Age-dependent changes in the gut environment restrict the invasion of the hindgut by enteric neural progenitors

The enteric nervous system (ENS) is formed primarily from neural crest cells (NCCs) from the post-otic hindbrain (vagal NCCs) and by a smaller contribution from the sacral neural tube (sacral NCCs)(Burns et al., 2000). Vagal NCCs migrate to the anterior portion of the fetal gut and enter the pharynx to become enteric neural crest-derived cells (ENCCs). ENCCs subsequently advance caudally as multicellular strands along the entire length of the developing gut. The resulting network of ganglia controls motor activity, secretory functions and microcirculation, and regulates immune and inflammatory responses in the gastrointestinal tract. When the ENS is absent from a gut region, the gut loses motor coordination and undergoes sustained contraction. The absence of such ganglia from the human distal colon is a diagnostic feature of Hirschsprung's disease (HSCR). Cases of HSCR are most frequently associated with mutations affecting RET or endothelin receptor type B (EDNRB)signaling (McCallion and Chakravarti,2001), but HSCR is also associated with mutations in a number of other genes (McCallion et al.,2003; Gershon and Ratcliffe,2004; Anderson et al.,2006; Heanue and Pachnis,2007).

HSCR results from the failed advance of ENCCs through the hindgut, but the cellular and molecular defects resulting in colonic aganglionosis are not clear. Normal ENS development initially proceeds with the production of a sufficient pool of neural crest progenitor cells capable of advancing into the foregut. Aganglionosis results when an inadequate number of pre-enteric NCCs are generated either experimentally by neural tube ablation and/or grafting(Yntema and Hammond, 1954; Peters van der Sanden et al.,1993; Barlow et al.,2008) or genetically by mutation(Kapur et al., 1996; Maka et al., 2005; Stanchina et al., 2006). Once in the gut, aganglionosis can also result from ENCC defects in proliferation(Sidebotham et al., 2002; Young et al., 2005; Landman et al., 2007; Simpson et al., 2007) and/or differentiation (Young et al.,2004; Anderson et al.,2006). Defects in ENCC migration can also cause aganglionosis(Breau et al., 2006). The migratory properties of ENCCs in situ has been studied by tracking ENCCs in fetal mouse gut (Young et al.,2004; Druckenbrod and Epstein,2005) or individually labeled ENCCs in chick gut(Druckenbrod and Epstein,2007). However, it has been difficult to determine the intrinsic migratory capability of individual ENCCs independent of the influence of proliferation and differentiation, or the local gut microenvironment through which they advance.

Much of our information on HSCR comes from studies of animal models. There are multiple mouse genes, which, when mutated, result in aganglionosis restricted to the colon (Lane,1966; Webster,1973; Baynash et al.,1994; Herbarth et al.,1998; Hosoda et al.,1994; Yanagisawa et al.,1998; de Graaff et al.,2001; Breau et al.,2006). Two mouse models that have been used extensively and closely resemble the human phenotype are piebald-lethal mice that lack Ednrb expression (Ednrb^sl/sl) and lethal-spotted mice that lack Edn3 expression (Edn3^ls/ls)(Hosoda et al., 1994; Baynash et al., 1994; Gariepy et al., 1998). Studies have found associated defects in Edn3^ls/ls ENCCs,including decreased proliferation, increased differentiation and reduced expression of the neural crest stem cell marker SOX10(Barlow et al., 2003; Bondurand et al., 2006). Other studies have found changes in the gut environment associated with aganglionosis. In Edn3^ls/ls mice, increased laminin is found in aganglionic colons, a result suggesting that laminin reduces the number of ENCCs available for colonization by increasing ENCC differentiation(Jacobs-Cohen et al., 1987; Wu et al., 1999). Despite these studies, the cellular defects responsible for the failure of ENCC advance into the hindgut remain unknown.

To date, the properties and migratory behavior of ENCCs along the colon during the last stages of colonization have not been directly examined in HSCR animal models. In order to further our understanding of failed ENCC advance in the hindgut, we studied ENS development in mice with an NCC-specific deletion of Ednrb (flex3 allele). Control heterozygous mice(Ednrb^flex3/+) are viable and phenotypically indistinguishable from wild-type pups or fetuses(Druckenbrod et al., 2008). However, mutant homozygous pups (Ednrb^flex3/flex3) show the characteristic aganglionic colon of HSCR and die within 5 weeks from megacolon. We utilized NCC-specific expression of YFP to isolate and visualize the in situ properties of mutant ENCCs in gut fated to become aganglionic(Danelian et al., 1997; Srinivas et al.,2001; Druckenbrod and Epstein,2005). We investigated ENCC proliferation, differentiation and migration in Ednrb^flex3/+ and Ednrb^flex3/flex3 fetuses, referred to as control and mutant, respectively, throughout the remainder of the text. Our data indicate that mutant ENCCs are delayed in their advance along the gut by almost 24 hours relative to wild-type or control preparations. Apart from this delay,mutant ENCCs migrate at similar individual speeds to their control counterparts until they reach the proximal colon at E14.5, when they begin to display defective migratory behavior that is independent of cell proliferation or differentiation. We constructed tissue grafts to determine whether the defect was located in the ENCCs or the aganglionic colon. Our results suggest that a time-dependent change in the mutant hindgut environment restricts ENCC migration and that this delayed ENCC advance is sufficient to result in colonic aganglionosis in mutant gut.

The following transgenic and recombinant mice were bred on a C57BL/6 background: Tg^Wnt1-Cre(Danielian et al., 1997), Rosa26^YFPStop(Srinivas et al., 2001) and Ednrb^flex3(Druckenbrod et al., 2008). Mice with the genotype Tg^Wnt1-Cre/+Ednrb^flex3/+were bred with mice containing Rosa26^{YFPStop/YFPStop}Ednrb^flex3/flex3 alleles to generate fetuses and animals with YFP-positive NCCs lacking EDNRB. Mice and fetuses were genotyped by PCR as described previously(Druckenbrod and Epstein,2005; Druckenbrod et al.,2008). Timed mating was carried out; pregnant mice were killed by cervical dislocation after treatment with isoflurane and the day of the vaginal plug was considered as embryonic day (E) 0.5. The University of Wisconsin Animal Care Committee approved these procedures.

The gastrointestinal tracts from YFP-positive fetuses were placed in culture media [DMEM/F12 (Cellgro), heat-inactivated 10% FBS (Gibco), 2 mM L-glutamine (HyClone), 50 mM D-glucose, 30 mM NaHCO₃, 50 U penicillin, 50 μg streptomycin; pH 7.4] and incubated in 5% CO₂at 37°C. The EDNRB antagonist BQ-788 (American Peptide) was constituted in DMSO and used at 5-10 μM. Time-lapse imaging was done as described previously (Druckenbrod and Epstein,2005).

Culture dishes were filled with 2% agarose prepared by diluting melted 4%agarose in water with an equal volume of culture media, and were equilibrated in media for 12 hours at 37°C. Media was removed and a small piece of black Millipore paper was placed on the gel. Previously identified donor and recipient colon were cut with a tungsten needle, and the cut ends were quickly placed in apposition on top of the filter paper in their correct rostral-caudal orientation. Donor segments were cut near the wavefront. After 1-2 hours of incubation at 37°C, media was added to the dishes. The next day, steel pins were used to secure the distal ends of the grafted segments. After 6 days, cultures were fixed and immunostained for YFP. The lengths of each continuous YFP strand extending from the graft site were measured and averaged. To measure the area of grafted cells(Fig. 3B) and the percentage of Hu-positive cells (Fig. 5),equal thresholds were set for the YFP signal and were used to produce a binary mask for quantification.

Cell speeds were the average distance each single ENCC (isolated or at the front of a strand) moved over 7- to 14-minute intervals for 2-14 hours. On average, each age/genotype pair contained 35 tracks and 1253 data points. Directionality is the length of a line between the start and end of each track divided by the actual length of the track based on the 7- to 14-minute positional plots, whereby 1 represents a straight-line path. Directionalities were averaged for each condition. Wavefront advance is the change in position of the most caudal ENCC strand before and after 12 hours of incubation at 37°C in 5% CO₂. Images were processed and analyzed with Metamorph (Molecular Devices), ImageJ (NIH), SigmaPlot (Systat Software) and Excel (Microsoft). Adobe Photoshop was used to create montages, and to add color and symbols.

Whole-mount tissue was fixed in 4% paraformaldehyde for 1-2 hours, treated with 1% Triton X-100, washed and incubated overnight with primary, and then secondary, antibodies. For sections, tissue was equilibrated in OCT Compound,frozen on dry ice and sectioned at 20 μm on a cryostat. Immunostained tissues were imaged with a Bio-Rad MRC 1024 confocal microscope. Primary and secondary antibodies are listed in Table 1.

Total RNA from purified ENCCs was isolated using RNeasy Protect Mini Kit(Qiagen). RNA (100 ng) was primed with Ednrb-specific primers:forward, 5′-ACCGTGGTTTAACGCCATAG-3′; reverse,5′-CAGCTCTCTCGGAGGCATAC-3′ (Integrated DNA), and transcribed with SuperScript II Reverse Transcriptase (Invitrogen). The same primers were used for end-point PCR, using a cDNA template and transcribed by Taq polymerase(New England Biolabs), and products were run on a 15% acrylamide gel stained with ethidium bromide.

Gastrointestinal tracts were removed from YFP-positive E14.5 animals and identified as Ednrb^flex3/+ or Ednrb^flex3/flex3 by the location of the ENCC wavefront. Tissues were dissociated in a mixture of 5 mg/ml collagenase type III(Worthington Biochemicals) and 2 mg/ml DNase type I (Sigma) for 10-20 minutes at 37°C, centrifuged at 100 g, washed in PBS and triturated with fire-polished siliconized glass pipettes of decreasing diameter. Dissociates were centrifuged at 100 g, resuspended in DMEM/F12 buffered with HEPES, filtered through 20 μm Nitex and sorted into YFP-positive and -negative pools with a Becton Dickinson Vantage SE fluorescence activated cell sorter. The sorted cells were collected at 4°C in siliconized tubes and were then prepared for isolation of RNA or immunostaining.

To study the proliferation of ENCCs, BrdU (10 mg/ml) was injected intraperitoneally into pregnant mice at a dose of 50 μg/g of body weight. E14.5 fetuses were harvested 2-3 hours after injection and YFP-positive guts were dissociated as described above. The dissociated cells were fixed and then immunostained for YFP and BrdU. Cells were sorted and counted using an LSRII Becton Dickinson benchtop analyzer.

We developed a novel knockout mouse in which the expression of a null Ednrb allele and yellow fluorescent protein (YFP) is confined to NCCs. To make the null allele (flex3), exon 3 of Ednrb was flanked by loxP sequences and was selectively excised by Cre recombinase(Druckenbrod et al., 2008). In order to excise Ednrb specifically in ENCCs, Cre recombinase expression was regulated by the Wnt1 promoter(Tg^Wnt1-Cre), which is activated specifically in NCCs prior to their emigration from the neural tube(Danielian et al., 1997; Jiang et al., 2000). In addition, we bred Ednrb^flex3 mice with transgenic mice that constitutively express YFP (Rosa26^YFPStop) in the presence of Cre recombinase (Srinivas et al., 2001). Mating these lines of mice enabled us to obtain animals with the Ednrb mutation confined to NCCs, which in turn are all fluorescent.

To confirm Ednrb disruption in ENCCs, YFP-positive (i.e. ENCCs)and YFP-negative (i.e. smooth muscle precursors and enterocytes) cells were isolated by FACS from Ednrb^flex3/+ and Ednrb^flex3/flex3 tissue for RT-PCR. RT-PCR of a sequence within Ednrb exon 7 produced a product using template cDNA derived from Ednrb^flex3/+, but not Ednrb^flex3/flex3, ENCCs(Fig. 1). Immunostained E15.5 gut cross-sections showed that YFP-negative smooth muscle progenitors in Ednrb^flex3/+ and Ednrb^flex3/flex3 gut both contain EDNRB, and that only Ednrb^flex3/+ ENCCs colocalize with EDNRB (Fig. 2). Together, these data and previous findings(Druckenbrod et al., 2008)indicate that EDNRB is deleted specifically in Ednrb^flex3/flex3 ENCCs.

To characterize the development of the aganglionosis, we measured ENCC advance through control Ednrb^flex3/+ and mutant Ednrb^flex3/flex3 gut from E10.5 to E14.5(Fig. 3). At E10.5, mutant ENCCs were reduced in number, were not as caudally advanced and were less dense compared with the corresponding regions of control gut. Occasionally there were no ENCCs in E10.5 mutant gut (data not shown). Between E10.5 and E11.5, mutant ENCCs increased in number and in their advance, but remained∼24 hours behind those in identically aged control gut. In contrast to controls, mutant ENCCs did not invade the cecal body in their initial advance but did so later in development (Fig. 3D; see Movies 1 and 2 in the supplementary material). By E14.5,the control ENCC wavefront had reached the terminal hindgut, whereas the mutant wavefront had reached only the proximal colon. The absence of ENCCs in the mutant hindgut allowed us to identify sacral neural crest cells and extrinsic nerve fibers, which are normally obscured and indistinguishable from vagal neural crest from E14.5 onwards (Fig. 3D, inset). Although the number of sacral-derived crest cells was small, at E14.5 they extended rostrally in the mutant hindgut. In the fetal gut, sacral crest cells are made up of long, thin strands that project rostrally and contain fewer, more widely spaced cell bodies than do strands from the vagal crest. It is not clear what properties result in their advance or failed expansion, but recent studies suggest a difference in RET expression between vagal- and sacral-derived neural crest may be crucial(Delalande et al., 2008).

Time-lapse recordings of control and mutant ENCCs were made to determine whether differences existed in their speed or directionality through similar gut regions, and separate experiments were performed to measure the change in wavefront position after 24 hours in culture (see Fig. S1A-C in the supplementary material). The only significant difference found between E10.5 and E14.5 was the reduced speed of mutant ENCCs near the cecal base. This difference could relate to the lack of EDNRB, or to the fact that mutant ENCCs do not directly invade the cecum. The overall absence of differences in speed and directionality seems surprising but is consistent with previous findings that indicate wavefront advance occurs at a relatively constant rate once a critical number of ENCCs is reached (Young et al., 2004; Druckenbrod and Epstein, 2005).

At E14.5, an age when both wild-type and Ednrb^flex3/+guts are fully occupied by ENCCs, we find that mutant ENCCs begin to display behaviors that indicate a defective advance. As E14.5 mutant ENCCs advanced through the proximal colon, their network became denser but had more narrow strands compared with those in control gut(Fig. 4). Instead of following the general rostral-caudal trajectory found in wild-type and Ednrb^flex3/+ colon(Fig. 4A,C; see Movie 3 in the supplementary material), most mutant ENCCs migrated along paths perpendicular to the rostral-caudal axis of the gut (Fig. 4B,D; see Movies 4 and 5 in the supplementary material). A few strands advanced caudally, but these also displayed reduced speed and had erratic trajectories (compare Fig. 4Aa with 4Bc). Quantification of these behaviors showed that mutant ENCC speed, directionality and overall wavefront advance dropped sharply at E14.5 (see Fig. S1A-C in the supplementary material). Between E15.5 and E16.5, very few ENCC strands showed detectable migration, but a few ENCCs separated from their strands and migrated erratically through the growing gut (see Fig. S2 and Movie 6 in the supplementary material).

Because the advance of the wavefront depends on maintaining the ENCC number above a certain threshold, the inability of mutant ENCCs to advance through E14.5 gut could result from impaired proliferation, premature differentiation or defective migration. To investigate whether there is a proliferative defect, we used FACS cytometry on E14.5 gut taken from BrdU-injected mothers(see Fig. S3 in the supplementary material). Although the average number of ENCCs in the E14.5 mutant was 43% (±2.5%) of those counted in control small and large intestine, BrdU experiments revealed no significant differences between control and mutant ENCC proliferation at E14.5, the stage at which mutant ENCC advance becomes defective. To investigate whether mutant ENCCs were prematurely differentiating, we compared the organization and percentage of ENCCs expressing the pan-neuronal marker Hu at the wavefront of control and mutant gut (Fig. 5). The exclusive expression of neuronal Hu proteins in the cell body makes this marker useful for the enumeration of neurons. The control nascent ganglia were generally larger, and more rounded and spaced, with some regularity along strands (Fig. 5A). By contrast, mutant ganglia were smaller, poorly formed and oriented perpendicularly to the rostral-caudal axis along disconnected strands(Fig. 5B). However, no significant difference in the percentage of Hu-positive ENCCs was found between these preparations or those at earlier ages(Fig. 5C). Together, these data do not implicate altered proliferation or differentiation in the defective advance of ENCCs, which leads to aganglionosis.

Application of the EDNRB-specific antagonist BQ-788 leads to colonic aganglionosis in fetal organ cultures(Sidebotham et al., 2002; Nagy and Goldstein, 2006). Why do otherwise normally advancing ENCCs, once in the presence of BQ-788, fail to reach the terminal colon? We wanted to know if the defective advance produced by acute EDNRB inhibition occurred on the timescale of proliferation/differentiation, or whether it was associated with more rapid changes in cell behavior. Application of 5 μM BQ-788 to E11.5-E12.5 mutant preparations did not elicit a response (data not shown). However, BQ-788 produced a pronounced dose-dependent response in wild-type ENCC morphology and migration. Upon application of 5 μM BQ-788, wild-type ENCC strands and processes retracted, resulting in reduced motility (see Movie 7 in the supplementary material). After application of 10 μM BQ-788, ENCCs rapidly rounded up and strands dissociated within 2 hours(Fig. 6A). After 6-12 hours,ENCCs reassociated and resumed their advance. This temporary dissociation of ENCC strands effectively delayed the wavefront and the distance it advanced over 12 hours (Fig. 6B). Therefore, application of BQ-788 results in the wavefront reaching the hindgut later than it normally would, preventing the colonization of the otherwise normal hindgut. In addition, application of BQ-788 to E10.5-E11.5 mutant preparations had no observable effect on ENCC morphology or advance,suggesting that the expression of EDNRB in ENCCs is necessary for a response to BQ-788.

To determine whether the migratory defect in E14.5 mutant gut is ENCC-autonomous or mediated by the environment, we constructed grafts to compare the abilities of control and mutant ENCCs to invade aganglionic gut of different ages and genotype (Fig. 7; see Figs S4 and S5 in the supplementary material). Control donor ENCCs invaded E13.5 mutant colon extensively(Fig. 7A), and E14.5 mutant donor ENCCs showed a similar capacity for invasion when grafted to E13.5 control recipient colon (Fig. 7C). Donor ENCCs of either genotype made little or no invasion of E14.5 Ednrb^flex3/flex3 colon(Fig. 7B) or older (our unpublished results). Donor ENCCs never invaded recipient gut already occupied by ENCCs, so experiments using E14.5 Ednrb^flex3/+recipient gut were uninformative (our unpublished results). These results suggest that it is the age of the recipient tissue, not the genotype, that restricts the invasion of donor ENCCs.

We propose that, after E14.5, a time-dependent change in Ednrb^flex3/flex3 colon restricts invasion by ENCCs from either genotype. What accounts for the failure of ENCCs to invade post-E14.5 Ednrb^flex3/flex3 colon? Laminin is a candidate molecule suspected to play a role in aganglionosis(Jacobs-Cohen et al., 1987; Wu et al., 1999). To investigate the pattern and development of laminin expression in the colon, we imaged the wavefronts of control and mutant whole-mount guts from E13.5 to E14.5, the developmental window when the colon changes from a permissive to a nonpermissive environment. In E13.5 control and mutant wavefronts, laminin was sparsely distributed in punctate deposits(Fig. 8A,B). By contrast, both genotypes showed increased laminin at E14.5(Fig. 8C,D). However, whereas the control ENCCs had reached the terminal colon, the mutant ENCCs had not. Future work will determine whether laminin has a direct role in limiting ENCC strand invasion of post-E14.5 mutant colon in situ.

Colonic aganglionosis is generally considered to be caused by defects in either NCCs or the gut microenvironment(Heanue and Pachnis, 2007; Alzahem and Cass, 2008). Our results suggest that both components underlie the disease process, with an early delay in the onset of ENCC advance and a later failure of ENCCs to invade the terminal colon. It is clear that the absence of cells within the terminal portion of the bowel can result from defects in a number of steps required for colonization of the entire gut. The process begins with the production of a sufficient pool of pre-enteric progenitor NCCs that emigrate from the neural tube to the foregut. Upon entry into the gut, the coordination of proliferation, differentiation and migration of ENCCs is required for their advance along the midgut and colon. Here we show that the Ednrb^flex3/flex3 ENCC wavefront is already delayed shortly after NCCs enter the gut at E10.5, a finding suggesting that a defect limits the number of NCCs available to enter the gut. A pre-enteric NCC defect resulting in NCC apoptosis has been described for complete gut aganglionosis in SOX10-deficient mice (Kapur et al.,1996) and less severe colonic aganglionosis in Sox8,Sox10:Edn3 and Sox10:Ednrb mutant mice(Maka et al., 2005; Stanchina et al., 2006). However, more subtle abnormalities in ENS development might also be associated with pre-enteric NCC defects that have not yet identified. Although delayed by∼24 hours, we found Ednrb^flex3/flex3 mutant ENCCs still advance at equivalent speeds through the ileum and proximal colon, and it is only after they enter the proximal colon at E14.5 that migration is altered. This defective ENCC invasion is independent of their proliferation and differentiation, and is instead the result of a time-dependent change in the mutant gut microenvironment after E14.5. This change may reflect a phenomenon that occurs in the course of ENS development, so that an initial 24-hour delay in ENCC advance is sufficient to produce colonic aganglionosis.

An explanation for the failure of ENCCs to advance in the mutant colon appears to reside in the gut microenvironment, as our tissue grafting experiments revealed that both mutant and control ENCCs can invade E13.5 recipient colon of either genotype, but not E14.5 mutant recipient colon. Put simply, the age of the recipient aganglionic colon, and not the donor ENCC genotype, is crucial to ENCC invasion. This hypothesis is strengthened by our analysis that shows no change in ENCC proliferation within the intestine or differentiation at the migratory wavefront. These data differ from previous studies performed on Edn3^ls/ls mice, which also show delayed migration of ENCCs along the gut tube from E10.5(Barlow et al., 2003). Bondurand et al. (Bondurand et al.,2006) reported regional differences in ENCC proliferation and differentiation in Edn3^ls/ls mice. A small but significant decrease in ENCC proliferation was identified at the E11.5-E12.0 Edn3^ls/ls wavefront. Using automated cytometry, we were unable to detect any difference between mutant and control ENCCs from E14.5 dissociated gut. Small regional changes in proliferation may have gone undetected because we analyzed the BrdU-positive ENCCs as a single population. In agreement with data from Bondurand et al., we observed a reduction in ENCC number in E14.5 Ednrb^flex3/flex3 gut. This reduction must be due in part to the absence of mutant ENCCs in the distal colon and cecum. However, it is also likely to reflect a decreased number of mutant NCCs entering the gut, and/or a small ongoing reduction in ENCC proliferation. With regards to neuronal differentiation, Bonderand et al.(Bondurand et al., 2006) found a 2-fold increase in the percentage of ENCCs expressing βIII tubulin at the wavefront in Edn3^ls/ls compared with their wild-type counterparts, despite the fact that the neuronal differentiation within the intestines was reduced overall. By contrast, we observed no difference in the expression of the pan-neuronal marker Hu between the ENCC wavefronts of control and mutant gut. This disparity might be attributable to the properties of the two different neuronal markers used in these studies. Regardless of these changes in cell proliferation and/or differentiation, the ENCC wavefront in Edn3^ls/ls mice is still capable of advancing from E10.5 to E14.5, such that the extent of migration delay does not increase with time when compared with control guts. However, both Edn3^ls/lsand Ednrb^flex3/flex3 hindguts fail to be colonized further by ENCCs, a result which suggests that there must be an ensuing process that prevents the wavefront from `catching up' over time and invading the terminal hindgut during the remaining time in gestation (i.e. ∼7 days in mouse and∼7 months in human).

Our time-lapse imaging studies provide further support for a colonic microenvironment that is nonpermissive in the E14.5 mutant. Until now, live imaging of ENCCs in intestine fated to become aganglionic had not been carried out. We found that ENCCs in E14.5 mutant colons show alterations in direction,duration and speed, and ultimately stop moving. By contrast, mutant and control ENCCs do not show any differences in these properties as they migrate in the midgut prior to E14.5. Furthermore, from E10.5 to E14.5 the difference in the position of control and mutant ENCCs remains separated by a 24-hour gap and thus does not change with developmental age. Although these data do not distinguish between a defect in the ENCCs or the gut environment, they do invite a comparison of our results with those of Breau et al.(Breau et al., 2006). Their mice that lack β1 integrin expression specifically in NCCs show a delay in ENCC migration that is independent of cell survival, proliferation and differentiation. This finding is instructive because they concluded that aganglionosis resulted from an ENCC defect, leading to increased aggregation and consequently a reduced ability to migrate. However, in contrast to our mutants, the difference in the position of the wavefront between the mutant and control guts does increase with fetal age, and mutant ENCCs migrate shorter distances in grafts than controls. These differences support the idea that ENCCs lacking β1 integrin fail to invade the colon because they have reduced migratory capability, whereas ENCCs lacking EDNRB fail to colonize the Ednrb^flex3/flex3 hindgut because of the environment.

Our experiments suggest that a change within the mutant microenvironment beyond E14.5 restricts the continued migration of ENCCs and eventually results in colonic aganglionosis. The studies of others (e.g. Wu et al., 1999) have implicated laminin as a molecule with a putative inhibitory role in the colonization of the hindgut in Edn3^ls/ls mice. We found no difference in the expression of laminin between E13.5 control and mutant colons. Rather, laminin increased between E13.5 and E14.5 in both control and mutant hindgut, a finding which suggests a role for laminin in restricting ENCC strand invasion of post-E14.5 colon. These data imply that the temporal change in permissiveness to ENCC invasion within the colonic microenvironment might not just be restricted to the mutant gut but might be a phenomenon that occurs during normal development. Further studies will need to be performed to see whether this is indeed the case, and to determine if laminin acts directly on ENCCs to alter their invasion of the gut wall.

A link between the delay in ENCC migration and aganglionosis comes from experiments combining in situ time-lapse analysis with the EDNRB-specific antagonist BQ-788. Application of BQ-788 to cultured mouse and chick preparations results in aganglionic colon. We found that the antagonist elicited a pronounced migratory and morphological response from wild-type, but not mutant, ENCCs. These responses ultimately caused a temporary cessation of ENCC advance. The rapidity of the response argues for an effect on migration rather than proliferation or differentiation. In addition, the resulting aganglionosis following BQ-788 treatment of wild-type gut suggests that the same time-dependent changes restricting ENCC entry in mutants are likely to occur in control colon. The molecular role of EDNRB signaling in ENCC strand invasion and its failure in the mutant are unknown. However, studies in cancer cells may prove to be informative. For example, interaction and communication between advancing melanoma cells is disrupted by an EDNRB-specific antagonist(Bagnato et al., 2004; Lahav, 2005), whereas EDNRB signaling counteracts cell dissociation in vitro by regulating focal adhesion and cytoskeletal molecules (Lange et al.,2007). Therefore, changes in ENCC morphology and migration following BQ-788 application to wild-type colon might result from alterations in the ability of ENCCs to associate with one another. It is unknown how the effect produced by acute pharmacological inhibition of EDNRB signaling relates to the absence of EDNRB seen in transgenic mice. Future experiments will be necessary to clarify the mechanism behind the effect of BQ-788 on ENCC behavior.

Although the mutant vagal ENCCs fail to reach the terminal hindgut, the mutant sacral-derived neural crest cells are present at E14.0(Fig. 3). Their number is initially small and does not appear to increase substantially. The fate of these sacral-derived cells is not clear but they are absent in the 10-day-old postnatal mouse, with the result that the terminal hindgut is aganglionic. Their presence in the terminal hindgut would appear to argue against our hypothesis of an E14.5 nonpermissive hindgut. However, putative changes in the gut environment might differ between the rostral and terminal portions of the middle region of colon so that vagal ENCCs do not progress, whereas the sacral crest cells might move. Nevertheless, the advance and morphology of sacral crest cells are not comparable to vagal ENCCs. The sacral-derived neural crest cells do not appear to invade the gut as multicellular strands but as single cells with long processes. Future investigations are necessary to clarify the migration and fate of these cells.

In conclusion, our studies show that the mutant colon becomes nonpermissive to ENCC invasion at a certain stage in development. Although our results clearly indicate a cell-autonomous role for EDNRB signaling in ENCCs, this does not preclude other non-autonomous effects or indirect consequences. Although the mutation is targeted only to NCCs, it is possible that the post-E14.5 environmental change is a consequence of this mutation and is not a normal physiological transition. Therefore, it will be important to determine whether aganglionic wild-type colon undergoes a similar change. Future experiments combining wild-type, and Ednrb conditional and total knockout mice will be used to determine whether the transition of the hindgut to a nonpermissive environment is a normal process.