Neurog3 gene dosage regulates allocation of endocrine and exocrine cell fates in the developing mouse pancreas

Developmental Biology 339 (2010) 26–37

Contents lists available at ScienceDirect

Developmental Biology j o u r n a l h o m e p a g e : w w w. e l s e v i e r. c o m / d e v e l o p m e n t a l b i o l o g y

Neurog3 gene dosage regulates allocation of endocrine and exocrine cell fates in the developing mouse pancreas Sui Wang, Jingbo Yan, Daniel A. Anderson, Yanwen Xu, Maneesh C. Kanal, Zheng Cao, Christopher V.E. Wright, Guoqiang Gu ⁎ Program in Developmental Biology and Department of Cell and Developmental Biology, Center for Stem Cell Biology, Vanderbilt University Medical Center, 465 21st Avenue South, Rm 4128, Vanderbilt Medical Center, Nashville, TN 37232, USA

a r t i c l e

i n f o

Article history: Received for publication 1 July 2009 Revised 5 December 2009 Accepted 7 December 2009 Available online 16 December 2009 Keywords: Diabetes Islet Neurog3 Ngn3 Lateral inhibition Notch Endocrine progenitor Dosage Haploinsufficiency Pancreas

a b s t r a c t The basic helix–loop–helix transcription factor Neurog3 (Neurogenin3 or Ngn3) actively drives endodermal progenitor cells towards endocrine islet cell differentiation during embryogenesis. Here, we manipulate Neurog3 expression levels in endocrine progenitor cells without altering its expression pattern using heterozygosity and a hypomorph. Lowered Neurog3 gene dosage in the developing pancreatic epithelium reduces the overall production of endocrine islet cells without significantly affecting the proportions of various islet cell types that do form. A reduced Neurog3 production level in the endocrine-directed pancreatic progenitor population activates the expression of Neurog3 in an increased number of epithelial progenitors. Yet a significant number of these Neurog3+ cells detected in heterozygous and hypomorphic pancreata, possibly those that express low levels of Neurog3, move on to adopt pancreatic ductal or acinar fates. These data directly demonstrate that achieving high levels of Neurog3 expression is a critical step for endocrine commitment from multipotent pancreatic progenitors. These findings also suggest that a high level of Neurog3 expression could mediate lateral inhibition or other unknown feedback mechanisms to regulate the number of cells that initiate Neurog3 transcription and protein production. The control of Neurog3+ cell number and the Neurog3 threshold-dependent endocrine differentiation mechanism combine to select a specific proportion of pancreatic progenitor cells to adopt the islet cell fate. © 2009 Elsevier Inc. All rights reserved.

Introduction During early embryogenesis, the seemingly equivalent and likely multipotent pancreatic precursors give rise to both exocrine (acinar and ductal) or endocrine islet cells, in generally reproducible proportions (Gu et al., 2002; Kawaguchi et al., 2002; Seymour et al., 2007; Zhou et al., 2007). Neurog3 (Neurogenin3 or Ngn3) is essential for endocrine differentiation. All endocrine islet cells are derived from pancreatic progenitors that transiently express high levels of Neurog3 (Gu et al., 2002; Mellitzer et al., 2004; Schonhoff et al., 2004). Loss of Neurog3 activity nearly abolishes endocrine islet cell differentiation (Gradwohl et al., 2000; Wang et al., 2008). Conversely, ectopic expression of Neurog3 results in precocious endocrine differentiation from embryonic endodermal cells, with a differential stage-dependent output of islet cell types (Apelqvist et al., 1999; Grapin-Botton et al., 2001; Schwitzgebel et al., 2000). Remarkably, co-expression of Neurog3 with MafA and Pdx1 converts adult acinar cells into β-like islet cells (Zhou et al., 2008). Thus, understanding Neurog3 expression and function is essential for elucidating islet cell neogenesis, and may

⁎ Corresponding author. Fax: +1 615 936 5673. E-mail address: [email protected] (G. Gu). 0012-1606/$ – see front matter © 2009 Elsevier Inc. All rights reserved. doi:10.1016/j.ydbio.2009.12.009

be relevant for both in vitro differentiation of ES cells and in vivo exocrine cell transdifferentiation, towards endocrine islets [(Yang and Wright, 2009; Zaret, 2009) and references therein]. Previous studies have suggested that Neurog3+ cells are selected by Notch-mediated lateral inhibition from a set of multipotent pancreatic progenitor cells (MPCs) that maintain active Notch signaling (Apelqvist et al., 1999; Jensen et al., 2000b). Specifically, activated Notch converts transcriptional repressor RBP-J{kappa}/ CBF1 to an activator to up-regulate the expression of Hes (orthologs of Hairy enhancer of split) or Hairy-related (Hey) genes (Fujikura et al., 2006). Protein products of the Hes and Hey gene family members directly repress Neurog3 transcription (Jensen et al., 2000b) to allow for MPC maintenance. Stochastic events or asymmetries developed during cell division cause a subset of MPCs to reduce the strength of Notch signaling and raise Neurog3 level above a specific threshold level for islet cell specification and commitment (Apelqvist et al., 1999; Jensen et al., 2000b). Neurog3 directly activates the expression of several transcription factor genes (Mellitzer et al., 2006; Petri et al., 2006; White et al., 2008). Subsequently, these transcription factors cooperatively activate the expression of endocrine hormones and other necessary genes that are involved in islet cell development and maturation (Artner et al., 2007; Huang et al., 2000; Wang et al., 2007). During the

S. Wang et al. / Developmental Biology 339 (2010) 26–37

differentiation process, short-lived intermediate cell types that only express these transcription factors but not hormones have been detected and are proposed to be different-staged endocrine progenitors (Artner et al., 2007; Chiang and Melton, 2003; Jensen et al., 2000a). Whether the level of Neurog3 expression or the activation of any specific Neurog3 targets is the limiting step for producing mature and functional islet cells remains unclear. Other than activating the endocrine differentiation program, Neurog3 was proposed to mediate lateral inhibition to properly allocate the specific proportions of pancreatic MPCs to endocrine and exocrine fates in a non-cell autonomous fashion (Skipper and Lewis, 2000). By analogy to studies of neuronal development (Kageyama et al., 2008), it was proposed that Neurog3 upregulates the expression of Notch ligand(s), raising nuclear Notch signaling in adjacent cells non-cell-autonomously and preventing them from activating Neurog3 expression (Lewis, 1998; Meinhardt and Gierer, 1974). A result of lateral inhibition is the preservation of mitotically active pancreatic MPCs, which is critical for generating adequate endocrine mass because most Neurog3 expressing cells are mitotically quiescent and differentiated islet cells have limited proliferation capability (Gu et al., 2002; Jensen et al., 2000a; Stanger et al., 2007). This model predicts that the level of Neurog3 expression in individual pancreatic progenitor cells is critical for controlling the proportion of progenitors that activate Neurog3 expression. Here, we demonstrate that a low level of Neurog3 expression in individual pancreatic cells is not sufficient to cause full differentiation and end-stage endocrine hormone production. Consequently, a portion of the Neurog3-expressing cells, possibly those expressing low levels of Neurog3, adopts exocrine (both acinar and ductal) cell fates. This finding suggests that pancreatic progenitor cells that turn on a low level of Neurog3 maintain a substantial degree of plasticity to revert to exocrine progenitors: we propose that they are therefore endocrine-biased and metastable. We further show that lowering Neurog3 production activates Neurog3 in a larger proportion of pancreatic progenitor cells, although the transcription levels of Notch ligands in cells that express different levels of Neurog3 are not significantly altered. This latter finding suggests that Neurog3 mediates lateral inhibition with a Notch ligand transcriptionindependent pathway(s) to control the abundance of Neurog3+ cells. Materials and methods

27

anti-amylase, and rabbit anti-Ki67 were from ABCam, Cambridge, MA. Mouse anti-mouse alpha tubulin and Dolichos Biflorus Agglutinin (DBA) were from Sigma Aldrich, St Louis, MI. Secondary antibodies used were: Cy3-conjugated donkey anti-rabbit IgG, Cy3-conjugated donkey anti-mouse IgG, FITC-conjugated donkey anti-guinea pig IgG, Cy3-conjugated donkey anti-goat, Cy5-conjugated donkey antiguinea pig IgG; and Cy5-conjugated donkey anti-rabbit (Jackson Immunoresearch, West Grove, PA). All antibodies utilized 1:500– 1:2000 dilutions, depending on the amount of tissue on each slide. Assays and quantification Intraperitoneal glucose tolerance test (IPGTT), serum insulin assays, and insulin secretion from isolated islets were performed as described (Wang et al., 2007). Western blot utilized α-tubulin as loading control. 20 or 10 μg of total protein was loaded for each lane. In order to assay islet mass, pancreata of each stage were weighted, fixed, and prepared as 6 μm paraffin sections. One of every 10 sections was collected and stained with a mix of antibodies against insulin, glucagon, PP, and SS, respectively, as HRP signals. Slides were counterstained with hematoxylin. Sections were then scanned by Ariol SL-50 (Genetix, Hampshire, UK) for quantitative IHC analysis to calculate the ratio of the stained endocrine area over that of the total pancreas. The islet mass was calculated based on the endocrine/exocrine ratio and pancreas weight. In order to examine the mitotic index of different pancreatic cells, animals were injected IP with multiple doses of BrdU (at 50 mg/kg body weight). For P1 animals, two doses at 2-h interval were injected. For P10 animals, 4 doses at 2-h interval were injected. Pancreata were then collected 2 h after the last BrdU injection. Pancreata were then prepared as paraffin sections to assay for BrdU+ cells, which were visualized by IF and confocal imaging. Multiple images/sections from each pancreas were counted to obtain representative labeling indices. Microscopy Confocal microscopy (Zeiss LSM500) was utilized to visualize fluorescent images. Samples were processed side by side. Images were captured using identical confocal parameters as Z-sections of one μm thickness. Images were then projected onto a single plane for side-byside comparison for protein level/cell quantification.

Mouse strains and care Fluorescence activated cell sorting and real time RT-PCR (QRTPCR) Mouse strain derivation/maintenance, crosses and treatments follow protocols approved by the Vanderbilt Medical Center IACUC. The noon of vaginal plug appearance was counted as embryonic day 0.5 (E0.5). Routine mouse embryo production utilized CD1 mice (Charles River Laboratories, Inc. Wilmington, MA). The Neurog3F and Neurog3− mouse strains were described previously (Wang et al., 2008). Neurog3TgBAC-Cre; R26REYFP and Neurog3EGFP knockin mice were generous gifts from A. Leiter (Schonhoff et al., 2004), F. Costantini (Srinivas et al., 2001), and K. Kaestner (Lee et al., 2002), respectively. Genotyping follows published methods with minor modification. Briefly, tissue from an ear puncture was collected, boiled in 40 mM NaOH for 30 min. An equal volume of 0.1 mM Tris (pH: 5.0) was added to neutralize. 1–3 μl of lysate was used in PCR genotyping. Immunodetection Immunofluorescence (IF) and immunohistochemistry followed established protocols. Tissues were prepared either as frozen or paraffin sections. For frozen sections, dissected tissues were immediately frozen in OCT and sectioned on the day of staining. Guinea pig anti-glucagon, guinea pig anti-insulin, guinea pig anti-PP, and rabbit anti-SS were from Dako, Carpinteria, CA. Rat anti-BrdU (BU1/75), goat

E14.5 pancreata were dissected and genotyped. Batches of 8–10 individual EGFP+ pancreata were dissociated into single cells using a mix of trypsin (0.05%) and Type IV collagenase (1 mg/ml). EGFP+ cells were collected using FACS. Before further analysis, a portion of collected cells were spun onto glass slides utilizing Cytospin (Thermo Shandon, Oak Park, MI) and examined for purity (by analyzing the expression of hormones, and the presence of EGFP). The rest of the samples were utilized for RT-PCR, with two duplicate cDNA preparations and four QRT-PCR (quantitative RT-PCR) reactions to assay the mRNA levels. For each genotype, a minimum of three independent batches of embryos was processed independently (i.e. the mRNA level for each gene will be an average of 12 assays). Reverse-transcription PCR (RT-PCR) follows standard protocol. cDNA synthesis and RT-PCR were done by using a High Capacity cDNA Reverse Transcription Kit (Applied Biosystems, Foster City, CA). QRT-PCR was performed using mouse-specific primers (Table 1) with iQ SYBR Green Supermix on an iCycler Thermal Cycler. The abundance of all transcripts was normalized versus that of GAPDH. All statistical analyses utilized the standard Student t-test. A P-value of b0.05 was considered significant. Quantification data were presented as the mean ± standard error of the means.

28

S. Wang et al. / Developmental Biology 339 (2010) 26–37

Results Reduced Neurog3 dosage impairs endocrine islet function While analyzing the function of Neurog3 in differentiated endocrine islet cells utilizing a conditional floxed Neurog3 allele (Neurog3F), we noticed that Neurog3+/− and Neurog3F/−adult animals displayed a significantly compromised glucose-clearance capacity over that of wild type littermates (+/+), suggesting impaired pancreatic endocrine function (Wang et al., 2009). In the Neurog3− allele, a CreERT open reading frame replaced 150 nucleotides immediately downstream from the ATG start codon of endogenous Neurog3. Thus, this allele produces CreERT but not Neurog3, and is expected to behave as a null allele. In the Neurog3Fallele, LoxP sites flank the Neurog3 coding region, with a CreERT cDNA lying farther 3′ (Wang et al., 2008). No DNA sequence was deleted around the endogenous Neurog3 locus in the Neurog3F allele. Thus, we expect the Neurog3 expression pattern (i.e. the specific cells that express Neurog3) in Neurog3+/−and Neurog3F/− pancreata to be equivalent to that of wild type littermates. Thus, the compromised endocrine function in Neurog3+/−and Neurog3F/− animals over the wild type littermates likely reflects some form of gene dosage or threshold effect of Neurog3 on endocrine differentiation. We therefore systematically examined how Neurog3 gene dosage affected islet cell differentiation and function. We obtained mice of different genotypes and assayed their endocrine function with IPGTT assays at different ages. The fasting blood glucose levels of the Neurog3+/+, Neurog3+/−, and Neurog3F/− adult mice did not vary significantly (Fig. 1A. Blood glucose levels at “0” time point). Yet both male and female Neurog3+/− and Neurog3F/− mice displayed significantly compromised glucose-clearance capa-

bility at all ages tested, including 1 and 2 months after birth (Fig. 1A). The Neurog3F/− mice showed a trend of lower glucose clearing capability at 1 month of age than Neurog3+/− animals. This trend became undetectable after 2 months of age (Fig. 1A). Despite being glucose intolerant, the Neurog3+/− and Neurog3F/− animals do not develop frank diabetes (ad lib-feeding blood glucose over 300 mg/dl) even at 1 year of age (not shown). In order to investigate whether the glucose-clearing defect in Neurog3+/− and Neurog3F/− animals is due to either compromised insulin secretion from β cells or reduced response of peripheral tissues to insulin, we examined insulin sensitivity. Overnight fasted adult mice were administered with insulin by intraperitoneal injection. Their blood glucose level was assayed at different time points after the insulin injection. All three groups of animals, when examined at 2 months of age, displayed similar profiles of blood glucose reduction in response to insulin administration (Fig. 1B), demonstrating an unaltered insulin sensitivity. Consistent with this insulin tolerance result and the IPGTT defects above, the fasting serum insulin levels did not vary significantly between these three genotypes (Fig. 1C). Yet 30 min after glucose challenge, the Neurog3+/− and Neurog3F/− animals showed significantly reduced serum insulin up-regulation compared with Neurog3+/+ control animals (Fig. 1C). Therefore, Neurog3+/− and Neurog3F/− animals have compromised insulin secretion in response to nutritional stimulation. Neurog3 production is reduced in Neurog3+/− and Neurog3F/− pancreata We then determined the underlying mechanisms for the compromised endocrine function in Neurog3+/− and Neurog3F/− animals. We first examined Neurog3 production in pancreata at several

Fig. 1. Neurog3 dosage reduction impairs endocrine islet function. (A) IPGTT results of one and 2-month-old animals (including both male and female mice). (B) Insulin sensitivity in 2-month-old mice (both male and female mice included). The “Y” axis represents the ratio of glucose level at point of assay over glucose level prior to insulin injection. (C) Serum insulin level before and 30 min after glucose challenge. Only male mice were utilized in C. ⁎P b 0.05.

S. Wang et al. / Developmental Biology 339 (2010) 26–37

embryonic stages, E13.5, E15.5, and E16.5, using semi-quantitative immunofluorescence (IF), western blot, and real time RT-PCR. For IFstaining-based assays, pancreata of different genotypes were dissected and processed identically, side-by-side. After IF-staining, confocal image series (as Z-sections of 1 micron thickness) were captured under identical parameters and stacked as single images, so that the fluorescence of multiple nuclear focal planes was integrated to estimate the total fluorescence intensity within individual nuclei (Materials and methods). The average Neurog3 levels in each progenitor cell were visibly reduced in Neurog3+/− and Neurog3F/− pancreata (Supplementary Fig. 1), as were the number of cells that express high levels of Neurog3, with the caveat that this type of nuclear abundance estimation is semi-quantitative (Fig. 2A). Yet consistent with the IF-based assays, western blot analysis estimated that the total amount of Neurog3 protein in Neurog3+/− and Neurog3F/− pancreata was significantly reduced compared to Neurog3+/+ littermates (Fig. 2B), a result that is consistent with the abundance of Neurog3 mRNA levels in Neurog3+/+, Neurog3+/−, and Neurog3F/− pancreata (E15.5) (Fig. 2C). The lowered Neurog3F mRNA level could be a result of its destabilization by the presence of a long 3′-UTR (Amrani et al., 2006). Reduced Neurog3 dosage decreases endocrine islet mass We also investigated how the level of Neurog3 produced affects endocrine function. In the adult pancreas, overall islet architecture did

29

not vary among the various genotypes, with β cells mostly localized within the islet core and α, δ, and PP cells localized peripherally (data not shown). Significantly, the average islet size and the abundance of sectioned islets appeared reproducibly reduced in the order of WT, Neurog3+/−, and Neurog3F/−(Fig. 3A–C). We estimated the total islet mass by measuring islet area in several animals of various genotypes in a subset of pancreatic sections (see Materials and methods). Total islet mass was substantially reduced at P1 (postnatal day 1) and P28 in Neurog3+/− and Neurog3F/−animals than in Neurog3+/+ animals (Fig. 3D). In order to examine whether a reduction of Neurog3 expression affects endocrine function of each islet cell after differentiation, we examined the expression of several factors that are critical for endocrine function after birth. We found that the hormone expressing islet cells in all three groups of animals expressed similar levels of Pdx1, Nkx6.1, Glut2, MafA, and MafB on per cell basis (Supplementary Fig. 2). This finding suggests that the Neurog3 level per cell does not appreciably affect endocrine differentiation, provided that the Neurog3 level is high enough to allow endocrine commitment (and differentiation) to be completed. In addition, we tested the insulin secretion capability of islets isolated from 6-weekold pancreata from the three groups of animals. To our surprise, the straightforward Neurog3+/− heterozygous islets displayed a trend, although statistically insignificant, of increased insulin secretion over that of wild-type islets, in response to glucose stimulation. In accordance with this trend, Neurog3F/− islets showed significantly

Fig. 2. Neurog3 production is reduced in Neurog3+/− and Neurog3F/− pancreata. (A) IF (red) detection of Neurog3 in pancreata at E14.5 and E16.5. Shown are projected Z-stack images captured at identical confocal parameters. White arrows, Neurog3HI cells. Green arrows, Neurog3LO cells. Scale bars = 20 μm. (B) Western blot detection of Neurog3 protein in E15.5 pancreata. Each sample was loaded twice, with different amount of total protein loaded. α-tubulin was used as loading control. Note that Neurog3 protein migrated as two visible bands under this condition. (C) Real time RT-PCR detection of Neurog3 mRNA levels in E15.5 pancreata. Presented are relative mRNA levels normalized against amylase mRNA. ⁎P b 0.05.

30

S. Wang et al. / Developmental Biology 339 (2010) 26–37

Fig. 3. Reduced Neurog3 production decreases pancreatic islet cell production. (A–C) Representative images of islets in 4-week-old pancreata in different genotyped animals. Islets were visualized as brown cell clusters by HRP-based IHC assays against a combination of insulin, glucagon, SS and PP antibodies. Scale bar = 50 μm. (D) Quantification of islet mass at 1 day (P1) and 4 weeks after birth (P28). The islet masses in wild type animals were artificially set at 100%. (E) Insulin secretion in isolated islets of 2-month-old male mice. The X-axis represents the % of total insulin in isolated islets that is released within 1 h under 20 mM glucose stimulation. ⁎P b 0.05.

increased glucose-stimulated-insulin-secretion (Fig. 3E. P = 0.03). This observation could explain why Neurog3F/−animals, which have reduced islet mass compared with Neurog3+/−animals, have statistically similar IPGTT profiles to their Neurog3+/− littermates. It is not clear at present what causes the enhanced insulin secretion in islets isolated from Neurog3F/−animals. It could represent an unknown aspect of improper differentiation. Alternatively, it could represent an adaptive behavior of the islets, which up-regulate glucose-induced insulin secretion to compensate for insulin insufficiency. Such a process is known to occur prior to phenotypic manifestation of Type II Diabetes in human patients (Prentki and Nolan, 2006). The Neurog3 expression level does not affect pancreatic cell proliferation In order to examine whether a reduced Neurog3 production compromises pancreatic cell proliferation, we assayed pancreatic weight and cell division at various stages. At P1, P10, and P16, the pancreatic mass normalized to body weight remains unchanged in wild type, Neurog3+/−, and Neurog3F/− animals (Supplementary Fig. 3). We assayed both the mitotic indices with BrdU incorporation in the exocrine (ducts, acini) and the four major endocrine cell types (α, β, δ, and PP). At E17.5 and P10, two representative stages when pancreatic cells are actively dividing, the mitotic indices of these specific pancreatic cells in wild type, Neurog3+/−, and Neurog3F/− pancreata were similar (Fig. 4 and Supplementary Fig. 4). These

combined data suggest that Neurog3 dosage does not affect cell division and that the observed endocrine mass reduction in Neurog3+/− and Neurog3F/− animals are likely caused by reduced neogenesis (i.e. direct differentiation of Neurog3+ progenitors to form islet endocrine cells) during embryogenesis, rather than compromised cell proliferation during fetal or postnatal animal growth. Along this line, we examined whether the fate or number of Neurog3+ cells was altered when Neurog3 dosage was reduced. Lowered Neurog3 dosage shunts Neurog3+ cells to an exocrine cell fate Neurog3CreERT transgene or knock-in mouse-based lineage studies with the Z/AP reporter line (Lobe et al., 1999) have shown that most, if not all, pancreatic progenitors that express a high level of Neurog3 will adopt an endocrine islet cell fate (Gu et al., 2002; Johansson et al., 2007). Neurog3 TGBAC-Cre transgenic-BAC-based studies suggest that a low but significant fraction of pancreatic progenitors that express Neurog3, detected with the more sensitive Cre reporter R26R, gives rise to exocrine cells (Schonhoff et al., 2004; CW, unpublished observations). Based on these latter findings, we hypothesized that the level of Neurog3 is critical for moving pancreatic progenitors to an endocrine-biased condition. We tested to see if lowering Neurog3 expression level in individual cells would allow more Neurog3+ cells to divert to an exocrine fate. This possibility could explain why endocrine islet cell production is reduced in Neurog3+/− and NeurogF/− animals.

S. Wang et al. / Developmental Biology 339 (2010) 26–37

31

Fig. 4. Pancreatic cell division is not affected by Neurog3 dosage reduction. (A–C) Representative images of BrdU labeling in E17.5 pancreata. (D–F) BrdU labeling in P10 pancreata. DAPI was used to visualize nuclei and insulin signals mark β cells. Scale bar = 20 μm. (G and H) Mitotic indices in E17.5 and P10 pancreata. The indices in β cells and exocrine cells (including both duct and acinar cells) are presented separately.

We utilized Neurog3TGBAC-Cre transgenic mice and R26REYFP reporter mice to trace the fate of Neurog3+ cells in both wild type and heterozygous backgrounds. The advantage of this binary-switch approach is that we can detect all cells that express Neurog3 at a level above the threshold for Cre activity, because the activation of R26REYFP leads to a similarly strong level of EYFP, irrespective of cell fate, and therefore their unambiguous detection. The lineage of Neurog3+ progenitor cells was monitored at embryonic and newborn stages. In the E14.5 Neurog3TGBAC-Cre; R26REYFP wild-type pancreas, most EYFP+ cells appeared as clusters of endocrine cells (Fig. 5A1-A2), with only occasional EYFP+ cells localized in duct (Fig. 5A3) or acinar cells (Fig. 5A4). By contrast, in the E14.5 Neurog3+/−; Neurog3TGBAC-Cre; R26REYFP pancreas, a substantial number of EYFP+ cells appeared in exocrine acinar and ductal cells (Fig. 5B). The similar EYFP+ cell distribution between endocrine and exocrine tissues was observed at newborn and P10 (data not shown): that is, more exocrine cells express EYFP in Neurog3+/− pancreas, and are thus derived from Neurog3-expressing progenitors. Quantitatively, while about 15% of Neurog3TGBAC-Cre-labeled cells were found to express exocrine cell markers in the wild-type background, about 45% of the labeled cells were found within the exocrine lineage in the heterozygous Neurog3+/− pancreas: more Neurog3+ precursors adopt an exocrine fate with Neurog3 dosage reduction (Fig. 5D).

These data strongly suggest that Neurog3 expression must reach a threshold (“high”) level in order to trigger the commitment of progeny towards endocrine differentiation. At lower Neurog3 levels, many more progeny of Neurog3-expressing cells maintain multipotent-progenitor-like activity and can “reallocate” to the exocrine acini or duct fates. Consistent with this hypothesis, total inactivation of Neurog3 did not eliminate the presence of cells that activated the Neurog3 promoter. Instead, the cells that activated the Neurog3 promoter (and therefore also induced to express Cre from Neurog3TgBAC-Cre) were relocated to the exocrine tissues, including exocrine acinar and pancreatic ducts (Fig. 5C). We also examined the fate of Neurog3+ cells in the Neurog3F/− compound heterozygous animals (hypomorphic floxed allele over null) using the Neurog3TgBAC-Cre-based lineage tracing. Neurog3F/−; Neurog3TgBAC-Cre; R26REYFP animals were derived. In these mice, Neurog3 was immediately inactivated when its own transcription was activated. As a result, the Neurog3F/−; Neurog3TgBAC-Cre; R26REYFP pancreata contained nearly no endocrine cells and most of the EYFPexpressing cells were located in the exocrine tissue (not shown). However, these data were difficult to interpret, since we do not know the time frame during which Neurog3TgBAC-Cre inactivates the Neurog3 F allele. We therefore did not further investigate the phenotype in Neurog3F/−; Neurog3TgBAC-Cre; R26REYFP animals.

32

S. Wang et al. / Developmental Biology 339 (2010) 26–37

Lowered Neurog3 dosage results in wider Neurog3 expression in pancreatic progenitors We counted the total number of pancreatic cells derived from Neurog3+ progenitors (i.e. the cells that activated the R26REYFP reporter) at P1. The total number of EYFP+ pancreatic cells (acinar, duct, and islet cells) was significantly increased in Neurog3+/− pancreata over that in wild-type animals (from 4% to 8% of the total

pancreatic cells with no variation of total pancreatic mass. Supplementary Fig. 5). At least three sources could account for this increase in cells activating Neurog3 expression. First, Neurog3 could be ectopically activated in differentiated acinar cells. Because we could not detect Neurog3 expression in exocrine acinar cells with immunodetection or Neurog3EGFP reporter expression (Fig. 6 and data not shown), this possibility is unlikely to occur. Second, if more Neurog3+ progenitors were allocated into exocrine (acinar and duct) pancreas

Fig. 5. Lowered Neurog3 activity shunts more Neurog3+ cells to exocrine fate. (A1, B1, and C1) EYFP expression (white) in E14.5 Neurog3TgBAC-Cre; R26REYFP pancreata under different genetic background. Shown are whole mount pictures. Dotted lines circle the pancreata. (A2–C4) IF showing the identities of EYFP+ pancreatic cells (E14.5). Panels marked with A, B, or C showed images from one individual pancreas, respectively. Panels labeled with 2, 3, and 4 showed the identity of EYFP+ cells with different antibodies (Amy: amylase; Horm: mix of all hormones). Scale bars = 50 μm. (D) % of EYFP+ cells that are of islet, acinar, and duct cells (n = 6). Neonatal tissues were used in D.

S. Wang et al. / Developmental Biology 339 (2010) 26–37

33

Fig. 6. More pancreatic progenitor cells activate Neurog3 expression with reduced Neurog3 dosage at E14.5. (A and B) Immuno-characterization of EGFP+ cells in Neurog3EGFP/+ and Neurog3EGFP/F pancreata and sorted cells. DAPI was used to mark nuclei. (A1 and B1) Expression analysis of unsorted pancreatic cells. Note the large number of EGFP+ cells (green) and amylase (blue) or hormone expressing (red) cells. (A2, A3, B2, and B3) Characterization of sorted EGFP+ cells. Note that the sorted cells do not express amylase and only few EGFP+ cells express hormone (white arrow in A3). A4 and B4 show the same views as A3 and B3, respectively, yet with only the green channel shown. (C) Average number of EGFP+ cells in Neurog3EGFP/+ and Neurog3EGFP/F pancreata. (D) Ki67 expression in Neurog3EGFP/+ and Neurog3EGFP/F pancreata. Note the lack of Ki67 signal in endocrine progenitor cells (that express Neurog3, green) or endocrine cells (that express hormones, blue) at this stage. Scale bar=40 μm.

in Neurog3+/− animals, the faster division of exocrine cells could lead to more EYFP+ descendents. This possibility is known to play a role for the increased EYFP+ pancreatic cells in Neurog3+/− mice because acinar cells divide twice as frequently as endocrine islet cells (Fig. 4 and supplementary Fig. 4). However, the fact that more Neurog3+ cells were allocated to the exocrine progenitor pool also suggests the third possibility: more pancreatic progenitors might have activated Neurog3 expression when its gene dosage was reduced, because reduced number of Neurog3HI cells had likely weakened Neurog3mediated lateral inhibition. In order to examine this third possibility, we directly examined the number of Neurog3+ cells in the embryonic pancreas. The number of Neurog3+ cells was counted utilizing the Neurog3EGFP knockin mice (Lee et al., 2002). In this allele, EGFP replaces the endogenous Neurog3 coding sequence. The selection of this line instead of the Neurog3EGFP-tg line (a transgenic line with a relatively short promoter region; Gu et al., 2004) was made to try to ensure that EGFP production recapitulates that of the endogenous Neurog3 gene. The limitation is that we could not assay the number of Neurog3+ cells in the wild-type background. Yet this deficiency did

not prevent us from examining the basic principle of whether or not a reduced Neurog3 gene function would result in an increase in the number of Neurog3+ cells. We did not utilize antibodies to directly assay the number of Neurog3+ cells, because of the inherent problem of immunostaining in defining the signal-to-background ratios between different samples, especially when the gene expression levels are low (See Supplementary Fig. 1 for an example). At E14.5, when large numbers of endocrine progenitor cells appeared and exocrine differentiation started to occur, Neurog3EGFP/+ and Neurog3EGFP/F pancreatic buds were isolated and dissociated into single cells. The samples were subjected to analysis by fluorescenceactivated cell sorting, which counted and simultaneously purified the EGFP+ cells. Collected cells were spun onto slides for purity assays. Over 97% of the FACS-sorted cells emitted detectable green fluorescence (with 1000–2000 cells counted for each sample. Figs. 6A and B). We could not detect any amylase+ cells in the sorted samples. Only occasionally could we detect cells that express endocrine hormones (Figs. 6A3 and B3). Indeed, we found that the number of Neurog3+ cells was substantially greater in the Neurog3EGFP/Fthan in the Neurog3EGFP/+ animals (Fig. 6C).

34

S. Wang et al. / Developmental Biology 339 (2010) 26–37

Next, we examined whether or not Neurog3+ progenitor cells have a higher division rate in Neurog3EGFP/+ than in Neurog3EGFP/F pancreata. Ki67, which is expressed in all stages of cell cycles of proliferating cells, was chosen to examine the proliferation status of Neurog3+ cells at E13.5, E14.5, and E15.5, which are the three stages when large numbers of Neurog3+ cells were produced. We again utilized EGFP production in both Neurog3EGFP/+ and Neurog3EGFP/F animals as a surrogate for Neurog3 in order to avoid artificially defining an immunodetection Neurog3 signal. As reported in wildtype Neurog3+ cells (Jensen et al., 2000a), most of the EGFP+ cells in the Neurog3EGFP/+ and Neurog3EGFP/F pancreata did not express Ki67 (Fig. 6D and data not shown). These findings suggest that the increased number of Neurog3+ cells in Neurog3EGFP/Fthan in the Neurog3EGFP/+ pancreas results from increasing the numbers of pancreatic progenitors that activate Neurog3 expression, rather than from an increased mitotic division of the Neurog3+ cells. Reducing Neurog3 activity does not reduce Notch ligand expression at a global level Overall, the findings above demonstrate that the Neurog3 protein level within specific cells can feed back to the epithelium to regulate the production of Neurog3+ cells, which is consistent with the hypothesis that Neurog3 expression is linked to a lateral inhibition process (Skipper and Lewis, 2000). In the classical Notch-based lateral inhibition model, down-regulation of nuclear Notch activity lowers the Hes/Hey expression that is responsible for activating the cellautonomous proneural gene expression (which is the pro-endocrine gene Neurog3 in endocrine pancreatic development). One function of the proneural gene is to raise the expression of Notch ligands, Delta or Jagged, which consequently potentiate Notch activity in neighboring progenitors, and repress expression of proneural genes in these neighbors. Because our above finding is consistent with Neurog3 normally being intimately linked to lateral inhibition, such that its expression is prevented from spreading to neighboring cells, we tested for evidence of which Notch ligand was utilized. We again utilized the knockin mouse line, Neurog3EGFP, to isolate EGFP+ cells in both the Neurog3EGFP/+ (heterozygous) and Neurog3EGFP/F (Hypomorphic) backgrounds. Since we expect the Notch pathway to operate via a similar mechanism in Neurog3+/+,

Neurog3EGFP/+, or Neurog3EGFP/F cells, we omit wild-type Neurog3+ cells (which are untagged and therefore not sortable) for this analysis, which, in principle, should not affect our data interpretation. EGFP+ cells were flow-cytometrically isolated from E14.5 embryonic Neurog3EGFP/+ and Neurog3EGFP/F pancreata to over 95% purity. QRT-PCR analysis of several well-established Notch components was carried out [(Fig. 7). Hes3, Hes5, Notch2, and Notch3 expression levels were below the detection limits and were not investigated in detail]. As expected, lowered Neurog3 production in the Neurog3EGFP/Fpancreas resulted in higher Notch activity, that is, EGFP+ cells from Neurog3EGFP/Fpancreata had higher levels of Hes1, Hey1, and Notch1 expression than EGFP+ cells from Neurog3EGFP/+ pancreata (Fig. 7). The expression of Hes6, although detected at high levels in EGFP+ cells, did not vary between Neurog3EGFP/+ and Neurog3EGFP/Fpancreata (Fig. 7). Consistent with a recent report that Mfng expression depends on Neurog3 activity (Golson et al., 2009; Svensson et al., 2009), its expression was significantly reduced in Neurog3EGFP/Fpancreata from that in Neurog3EGFP/+cells. Surprisingly, the EGFP+ cells from the Neurog3EGFP/+and Neurog3EGFP/F pancreata produced similar levels of mRNA for Delta1, Delta3, Delta4, and Jagged2. Contrary to our prediction, the expression of another Notch ligand, Jagged1, was significantly increased with lowered Neurog3 dosage. These somewhat confounding results suggest, but do not prove, that Neurog3 does not utilize increases in the classical Notch ligand, at the transcription level, to mediate lateral inhibition in pancreatic progenitors. We also assayed the expression of several pancreatic progenitor markers, including Ptf1a (Kawaguchi et al., 2002), Mist1 (Pin et al., 2001), and Sox9 (Seymour et al., 2007). We found that both Ptf1a and Mist1 mRNAs were detected at low levels in sorted EGFP+ cells. Their expression levels did not vary in the Neurog3+ cells of Neurog3EGFP/+ and Neurog3EGFP/F pancreata. This finding suggests that the reduction of Neurog3 dosage does not directly activate the exocrine differentiation or repress the pancreatic progenitor program in Neurog3+ cells. In contrast, the Neurog3+ cells in Neurog3EGFP/F animals had a14-fold increase of Sox9 mRNA than those in the Neurog3EGFP/+ animals. This finding suggests that either the Sox9+ progenitors activate Neurog3 expression prematurely or that the Neurog3+ cells maintain Sox9 expression more highly in Neurog3EGFP/F pancreata than in Neurog3EGFP/+ pancreata.

Fig. 7. Neurog3LO progenitor cells have a higher Notch activity but not-reduced Notch ligand transcription. QRT-PCR assay of mRNA levels for several well established Notch components and pancreatic progenitor markers in Neurog3EGFP+ cells. ⁎P b 0.05.

S. Wang et al. / Developmental Biology 339 (2010) 26–37

35

Discussion Endocrine islet cells have a limited capacity to divide under normal physiological conditions, and this process may be restricted further by isotype-specific proliferation [e.g., beta cells only producing beta cells (Herrera, 2000; Dor et al., 2004)]. Therefore, islet mass in individual animals is predominantly determined by the controlled allocation of pancreatic progenitors to the endocrine lineage during organogenesis (Stanger et al., 2007). While it is known that activation of Neurog3 in pancreatic progenitors is essential for cells to adopt the endocrine fate, it remains unclear how only a specific portion of pancreatic progenitors activates Neurog3 expression, or how the Neurog3 level within each cell affects endocrine differentiation. Here, we show that the high level of Neurog3 expression within individual progenitor cells is critical for the allocation of pancreatic progenitor cells towards the endocrine cell lineage. We further show that although reduced Neurog3 expression causes more pancreatic progenitors to activate low levels of Neurog3 expression (Neurog3LO), most of these Neurog3LO cells are incapable of becoming endocrine cells, and many move to exocrine fates. Thus, a reduction in Neurog3 dosage results in a net reduction of endocrine differentiation. High Neurog3 activity drives endocrine commitment Examination of Neurog3 expression at various stages of pancreas organogenesis detects cells with low (Neurog3LO) or high (Neurog3HI) levels of Neurog3 (Apelqvist et al., 1999; Gradwohl et al., 2000; Gu et al., 2002; Schwitzgebel et al., 2000; Villasenor et al., 2008) (D. Boyer and CW, submitted). It is possible that the Neurog3LO and Neurog3HI cells represent different time-points in a process whereby equivalent progenitor cells move relatively rapidly from Neurog3− (progenitor status) to Neurog3LO, up to Neurog3HI then back to Neurog3LO/ Neurog3− (these cells now having adopted a committed-endocrinestatus) (Fig. 8A). In this case, all Neurog3+ cells would be expected to reach identical fates. Alternatively, some Neurog3LO progenitor cells may “back-up” or become diverted (even quite directly) into the exocrine fate, or they might die. Direct lineage tracing has begun to shed light on these two possibilities. When a CreERT-based lineage study was performed in NeurogCreERT knock-in animals, most, if not all, CreERTexpressing cells give rise to endocrine islet cells (Johansson et al., 2007). Because the limited CreERT cell-labeling efficiency only allowed for the marking of Neurog3HI cells, this observation demonstrated that most, if not all, Neurog3HI cells move forward into the endocrine program. Conversely, Neurog3TgBAC-Cre-based lineage tracing showed that, even in the wild-type situation, a portion of Neurog3-expressing precursors gives rise to exocrine cells (Schonhoff et al., 2004). These combined findings lead to the hypothesis that the Neurog3LO progenitor cells maintain some degree of plasticity towards exocrine fates. Our findings here demonstrate that, indeed, Neurog3 activity must reach a high level for endocrine differentiation to occur. Most Neurog3LO cells, if they fail to activate high Neurog3 expression, adopt exocrine (acinar or ductal) cell fates. Therefore, pancreatic progenitors maintain their multipotency, i.e. the ability of a group of cells to give rise to multiple cell types, until their respective Neurog3 expression reaches a specific level. In other words, pancreatic progenitors are not pre-selected to become endocrine progenitors before Neurog3 activation. This Neurog3 level “check point” provides pancreatic MPCs with some tolerance to the fluctuation of endocrine-inducing signals: if low levels of Neurog3 were to be induced in most pancreatic progenitors either due to stochastic events, gene expression fluctuation, or even in the presence of certain gene mutations, progenitors would not be forced to differentiate into endocrine islet cells. This mechanism would provide a way to prevent premature pancreatic MPCs exhaustion because of improper endocrine differentiation. Why a low level of Neurog3 expression is not sufficient to activate the endocrine differentiation program, or repress exocrine

Fig. 8. A model on Neurog3 expression level and pancreatic cell differentiation. (A) Pancreatic progenitors do not express detectable levels of Neurog3. Stochastic events or other processes activate low Neurog3 expression to start a possible journey towards endocrine differentiation. Neurog3LO cells could go on to become Neurog3HI cells, under the direction of unknown molecules, and differentiate into endocrine islet cells. A portion of Neurog3LO cells could also revert to Neurog3− (OFF) progenitor status or directly adopt exocrine cell fate. (B) Compensatory islet cell production with limited Neurog3 expression reduction. In wild type pancreas, most Neurog3+ cells produce high enough Neurog3 activity to allow for endocrine differentiation. In Neurog3+/− pancreas, the average level of Neuog3 is lowered. However, a weakened lateral inhibition increases the number of progenitor cells that express a medium-high level of Neurog3, which partially compensates for the loss of progenitors that express a very high level of Neurog3.

differentiation, is not clear. It is likely that the various Neurog3 target genes possess different Neurog3 recognition sequences or combinations of different motifs with different affinities to Neurog3. As a result, high levels of Neurog3 are required to activate a set of targets, which are all necessary for endocrine differentiation. At low levels, Neurog3 might activate only some of its targets, a situation insufficient to commit and move from the progenitor state. Our recent finding of Neurog3 function in differentiated β cells is consistent with this type of speculation (Wang et al., 2009). In that study, we found that differentiated islet cells maintain a low yet detectable level of Neurog3 expression. We further showed that a selective deletion of Neurog3 in newly born or mature adult β cells compromised the expression of several, but not all, known Neurog3 targets (Wang et al., 2009). This observed Neurog3 requirement for endocrine gene expression in differentiated islet cells does not appear to be a consequence of compromised endocrine differentiation in Neurog3F/− animals. As we have shown here, the β cells of the Neurog3F/− pancreas have enhanced insulin secretion in response to glucose stimulation. Nevertheless, whether different levels of Neurog3 activate the expression of different endocrine gene sets could only be directly examined with the future availability of high affinity/specificity Neurog3 antibodies. Neurog3 mediates lateral inhibition or unknown mechanisms to regulate the number of Neurog3 expressing cells Reduced Neurog3 dosage results in Neurog3 expression in more pancreatic progenitors, demonstrating the reverse correlation between the Neurog3 level per cell and the number of Neurog3+ cells in the developing pancreas. This type of feedback mechanism likely promotes the production of adequate number of endocrine islet cells. To this end, Neurog3+/− pancreas tends to produce less Neurog3HI

36

S. Wang et al. / Developmental Biology 339 (2010) 26–37

cells. Yet the reduction in the number of Neurog3HI cells leads to more pancreatic epithelial cells to activate Neurog3 expression. As a result, these “extra” Neurog3+ (most likely expressing a ‘medium level’ of Neurog3) will partially compensate for the loss of some progenitors that would have normally expressed the highest levels of Neurog3 to adopt endocrine islet cell fate (Fig. 8B). Surprisingly, a lowered Neurog3 activity does not significantly reduce the transcription levels of any Notch ligand that has been tested, a finding seemed to contradict the hypothesis that Neurog3 normally up-regulates Notch ligand expression to mediate the classical lateral inhibition (Skipper and Lewis, 2000). It is possible that the real time RT-PCR assays are not sensitive enough to detect a small reduction in the transcription of individual Notch ligands (e. g., Delta1 in Fig. 7). Alternatively, Neurog3 could utilize post-transcriptional mechanism to regulate Notch activity. For example, Neurog3 could regulate the sub-cellular localization of Notch ligands. It is also possible Neurog3 could modulate Notch activity through modifiers such as Mfng (whose expression is reduced when Neurog3 expression level is reduced), as demonstrated by biochemical analysis (Panin et al., 2002). Lastly, it is also possible that the classical lateral inhibition does not apply in the Notch-Ngn3 pathway. Instead, the altered endocrine differentiation could in turn affect Neurog3 expression indirectly through non-Notch ligands (eg., some unknown soluble molecules) or non-Notch receptor dependent pathways. It would be of interest in the future to explore these latter possibilities. In summary, our analysis revealed a previously unrecognized impact of Neurog3 level on endocrine production. These findings underscore the importance of understanding Neurog3 expression control, including its expression pattern and its expression level in individual cells, for both endocrine differentiation and diabetes development for future studies. Acknowledgments We thank Dr. Jim Goldenring and Frank Revetta for their help in using the Ariol SL-50 system. We also thank Palle Serup and Yuval Dor for their participation in useful discussions. Finally, we are grateful to Klaus Kaestner for the Neurog3EGFP mouse line and to Roland Stein for the MafB antibodies. This research was supported by grants from the NIH (DK065949 to GG and DK42502to CW). Cell Imaging Core Services performed through Vanderbilt University Medical Center's Digestive Disease Research Center supported by NIH grant P30DK058404. Flow Cytometry experiments were performed in the VMC Flow Cytometry Shared Resource. The VMC Flow Cytometry Shared Resource is supported by the Vanderbilt Ingram Cancer Center (P30 CA68485) and the Vanderbilt Digestive Disease Research Center (DK058404). Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.ydbio.2009.12.009. References Amrani, N., Dong, S., He, F., Ganesan, R., Ghosh, S., Kervestin, S., Li, C., Mangus, D.A., Spatrick, P., Jacobson, A., 2006. Aberrant termination triggers nonsense-mediated mRNA decay. Biochem. Soc. Trans. 34, 39–42. Apelqvist, A., Li, H., Sommer, L., Beatus, P., Anderson, D.J., Honjo, T., Hrabe de Angelis, M., Lendahl, U., Edlund, H., 1999. Notch signalling controls pancreatic cell differentiation. Nature 400, 877–881. Artner, I., Blanchi, B., Raum, J.C., Guo, M., Kaneko, T., Cordes, S., Sieweke, M., Stein, R., 2007. MafB is required for islet beta cell maturation. Proc. Natl. Acad. Sci. U. S. A. 104, 3853–3858. Chiang, M.K., Melton, D.A., 2003. Single-cell transcript analysis of pancreas development. Dev. Cell 4, 383–393. Dor, Y., Brown, J., Martinez, O.I., Melton, D.A., 2004. Adult pancreatic beta-cells are formed by self-duplication rather than stem-cell differentiation. Nature 429, 41–46. Fujikura, J., Hosoda, K., Iwakura, H., Tomita, T., Noguchi, M., Masuzaki, H., Tanigaki, K.,

Yabe, D., Honjo, T., Nakao, K., 2006. Notch/Rbp-j signaling prevents premature endocrine and ductal cell differentiation in the pancreas. Cell Metab. 3, 59–65. Golson, M.L., Le Lay, J., Gao, N., Bramswig, N., Loomes, K.M., Oakey, R., May, C.L., White, P., Kaestner, K.H., 2009. Jagged1 is a competitive inhibitor of Notch signaling in the embryonic pancreas. Mech. Dev. 126, 687–699. Gradwohl, G., Dierich, A., LeMeur, M., Guillemot, F., 2000. 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. Grapin-Botton, A., Majithia, A.R., Melton, D.A., 2001. Key events of pancreas formation are triggered in gut endoderm by ectopic expression of pancreatic regulatory genes. Genes Dev. 15, 444–454. Gu, G., Dubauskaite, J., Melton, D.A., 2002. Direct evidence for the pancreatic lineage: NGN3+ cells are islet progenitors and are distinct from duct progenitors. Development 129, 2447–2457. Gu, G., Wells, J.M., Dombkowski, D., Preffer, F., Aronow, B., Melton, D.A., 2004. Global expression analysis of gene regulatory pathways during endocrine pancreatic development. Development 131, 165–179. Herrera, P.L., 2000. Adult insulin- and glucagon-producing cells differentiate from two independent cell lineages. Development 127, 2317–2322. Huang, H.P., Liu, M., El-Hodiri, H.M., Chu, K., Jamrich, M., Tsai, M.J., 2000. Regulation of the pancreatic islet-specific gene BETA2 (neuroD) by neurogenin 3. Mol. Cell. Biol. 20, 3292–3307. Jensen, J., Heller, R.S., Funder-Nielsen, T., Pedersen, E.E., Lindsell, C., Weinmaster, G., Madsen, O.D., Serup, P., 2000a. Independent development of pancreatic alpha- and beta-cells from neurogenin3-expressing precursors: a role for the notch pathway in repression of premature differentiation. Diabetes 49, 163–176. Jensen, J., Pedersen, E.E., Galante, P., Hald, J., Heller, R.S., Ishibashi, M., Kageyama, R., Guillemot, F., Serup, P., Madsen, O.D., 2000b. Control of endodermal endocrine development by Hes-1. Nat. Genet. 24, 36–44. Johansson, K.A., Dursun, U., Jordan, N., Gu, G., Beermann, F., Gradwohl, G., GrapinBotton, A., 2007. Temporal control of neurogenin3 activity in pancreas progenitors reveals competence windows for the generation of different endocrine cell types. Dev. Cell 12, 457–465. Kageyama, R., Ohtsuka, T., Shimojo, H., Imayoshi, I., 2008. Dynamic Notch signaling in neural progenitor cells and a revised view of lateral inhibition. Nat. Neurosci. 11, 1247–1251. Kawaguchi, Y., Cooper, B., Gannon, M., Ray, M., MacDonald, R.J., Wright, C.V., 2002. The role of the transcriptional regulator Ptf1a in converting intestinal to pancreatic progenitors. Nat. Genet. 32, 128–134. Lee, C.S., Perreault, N., Brestelli, J.E., Kaestner, K.H., 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. Lewis, J., 1998. Notch signalling and the control of cell fate choices in vertebrates. Semin. Cell Dev. Biol. 9, 583–589. Lobe, C.G., Koop, K.E., Kreppner, W., Lomeli, H., Gertsenstein, M., Nagy, A., 1999. Z/AP, a double reporter for cre-mediated recombination. Dev. Biol. 208, 281–292. Meinhardt, H., Gierer, A., 1974. Applications of a theory of biological pattern formation based on lateral inhibition. J. Cell Sci. 15, 321–346. Mellitzer, G., Martin, M., Sidhoum-Jenny, M., Orvain, C., Barths, J., Seymour, P.A., Sander, M., Gradwohl, G., 2004. Pancreatic islet progenitor cells in neurogenin 3-yellow fluorescent protein knock-add-on mice. Mol. Endocrinol. 18, 2765–2776. Mellitzer, G., Bonne, S., Luco, R.F., Van De Casteele, M., Lenne-Samuel, N., Collombat, P., Mansouri, A., Lee, J., Lan, M., Pipeleers, D., et al., 2006. IA1 is NGN3-dependent and essential for differentiation of the endocrine pancreas. EMBO J. 25, 1344–1352. Panin, V.M., Shao, L., Lei, L., Moloney, D.J., Irvine, K.D., Haltiwanger, R.S., 2002. Notch ligands are substrates for protein O-fucosyltransferase-1 and Fringe. J. Biol. Chem. 277, 29945–29952. Petri, A., Ahnfelt-Ronne, J., Frederiksen, K.S., Edwards, D.G., Madsen, D., Serup, P., Fleckner, J., Heller, R.S., 2006. The effect of neurogenin3 deficiency on pancreatic gene expression in embryonic mice. J. Mol. Endocrinol. 37, 301–316. Pin, C.L., Rukstalis, J.M., Johnson, C., Konieczny, S.F., 2001. The bHLH transcription factor Mist1 is required to maintain exocrine pancreas cell organization and acinar cell identity. J. Cell Biol. 155, 519–530. Prentki, M., Nolan, C.J., 2006. Islet beta cell failure in type 2 diabetes. J. Clin. Invest. 116, 1802–1812. Schonhoff, S.E., Giel-Moloney, M., Leiter, A.B., 2004. Neurogenin 3-expressing progenitor cells in the gastrointestinal tract differentiate into both endocrine and non-endocrine cell types. Dev. Biol. 270, 443–454. Schwitzgebel, V.M., Scheel, D.W., Conners, J.R., Kalamaras, J., Lee, J.E., Anderson, D.J., Sussel, L., Johnson, J.D., German, M.S., 2000. Expression of neurogenin3 reveals an islet cell precursor population in the pancreas. Development 127, 3533–3542. Seymour, P.A., Freude, K.K., Tran, M.N., Mayes, E.E., Jensen, J., Kist, R., Scherer, G., Sander, M., 2007. SOX9 is required for maintenance of the pancreatic progenitor cell pool. Proc. Natl. Acad. Sci. U. S. A. 104, 1865–1870. Skipper, M., Lewis, J., 2000. Getting to the guts of enteroendocrine differentiation. Nat. Genet. 24, 3–4. Srinivas, S., Watanabe, T., Lin, C.S., William, C.M., Tanabe, Y., Jessell, T.M., Costantini, F., 2001. Cre reporter strains produced by targeted insertion of EYFP and ECFP into the ROSA26 locus. BMC Dev. Biol. 1, 4. Stanger, B.Z., Tanaka, A.J., Melton, D.A., 2007. Organ size is limited by the number of embryonic progenitor cells in the pancreas but not the liver. Nature 445, 886–891. Svensson, P., Bergqvist, I., Norlin, S., Edlund, H., 2009. MFng is dispensable for mouse pancreas development and function. Mol. Cell. Biol. 29, 2129–2138.

S. Wang et al. / Developmental Biology 339 (2010) 26–37 Villasenor, A., Chong, D.C., Cleaver, O., 2008. Biphasic Ngn3 expression in the developing pancreas. Dev. Dyn. 237, 3270–3279. Wang, S., Hecksher-Sorensen, J., Xu, Y., Zhao, A., Dor, Y., Rosenberg, L., Serup, P., Gu, G., 2008. Myt1 and Ngn3 form a feed-forward expression loop to promote endocrine islet cell differentiation. Dev. Biol. 317, 531–540. Wang, S., Zhang, J., Zhao, A., Hipkens, S., Magnuson, M.A., Gu, G., 2007. Loss of Myt1 function partially compromises endocrine islet cell differentiation and pancreatic physiological function in the mouse. Mech. Dev. 124, 898–910. Wang, S., Jensen, J.N., Seymour, P.A., Hsu, W., Dor, Y., Sander, M., Magnuson, M.A., Serup, P., Gu, G., 2009. Sustained Neurog3 expression in hormone-expressing islet cells is required for endocrine maturation and function. Proc. Natl. Acad. Sci. U. S. A. 106, 9715–9720.

37

White, P., May, C.L., Lamounier, R.N., Brestelli, J.E., Kaestner, K.H., 2008. Defining pancreatic endocrine precursors and their descendants. Diabetes 57, 654–668. Yang, Y.P., Wright, C., 2009. Chemicals turn human embryonic stem cells towards beta cells. Nat. Chem. Biol. 5, 195–196. Zaret, K.S., 2009. Using small molecules to great effect in stem cell differentiation. Cell Stem Cell 4, 373–374. Zhou, Q., Law, A.C., Rajagopal, J., Anderson, W.J., Gray, P.A., Melton, D.A., 2007. A multipotent progenitor domain guides pancreatic organogenesis. Dev. Cell 13, 103–114. Zhou, Q., Brown, J., Kanarek, A., Rajagopal, J., Melton, D.A., 2008. In vivo reprogramming of adult pancreatic exocrine cells to beta-cells. Nature 455, 627–632.