Notch signaling is required for the maintenance of enteric neural crest progenitors

The neural crest is a pluripotent cell population derived from the neural plate during the early stages of embryogenesis. Neural crest cells (NCCs)disperse from the dorsal surface of the neural tube and migrate extensively throughout the embryo. This gives rise to a wide variety of differentiated cell types, including neurons, glial cells, pigment cells, facial cartilage and bone, and connective tissues (Le Douarin et al., 2004). The fate of individual NCCs largely depends upon the locations to which they migrate. Some NCCs contribute to the enteric nervous system (ENS), which yields neurons and glial cells. The ENS arises from vagal and sacral NCCs.

At E8.5, the vagal NCCs detach from the dorsal neural tube at somite level 1-7, colonize the foregut at E9.5, then migrate in a caudal direction and reach the rectum by E14 (Young et al.,1998). Sacral NCCs are delaminated from the neural tube that is posterior to somite 28. Some of these cells give rise to pelvic ganglia,whereas others contribute to the ENS by colonizing the hindgut in a caudal-rostral migratory path (Druckenbrod and Epstein, 2005; Kapur,2000).

The ENS constitutes a part of the autonomic nervous system that regulates peristalsis, secretion, blood supply and the immune response in the intestinal tract. Defects in neural crest development are a significant cause of Hirschsprung disease in humans, which is characterized by distal intestinal aganglionosis and which occurs about once in every 5000 live births(Swenson, 2002). Many genes,including Edn3, Ednrb, Sox10, Ret, Gfra1 and Gdnf, have been implicated in the pathogenesis of congenital megacolon, and these genes control the development of the ENS through their coordinated activities(Barlow et al., 2003; Baynash et al., 1994; Enomoto et al., 1998; Herbarth et al., 1998; Hosoda et al., 1994; Schuchardt et al., 1994; Stanchina et al., 2006). During the development of the ENS, the transcriptional regulator Sox10 is essential for the maintenance of neural crest progenitors(Paratore et al., 2002). The expression of Sox10, a member of the SRY-like HMG-box family of transcription factors, is initiated in NCCs when they detach from the neural tube and is maintained during NCC migration. Eventually, Sox10 expression is turned off in the NCC derivatives, except for the glial and melanocyte lineages(Kuhlbrodt et al., 1998; Pusch et al., 1998). Two mutations of Sox10 have been examined in mice: a null mutation generated by targeted insertion of lacZ(Sox10^lacZ), and a spontaneous mutation referred to as Dominant megacolon (Sox10^Dom)(Britsch et al., 2001; Herbarth et al., 1998; Southard-Smith et al., 1998). In the heterozygous state, both mutants lack enteric ganglia in a variable length of the colon and exhibit pigmentation defects. Homozygous mutants display severe deficits in several neural crest derivatives, including complete absence of the ENS (Britsch et al., 2001; Kapur,1999). By contrast, the overexpression of Sox10 in neural crest stem cells (NCSCs) preserves both glial and neuronal differentiation potentials (Kim et al.,2003).

In addition to these mutants, it is known that Mash1(Ascl1 - Mouse Genome Informatics) mutant mice also show defects in ENS development. Mash1 encodes a member of the basic helix-loop-helix family of transcriptional regulators and is a mammalian homolog of the Drosophila proneural genes of the achaete-scute complex; Mash1 is expressed in both the central nervous system (CNS) and the peripheral nervous system and is induced by bone morphogenetic protein 2(Bmp2) in NCSCs (Shah et al.,1996). Mash1 mutant mice show defects in the development of olfactory and peripheral autonomic neurons and a loss of part of the enteric neuron lineages (Blaugrund et al.,1996; Guillemot et al.,1993).

Notch is one of the transmembrane receptors activated by the juxtacrine interactions of specific ligands with Notch receptors(Artavanis-Tsakonas et al.,1999). Once activated, the Notch receptors then undergo a series of proteolytic cleavages, resulting in the release of the Notch intracellular domain (NICD) (Mumm et al.,2000). Subsequently, the NICD translocates into the nucleus,associates with the transcription factor RBP-Jκ (Rbpj - Mouse Genome Informatics) to generate the transactivation complex, and thereby initiates the transcription of target genes such as Hes1(Kageyama et al., 2000). In addition to the four Notch receptors (Notch1-4) and five ligands (delta-like 1, 3, 4 and jagged 1, 2) that have been identified in vertebrates, there are many auxiliary factors through which Notch signaling is tightly regulated(Bray, 2006). One of these,protein O-fucosyltransferase 1 (Pofut1), which transfers O-fucose to Notch EGF repeats, is an essential component of Notch signaling and functions upstream of NICD, where it regulates Notch-Delta interactions and trafficking of the Notch receptor(Okajima and Irvine, 2002; Okajima et al., 2003; Okajima et al., 2005; Okamura and Saga, 2008; Sasamura et al., 2007; Sasamura et al., 2003; Shi and Stanley, 2003). Although Pofut1 is ubiquitously expressed in the mouse, its removal leads to global Notch signaling defects similar to those associated with the loss of RBP-Jk, the common mediator of Notch signaling in mice(Oka et al., 1995). This indicates that Pofut1 functions are largely restricted to the Notch signaling pathway.

Notch signaling is involved in various aspects of neurogenesis. In the CNS it is essential for the maintenance of neural stem cells(Hitoshi et al., 2002). Transient activation of Notch signaling is not sufficient to maintain NCSCs. Rather, it acts positively to promote switching from neurogenesis to gliogenesis in vitro (Morrison et al.,2000) and in vivo (Wakamatsu et al., 2000). In addition, Notch signaling regulates sympathetic neuron development (Tsarovina et al.,2008), cardiac neural crest differentiation(High et al., 2007),melanocyte development (Moriyama et al.,2006) and mesencephalic neural crest differentiation(Ijuin et al., 2008). However,little is known regarding the role of Notch signaling in enteric neural crest development in vivo.

Here, we have examined the role of Notch signaling in the ENS by specifically eliminating the Pofut1 gene in NCCs. These Pofut1 conditional knockout (cKO) embryos showed a reduction in enteric neural crest cells (ENCCs) and premature neurogenesis, and this was accompanied by the loss of Sox10 expression and by an increase in the number of Mash1-positive ENCCs. These results suggest that Notch signaling is necessary for the maintenance of enteric neural crest progenitors via the regulation of Sox10 and Mash1 expression.

The generation of a conditional allele of the Pofut1 locus has recently been described (Okamura and Saga,2008). The generation of CAG-CAT-EGFP, Wnt1-Cre and R26R reporter mice has also been reported(Ahmed et al., 2002; Chai et al., 2000; Jiang et al., 2000; Soriano, 1999). Mouse tail or embryo yolk sac DNA was used for genotyping. Tissue samples were heated at 95°C for 5 minutes and then incubated in 100 mM Tris (pH 8), 5 mM EDTA,300 μg/ml proteinase K at 55°C overnight. After heat denaturation at 95°C for 5 minutes, tissue suspensions were used as PCR templates. PCR primers used for genotyping were as follows: Pofut1 mice, LA-4.9(5′-CCATTGTGCGGTGCATTGTG-3′), cprb-L1(5′-GCACTCTGGGGCTCTGCCGT-3′) and SA-R0.01(5′-CTGTACCCAGGCAAGTAGGG-3′); Wnt1-Cre mice, Cre1(5′-GGACATGTTCAGGGATCGCCAGGCG-3′) and Cre2(5′-GCATAACCAGTGAAACAGCATTGCTG-3′); Rosa26 reporter, Rosa-2(5′-GCGAAGAGTTTGTCCTCAACC-3′), Rosa-3(5′-GGAGCGGGAGAAATGGATATG-3′), Rosa-4(5′-AAAGTCGCTCTGAGTTGTTAT-3′); CAG-CAT-EGFP mice, CAT2(5′-CAGTCAGTTGCTCAATGTACC-3′), CAT3(5′-ACTGGTGAAACTCACCCA-3′).

Whole-mount X-Gal staining of embryos was as described(Saga et al., 1992). The number of X-Gal-positive ENCCs was counted and quantified as the number per unit area (100 μm²). Whole-mount immunostaining using monoclonal anti-neurofilament antibody 2H3 (Developmental Studies Hybridoma Bank) was performed as described previously (Matsuo et al., 1995). For section immunohistochemistry, embryos were fixed in 4% paraformaldehyde (PFA) in PBS at 4°C, cryoprotected in 30%sucrose in PBS at 4°C, embedded in OCT compound (Tissue-Tek) and sectioned(7 μm). The sections were blocked in PBS containing 10% normal donkey serum(Chemicon) for 1 hour at room temperature, then incubated overnight at 4°C in a mixture of the following primary antibodies in blocking solution:anti-GFP (Nacalai Tesque), anti-activated caspase 3 (Cell Signaling Technology), anti-β-tubulin isotype III (TuJ1) (Sigma), anti-BFABP (a gift from T. Müller, Max-Delbruck Center, Berlin, Germany), anti-Sox10(Santa Cruz Biotechnology), anti-Hes1 (a gift from N. Brown, Cincinnati Children's Hospital Medical Center, Cincinnati, OH) and anti-Mash1 (BD Pharmingen). After washing, these were incubated with a mixture of Alexa Fluor 488- and 594- (Molecular Probes) or Cy3- (Jackson Laboratories) conjugated secondary antibodies.

For BrdU labeling, pregnant mice were injected intraperitoneally with BrdU(Sigma) at 10 mg/kg body weight and sacrificed 2 hours later. Embryos were handled as described above and sectioned (7 μm). After blocking, these sections were incubated overnight at 4°C in anti-GFP or anti-Sox10 antibody in blocking solution. After washing, the samples were further incubated with Alexa Fluor 488-conjugated secondary antibody, washed in PBS and fixed in 4% PFA in PBS at 4°C. The specimens were then placed in 2M HCl for 30 minutes at 37°C, then washed in PBS. After blocking, the sections were incubated with anti-BrdU antibody (Sigma), washed in PBS and incubated with Cy3-conjugated secondary antibody.

For double staining using immunohistochemistry and in situ hybridization,frozen sections (12 μm) were incubated with anti-p75^NTR primary antibody (1:200, Promega) for the detection of ENCCs, followed by a biotinylated goat anti-rabbit IgG secondary antibody (1:200, Vector Laboratories). These sections were then hybridized with digoxigenin(DIG)-labeled antisense cRNA probes (Roche). The hybridized probes were detected using horseradish peroxidase-conjugated anti-DIG sheep antibody(Roche) and Cyanine 3-Tyramid Signal Detection Reagent (Perkin Elmer). p75^NTR was detected using horseradish peroxidase-conjugated Streptavidin (Roche) and fluorescein isothiocyanate-conjugated Tyramid signal detection (Perkin Elmer). Sections were counterstained with 0.5 μg/ml 4′,6-diamino-2-phenylindole (DAPI) for 10 minutes and examined using a scanning-laser confocal imaging system (Zeiss LSM510) or with an Olympus BX61 fluorescence microscope system with an ORCA-ER digital camera (Hamamatsu Photo). Subsequent analysis was performed using MetaMorph software (Universal Imaging). More than six sections from the stomach or intestine regions of three different embryos were counted in each experiment. Statistical analysis was performed using a Student's t-test. Differences were considered to be significant if the P-value was less than 0.05.

To achieve a conditional deletion of the Pofut1 locus in NCCs, we employed Wnt1-Cre transgenic mice that express Cre recombinase under the control of the Wnt1 promoter. The Cre recombinase is active in the early migratory neural crest population of these mice at all axial levels(Chai et al., 2000; Jiang et al., 2000). We also utilized the Rosa26 reporter (R26R) and CAG-CAT-EGFP, which mark the Cre-expressed cell lineages via β-galactosidase and EGFP expression, respectively, after Cre-mediated excision of a floxed PGK-neo or CAT cassette. To generate a Pofut1 deficiency in mouse NCCs, we crossed Pofut1^+/-; Wnt1-Cre mice with Pofut1^flox/flox mice harboring a R26R or CAG-CAT-EGFP allele. From this cross, we obtained two genotypes: Pofut1^+/flox; Wnt1-Cre (R26Ror CAG-CAT-EGFP), referred to hereafter as control; and Pofut1^-/flox; Wnt1-Cre (R26Ror CAG-CAT-EGFP), referred to as Pofut1 cKO.

The Pofut1 cKO mice died within 1 day of birth, but showed no obvious morphological abnormalities other than a relatively small body size(Fig. 1A). However, 18/20 newborn Pofut1 cKO mice had no milk in their stomachs(Fig. 1A, arrowheads). Since NCCs contribute to the formation of the ENS, we speculated that this phenotype might be due to defective ENS development. We examined enteric neuronal development in the gastrointestinal tract using anti-neurofilament antibody 2H3, which labels both extrinsic and intrinsic nerve fibers. The ganglion structure of the ENS was disrupted and dispersed and the nerve bundles in the Pofut1 cKO embryos were relatively thick by comparison with the dense reticulate pattern of the enteric ganglia in the control embryos(Fig. 1B-E). These results suggest that Notch signaling participates in ENS development.

We examined enteric neural crest development by lineage tracing with lacZ in R26R reporter mice(Fig. 2). In control;R26R embryos, the ENCCs migrated to and colonized the foregut at around E9.5, then continued to migrate through the gut mesenchyme in a caudal direction, completing their migration at around E14(Fig. 2A-D). Similar to control;R26R embryos, the ENCCs in Pofut1 cKO;R26Rembryos colonized the foregut and migrated to the hindgut by E11.5(Fig. 2E,F). However, the number of ENCCs in these embryos was greatly reduced after E12.5(Fig. 2G,H). Compared with control embryos, the number of ENCCs at E12.5 was estimated to be reduced to∼53% in the stomach, ∼47% in the foregut and ∼49% in the midgut and hindgut. Interestingly, specific regions in Pofut1cKO;R26R embryos showed a slightly elevated number of ENCCs as compared with other regions (see Fig. S1 in the supplementary material). In spite of their reduction in Pofut1 cKO;R26R embryos, the ENCCs migrated on a normal timetable and contributed throughout the gastrointestinal tract by E14 (Fig. 2E-H), indicating that ENCC migration is not overtly disrupted by the loss of Pofut1. Hence, Pofut1 cKO embryos exhibited an abnormal ENS network, for which a reduction in the number of ENCCs appeared responsible.

To determine the underlying cause of the reduction of ENCCs in Pofut1 cKO embryos, we examined the apoptotic and proliferation status of these cells (Fig. 3). For this analysis, we used a CAG-CAT-EGFP reporter to facilitate the immunohistological detection of ENCCs among other molecular markers. Apoptotic cells were not observed in control or Pofut1 cKO ENCCs at E11.5 and E12.5 (Fig. 3A-H). This indicates that the loss of ENCCs in Pofut1 cKO embryos is not due to an increase in apoptosis. However, our data do not exclude the possibility that non-apoptotic cell death contributes to the loss of ENCCs, as it has recently been shown that the conditional ablation of Gfra1 in mouse induces a loss of enteric NCCs via a caspase-independent mechanism(Uesaka et al., 2007).

We next examined proliferating cells in the gastrointestinal tract by BrdU-incorporation analyses at E11.5 and E12.5. The proliferation rate of ENCCs in Pofut1 cKO embryos was found to be significantly reduced in the intestinal tract at E12.5 (Fig. 3I-R). Interestingly, consistent with the observation that slightly increased numbers of ENCCs exist in the stomach and cecum (see Fig. S1 in the supplementary material), the proliferating ENCCs in Pofut1cKO embryos at E12.5 were mainly located in the stomach and in the vicinity of the cecum, whereas these cells were uniformly detected throughout the gastrointestinal tract in control embryos(Fig. 3K,L,O,P; see Fig. S2 in the supplementary material). These regions are known to be sources of Gdnf and Edn3 (Burns and Thapar, 2006). Gdnf is known to promote the proliferation of ENS progenitors(Chalazonitis et al., 1998; Hearn et al., 1998; Heuckeroth et al., 1998; Taraviras et al., 1999),although the role of Edn3/Ednrb signaling remains controversial(Bondurand et al., 2006; Kruger et al., 2003). In either case, the localized proliferation we observed in Pofut1 cKO embryos might well be influenced by such factors. Furthermore, despite their significant reduction in number, NCCs still contributed to development throughout the gastrointestinal tract in Pofut1 cKO embryos. This is likely to be due to the presence of a few proliferating ENCCs. Our results thus suggest that the loss of ENCCs in Pofut1 cKO embryos might be partially, although not solely, due to a decrease in their proliferation.

Since the decreased number of ENCCs in Pofut1 cKO embryos could also be explained by premature differentiation, we examined the differentiation status of ENCCs in these embryos using the post-mitotic neuron marker TuJ1 (Tubb3 - Mouse Genome Informatics)(Fig. 4). It has been reported that ENCCs produce neurons and glia during their migration. About 10-15% of the ENCC population commences differentiation into neurons at E10.5, and this gradually increases to ∼25% by E12.5(Young et al., 2002). As reported previously, differentiation was already detected at E10.5 (8% in the intestine), and increased to ∼37% by E11.5 in our control embryos(Fig. 4A-C,I,J). By contrast,although the proportion of Tuj1-positive cells in the Pofut1 cKO embryos was not significantly different from that of control embryos at E10.5(Fig. 4D,I), differentiation was strongly induced thereafter in ∼77% of the ENCCs in the Pofut1 cKO embryos at E12.5 (Fig. 4F). Premature neurogenesis could already be observed at E11.5(Fig. 4E,J), which is before a reduction in the number of these cells was detectable(Fig. 2C,G).

In addition to neural differentiation, glial differentiation was examined using an anti-BFABP (Fabp7 - Mouse Genome Informatics) antibody(Fig, 4G,H,K), which is a marker for glial progenitors. Glial differentiation is initiated around E11.5(Young et al., 2003). In Pofut1 cKO embryos, however, the generation of glial progenitors was reduced at E11.5 (Fig. 4H,K). Therefore, premature neuronal differentiation is likely to be the major reason for the observed reduction in the ENCC number in the Pofut1 cKO embryos. Hence, we speculated that this premature differentiation might be due to a defect in the maintenance of progenitor cells, which would indicate that Notch signaling is required for the maintenance of ENS progenitors.

We next addressed the question of how Notch signaling regulates the pluripotency of ENS progenitors by analyzing the expression of Sox10, which is essential for the maintenance of neural crest progenitors during ENS development (Bondurand et al.,2006; Paratore et al.,2002). The Sox10 expression profile did not differ significantly between control and Pofut1 cKO embryos at E10.5(Fig. 5A,B,E,F,I). However, the number of ENCCs expressing Sox10 was significantly reduced, to ∼22% of the total population (in the intestine), in Pofut1 cKO embryos at E11.5(Fig. 5G,H,J). By contrast,Sox10 expression was observed in 62% of ENCCs (in the intestine) in control embryos (Fig. 5C,D). After E12.5, however, virtually no ENCCs expressed Sox10 in Pofut1 cKO embryos (data not shown). This suggests that Notch signaling is necessary for the maintenance of Sox10 expression in ENCCs after E11.5.

To investigate the relationship between the decrease in Sox10 expression and the proliferative ability of ENCCs, we examined the proliferation rate of Sox10-positive ENCCs. At E11.5, this did not differ significantly between Pofut1 cKO (11% in the stomach and the intestine) and control (10% in the stomach, 14% in the intestine) embryos(Fig. 5K). The proliferation rate of Sox10-negative ENCCs, estimated from that of total ENCCs and Sox10-positive ENCCs (Fig. 3Qand Fig. 5J), was ∼26% in the stomach and ∼30% in the intestine of control embryos, but only∼14% in the stomach and ∼16% in the intestine of Pofut1 cKO embryos. Although these are only estimates of the proliferation rate, the rate for Sox10-negative ENCCs was nonetheless significantly reduced in Pofut1 cKO embryos. Since Sox10-negative ENCCs may represent cells committed to the neural lineage, these results suggest that proliferation defects are indeed secondary to the premature appearance of cells committed to the neural lineage. Hence, the decreased proliferation and premature neurogenesis observed in Pofut1 cKO embryos are likely to be a consequence of the downregulation of Sox10 expression, which could then lead to a diminished progenitor population that eventually results in the loss of NCCs after E12.5. However, it was reported recently that Edn/Ednrb signaling is also required for the maintenance of ENS progenitors(Bondurand et al., 2006) and that the expression of Ednrb in ENCCs is regulated by Sox10(Zhu et al., 2004). In this context, it is noteworthy that we observed impaired Ednrb expression in Pofut1 cKO embryos (see Fig. S3 in the supplementary material),which is consistent with the downregulation of Sox10 expression in these embryos. Hence, the progenitor maintenance defect in the Pofut1 cKO embryo might also be due to the loss of Edn/Ednrb signaling.

Since the downregulation of Sox10 expression was observed at E11.5 in Pofut1 cKO embryos, we speculated that Notch signaling might be activated at around E11.5. However, there have been no previous reports regarding the involvement of Notch ligands or their receptors in ENCC development. We examined the expression of all the known Notch receptors and ligands (Notch1-4, Dll1, Dll3, Dll4, Jag1 and Jag2) by in situ hybridization (Fig. 6). The expression of these factors in ENCCs was confirmed by immunohistochemistry with the NCC marker p75^NTR (Ngfr - Mouse Genome Informatics)concomitant with in situ hybridization analysis (see Materials and methods). Notch1, Notch4, Dll1, Dll3, Dll4 and Jag1 were found to be expressed in the gastrointestinal tract. A number of ENCCs expressed Notch1, Dll1 and Dll3, but only a few expressed Notch4,Dll4 and Jag1 (Fig. 6; see Fig. S4 in the supplementary material).

Since ENCCs are known to migrate within the gut mesenchyme as chains of cells (Druckenbrod and Epstein,2005; Young et al.,2004), we speculated that Notch signaling might be activated through cell-cell contact among NCCs. This idea might be supported by our current finding that both Notch receptors and ligands are expressed in ENCCs. However, some other cell types neighboring the ENCCs were also found to express Dll4 and Jag1(Fig. 6; see Fig. S4 in the supplementary material) and the activation of Notch signaling through neighboring cells might therefore also occur. To investigate ligand redundancy and Notch activation through neighboring cells, we examined Sox10 expression in Dll1 mutant embryos. Dll1 mutant embryos did not show clear downregulation of Sox10 expression in post-migratory ENCCs at E11.5 (see Fig. S5 in the supplementary material), indicating that other ligands might also be functional during the maintenance of ENS progenitors.

A question that arises from our current data is how Notch signaling regulates Sox10 expression. We speculated that Sox10 might be a direct target of Notch signaling. However, it has been shown that a subset of NCCs expressing Sox10 induces Mash1, which in turn represses Sox10(Kim et al., 2003). In addition to this negative-feedback mechanism, Mash1 is known to be repressed by Hes1, one of the targets of Notch signaling in the CNS(Chen et al., 1997). These relationships give rise to the possibility that Sox10 expression is maintained by the repression of Mash1 through Notch signaling. This likelihood can be further evaluated by examining Hes1 and Mash1 expression.

If Notch signaling directly activates Sox10 expression, we predicted that Mash1 expression would not be altered or downregulated in Pofut1 cKO embryos. However, if Notch signaling maintains Sox10 expression by suppressing Mash1 via Hes1, then Hes1 expression would be expected to be downregulated and Mash1 expression to be upregulated in the Pofut1 cKO mice. We examined these expression patterns in the ENS at E10.5, before premature neurogenesis occurs, in Pofut1 cKO embryos(Fig. 7). In control embryos,Hes1 expression was observed in ∼16% of ENCCs in the intestine and also in other non-ENCCs (Fig. 7A-C). By contrast, Hes1 expression in ENCCs was found to be downregulated to ∼7% of the total population in the intestine of Pofut1 cKO embryos(Fig. 7D-F,M). Mash1 expression is restricted in ENCCs and ∼48% of ENCCs were indeed found to be Mash1-positive in the intestine of control embryos at this stage(Fig. 7G-I). However,Mash1-positive ENCCs were significantly increased in Pofut1 cKO embryos (up to ∼70% of the intestinal population)(Fig. 7J-L,N), indicating that Notch signaling indirectly regulates Sox10 expression through the suppression of Mash1.

To confirm this scenario, we performed double-immunostaining experiments for Sox10 and Mash1 (Fig. 8)and from their expression patterns the ENCCs were classified into three populations: Sox10 or Mash1 single-positive, and Sox10/Mash1 double-positive. In control embryos, the Sox10/Mash1 double-positive ENCCs (58% of the population in the intestine) were observed from E10.5 and continually thereafter during embryogenesis (data not shown), but the proportion of Mash1 single-positive cells was low (12% in the intestine) compared with that of Sox10 single-positive cells (30% in the intestine)(Fig. 8C). As expected,however, in Pofut1 cKO embryos, the abundance of Mash1 single-positive ENCCs was increased (28% in the intestine) and Sox10 single-positive ENCCs were decreased (13% in the intestine)(Fig. 8F,G). The proportion of Sox10/Mash1 double-positive progenitor cells was not significantly different between the control (58% in the intestine) and Pofut1 cKO (59% in the intestine) embryos (Fig. 8G),indicating that Notch signaling may not participate in the induction of Mash1 expression itself. Hence, Notch signaling appears to regulate the population of Mash1 single-positive ENCCs and be required for the maintenance of enteric neural crest progenitors via the regulation of a Sox10/Mash1 expression balance. Importantly, the balance of transcription factor expression in Pofut1 cKO embryos was found to have shifted at E10.5, which is prior to the occurrence of the premature differentiation observed only after E11.5. This may indicate that the decrease in the proliferation rate in Pofut1 cKO embryos is secondary to the premature appearance of ENCCs that are committed to the neuronal lineage. In other words, altering the balance of expressed transcription factors at E10.5, which results in the upregulation of Mash1, may cause premature differentiation and a decrease in proliferation at E11.5 and, eventually, a reduction in the number of ENCCs at E12.5.

We performed a comprehensive series of expression analyses in the Pofut1 cKO mouse in combination with a lineage tracer and demonstrate that Notch signaling plays a crucial role during ENS development in vivo. In general, this strategy can address the function of Notch signaling in all neural crest derivatives. Mice lacking Pofut1 in NCCs had a slightly smaller body size than their control littermates. This was probably caused by a defect in a subset of the neural crest derivatives in which Notch signaling plays some role. However, the Pofut1 cKO embryos did not exhibit obvious abnormalities in regions such as the craniofacial structures and heart arteries. Hence, the smaller body size might be a secondary effect of Notch signaling impairment in neural crest derivatives that were not included in our current analysis.

Mice lacking Pofut1 in their NCCs showed a reduction in ENCCs at E12.5,although these cells retained their capacity to contribute to development throughout the gastrointestinal tract. This phenotype appears to be different from the defects observed in mouse models of human congenital megacolon(Hirschsprung disease), including Sox10, Edn3, Ednr, Gdnf, Ret and Gfra1 knockout mice (Enomoto et al., 1998; Herbarth et al.,1998; Schuchardt et al.,1994; Southard-Smith et al.,1998; Stanchina et al.,2006). Each of these mutant mice shows a complete loss of enteric neurons in the gut region that is distal to the stomach. In addition, some of them, including Gdnf, Ret and Gfra1 knockout mice, show milk in their esophagus but die within 1 day of birth.

It has been reported that Mash1-null mutants also die within 1 day of birth but without milk in their stomachs(Guillemot et al., 1993). Interestingly, Mash1 mutants lack neurons in the esophagus but possess these cells in the stomach and intestine. There might be a correlation, therefore, between milk in the stomach and the presence of neurons in the esophagus. However, Pofut1 cKO embryos, in which neurons are present in the esophagus, but are reduced in number in the stomach and intestine, had no milk in their stomach. The reason for this is currently unknown.

Recently, experiments that are similar to those of our current study were conducted using RBP-Jk^flox and Wnt1-Cre mice. The authors of this earlier study concluded that Notch signaling promotes gliogenesis in the ENS, based upon their finding that neurogenesis was less affected, but gliogenesis severely affected, by the gene knockout(Taylor et al., 2007). This apparent inconsistency with our current data might be due to differences in the methods used and in the time windows analyzed, as we obtained results similar to those reported here when we performed comparable experiments using RBP-Jk^flox and Wnt1-Cre mice (data not shown). We also employed a linage tracer to analyze specific gene expression patterns in ENCCs, and this might have affected the interpretation of results even though similar probes were used. In our lineage analysis, we observed a clear reduction in the number of ENCCs at E12.5 in Pofut1 cKO embryos(Fig. 2). However, Taylor et al. (Taylor et al., 2007)counted differentiated ENCCs in RBP-Jk cKO embryos at E10.5, E14.5 and E18.5, by which time decreased proliferation and any premature neurogenesis occurring at E11.5 could have been missed. Since Sox10 is expressed not only in the ENS progenitors, but also in the glial cell lineage,the downregulation of Sox10 expression observed in the Pofut1 cKO embryo at E11.5 must lead to the loss of glial progenitors thereafter. Hence,we support the notion that both the maintenance of ENS progenitors and the formation of glial progenitors are accounted for by the maintenance of Sox10 expression. Taken together, we conclude from the available data, including our current results, that Notch signaling is required for the maintenance of ENS progenitors rather than for the promotion of glial cell differentiation.

The ENCCs in Pofut1 cKO embryos exhibited premature differentiation without increased apoptosis and with a slight decrease in proliferation during ENS development. These results suggest that Notch signaling is necessary for the maintenance of ENS progenitors. Since this finding is consistent with a role for Notch signaling in the CNS, it may well reflect a common role for Notch during neurogenesis.

Previous studies have shown that the lack of some progenitors or neurons is simply explained by the lack of particular transcription factors, which include the maintenance of NCC progenitors by Sox10 and the induction of some of the enteric neuron lineages by Mash1. However, our current study demonstrates for the first time that progenitor maintenance and the subsequent differentiation of ENCCs are regulated by the balance of Sox10 and Mash1 expression, which might enable a continuous derivation of mature neurons while maintaining a constant progenitor population. We also demonstrate herein that Notch signaling plays a crucial role in maintaining this balance by suppressing Mash1 expression (Fig. 9).

Further interesting aspects of Mash1-expressing cells have emerged from previous studies. In the developing gut, catecholaminergic (TC) cells have been transiently detected and are thought to be the precursors of a subset of enteric neurons as they show proliferative ability(Baetge et al., 1990). Both the selective elimination of TC cells and a Mash1-null mutation prevent the generation of a subset of enteric neurons. However, the remaining ENCCs can still produce another class of enteric neurons(Blaugrund et al., 1996). From this observation, it has been proposed that the ENS comprises two progenitor populations. One comprises the TC cells that give rise to early-born enteric neurons via a Mash1-dependent pathway, and the other contains a population of non-TC cells that generate late-born enteric neurons via a Mash1-independent pathway (Blaugrund et al.,1996). Since the ENCCs that express Sox10 are a cell population that is distinct from TC cells (Young et al., 2002), this suggests that non-TC cells can be formed by a Sox10-dependent pathway. In addition, Sox10/Mash1 double-positive ENCCs were observed in control embryos throughout embryogenesis(Fig. 8 and data not shown). Hence, it is intuitive that the expression of these transcription factors must be balanced in ENCCs, even though one of the functions of Mash1 is to suppress Sox10 expression. We speculate that Notch signaling is involved in regulating the balance between Sox10 and Mash1 expression, which may in turn generate two progenitor populations, one Mash1-dependent and the other Sox10-dependent(Fig. 9).

Although many genes involved in ENS development have been identified in recent years, their regulatory mechanisms and the relationships between the different gene functions remain poorly understood. Our current study provides new insights into these pathways by demonstrating that the cellular properties of ENCCs are regulated by a cell-cell communication mechanism that is mediated by Notch signaling. Further analyses of the molecular mechanisms underlying this cell-cell communication might provide a more precise understanding of the developmental processes within the ENS that might be applied in future therapeutic strategies against Hirschsprung disease.

We thank A. McMahon for providing the Wnt1-Cre mice; T. Müller and N. Brown for generously donating the antibodies against BFABP and Hes1, respectively; and our colleagues who provided cDNA probes, including R. A. Conlon (Notch1), T. A. Mitsiadis (Jag1), T. Gridley(Notch4, Jag2) and M. Hirashima (Dll4). This work was supported in part by Grants-in-Aid for Scientific Research on Priority Areas, Dynamics of Extracellular Environments, and for the National BioResource Project from the Ministry of Education, Culture, Sports, Science and Technology, Japan.