine development by Hes-1. Nat Genet, 24, 36–44. | Article | PubMed | ISI | ChemPort |

Jost B, Duluc I, Vilotte JL and Freund JN (1998) Lactase is unchanged in suckling mice fed with lactose-free milk. Gastroenterol Clin Biol, 22, 863–867. | PubMed | ISI | ChemPort |

Karam SM and Leblond CP (1993) Dynamics of epithelial cells in the corpus of the mouse stomach. V. Behavior of entero-endocrine and caveolated cells: general conclusions on cell kinetics in the oxyntic epithelium. Anat Rec, 236, 333–340. | PubMed | ISI | ChemPort |

Karam SM, Li Q and Gordon JI (1997) Gastric epithelial morphogenesis in normal and transgenic mice. Am J Physiol, 272, G1209–G1220. | PubMed | ISI | ChemPort |

Larsson LI, St-Onge L, Hougaard DM, Sosa-Pineda B and Gruss P (1998) Pax 4 and 6 regulate gastrointestinal endocrine cell development. Mech Dev, 79, 153–159. | Article | PubMed | ISI | ChemPort |

Lee CS, Perreault N, Brestelli JE and Kaestner KH (2002) Neurogenin 3 is essential for the proper specification of gastric enteroendocrine cells and the maintenance of gastric epithelial cell identity. Genes Dev, 16, 1488–1497. | Article | PubMed | ISI | ChemPort |

Lee JC, Smith SB, Watada H, Lin J, Scheel D, Wang J, Mirmira RG and German MS (2001) Regulation of the pancreatic pro-endocrine gene neurogenin3. Diabetes, 50, 928–936. | PubMed | ISI | ChemPort |

Lopez MJ, Upchurch BH, Rindi G and Leiter AB (1995) Studies in transgenic mice reveal potential relationships between secretin-producing cells and other endocrine cell types. J Biol Chem, 270, 885–891. | Article | PubMed | ISI | ChemPort |

Ma Q, Fode C, Guillemot F and Anderson DJ (1999) Neurogenin1 and neurogenin2 control two distinct waves of neurogenesis in developing dorsal root ganglia. Genes Dev, 13, 1717–1728. | PubMed | ISI | ChemPort |

Marshman E, Booth C and Potten CS (2002) The intestinal epithelial stem cell. Bioessays, 24, 91–98. | Article | PubMed | ISI |

Massari ME and Murre C (2000) Helix–loop–helix proteins: regulators of transcription in eucaryotic organisms. Mol Cell Biol, 20, 429–440. | Article | PubMed | ISI | ChemPort |

Mutoh H, Naya FJ, Tsai MJ and Leiter AB (1998) The basic helix–loop–helix protein BETA2 interacts with p300 to coordinate differentiation of secretin-expressing enteroendocrine cells. Genes Dev, 12, 820–830. | PubMed | ISI | ChemPort |

Naya FJ, Huang HP, Qiu Y, Mutoh H, DeMayo FJ, Leiter AB and Tsai MJ (1997) Diabetes, defective pancreatic morphogenesis and abnormal enteroendocrine differentiation in BETA2/neuroD-deficient mice. Genes Dev, 11, 2323–2334. | PubMed | ISI | ChemPort |

Norlen P et al. (2001) Cell-specific processing of chromogranin A in endocrine cells of the rat stomach. J Histochem Cytochem, 49, 9–18. | PubMed | ISI | ChemPort |

Parras CM, Schuurmans C, Scardigli R, Kim J, Anderson DJ and Guillemot F (2002) Divergent functions of the proneural genes Mash1 and Ngn2 in the specification of neuronal subtype identity. Genes Dev, 16, 324–338. | Article | PubMed | ISI | ChemPort |

Poulsom R, Chinery R, Sarraf C, Van Noorden S, Stamp GW, Lalani EN, Elia G and Wright NA (1993) Trefoil peptide gene expression in small intestinal Crohn's disease and dietary adaptation. J Clin Gastroenterol, 17, S78–S91. | PubMed | ISI |

Ratineau C, Petry MW, Mutoh H and Leiter AB (2002) Cyclin D1 represses the basic helix–loop–helix transcription factor, BETA2/NeuroD. J Biol Chem, 277, 8847–8853. | Article | PubMed | ISI | ChemPort |

Rindi G, Ratineau C, Ronco A, Candusso ME, Tsai M and Leiter AB (1999) Targeted ablation of secretin-producing cells in transgenic mice reveals a common differentiation pathway with multiple enteroendocrine cell lineages in the small intestine. Development, 126, 4149–4156. | PubMed | ISI | ChemPort |

Roth KA and Gordon JI (1990) Spatial differentiation of the intestinal epithelium: analysis of enteroendocrine cells containing immuno reactive serotonin, secretin and substance P in normal and transgenic mice. Proc Natl Acad Sci USA, 87, 6408–6412. | PubMed | ChemPort |

Roth KA, Hertz JM and Gordon JI (1990) Mapping enteroendocrine cell populations in transgenic mice reveals an unexpected degree of complexity in cellular differentiation within the gastrointestinal tract. J Cell Biol, 110, 1791–1801. | PubMed | ISI | ChemPort |

Schwitzgebel VM, Scheel DW, Conners JR, Kalamaras J, Lee JE, Anderson DJ, Sussel L, Johnson JD and German MS (2000) Expression of neurogenin3 reveals an islet cell precursor population in the pancreas. Development, 127, 3533–3542. | PubMed | ISI | ChemPort |

Skipper M and Lewis J (2000) Getting to the guts of enteroendocrine differentiation. Nat Genet, 24, 3–4. | Article | PubMed | ISI | ChemPort |

Smith RL, Geller AI, Escudero KW and Wilcox CL (1995) Long-term expression in sensory neurons in tissue culture from herpes simplex virus type 1 (HSV-1) promoters in an HSV-1-derived vector. J Virol, 69, 4593–4599. | PubMed | ISI | ChemPort |

Solcia E, Rindi G, Buffa R, Fiocca R and Capella C (2000) Gastric endocrine cells: types, function and growth. Regul Pept, 93, 31–35. | Article | PubMed | ISI | ChemPort |

Upchurch BH, Fung BP, Rindi G, Ronco A and Leiter AB (1996) Peptide YY expression is an early event in colonic endocrine cell differentiation: evidence from normal and transgenic mice. Development, 122, 1157–1163. | PubMed | ISI | ChemPort |

Wang G, Lee HM, Englander E and Greeley GH,Jr (2002) Ghrelin—not just another stomach hormone. Regul Pept, 105, 75–81. | Article | PubMed | ISI | ChemPort |

Yamaoka T and Itakura M (1999) Development of pancreatic islets. Int J Mol Med, 3, 247–261. | PubMed | ISI | ChemPort |

Yang Q, Bermingham NA, Finegold MJ and Zoghbi HY (2001) Requirement of Math1 for secretory cell lineage commitment in the mouse intestine. Science, 294, 2155–2158. | Article | PubMed | ISI | ChemPort |