The importance of mesenchymal-epithelial interactions in the proliferation of pancreatic progenitor cells is well established. Here, we provide evidence that the mesenchyme also controls the timing of β-cell differentiation. When rat embryonic pancreatic epithelium was cultured without mesenchyme, we found first rapid induction in epithelial progenitor cells of the transcription factor neurogenin3 (Ngn3), a master gene controlling endocrine cell-fate decisions in progenitor cells; then β-cell differentiation occurred. In the presence of mesenchyme, Ngn3 induction was delayed, and few β-cells developed. This effect of the mesenchyme on Ngn3 induction was mediated by cell-cell contacts and required a functional Notch pathway. We then showed that associating Ngn3-expressing epithelial cells with mesenchyme resulted in poor β-cell development via a mechanism mediated by soluble factors. Thus, in addition to its effect upstream of Ngn3, the mesenchyme regulated β-cell differentiation downstream of Ngn3. In conclusion, these data indicate that the mesenchyme controls the timing of β-cell differentiation both upstream and downstream of Ngn3.

The final size of a specific tissue is largely dependent on the number of stem-cell divisions before and during differentiation. If differentiation occurs too early, the stem-cell pool remains small, and the final number of differentiated cells is abnormally low. For instance, in the brain of mice deficient in specific members of the basic helix-loop-helix Hairy and Enhancer-of-Split family (Hes), virtually the entire pool of neural stem cells differentiates simultaneously and prematurely into neurons, without generating the later-born cell types (1). Similarly, Hes1-deficient mice exhibit severe pancreatic hypoplasia. It is caused by depletion of pancreatic epithelial precursors due to accelerated differentiation generating postmitotic endocrine cells expressing glucagon (2). No data are available on the mechanisms by which extracellular signals regulate the combined timing of both progenitor-cell proliferation and cell differentiation. Here, we used the pancreas as a model to investigate this question.

The pancreas is an ideal model for studying the coordinate control between proliferation and differentiation of epithelial progenitor cells and for understanding how these two processes are kept in balance during development. The mature pancreas contains two types of tissue: endocrine islets composed of cells that produce hormones such as insulin (β-cells) and glucagon (α-cells) and exocrine tissue composed of acinar cells that produce enzymes (e.g., carboxypeptidase) secreted via pancreatic ducts into the intestine. The pancreas originates from the dorsal and ventral regions of the foregut endoderm directly posterior to the stomach. Pancreas development is successively controlled by permissive signals derived from adjacent mesodermal structures such as the notochord and the dorsal aorta (3). Subsequently, the mesenchyme condenses around the underlying committed endoderm, and the epithelial buds grow larger (4). Inductive signals originating in the mesenchyme play an essential role in the proliferation of pancreatic epithelial cells (5,6), and recent evidence indicates that a member of the fibroblast growth factor (FGF) family, FGF10, is pivotal in the proliferation of epithelial progenitor cells (7).

A few years ago, we developed a simple in vitro model to further study the role of the mesenchyme in pancreatic differentiation. When rat embryonic pancreatic epithelium was cultured without mesenchyme, endocrine cell differentiation occurred (8,9), supporting previous in vivo data (10). In addition, the number of endocrine cells that developed in cultured epithelium during a 7-day period was greater without than with mesenchyme (8). However, the mechanisms by which the mesenchyme controlled endocrine-cell differentiation remained unknown.

In the last few years, important findings have shed light on the processes controlling pancreatic endocrine-cell development. Studies of genetically engineered mice identified a hierarchy of transcription factors regulating pancreas organogenesis and islet-cell differentiation (11,12). The endodermal region committed to a pancreatic fate expresses the transcription factor pancreatic duodenal homeobox-1 (13,14). The basic helix-loop-helix factor neurogenin3 (Ngn3) is then expressed in epithelial pancreatic progenitor cells before endocrine differentiation (15). Ngn3 is necessary for pancreatic endocrine cell development, and Ngn3-deficient mice lack pancreatic endocrine cells (16). Lineage tracing experiments have also provided direct evidence that Ngn3-expressing cells are islet progenitors (17). Thus, Ngn3 represents a marker of choice to follow the onset of pancreatic endocrine cell differentiation.

In this study, we performed in vitro experiments to determine whether the mesenchyme controlled the timing of endocrine cell differentiation, in addition to its effect on pancreatic progenitor cell proliferation. We found that the mesenchyme acted both upstream and downstream of Ngn3. By acting upstream of Ngn3 and via direct cell-cell contact, the mesenchyme delayed Ngn3 induction. By acting downstream of Ngn3 and via secreted soluble factors, the mesenchyme repressed the differentiation of Ngn3-positive cells into β-cells.

Pregnant Wistar rats were purchased from the Janvier breeding center (CERJ, Le Genest, France). The animals had free access to food pellets and water. The 1st day postcoitum was taken as embryonic day (E) 0.5. Pregnant female rats were killed by CO2 asphyxiation according to the guidelines of the French Animal Care Committee.

Dissection of pancreatic rudiments.

The embryos were harvested on E13.5 and dissected. The dorsal pancreatic bud was dissected as described previously (8,18). Briefly, the stomach, pancreas, and a small portion of the intestine were dissected together; then, either the pancreatic primordium was dissected or the mesenchyme was separated from the pancreatic epithelium as follows: the digestive tract was incubated with 0.5 mg/ml collagenase A (Roche, Meylan, France) at 37°C for 30 min then washed several times with Hank’s balanced salt solution (Invitrogen, Cergy Pontoise, France) at 4°C, and the epithelium was mechanically separated from the surrounding mesenchyme using needles on 0.25% agar, 25% Hank’s balanced salt solution, and 75% RPMI (Gibco) gel in a Petri dish.

Organ culture.

Dorsal pancreatic rudiments with or without their mesenchyme were embedded in 500 μl collagen gel (10% RPMI 10× [Sigma-Aldrich], 80% type I rat-tail collagen [3 mg/ml; Sigma-Aldrich], and 10% sodium bicarbonate in 0.1 mol/l NaOH) in four-well plates, as previously described (8). After gel polymerization, 500 μl RPMI-1640 (Invitrogen) containing 100 units/ml penicillin, 100 μg/ml streptomycin, 10 mmol/l HEPES, 2 mmol/l l-glutamine, 1× nonessential amino acids (Gibco), and 1% heat-inactivated calf serum (Hyclone) was added. Cultures were maintained at 37°C in humidified 95% air 5% CO2. In some experiments, the epithelium was separated from its surrounding mesenchyme then reassociated with two mesenchymes. For coculture experiments, the pancreatic epithelium was separated from the mesenchyme by a filter (size of the pores, 0,4 μm; Millicell, Millipore, France) and cultured in 20 μl of the above-described medium in Terazaki plates (Nunc, France); the epithelium was placed on the filter and the mesenchyme under the filter. Epithelia alone and whole pancreata were cultured on the filters under the same conditions as the controls.

FGF7 (R&D Systems) and epidermal growth factor (EGF) (Sigma-Aldrich) were used at a concentration of 50 ng/ml. The γ-secretase inhibitor XVIII (compound E; Calbiochem) was used as previously described (19) at a concentration of 250 nmol/l.

Immunohistochemistry.

At the end of the culture period, the pancreatic rudiments were photographed, fixed in 10% formalin, pre-embedded in agarose gel (4% of type VII low gelling-temperature agarose [Sigma] in H20), and embedded in paraffin. Sections (4 μm thick) were collected and processed for immunohistochemistry, as described previously (18). The antibodies were used at the following dilutions: mouse anti-human insulin (1/2000; Sigma), rabbit anti–carboxypeptidase A (1/600; Biogenesis). The fluorescent secondary antibodies were fluorescein isothiocyanate anti-rabbit antibody (1/200; Jackson Immunoresearch, Baltimore, MD) and Texas red anti-mouse antibody (1/200; Jackson). Photographs were taken using a fluorescence microscope (Leitz DMRB; Leica) and digitized using a Hamamatsu (Middlesex, NJ) C5810 cooled 3CCD camera.

In situ hybridization.

Tissues were fixed at 4°C in 4% paraformaldehyde in PBS, cryoprotected in 15% sucrose-PBS at 4°C overnight, embedded in 15% sucrose-7.5% gelatin in PBS, and frozen at −50°C in isopentane. Cryosections 14 μm in thickness were prepared. The Ngn3 probe (726 bp) was prepared as previously described (20). Hes1 probe was provided by F. Guillemot (21). Plasmids were linearized and used as templates for synthesizing sense or antisense riboprobes using T7 or SP6 RNA polymerase (Roche), in the presence of digoxygenin-UTP (Roche Diagnostic). In situ hybridization was done as previously described (22), and colorimetric revelation was performed with 5-bromo-4-chloro-3-indolyl phosphate (Promega) and nitroblue tetrazolium (Roche) to obtain a blue precipitate. Photographs were digitized using a Hamamatsu (Middlesex, NJ) C5810 cooled 3CCD camera. No signal was obtained when a sense riboprobe was used.

Quantification.

To quantify the number of Ngn3-positive cells present in E13.5 rat embryonic pancreatic epithelium before culture or after 24 h of culture with or without mesenchyme, tissues (nine for each condition) were fixed and frozen. Cryosections 14 μm in thickness were prepared, and all sections were hybridized using a Ngn3 antisense riboprobe. Ngn3-positive cells were counted on all sections. Statistical significance was determined using Student’s t test.

Expression of Ngn3 in cultured embryonic pancreatic epithelium.

We and others have previously shown that, around midgestation in rodents, the pancreatic mesenchyme is not necessary for normal differentiation of pancreatic progenitor cells into endocrine cells (8,10). As shown in Fig. 1A, insulin-expressing cells were rare at E13.5 in rat pancreas. When immature E13.5 pancreatic epithelium was cultured without its surrounding mesenchyme, β-cell differentiation occurred, and the number of β-cells increased, as shown by in situ hybridization (Fig. 1A–E). Here, we investigated the same developmental stages in vitro, examining the expression of Ngn3, the earliest transcription factor specifically expressed by pancreatic progenitor cells and necessary for pancreatic endocrine-cell differentiation (16,17). On E13.5, very few cells expressed Ngn3 in the rat pancreatic epithelium (Fig. 1F). After 1 day of culture, Ngn3 expression increased dramatically, reaching a peak followed by a plateau until day 3 and then by a decrease on days 5 and 7 (Fig. 1F–J). Thus, in vitro, in the absence of mesenchyme, the pattern of Ngn3 expression resembles that found in vivo. Ngn3 expression precedes insulin expression and is transient.

Signals derived from the mesenchyme control Ngn3 expression in the pancreatic epithelium.

In vitro, in the absence of mesenchyme, a large number of β-cells developed (Fig. 2A). On the contrary, when entire pancreas (epithelium with mesenchyme) was cultured under the same conditions, very few insulin-positive cells developed (Fig. 2B). At the same time, in the presence of mesenchyme, acinar cell development is favored, and a large number of cells express carboxypeptidase A (Fig. 2B). To determine whether the difference in terms of β-cell development between epithelium grown without or with mesenchyme was dependent on the experimental procedure used to separate the epithelium from its mesenchyme, we dissected epithelium then placed it directly in contact with mesenchyme. As shown in Fig. 2C, when this reassociated tissue was cultured for 7 days, no β-cells developed as observed with the culture of an entire pancreas (Fig. 2B).

The next step was to determine how the mesenchyme controls β-cell development. As shown above, Ngn3 expression was activated in vitro in the absence of mesenchyme (Fig. 1). We therefore investigated whether the mesenchyme perturbs β-cell differentiation upstream or downstream of Ngn3. Pancreatic epithelia were cultured with or without mesenchyme for 1 day, and Ngn3 expression was examined by in situ hybridization. On E13.5 before culture, a few Ngn3-positive cells (15 ± 6 Ngn3-expressing cells per epithelium) were detected in the pancreatic epithelium. After 1 day of culturing without mesenchyme, the absolute number of Ngn3-positive cells was increased 26-fold (Fig. 3A, a and b, and B for quantification). In contrast, after 1 day of culture with mesenchyme, the increase in absolute Ngn3-positive cells was of only 2.4-fold and thus 11 times smaller than the increase noted without mesenchyme (Fig. 3A, ac, and B). Interestingly, Ngn3-positive cells that developed in the absence of mesenchyme did not incorporate bromodeoxyuridine (data not shown). This suggests that the increase in the number of Ngn3-positive cells observed in epithelia cultured during 24 h in the absence of mesenchyme is due to differentiation of Ngn3-negative cells into Ngn3-positive cells and that the mesenchyme perturbs the activation of Ngn3 expression in pre-Ngn3 cells. Taken together, our data indicate that pancreatic mesenchyme in E13.5 rats modulates Ngn3 expression and consequently β-cell development. The next step was to define whether Ngn3 repression by the mesenchyme was mediated by direct contact between epithelial and mesenchymal cells or by soluble factors.

The effect of the mesenchyme on Ngn3 is mediated by direct cell-cell contact and requires an intact Notch pathway.

To determine the nature of the mesenchymal signals that control Ngn3 expression, we first performed experiments involving reassociation of epithelium and mesenchyme. Epithelia from E13.5 pancreata were reassociated with mesenchymes prepared from E13.5 pancreata. Ngn3-positive epithelial cells were numerous after 1 day of culturing without mesenchyme; in contrast, they were scarce 1 day after mesenchyme reassociation (Fig. 3A, b and d, and B for quantification), and no β-cells were present after 7 days (Fig. 2C). We next compared Ngn3 expression in epithelium cultured in direct contact with mesenchyme or separated from mesenchyme by a filter that precluded direct cell-cell contact but allowed interactions mediated by soluble factors. When epithelia and mesenchyme were cultured on two opposite sides of a filter, Ngn3 expression was activated after 1 day of culture, as was the case when epithelium was cultured alone, whereas Ngn3 expression was not detected when epithelium and mesenchyme were in direct contact (Fig. 4A, compare a and b with c). This result strongly suggests that direct cell-cell contact is necessary for the repressive effect from the mesenchyme. However, we cannot completely exclude that the repressive effect of the mesenchyme is due to soluble factors that only travel at a short distance.

In addition, when epithelium was cultured for 24 h with EGF or FGF7, two soluble factors previously shown to decrease endocrine cell differentiation to the detriment of progenitor cell proliferation (23,24), epithelial cell proliferation was induced (data not shown), but induction of Ngn3 expression was not modified (Fig. 4B, ac). Taken together, these results strongly suggest that the repressive effect of the mesenchyme on Ngn3 expression requires cell-cell contact between epithelium and mesenchyme.

Ngn3 is known to be tightly controlled by the Notch signaling pathway, which determines which cells of the developing pancreas will activate the endocrine differentiation program (2,15). Here, we used the γ-secretase inhibitor XVIII to block the Notch activation (19) and analyzed the effect of the presence of the mesenchyme on Ngn3 expression. We cultured entire E13.5 pancreata (epithelium+mesenchyme) for 1 day in the presence of γ-secretase inhibitor XVIII. Under such conditions, a dramatic decrease in Hes1 mRNA, a direct target of activated Notch, was observed (data not shown). At the same time, Ngn3 expression detected by in situ hybridization dramatically increased in pancreata treated with the inhibitor compared with nontreated pancreata (Fig. 5B and C). Taken together, such data indicate that an intact Notch pathway is necessary for the effect of the mesenchyme on Ngn3. To further investigate the role of the mesenchyme in regulating the Notch signaling, we examined Hes1 expression by in situ hybridization in the pancreatic epithelium cultured in the presence or absence of mesenchyme. Before culturing, Hes1 expression was distributed in all epithelial cells and was lightly expressed in the mesenchyme (Fig. 6A). After 1 day of culturing of whole pancreata, Hes1 expression remained broadly expressed in the epithelium (Fig. 6B). On the contrary, when epithelium was cultured in the absence of mesenchyme for 24 h, Hes1 was found only in a few isolated epithelial cells (Fig. 6C). This result indicates that the mesenchyme maintains Hes1 expression in the pancreatic epithelium.

Signals derived from the mesenchyme are also involved in β-cell development downstream of Ngn3.

As previously shown, in epithelia grown in the absence of mesenchyme, Ngn3 induction is followed by β-cell differentiation (Fig. 5, A and D). On the other hand, neither Ngn3 nor insulin was induced in the presence of mesenchyme, whereas acinar tissue that stained positive for carboxypeptidase A developed (Fig. 5, B and E). However, we found that when whole pancreata (epithelium+mesenchyme) were treated with γ-secretase inhibitor XVIII while Ngn3 was induced, insulin-positive cells did not develop (Fig. 5, C and F). This suggests that in addition to its action upstream of Ngn3, the mesenchyme acted downstream of Ngn3. To further support this point, we cultured embryonic pancreatic epithelia for 1 day to activate Ngn3 expression. On day 1, these epithelia were reassociated with fresh mesenchymes dissected from E13.5 rat pancreata and cultured for 2 or 6 additional days. β-Cells developed normally in epithelia not reassociated with mesenchyme (Fig. 7A, a and c) but did not develop when epithelia expressing Ngn3 were reassociated with mesenchymes after 1 day of culturing (Fig. 7A, b and d). This result shows that, in addition to its effect upstream of Ngn3, the mesenchyme also acts downstream of Ngn3 to inhibit β-cell differentiation.

We next investigated whether Ngn3+ cells that did not differentiate into β-cells upon mesenchyme reassociation at day 1 developed into glucagon-expressing cells. We found that reassociated tissues never developed glucagon-expressing cells (data not shown). We next investigated whether Ngn3+ cells were maintained as endocrine progenitors expressing NeuroD, a direct target of Ngn3, or Pax4. Real-time PCR experiments showed that expression of NeuroD and Pax4 were dramatically reduced in epithelium reassociated with mesenchyme on day 1 and cultured for 2 additional days compared with epithelium alone cultured for 3 days (data not shown). Altogether, such findings show that the mesenchyme exerts a strong repressive effect immediately downstream of Ngn3.

To define the type of mesenchymal signals that control β-cell development downstream of Ngn3, E13.5 epithelia were cultured either alone or with mesenchyme, either in direct contact or separated by a filter. When epithelium was cultured alone, Ngn3 induction was followed by β-cell development (Figs. 4A, a, and 7B, a). When epithelium and mesenchyme were separated by a filter that only permits interactions mediated by soluble factors while Ngn3 was induced, β-cells did not develop (Figs. 4A, b and 7B, b). These results indicate that the mesenchyme modulates β-cell differentiation downstream of Ngn3 via soluble factors.

Induction of Ngn3 expression is accelerated in the absence of mesenchyme.

Having demonstrated that the mesenchyme controlled β-cell differentiation in vitro by acting both upstream and downstream of Ngn3 on E13.5 in rats, we compared these in vitro data with the in vivo pattern of expression of Ngn3. As shown in Fig. 8A–E, Ngn3 expression in vivo peaked between E16.5 and E18.5 in rat and decreased thereafter. This timing resembled that observed in vitro when epithelium was grown with its surrounding mesenchyme (Fig. 8F–J). On the other hand, in the absence of mesenchyme, Ngn3 induction was already apparent after 1 day of culture (Fig. 8K–O). When we examined earlier time points, we found that Ngn3 expression was induced as early as 6 h after mesenchyme removal (data not shown). These data indicate that the mesenchyme delays the timing of Ngn3-positive cell development.

In the present work, we report evidence that in addition to its role on the proliferation of pancreatic progenitor cells, the mesenchyme is crucial in controlling the timing of pancreatic β-cell differentiation. When we cultured rat embryonic pancreatic epithelium in the absence of its surrounding mesenchyme, we found that Ngn3 expression was turned on rapidly and then turned off a few days later, in keeping with the reported in vivo pattern (15,25). Our filter-separation experiments indicate that the mesenchyme-induced delay in Ngn3 induction requires direct contact between the epithelium and the mesenchyme. This result fits in with recent data indicating that direct cell-cell contact between epithelial and mesenchymal cells suppresses β-cell formation (26). We also found that Ngn3 expression occurred far earlier without mesenchyme, within a few hours compared with 3 days with mesenchyme either in vitro or in vivo. This acceleration in Ngn3 induction resembles the pancreatic phenotype of mice that lack the basic helix-loop-helix gene Hes1. In such mice, accelerated differentiation into postmitotic pancreatic endocrine cells results in depletion of the pool of pancreatic progenitor cells and in severe pancreatic hypoplasia (2). Thus, the mesenchyme could delay Ngn3 induction by activating the Notch pathway, and one simple hypothesis would be that the mesenchyme could be a source of Notch ligands. However, this hypothesis is not probable for the following reasons. There are five known ligands of the Notch receptors: Jagged-1; Jagged-2; and Delta-like-1, -3, and -4 (dll-1, -3, and -4). Recent data have shown that in rodents at E12–E14, Jagged-1 (25,27), Jagged-2 (2,28), and dll-1 (15,29) are highly enriched in the epithelium compared with the mesenchyme. No data were available in the literature on the expression of dll-3 and -4 in the embryonic pancreas. Our experiments performed by RT-PCR with mesenchyme- or epithelium-enriched fractions from E13.5 rat pancreas show that Jagged-1 and dll-1 mRNA are highly enriched in the epithelial fraction when compared with the mesenchyme. At the same time, dll-3 and -4 were barely detectable in both epithelium and mesenchyme (data not shown). Taken together, these data do not support the hypothesis that the mesenchyme is the source of Notch ligand that could act on the epithelium. However, Lammert et al. (29) reported that the Notch ligands Jagged-1 and -2 are expressed in pancreatic endothelial cells. Thus, we cannot formally exclude that Jagged-1 and -2 produced by endothelial cells may activate the Notch pathway in neighboring epithelial cells.

Our data also indicate that the repressive effect of the mesenchyme on Ngn3 expression requires an intact Notch pathway. In the present study and in a recent work in which the role of the notch pathway in kidney development was analyzed (30), γ-secretase inhibitor XVIII, known to interfere with Notch signaling in vivo and to mimic the effect of a dominant-negative form of Notch (31), was used to block the Notch pathway. We did not use genetic approaches with mice deficient in Notch1 or Notch2 expression, because Notch1 or Notch2 mutant mice die between E9.5 and E11.5 and thus cannot be used to analyze later stages of development (3234). Although it cannot be formally excluded that the effect we observed with γ-secretase inhibitor XVIII is mediated by the inhibition of proteolytic cleavage of other transmembrane proteins than Notch, the fact that expression of Hes1, a direct target of Notch, was decreased upon treatment with the inhibitor indicated that at least the Notch activation was inhibited. Moreover, in our experiments, the activated expression of Ngn3 observed in the presence of γ-secretase inhibitor resembles what is found in the pancreas of mice deficient in the Notch ligand Delta1. In such mice, in which the Notch receptor cannot be activated, Ngn3-positive cells developed earlier than in wild-type mice (15).

We also found here that the mesenchyme acted downstream of Ngn3. This repressive effect downstream of Ngn3 was observed in two types of experiments. First, when whole E13.5 pancreata (epithelium+mesenchyme) were cultured with an inhibitor of γ-secretase, Ngn3 was activated, but Ngn3-positive cells did not differentiate into β-cells. Next, when epithelium was cultured for 1 day in the absence of mesenchyme to induce Ngn3 expression and next reassociated with mesenchyme, again, β-cells did not develop. Thus, in addition to its repressive effect upstream of Ngn3, the mesenchyme acted downstream of Ngn3 by repressing the differentiation of Ngn3-positive cells into β-cells.

Our data indicated that the repression upstream of Ngn3 required direct cell-cell contact, whereas repression downstream of Ngn3 was mediated by soluble factors. This repressive effect mediated by soluble factors downstream of Ngn3 seems in contradiction with recent data by Li et al. (26), showing that the mesenchyme at a distance can stimulate β-cell development via soluble factors. However, the in vitro model and conditions used by Li et al. are different from those used in the present study. More specifically, using E11.5 mouse pancreas, Li et al. (26) showed that β-cells developed poorly in vitro in the absence of mesenchyme but developed in the presence of mesenchyme. On the other hand, using E13.5 rat pancreas, we showed that a large amount of β-cells developed in the absence of mesenchyme but not in the presence of mesenchyme (8,23). This difference in terms of development could be due to the age of the tissues used in each study. It is now established that the in vitro developmental potential of pancreata depends of the stage at which the pancreas is harvested, the age of the mesenchyme determining the differentiation state of the final explant (35). Thus, the difference in terms of repression mediated by soluble factors between the results from Li et al. (26) and the ones from the present study could be due to differences in the age of the mesenchyme.

The repressive effects that occur both upstream and downstream of Ngn3 during β-cell development could explain the difficulty to reproducibly generate β-cells from embryonic stem cells. Although procedures are available for generating other cell types, such as neurons (36) or cardiomyocytes (37), from embryonic stem cells, reproducible protocols to generate β-cells from embryonic stem cells are currently not available (38,39). Dissecting the repressive effects of the mesenchyme on β-cell development should be useful to define protocols to generate β-cells from embryonic stem cells.

In conclusion, we previously demonstrated that the mesenchyme surrounding the pancreatic epithelium during prenatal life was crucial for the normal proliferation of pancreatic progenitor cells (7). Here, we demonstrate an additional role for the mesenchyme, which consists of delaying the onset of endocrine-cell differentiation, thereby ensuring that pancreatic progenitor cells at a developmental state characterized by sensitivity to proliferative signals from the mesenchyme remain present until the late stages of embryonic development.

FIG. 1.

Expression of Ngn3 and insulin during in vitro development of rat embryonic pancreatic epithelium. Embryonic pancreatic epithelia obtained on E13.5 were dissected away from their surrounding mesenchyme and cultured for 0 (A and F), 1 (B and G), 3 (C and H), 5 (D and I), or 7 (E and J) days. Expression of Insulin (A–E) and Ngn3 (F–J) were detected by in situ hybridization. Scale bar = 50 μm.

FIG. 1.

Expression of Ngn3 and insulin during in vitro development of rat embryonic pancreatic epithelium. Embryonic pancreatic epithelia obtained on E13.5 were dissected away from their surrounding mesenchyme and cultured for 0 (A and F), 1 (B and G), 3 (C and H), 5 (D and I), or 7 (E and J) days. Expression of Insulin (A–E) and Ngn3 (F–J) were detected by in situ hybridization. Scale bar = 50 μm.

Close modal
FIG. 2.

Immunohistochemical study of pancreatic epithelia grown for 7 days with or without mesenchyme. Embryonic pancreatic epithelia were cultured for 7 days without (A) or with (B and C) mesenchyme. In B, the mesenchyme was not removed, whereas in C, the epithelia were first dissected then placed in direct contact with mesenchyme. β-Cell development was evaluated after anti-insulin (red) staining and development of acinar tissue using anti–carboxypeptidase A antibodies (in green). Scale bar = 50 μm.

FIG. 2.

Immunohistochemical study of pancreatic epithelia grown for 7 days with or without mesenchyme. Embryonic pancreatic epithelia were cultured for 7 days without (A) or with (B and C) mesenchyme. In B, the mesenchyme was not removed, whereas in C, the epithelia were first dissected then placed in direct contact with mesenchyme. β-Cell development was evaluated after anti-insulin (red) staining and development of acinar tissue using anti–carboxypeptidase A antibodies (in green). Scale bar = 50 μm.

Close modal
FIG. 3.

Ngn3 expression in epithelia cultured with or without mesenchyme. A: E13.5 rat embryonic pancreatic epithelia (a and b), whole pancreata (c), or pancreatic epithelia reassociated with two mesenchymes before culturing (d) were either fixed immediately (a) or cultured for 24 h (bd). In situ hybridization was then performed using an Ngn3 antisense riboprobe. Scale bar = 50 μm. B: Quantitative analysis of the number of Ngn3-positive cells in the E13.5 pancreatic epithelium before culturing (□) or after 24 h of culturing without mesenchyme (▪). Whole pancreata (□) and epithelia reassociated with two mesenchymes before culturing (▒) were also analyzed after 24 h of culturing. Each bar represents the means ± SE of nine experiments. a: Comparison of the number of Ngn3-positive cells in epithelia before or after a 24-h culture period (P < 0.001); b: comparison of the number of Ngn3-positive cells in epithelia cultured for 24 h without or with mesenchyme.

FIG. 3.

Ngn3 expression in epithelia cultured with or without mesenchyme. A: E13.5 rat embryonic pancreatic epithelia (a and b), whole pancreata (c), or pancreatic epithelia reassociated with two mesenchymes before culturing (d) were either fixed immediately (a) or cultured for 24 h (bd). In situ hybridization was then performed using an Ngn3 antisense riboprobe. Scale bar = 50 μm. B: Quantitative analysis of the number of Ngn3-positive cells in the E13.5 pancreatic epithelium before culturing (□) or after 24 h of culturing without mesenchyme (▪). Whole pancreata (□) and epithelia reassociated with two mesenchymes before culturing (▒) were also analyzed after 24 h of culturing. Each bar represents the means ± SE of nine experiments. a: Comparison of the number of Ngn3-positive cells in epithelia before or after a 24-h culture period (P < 0.001); b: comparison of the number of Ngn3-positive cells in epithelia cultured for 24 h without or with mesenchyme.

Close modal
FIG. 4.

The repression of Ngn3 by mesenchymal signals requires direct cell-cell contact. A: Epithelia were cultured on one side of a filter for 24 h without (a) or with (b and c) mesenchyme. The epithelium and mesenchyme were separated by a filter in b and were in direct contact in c. In situ hybridization was performed using an Ngn3 antisense riboprobe. Note that the mesenchyme repressed Ngn3 induction when the mesenchyme and epithelium were in direct contact but not when they were separated by a filter. The epithelium is circled in blue and the mesenchyme in red for clarity. B: Effect of EGF and FGF7 on Ngn3 expression. Epithelia were grown for 1 day in culture medium without growth factors (a), culture medium supplemented with EGF (b), or culture medium supplemented with FGF7 (c). In situ hybridization was performed using an Ngn3 antisense riboprobe. Note that neither EGF nor FGF7 modified Ngn3 expression. The epithelium is circled in blue. Scale bar = 50 μm.

FIG. 4.

The repression of Ngn3 by mesenchymal signals requires direct cell-cell contact. A: Epithelia were cultured on one side of a filter for 24 h without (a) or with (b and c) mesenchyme. The epithelium and mesenchyme were separated by a filter in b and were in direct contact in c. In situ hybridization was performed using an Ngn3 antisense riboprobe. Note that the mesenchyme repressed Ngn3 induction when the mesenchyme and epithelium were in direct contact but not when they were separated by a filter. The epithelium is circled in blue and the mesenchyme in red for clarity. B: Effect of EGF and FGF7 on Ngn3 expression. Epithelia were grown for 1 day in culture medium without growth factors (a), culture medium supplemented with EGF (b), or culture medium supplemented with FGF7 (c). In situ hybridization was performed using an Ngn3 antisense riboprobe. Note that neither EGF nor FGF7 modified Ngn3 expression. The epithelium is circled in blue. Scale bar = 50 μm.

Close modal
FIG. 5.

The γ-secretase inhibitor XVIII activates Ngn3 in E13.5 whole pancreata but is not sufficient for β-cell development. E13.5 rat embryonic pancreatic epithelia (A and D) or whole pancreata (B, C, E, and F) were cultured without (A, B, D, and E) or with (C and F) the γ-secretase inhibitor XVIII. A–C: At the end of day 1, in situ hybridization was performed using an Ngn3 antisense riboprobe. D–F: At the end of day 5, the tissues were analyzed by immunohistochemistry using anti-insulin (in red) and anti–carboxypeptidase A (in green). Note that treatment with the γ-secretase inhibitor XVIII is sufficient to suppress the repressive effect of the mesenchyme on Ngn3. On the other hand, it is not sufficient for β-cell differentiation. In A–C, the epithelium is circled in black. In D–F, the whole tissue is circled in white. Scale bar = 50 μm.

FIG. 5.

The γ-secretase inhibitor XVIII activates Ngn3 in E13.5 whole pancreata but is not sufficient for β-cell development. E13.5 rat embryonic pancreatic epithelia (A and D) or whole pancreata (B, C, E, and F) were cultured without (A, B, D, and E) or with (C and F) the γ-secretase inhibitor XVIII. A–C: At the end of day 1, in situ hybridization was performed using an Ngn3 antisense riboprobe. D–F: At the end of day 5, the tissues were analyzed by immunohistochemistry using anti-insulin (in red) and anti–carboxypeptidase A (in green). Note that treatment with the γ-secretase inhibitor XVIII is sufficient to suppress the repressive effect of the mesenchyme on Ngn3. On the other hand, it is not sufficient for β-cell differentiation. In A–C, the epithelium is circled in black. In D–F, the whole tissue is circled in white. Scale bar = 50 μm.

Close modal
FIG. 6.

The mesenchyme maintains Hes1 expression in the pancreatic epithelium. E13.5 rat embryonic whole pancreata (A and B) or pancreatic epithelia (C and D) were either fixed immediately (A) or cultured for either 24 h (B and C) or for 7 days (D). In situ hybridization was then performed using an Hes1 antisense riboprobe. Scale bar = 50 μm.

FIG. 6.

The mesenchyme maintains Hes1 expression in the pancreatic epithelium. E13.5 rat embryonic whole pancreata (A and B) or pancreatic epithelia (C and D) were either fixed immediately (A) or cultured for either 24 h (B and C) or for 7 days (D). In situ hybridization was then performed using an Hes1 antisense riboprobe. Scale bar = 50 μm.

Close modal
FIG. 7.

The mesenchyme represses endocrine cell development downstream of Ngn3. A: E13.5 rat embryonic pancreatic epithelia were cultured for 1 day. At the end of day 1, when Ngn3 expression is activated, they were associated with mesenchyme and cultured for 2 or 6 additional days. The tissues were analyzed by immunohistochemistry using anti-insulin (in red) and anti–carboxypeptidase A (in green) antibodies. a: Epithelium alone cultured for 3 days; b: epithelium reassociated with mesenchyme on day 1 and cultured for 2 additional days; c: epithelium alone cultured for 7 days; d: epithelium reassociated with mesenchyme on day 1 and cultured for 6 additional days. Note that β-cells did not develop when the mesenchyme was reassociated on day 1. B: Epithelia were cultured on one side of a filter for 3 days without (a) or with (b and c) mesenchyme. The epithelium and the mesenchyme were separated by a filter in b or were in direct contact in c. The tissues were examined by immunohistochemistry using antibodies to detect insulin (in red) and carboxypeptidase A (in green). Note that the mesenchyme repressed β-cell development both when separated from the epithelium by a filter and when in direct contact with the epithelium. Scale bar = 50 μm.

FIG. 7.

The mesenchyme represses endocrine cell development downstream of Ngn3. A: E13.5 rat embryonic pancreatic epithelia were cultured for 1 day. At the end of day 1, when Ngn3 expression is activated, they were associated with mesenchyme and cultured for 2 or 6 additional days. The tissues were analyzed by immunohistochemistry using anti-insulin (in red) and anti–carboxypeptidase A (in green) antibodies. a: Epithelium alone cultured for 3 days; b: epithelium reassociated with mesenchyme on day 1 and cultured for 2 additional days; c: epithelium alone cultured for 7 days; d: epithelium reassociated with mesenchyme on day 1 and cultured for 6 additional days. Note that β-cells did not develop when the mesenchyme was reassociated on day 1. B: Epithelia were cultured on one side of a filter for 3 days without (a) or with (b and c) mesenchyme. The epithelium and the mesenchyme were separated by a filter in b or were in direct contact in c. The tissues were examined by immunohistochemistry using antibodies to detect insulin (in red) and carboxypeptidase A (in green). Note that the mesenchyme repressed β-cell development both when separated from the epithelium by a filter and when in direct contact with the epithelium. Scale bar = 50 μm.

Close modal
FIG. 8.

The development of Ngn3-expressing cells is accelerated in the absence of mesenchyme. In situ hybridization was performed using a Ngn3 antisense riboprobe on the following tissues. AE: Pancreatic sections of rats at various stages of prenatal development (A, E13.5; B, E14.5; C, E16.5; D, E18.5; and E, E20). FJ: Whole E13.5 pancreata cultured for 0 (F), 1 (G), 3 (H), 5 (I), or 7 (J) days. KO: E13.5 pancreatic epithelium cultured for 0 (K), 1 (L), 3 (M), 5 (N), and 7 (O) days. Note the accelerated development of Ngn3-expressing cells in the absence of mesenchyme. In KO, the epithelium is circled. Scale bar = 50 μm.

FIG. 8.

The development of Ngn3-expressing cells is accelerated in the absence of mesenchyme. In situ hybridization was performed using a Ngn3 antisense riboprobe on the following tissues. AE: Pancreatic sections of rats at various stages of prenatal development (A, E13.5; B, E14.5; C, E16.5; D, E18.5; and E, E20). FJ: Whole E13.5 pancreata cultured for 0 (F), 1 (G), 3 (H), 5 (I), or 7 (J) days. KO: E13.5 pancreatic epithelium cultured for 0 (K), 1 (L), 3 (M), 5 (N), and 7 (O) days. Note the accelerated development of Ngn3-expressing cells in the absence of mesenchyme. In KO, the epithelium is circled. Scale bar = 50 μm.

Close modal

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

M.A. has received support from the French Ministry for Research and Technology. This work was supported by Juvenile Diabetes Research Foundation (JDRF Center for β-Cell Therapy in Europe), Fondation pour la Recherche Médicale, Institut National de la Santé et de la Recherche Médicale/Fondation pour la Recherche Médicale/Juvenile Diabetes Research Foundation (Grant 4DA03H), 6th European Union Framework Program (β-Cell Therapy Integrated Project), and Association Française des Diabétiques.

We are deeply grateful to Virginie Aiello for technical assistance.

1
Hatakeyama J, Bessho Y, Katoh K, Ookawara S, Fujioka M, Guillemot F, Kageyama R: Hes genes regulate size, shape and histogenesis of the nervous system by control of the timing of neural stem cell differentiation.
Development
131
:
5539
–5550,
2004
2
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
24
:
36
–44,
2000
3
Kim S, Hebrok M, Melton D: Notochord to endoderm signaling is required for pancreas development.
Development
124
:
4243
–4252,
1997
4
Pictet R, Rutter W: Development of the embryonic pancreas. In
Handbook of Physiology
. Vol. 1. Steiner DF, Freinkel N, Eds. Baltimore, MD, Williams and Wilkins,
1972
, p.
25
–66
5
Golosow N, Grobstein C: Epitheliomesenchymal interaction in pancreatic morphogenesis.
Dev Biol
4
:
242
–255,
1962
6
Wessels N, Cohen J: Early pancreas organogenesis, tissue interactions, and mass effects.
Dev Biol
15
:
237
–270,
1967
7
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
128
:
5109
–5117,
2001
8
Miralles F, Czernichow P, Scharfmann R: Follistatin regulates the relative proportions of endocrine versus exocrine tissue during pancreatic development.
Development
125
:
1017
–1024,
1998
9
Miralles F, Serup P, Cluzeaud F, Vandewalle A, Czernichow P, Scharfmann R: Characterization of beta cells developed in vitro from rat embryonic pancreatic epithelium.
Dev Dyn
214
:
116
–126,
1999
10
Gittes GK, Galante PE, Hanahan D, Rutter WJ, Debase HT: Lineage-specific morphogenesis in the developing pancreas: role of mesenchymal factors.
Development
122
:
439
–447,
1996
11
Edlund H: Transcribing pancreas.
Diabetes
47
:
1817
–1823,
1998
12
Wilson ME, Scheel D, German MS: Gene expression cascades in pancreatic development.
Mech Dev
120
:
65
–80,
2003
13
Ahlgren U, Jonsson J, Edlund H: The morphogenesis of the pancreatic mesenchyme is uncoupled from that of the pancreatic epithelium in IPF1/PDX1-deficient mice.
Development
122
:
1409
–1416,
1996
14
Offield M, Jetton T, Laborsky P, Ray M, Stein R, Magnuson M, Hogan B, Wright C: PDX-1 is required for pancreatic outgrowth and differentiation of the rostral duodenum.
Development
122
:
983
–995,
1996
15
Apelqvist A, Li H, Sommer L, Beatus P, Anderson DJ, Honjo T, Hrabe de Angelis M, Lendahl U, Edlund H: Notch signalling controls pancreatic cell differentiation.
Nature
400
:
877
–881,
1999
16
Gradwohl G, Dierich A, LeMeur M, Guillemot F: Neurogenin3 is required for the development of the four endocrine cell lineages of the pancreas.
Proc Natl Acad Sci U S A
97
:
1607
–1611,
2000
17
Gu G, Dubauskaite J, Melton DA: Direct evidence for the pancreatic lineage: NGN3+ cells are islet progenitors and are distinct from duct progenitors.
Development
129
:
2447
–2457,
2002
18
Duvillie B, Attali M, Aiello V, Quemeneur E, Scharfmann R: Label-retaining cells in the rat pancreas: location and differentiation potential in vitro.
Diabetes
52
:
2035
–2042,
2003
19
Weng AP, Nam Y, Wolfe MS, Pear WS, Griffin JD, Blacklow SC, Aster JC: Growth suppression of pre-T acute lymphoblastic leukemia cells by inhibition of notch signaling.
Mol Cell Biol
23
:
655
–664,
2003
20
Ravassard P, Chatail F, Mallet J, Icard-Liepkalns C: Relax, a novel rat bHLH transcriptional regulator transiently expressed in the ventricular proliferating zone of the developing central nervous system.
J Neurosci Res
48
:
146
–158,
1997
21
Ishibashi M, Ang SL, Shiota K, Nakanishi S, Kageyama R, Guillemot F: Targeted disruption of mammalian hairy and enhancer of split homolog-1 (HES-1) leads to up-regulation of neural helix-loop-helix factors, premature neurogenesis, and severe neural tube defects.
Genes Dev
9
:
3136
–3148,
1995
22
Herzog E, Bellenchi GC, Gras C, Bernard V, Ravassard P, Bedet C, Gasnier B, Giros B, El Mestikawy S: The existence of a second vesicular glutamate transporter specifies subpopulations of glutamatergic neurons.
J Neurosci
21
:
RC181
,
2001
23
Cras-Meneur C, Elghazi L, Czernichow P, Scharfmann R: Epidermal growth factor increases undifferentiated pancreatic embryonic cells in vitro: a balance between proliferation and differentiation.
Diabetes
50
:
1571
–1579,
2001
24
Elghazi L, Cras-Meneur C, Czernichow P, Scharfmann R: Role for FGFR2IIIb-mediated signals in controlling pancreatic endocrine progenitor cell proliferation.
Proc Natl Acad Sci U S A
99
:
3884
–3889,
2002
25
Jensen J, Heller RS, Funder-Nielsen T, Pedersen EE, Lindsell C, Weinmaster G, Madsen OD, Serup P: Independent development of pancreatic α- and β-cells from neurogenin3-expressing precursors: a role for the notch pathway in repression of premature differentiation.
Diabetes
49
:
163
–176,
2000
26
Li Z, Manna P, Kobayashi H, Spilde T, Bhatia A, Preuett B, Prasadan K, Hembree M, Gittes GK: Multifaceted pancreatic mesenchymal control of epithelial lineage selection.
Dev Biol
269
:
252
–263,
2004
27
Mitsiadis TA, Henrique D, Thesleff I, Lendahl U: Mouse Serrate-1 (Jagged-1): expression in the developing tooth is regulated by epithelial-mesenchymal interactions and fibroblast growth factor-4.
Development
124
:
1473
–1483,
1997
28
Jensen JN, Cameron E, Garay MV, Starkey TW, Gianani R, Jensen J: Recapitulation of elements of embryonic development in adult mouse pancreatic regeneration.
Gastroenterology
128
:
728
–741,
2005
29
Lammert E, Brown J, Melton DA: Notch gene expression during pancreatic organogenesis.
Mech Dev
94
:
199
–203,
2000
30
Cheng HT, Miner JH, Lin M, Tansey MG, Roth K, Kopan R: Gamma-secretase activity is dispensable for mesenchyme-to-epithelium transition but required for podocyte and proximal tubule formation in developing mouse kidney.
Development
130
:
5031
–5042,
2003
31
Francis R, McGrath G, Zhang J, Ruddy DA, Sym M, Apfeld J, Nicoll M, Maxwell M, Hai B, Ellis MC, Parks AL, Xu W, Li J, Gurney M, Myers RL, Himes CS, Hiebsch R, Ruble C, Nye JS, Curtis D: aph-1 and pen-2 are required for Notch pathway signaling, gamma-secretase cleavage of betaAPP, and presenilin protein accumulation.
Dev Cell
3
:
85
–97,
2002
32
Conlon RA, Reaume AG, Rossant J: Notch1 is required for the coordinate segmentation of somites.
Development
121
:
1533
–1545,
1995
33
Hamada Y, Kadokawa Y, Okabe M, Ikawa M, Coleman JR, Tsujimoto Y: Mutation in ankyrin repeats of the mouse Notch2 gene induces early embryonic lethality.
Development
126
:
3415
–3424,
1999
34
Swiatek PJ, Lindsell CE, del Amo FF, Weinmaster G, Gridley T: Notch1 is essential for postimplantation development in mice.
Genes Dev
8
:
707
–719,
1994
35
Rose MI, Crisera CA, Colen KL, Connelly PR, Longaker MT, Gittes GK: Epithelio-mesenchymal interactions in the developing mouse pancreas: morphogenesis of the adult architecture.
J Pediatr Surg
34
:
774
–779,
1999
36
Lee SH, Lumelsky N, Studer L, Auerbach JM, McKay RD: Efficient generation of midbrain and hindbrain neurons from mouse embryonic stem cells.
Nat Biotechnol
18
:
675
–679,
2000
37
Boheler KR, Czyz J, Tweedie D, Yang HT, Anisimov SV, Wobus AM: Differentiation of pluripotent embryonic stem cells into cardiomyocytes.
Circ Res
91
:
189
–201,
2002
38
Hansson M, Tonning A, Frandsen U, Petri A, Rajagopal J, Englund MC, Heller RS, Hakansson J, Fleckner J, Skold HN, Melton D, Semb H, Serup P: Artifactual insulin release from differentiated embryonic stem cells.
Diabetes
53
:
2603
–2609,
2004
39
Rajagopal J, Anderson WJ, Kume S, Martinez OI, Melton DA: Insulin staining of ES cell progeny from insulin uptake.
Science
299
:
363
,
2003