A Histone2BCerulean BAC transgene identifies differential expression of Phox2b in migrating enteric neural crest derivatives and enteric glia

Abstract

The mammalian enteric nervous system (ENS) derives from migratory enteric neural crest–derived cells (ENCC) that express the transcription factor Phox2b. Studies of these enteric progenitors have typically relied on immunohistochemical (IHC) detection. To circumvent complicating factors of IHC, we have generated a mouse BAC transgenic line that drives a Histone2BCerulean (H2BCFP) reporter from Phox2b regulatory regions. This construct does not alter the endogenous Phox2b locus and enables studies of normal neural crest (NC) derivatives. The Phox2b-H2BCFP transgene expresses the H2BCFP reporter in patterns that recapitulate expression of endogenous Phox2b. Our studies reveal Phox2b expression in mature enteric glia at levels below that of enteric neurons. Moreover, we also observe differential expression of the transgene reporter within the leading ENCC that traverse the gut. Our findings indicate that the wavefront of migrating enteric progenitors is not homogeneous, and suggest these cells may be fate-specified before expression of mature lineage markers appears. Developmental Dynamics 237:1119–1132, 2008. © 2008 Wiley-Liss, Inc.

INTRODUCTION

Formation of an intact, functional enteric nervous system (ENS) relies upon appropriate migration and differentiation of enteric neural crest–derived progenitors into and within the developing gut. Migration of neural crest derivatives to form enteric ganglia proceeds in two phases: the initial migration away from the neural tube and the subsequent migration of cells within the developing gastrointestinal (GI) tract to the terminal hindgut. Vagal NC cells leave the neural tube between 8.5 and 9 days post coitus (dpc). These progenitors migrate ventrocaudally and accumulate in an area ventral to the aorta and posterior to the brachial arches, then enter the gut wall by E9.5 (Baetge and Gershon,1989; Natarajan et al.,2002; Anderson et al.,2006). Within the GI tract, migration of enteric neural crest–derived cells (ENCC) progresses in a fairly specific time course. ENCC migrate through the foregut into the midgut and approach the ileocecal junction by 10.5dpc. These progenitors transition from the midgut and have passed around the caecal bulge by 11.5dpc, enter the hindgut by 12.5dpc, and reach the end of the hindgut by 14dpc (Kapur et al.,1992; Young et al.,1998; Anderson et al.,2006). Expression of neuronal markers emerges among ENCC not far behind the initial wave of progenitors that proceed down the gut (Young et al.,1998,1999,2002). Appearance of glial cell types marked by expression of brain-specific fatty acid binding protein (BFABP), calcium-binding protein S100β, and glial fibrillary acidic protein (GFAP) emerges among ENCC significantly later at 11dpc, 14dpc, and 16dpc, respectively, in the mouse (Young et al.,2003).

NC-derived cells from the sacral level of the neural tube, distal to somite 28, also contribute to the formation of enteric ganglia. However, the sacral NC derivatives (sENCC) give rise to a subpopulation of enteric ganglia in only the post-umbilical region of the gut (Burns and Douarin,1998). Initially, sENCC populate the pelvic mesenchyme around 9.5dpc and then pause in their migration outside the hindgut. Upon arrival of the vagal ENCC, sacral progenitors enter the hindgut at 13.5–14dpc (Kapur,2000). Sacral progenitors express many of the same genes as vagal ENCC including Sox10, Ret, Hu, NADPH-diaphorase, Dopamine β-hydroxylase, and Phox2b (Burns and Douarin,1998; Kapur,2000; Anderson et al.,2006).

Current models of events in ENS ontogeny are based on an initial wave of migrating undifferentiated ENCC, followed by a wave of neuronal differentiation just behind the migration wavefront with glial differentiation occurring secondarily in a somewhat slower wave of differentiation. But when are enteric neurons specified? Do enteric neurons and glia differ from other NC-derived lineages that are restricted soon after crest cells emerge from the neural tube (Henion and Weston,1997; Reedy et al.,1998; Luo et al.,2003), long before they reach their final destinations? Or, is ENS ontogeny analogous to proliferation of neuroepithelial cells in the central nervous system (CNS) that first give rise to neurons and later switch to glial production (Zhou and Anderson,2002; Pringle et al.,2003; Stolt et al.,2003; Rowitch,2004)? Studies of the processes that lead to production of enteric neurons and glia are needed to establish how these lineages are generated.

Phox2b is a master regulator of autonomic NC derivatives including enteric ganglia. Without Phox2b, autonomic ganglia do not form and normal specification of noradrenergic neurons and cranial motoneurons does not occur (Pattyn et al.,1999). In the CNS, Phox2b expression has been documented in proliferating neural precursors. Studies by Pattyn et al. (1997) suggested this transcription factor plays an early role in specifying neuronal phenotype and further analysis confirmed the role of Phox2b in neural determination (Brunet and Pattyn,2002), an early step of neuronal differentiation. Neuronal differentiation involves the transition of a cell from the proliferative progenitor state to a post-mitotic neuronal precursor. A final step in neural differentiation is the acquisition of generic neuronal features. All ENCC progenitors express Phox2b, and its expression is maintained in postnatal neurons. Cells expressing Phox2b are found within myenteric and submucosal plexi and are readily identified by strong nuclear labeling with an antibody to Phox2b (Pattyn et al.,1997). Phox2b+ cells exhibit large round nuclei consistent with enteric neuron morphology and are positive for neuron-specific enolase (NSE) (Young et al.,1998). Therefore, in addition to its function during development, Phox2b also contributes to the phenotype of mature enteric neurons. Complete ablation of Phox2b during development results in loss of enteric ganglia throughout the intestine (Pattyn et al.,1999), but phenotypes in heterozygous mutants that would suggest the function of this factor in mature enteric neurons have not been described.

Previous studies of ENS lineages have been limited by the tools available for imaging. Valuable information has been gained by relying on immunohistochemical (IHC) reagents, but analyses have been limited by species origin and specificity of primary antibodies as well as requirements for fixation/penetration to identify cell types. Beautiful dynamic images of enteric NC migration have been captured with fluorescent reporters. A Ret-TGM knock-in mouse model that drives expression of GFP from insertion of a tau-EGFP-myc cDNA into the first exon of the Ret locus (Enomoto et al.,2001) has been used to demonstrate the migration paths and behaviors of enteric NC progenitors (Young et al.,2004; Anderson et al.,2006). Similarly, images of a bigenic expression system that relies on Wnt1-Cre to activate a YFP Rosa reporter have identified regional differences in migratory behaviors of enteric NC (Druckenbrod and Epstein,2005,2007). Nonetheless, these systems are not ideal for lineage analysis. The Ret-TGM model suffers from low intensity of the GFP reporter, an inability to track glial lineages because Ret is downregulated in enteric glia, and cytoplasmic expression of the reporter that makes resolution of individual cells challenging. This system may also be susceptible to effects of Ret haploinsufficiency. The Wnt1-Cre/RosaYFP system requires that animals analyzed carry both the Cre and reporter alleles. In addition, the lineage specificity of the Wnt1-Cre system in the ENS has not been described. Given the recent report that astrocytes of the CNS do not drive anticipated expression of reporters from the Rosa locus (Casper and McCarthy,2006), further documentation of cell type–specific expression for the Rosa26R reporter is needed in the ENS before the Wnt1-Cre system can be relied upon for lineage analysis.

To facilitate studies of neuronal-glial lineage divergence in the ENS, we have applied in vitro modification of a bacterial artificial chromosome (BAC) clone spanning the Phox2b locus and established mouse transgenic lines driving expression of a Histone2B-Cerulean reporter (H2BCFP). Cerulean is a cyan spectral variant of green fluorescent protein that is brighter and more stable than prior blue fluorescent proteins (Rizzo et al.,2004). By linking this reporter to Histone2B, the reporter signal associates with chromatin throughout the cell cycle so that interphase and mitotic chromosomes are visible (Kanda et al.,1998; Hadjantonakis and Papaioannou,2004; Fraser et al.,2005) and individual cells are readily distinguishable. The Phox2b-H2BCFP transgene recapitulates expression of endogenous Phox2b and reveals previously unrecognized expression in NC derivatives outside the gut. Our initial studies with this construct reveal that expression of Phox2b is not confined strictly to enteric neurons but is also present at lower levels in enteric glia. Moreover, images of the earliest migrating progenitors entering the gut reveal that differential levels of Phox2b expression are observed long before markers of differentiated cell types have been described. Our observations suggest that enteric progenitors may be specified to distinct lineages well before characterized markers of mature cell types appear. The Phox2b-H2BCFP transgene will be a valuable tool for tracing migration of NC derivatives and lineage specification within the ENS and other aspects of the autonomic nervous system.

RESULTS

Generation and Analysis of Phox2b-H2BCFP Transgenic Mice

The Phox2b-Histone2BCerulean BAC construct (hereafter referred to as Phox2b-H2BCFP) was made by integrating a fluorescent fusion protein, Histone2B-Cerulean (H2BCFP), into a Phox2b BAC, clone 95M11 derived from the CHORI RP-24 C57BL/6J (B6) genomic library (Fig. 1A). In vitro homologous recombination methods were used to fuse the coding sequences of H2BCFP in frame with the ATG of the Phox2b coding region in exon 1 (Lee et al.,2001) so that no sequences from the Phox2b locus were deleted. As a consequence of the polyadenylation sequence at the end of the H2BCFP cassette, no Phox2b protein is produced by the BAC construct. The 95M11 BAC (207 kb, 137 kb 5′ flanking) contains no other neighboring genes. We anticipated that the final Phox2b-H2BCFP construct would contain necessary regulatory regions required for proper expression of the fluorescent reporter based on studies in transfected cell lines that have demonstrated that proximal promoter sequences are sufficient to drive expression in noradrenergic neurons (Jong Hong et al.,2004).

Figure 1. Phox2b-H2BCFP BAC transgene construct and expression in the developing nervous system. Schematic diagram of Phox2b-H2BCFP targeting construct and homologous recombination into the wild-type Phox2b 95M11 BAC (A) depicts Phox2b exons (black), initiator ATG (green), Histone2B-Cerulean reporter (teal) and polyadenylation sequence (red). Phox2b homology arms (approximately 500 bp) are shown on either side of H2BCerulean sequence. Phox2b-H2BCFP transgene expression in flat-mount preparations of midbrain-hindbrain from 10.5dpc embryos (B) exhibit CNS expression within the rhombencephalon, the met-mesencephalic domain in the oculomotor (III) and trochlear (IV) motor nuclei as well as the developing locus coeruleus (lc). Transgene expression was confirmed in the sympathetic chain (C) as well as whole mount fetal guts (D) at 14.5dpc. Phox2b-H2BCFP signal was seen in 14.5dpc adrenal and celiac ganglia (E, F, arrowhead) as well as βGal-stained tissue from 14.5dpc Phox2btm1Jbr/+ embryos. a, adrenal; k, kidney; g, gonad. Whole-mount image of Phox2b-H2BCFP embryo (G) cut transversely just below the mandible viewed into the base of the skull reveals CFP signal in submaxillary salivary glands (arrowheads). Flat-mount preparations of submaxillary gland demonstrates CFP signal within the ductal tree (H, fluorescence; I, brightfield). Equivalent staining of submaxillary gland was observed in Phox2b^tm1Jbr/+ embryos (J–L)

The resulting Phox2b-H2BCFP construct was microinjected into fertilized oocytes, and seven founders were identified that carried H2BCFP transgene sequences from a total of 100 pups. Five of the seven founders passed the transgene to their offspring and were bred to establish lines. Additional genotyping established that four of these lines (designated A–D) carried sequences for the flanking BAC T7 and Sp6 arms suggesting that the Phox2b-H2BCFP construct was intact. Further semi-quantitative genotyping with the internal and flanking markers by routine methods established that line A has multiple copies (∼4–8 transgene copies), line B has a large number of copies (>10), and lines C and D have low copy numbers of the construct (<2 copies). Expression across all the lines was documented in fetal gut at 13 dpc to rule out the possibility of internal deletions that can result in variable expression (Deal et al.,2006; Chandler et al.,2007). Lines A, B, and C all expressed the H2BCFP reporter in 13dpc gut consistent with the distribution of migrating ENCC (data not shown). Line D did not exhibit normal expression patterns in fetal gut nor did H2BCFP signal appear in gut muscle strips taken from adult animals where expression of Phox2b is normally seen in enteric ganglia. Therefore, line D was discontinued.

To assess transgene integrity and rule out the possibility of small deletions that can remove regulatory elements and impact transgene expression (Deal et al.,2006; Chandler et al.,2007), lines A, B, and C were out-crossed to C3HeB/FeJ (C3Fe) mice and third-generation progeny were screened with simple tandem repeat markers (STRs) that are polymorphic between C3Fe and B6 strains. The presence of B6 alleles detected by these STRs in genomic DNA from third-generation outcross mice for lines A and B indicated that the complete genomic interval derived from the B6-origin Phox2b BAC had integrated into the genome (data not shown). Line C had regions of the transgene where B6 alleles were not detected for all the STRs and further characterization of this line revealed loss of expression in the majority of enteric neurons of adult gut muscle. Subsequent analyses were carried forward with lines A and B because all molecular tests indicated the transgene was intact in these lines and patterns of expression in fetal and adult tissues appeared comparable.

Phox2b-H2BCFP BAC Is Appropriately Expressed in the CNS, PNS, and Other NC Derivatives

To establish that the Phox2b-H2BCFP transgene conferred expression on the Cerulean reporter analogous to that of endogenous Phox2b, we compared patterns of CFP fluorescence with published reports of Phox2b in situ (Pattyn et al.,1997) and with expression of B6D2. Phox2b^tm1Jbr/+, a LacZ knock-in allele for Phox2b (Pattyn et al.,1999). The Phox2b-H2BCFP transgene exhibited patterns of CFP fluorescence in the developing CNS that are analogous to in situ hybridization expression patterns previously reported by Pattyn et al. (1997). In midbrain-hindbrain flat-mount preparations from 10.5dpc Phox2b-H2BCFP embryos, CFP+ cells were present in the met-mesencephalic domain within two motor nuclei, oculomotor (III) and trochlear (IV), and in the forming locus coeruleus (Fig. 1B). Phox2b-H2BCFP midbrain-hindbrain preparations exhibited cells expressing the reporter in three obvious stripes within the rhombencephalon accompanied by very strong expression in rhombomere 4 analogous to what has previously been reported.

Phox2b is also expressed within the peripheral nervous system (PNS). Regions within the PNS that exhibit Phox2b expression include all autonomic ganglia (sympathetic, parasympathetic, and enteric) as well as distal ganglia of three cranial nerves: facial (VII), hypoglossal (IX), and vagal (X) (Pattyn et al.,1997; Brunet and Pattyn,2002). The cranial ganglia expression is not as evident in the dorsal views of midbrain-hindbrain preparations, but is clearly visible in lateral whole mounts of Phox2b-H2B transgenic embryos (see Supplementary Figure 1, which can be viewed at www.interscience.wiley.com/jpages/1058-8388/suppmat). We observed expression of the Phox2b-H2BCFP transgene very distinctly in cells of the sympathetic chain as well as enteric ganglia throughout the length of the gut in 14dpc whole mount samples (Fig. 1C, D).

We performed microscopic analyses of tissues from 14.5dpc embryos to identify additional sites of Phox2b-H2BCFP transgene expression. CFP fluorescence was observed in the adrenal medulla and in celiac ganglia immediately adjacent to the kidney pelvis (Fig. 1E). These sites are recognized NC derivatives and Phox2b transcripts have previously been detected in the adrenal by in situ hybridization (Visel et al.,2004). Moreover, β-galactosidase (βGal) expression was also seen in these sites in 14.5dpc tissues from Phox2b^tm1Jbr/+ embryos (Fig. 1F). Mesenchymal tissue comprising the salivary glands is of NC origin (Jaskoll et al.,2002), so we carefully examined this organ for any expression of the Phox2b BAC transgene. Phox2b-H2BCFP transgene fluorescence was identified in submaxillary salivary glands where CFP signal was seen along the ductal network at 14.5dpc (Fig. 1G–I). Similar localization was observed in the submaxillary gland for Phox2b^tm1Jbr/+ embryos (Fig. 1J–L) although the expression was faint and difficult to image due to the lower contrast of βGal stain. Close examination of transgene expression at higher magnification revealed CFP-labeled cells within the walls of the ductal network at 14.5dpc and in a tight cluster of cells that could represent a ganglion adjacent to the outflow tract of the salivary gland (Suppl. Fig. 2).

Phox2b Is Expressed at Low Levels in Enteric Glia

The requirement for Phox2b expression in the ENS and its expression within enteric neurons is well documented (Pattyn et al.,1999; Young et al.,2003). To establish that the Phox2b-H2BCFP transgene expression exhibits appropriate cell type–specific expression in the ENS, gut muscle strips that contain enteric ganglia were peeled away from the mucosa and imaged for CFP fluorescence. Surprisingly, two morphologically distinct types of nuclei were seen to express the transgene (Fig. 2A). Large round nuclei located within the center of the ganglia exhibited bright CFP reporter expression. The morphology of these cells is consistent with prior descriptions of enteric neurons (Young et al.,1998,2003). In addition, smaller spindle-shaped nuclei were also observed, which exhibited lower levels of CFP fluorescence. The size, shape, and location of these smaller CFP+ nuclei, both within and between ganglia, are suggestive of enteric glial cells. IHC for endogenous Phox2b on non-transgenic gut muscle strips revealed a similar pattern of cell localization and intensity (Fig. 2B) demonstrating that the transgene was accurately reflecting normal distribution and levels of Phox2b. This pattern of expression for the transgene and endogenous Phox2b protein was documented in both adult (10–13 weeks of age) and young (postnatal day 10) mice (data not shown).

Figure 2. Phox2b is expressed in cells other than neurons in the ENS. Confocal images of enteric gut muscle strips from the colon of Phox2b-H2BCFP transgenic adult mice (10–13 weeks old) show expression of the Phox2b-H2BCFP transgene (A) and endogenous Phox2b protein (B) in large round neuronal nuclei (arrowheads) and smaller stellate cells (arrows). Gut muscle strips from adult Phox2b-H2BCFP transgenic (C–E) and wild-type (F–H) mice stained with antibodies to Phox2b and the neuronal marker, Hu, reveal expression of Phox2b in additional cells (arrows) besides Hu+ enteric neurons (arrowheads).

Immunohistochemical co-localization with cell type–specific markers was undertaken to definitively establish the identity of the two cell types within enteric ganglia expressing Phox2b. Hu, a pan-neuronal marker (Fairman et al.,1995; Young et al.,2003), that identifies enteric neurons, was found to co-localize with the large, bright nuclei confirming the identity of these cells as neurons for both the Phox2b-H2BCFP transgene and for wild-type tissue exposed to Phox2b antibody (Fig. 2C–H). The small, spindle-shaped cells positive for Phox2b expression did not stain for Hu, further substantiating that Phox2b is present in a cell type distinct from enteric neurons. Several glial-cell-type-specific markers including Sox10, GFAP, and S-100β, were evaluated by IHC in Phox2b-H2BCFP transgenic and wild-type tissue. In each case, nuclei of these smaller, stellate cells exhibited co-localization of either endogenous Phox2b or the Phox2b-H2BCFP transgene with glial lineage markers (Fig. 3). Co-localization of Phox2b protein and S-100β in the same sample was not undertaken because antibodies for these two markers are derived from the same species, complicating IHC detection.

Figure 3. Phox2b protein and Phox2b-H2BCFP transgene expression colocalize with glial markers in the mature enteric nervous system. Gut muscle strips from adult, wildtype mice (10–13 weeks old) stained with antibodies to Phox2b, Sox10, and GFAP (A) reveal expression of Phox2b in neurons (arrowheads) and glia (arrows). Enteric muscle strips from adult, Phox2b-H2BCFP mice also stained with antibodies to Sox10, GFAP, and S-100β (B) demonstrate that Phox2b is present in both enteric neurons (arrowheads) and glia (arrows).

Two Populations of Cells That Differentially Express Phox2b Are Present at the Wavefront of Migrating Enteric Progenitors

Phox2b marks early migrating NC-derived progenitors in the developing intestine (Young et al.,1998,1999,2002; Young and Newgreen,2001). We rigorously evaluated temporal and spatial distribution of Phox2b-H2BCFP transgene expression in fetal gut to confirm it recapitulated the developmental profile of the endogenous Phox2b gene. CFP-labeled ENCC progenitors were observed migrating through the intestine on a timeline consistent with prior reports. Vagal progenitors had passed through the upper two-thirds of the hindgut and were approaching the end of the intestine at 13.5dpc (Fig. 4). At this stage, sacral NC-derived progenitors labeled by transgene expression were also evident in the tissue just outside the hindgut consistent with prior reports of Phox2b expression in this NC-derived lineage (Young et al.,1998; Young and Newgreen,2001). At 10.5dpc CFP+ vagal progenitors were just crossing from the midgut into the cecal bulge of the intestine (Fig. 5). Published reports have previously established that ENCC at 10.5dpc have migrated beyond the position of the future stomach and are well into the midgut at the junction of the future ileum with the cecum based on the imaging of a Ret transgene (Anderson et al.,2006), a Wnt1Cre transgene (Druckenbrod and Epstein,2005), and the ability of explanted gut segments from these regions to form neurons (Young et al.,1998). The extent of migration we observed for cells labeled by the Phox2b-H2BCFP transgene closely follows this timeline and pattern. At 9.5dpc, the earliest progenitors of the ENS labeled by CFP expression were observed entering the proximal foregut (Fig. 6).

Figure 4. Differential expression of the Phox2b-H2BCFP transgene among enteric neural crest–derived cells at the migration wavefront. A: Confocal image (200× magnification) of a 13.5dpc Phox2b-H2BCFP fetal gut from cecum to anus shows migration of enteric progenitors. B: Higher magnification image (400×) of the Phox2b-H2BCFP+ nuclei within the enclosed box in A reveals two visibly discernable cell types, bright (arrowhead) and dim (arrow) for transgene expression. CFP signal intensity for individual cells in B was determined using the region of interest tool in MetaMorph to encircle individual nuclei. Table inset lists region labels corresponding to individual cells. Bold values indicate bright cells. Scale bar = 200 μ.

Figure 5. Two populations of Phox2b+ cells exist at the wavefront of migrating enteric progenitors at 10.5dpc. A: Confocal image (200× magnification) of a 10.5dpc Phox2b-H2BCFP gut shows the migration of Phox2b+ cells through the foregut and into the cecal bulge. B: Cecal bulge (400× magnification) in the 10.5dpc gut from A showing both bright (arrowhead) and dim (arrow) cells in the wavefront of the migrating population. Fluorescent signal intensity analysis is provided in Supplementary Figure 4. Scale bar = 200 μ.

Figure 6. Distinct populations of Phox2b+ cells are evident as vagal ENCC enter the foregut. A: Confocal image (200× magnification) of a 9.5dpc Phox2b-H2BCFP gut showing the migration of enteric progenitors into the foregut. B: The 400× magnification of the migration wavefront reveals both bright (arrowhead) and dim (arrow) populations of cells. Fluorescent signal intensity analysis is provided in Supplementary Figure 5. Scale bar = 200 μ.

When higher magnification images of migrating vagal ENCC were captured, we noted that there were two subpopulations of cells within the wavefront, cells that were dimly labeled by CFP and cells that were notably brighter (Fig. 4B). Our imaging relied upon scanning the full depth of the fetal gut using 1-μ-slice intervals. While the CFP+ bright and dim cells were qualitatively apparent in compiled z-stacks, they were also notably evident even in consecutive single confocal sections. Analysis of CFP signal intensity was performed with MetaMorph image software to quantify levels of fluorescence within cells at the migration wavefront of ENCC (Fig. 4B). Average signal intensities were determined for all of the cells within 400 μ of the wavefront (Fig. 4, table inset). While two populations of cells are readily discernable by visual inspection of confocal images, we noted that the distribution of their signal intensities overlapped when plotted graphically (Suppl. Fig. 3).

As wavefront progenitors migrate from the midgut to the colon, they transition briefly from migrating in contact with one another as chains to migration as single cells and then revert back to contact migration in chains (Druckenbrod and Epstein,2005). This transition implies a difference in the microenvironment of the cells and, consequently, we questioned whether our observation of two subpopulations within CFP-labeled ENCC could be due to the hindgut environment. However, even at 10.5dpc before migrating ENCC labeled by the Phox2b-H2BCFP transgene expression have transitioned from the ileum to migration in the colon, we observed two populations of cells at the wavefront (Fig. 5 and Suppl. Fig. 4).

Vagal NC-derived progenitors initially enter the esophageal region of the foregut at 9–9.5dpc. Careful IHC studies have traced the initiation of Phox2b expression in the vagal and glossopharyngeal pathways (Anderson et al.,2006). Because background fluorescence from surrounding tissues complicates CFP imaging in whole mount samples, we subdissected 9.5dpc fetal guts to visualize entry and migration of ENCC precursors in the foregut. CFP+ cells were observed in the most rostral foregut just caudal to the brachial arches (Fig. 6). High-magnification imaging (400×) at this developmental stage once again identified two subpopulations of cells within Phox2b-H2BCFP labeled progenitors (Fig. 6B and Suppl. Fig. 5). Our analysis establishes heterogeneity within the earliest vagal NC-derived progenitors that enter the foregut.

The Phox2b-H2BCFP transgene reflects transcription from the Phox2b locus and does not monitor Phox2b protein levels. We considered the possibility that endogenous protein levels of Phox2b might not differ among wavefront progenitors. However, confocal images of IHC for Phox2b in 13dpc wild type gut samples similarly identified bright and dim populations of ENCC (Fig. 7).

Figure 7. Immunohistochemical detection of Phox2b confirms heterogeneity among wavefront progenitors in ENCC. Confocal image (400× magnification) of cells within the migration wavefront in the hindgut region of a fetal gastrointestinal tract at 13.5dpc stained with an antibody to Phox2b.

DISCUSSION

BAC transgenics are increasingly used for analysis of gene function and regulation in multiple model organisms. We present the construction and rigorous characterization of a mouse BAC transgenic line, Phox2b-H2BCFP, designed to facilitate analysis of lineage divergence in the autonomic nervous system. This line illuminates patterns of Phox2b expression that have previously not been appreciated in enteric glia and the earliest vagal NC-derived progenitors that populate the gastrointestinal tract. Our analysis of transgene expression and confirmation of similar patterns for the endogenous protein suggest a role for Phox2b in autonomic ganglia outside neuronal cell types and raises alternate scenarios for the generation of lineage divergence among ENCC.

Blue fluorescent reporters have not been as widely used for imaging as green fluorescent protein variants because they are typically not as stable. The Cerulean variant we have used is more stable than other CFP forms (Rizzo et al.,2004) and, when linked to the Histone2B moiety, results in strong nuclear fluorescence that is readily imaged. Moreover, because the Cerulean reporter is spectrally distinct from other GFP variants, it can be crossed to other transgenic lines for detailed analysis of lineage segregation among NC cell types.

The Phox2b-H2BCFP line recapitulates expression of Phox2b in the developing hindbrain, and multiple NC derivatives of the peripheral nervous system including the adrenal medulla and autonomic ganglia. Reporter CFP fluorescence in the hindbrain was anticipated, and its distribution among the rhombomeres is consistent with the role proposed for Phox2b in neuronal specification within these developmental compartments (Brunet and Pattyn,2002). CFP expression in chromaffin cells of the adrenal medulla is consistent with the neural crest origin of these cells (Ernsberger et al.,2005; Gut et al.,2005) and was also observed for the LacZ knock-in allele that expresses βGal from the Phox2b locus, Phox2b^tm1Jbr/+. Expression of CFP from the Phox2b-H2BCFP transgene recapitulates temporal, spatial, and cell type–specific levels of Phox2b in the ENS.

Our survey of tissues identified expression of Phox2b in developing submaxillary salivary glands. The localization of H2BCFP expression in this organ appears both in the ductal network that connects the lobes of the salivary gland as well as in what appears to be a small ganglia immediately adjacent to the duct outflow. Mesenchymal tissue of the salivary gland is a known derivative of NC (Jaskoll et al.,2002), and it has been suggested that neuroendocrine cells within the ducts of the salivary glands are of NC origin (Rollins et al.,1995; Modlin et al.,2006). The extremely low levels of βGal staining we observed may explain why this expression site has not previously been reported. The contrast of H2BCFP was much higher and easier to detect, illustrating one strength of this particular transgene model.

Visualization of Phox2b in the context of normal NC development was one of our primary aims. Enteric phenotypes have not yet been reported in mice haploinsufficient for Phox2b. However, impaired pupillary responses and abnormal pulmonary ventilation in heterozygous Phox2b^tm1Jbr/+ mice (Dauger et al.,2003; Cross et al.,2004) indicate reduced levels of this transcription factor can lead to subtle phenotypes that could complicate efforts to study normal NC processes in a knock-in model. The Phox2b-H2BCFP line avoids haploinsufficiency effects because the endogenous Phox2b gene is not altered.

Expression of Phox2b in enteric glia challenges the perception of Phox2b as strictly a neurogenic transcription factor. Prior studies suggested Phox2b might be expressed by other cell types in the ENS, but the initial finding was not definitively pursued (Young et al.,1998,2003). We demonstrate co-localization of Phox2b protein with Sox10 and GFAP in adult gut muscle strips, and the Phox2b-H2BCFP transgenic demonstrates co-localization of CFP reporter with Sox10, GFAP, and S100β. Our findings conclusively demonstrate that Phox2b expression is maintained in enteric glia. The presence of Phox2b in enteric glia suggests a role for this transcription factor in gliogenesis and glial maintenance that is open to future investigation.

We report notable heterogeneity in Phox2b expression at the wavefront of migrating enteric progenitors. Previous descriptions of enteric development that relied on IHC have concluded differentiation occurs at a distance behind the wavefront of migrating ENCC progenitors in the gut (Young et al.,1998; Conner et al.,2003). Our studies suggest that there is heterogeneity among Phox2b+ cells within the wavefront of ENCC, even upon first entry into the foregut. Careful examination of published images of Phox2b expression shows slight differences in expression of Phox2b protein among cells at the wavefront (Young et al.,1998,1999) that were likely not recognized as differential gene expression due to the modest effect. Our samples exhibited a very noticeable differential expression of Phox2b among enteric progenitors detected by IHC. The difference in degree of bright versus dim cells between the two studies could be due to technical details of our IHC sample processing that reduced the background and made the differential expression among progenitors more evident. The fact that the Phox2b-H2BCFP transgene also exhibits this variation in expression among migrating ENCC highlights the phenomena and demonstrates that the CFP reporter faithfully models expression of the endogenous gene.

We also noted that the ratio of bright:dim CFP+ cells at every developmental stage examined is such that the dim cells outnumber the bright cells. Interestingly, enteric glia outnumber enteric neurons in the mature ENS, although the exact glia:neuron ratio varies according to species and age of the animals examined (Gabella and Trigg,1984; Phillips et al.,2004; Ruhl et al.,2004). At 13dpc and 10dpc, counts of bright cells as compared to dim cells revealed that 20 and 24%, respectively, of the CFP+ cells at the furthest edge of migrating progenitors exhibited higher levels of the transgene. At 9dpc, approximately 40% of CFP+ cells appear brighter than immediately adjacent cells entering the foregut region. Analysis of transgene expression at even earlier stages when the density of ENCC progenitors in the foregut is lower would further substantiate our findings but is technically very demanding.

Given the role of Phox2b in neuronal determination (specification before differentiation) and the low levels of Phox2b present in enteric glia, we hypothesize that the two populations of enteric NC cells marked by differential expression of Phox2b represent specified neurons and glia. An alternate possibility is that the different levels of Phox2b expression observed in migrating ENCC are due to transient fluctuations in gene expression among cells that have not yet been specified. The Phox2b-H2BCFP transgene will be a valuable tool for capture and analysis of individual ENCC during the development of autonomic ganglia to distinguish between these and other scenarios of lineage diversification.

EXPERIMENTAL PROCEDURES

Generation of a pBOS-H2BCerulean Fusion

In order to create shuttle vectors, pBOS-H2BGFP was modified as follows to incorporate a new SacI site. Briefly, making a G to C substitution at the wobble position for L150 created a SacI site without altering the H2BGFP protein. This substitution was achieved by generating two PCR fragments, utilizing wild-type pBOS-H2BGFP (BD PharMingen, 559241) as a template. PCR primers 5′-GTCGACGGTACCGCCACCAT-3′ (7233, 5′H2BGFPKpn) and 5′-TACGTCGCCGTCGAGCTCGACCAG-3′ (7234, 3′H2BGFPSac) amplified a 480-bp fragment. Primers 5′-CTGGTCGAGCTCGACGGCGACGTA-3′ (7235, 5′H2BGFPSac) and 5′-AGTCGCGGCCGCTTTACTTG-3′ (7236, 3′H2BGFPNot) amplified a 687-bp fragment. The 5′ KpnI site and 3′ NotI are in italics, and the G-to-C substitution is indicated in bold in the above oligonucleotide sequences. Splice overlap extension-PCR utilizing these two fragments yielded “H2BGFP new SacI,” which was subcloned into Blunt TOPO (Promega). Sequencing of this H2BGFPSacITOPO vector confirmed the incorporation of a new SacI site. Both wild-type pBOS-H2BGFP and H2BGFPSacITOPO were digested with KpnI and NotI, and the vector and insert portion, respectively, were gel purified. These fragments were ligated to yield pBOS-H2BGFPSacI. Utilizing mCerulean-C1 (Rizzo et al.,2004) as template, primers 5′-CGAGCTCGACGGCGACGTAAACGG-3′ (7237, 5-FSacI) and 5′-CGCGGCCGCTTTACTTGTACAGCTCGTCCATGCCG-3′ (7238, 3-RNotI) amplified a 5′SacI-CeruleanFP-3′NotI fragment, which was subsequently subcloned into Blunt TOPO. Following sequence confirmation, the pBOS-H2BGFPSacI and 5′SacI-CeruleanFP-3′NotI were digested with SacI and NotI. Following gel purification, the Cerulean fragment was ligated into SacI/NotI-digested pBOS-H2BGFPSacI to yield pBOS-H2BCeruleanSacI. All PCR-generated fragments were synthesized utilizing Pfu Polymerase (Stratagene) and fully sequenced to validate amplified regions.

Phox2b-Histone2BCerulean Targeting Construct

Splice overlap extension-PCR was utilized to generate a 930-bp 5′UTR Phox2b/H2B fusion product from two templates, 95M11 BAC (CHORI library: RPCI-24 C57BL/6J strain) and pBOS H2BGFP. EcoRI and AscI sites at the 5′ end were engineered to facilitate subcloning and linearization of the final target sequence; an endogenous EcoRI site at the 3′ end was likewise used to subclone. PCR primers used were: 5′-GGAATTCGGCGCGCCGGCTTAAA- AGCTCTTGGAAATTGG-3′ (7245, 5′Phox2bUTR w/RI Asc); 5′-CCATCCAGACCTTTTCAATGCCAGAGCCAGCGAAGTCT-3′ (7246, 5′Phox2bH2B5spl); 5′-AGACTTCGCTGGCTCTGGCATTGAAAAGGTTCTGGATGG-3′ (7247, 5′Phox2bH2B3spl); and 5′-CCTCGCCCTTGCTCACCATGGTGG-3′ (7242, 3′H2Brev). The starting methionine shared by Phox2b and H2BGFP in the fusion construct is bolded within the oligo sequences above, while engineered restriction enzyme sites are underlined. EcoRI-flanked PCR fusion product was subcloned into EcoRI-digested pBOS H2BCFPSacI to yield pBOS-5′Phox2b-H2BCFPSacI. In parallel, PCR was used to generate the 3′ homology arm of Phox2b, utilizing the 95M11 BAC as template. This 510-bp arm initiated with the second codon of Phox2b to avoid the possible deletion of potential regulatory domains. AatII and SpeI sites at the 5′ end and AscI and AatII sites at the 3′ end were introduced to facilitate subcloning and linearization of the final target sequence. PCR primers used were 5′-TGACGTCACTAGTTATAAAATGGAATATTCTTACCTC-3′ (7248, 5′Phox2bAatSpe) and 5′-TGACGTCGGCGCGCCATCGCAAAGTAAATAAATTATGCC-3′ (7249, 3′Phox2bAscAat). Engineered restriction enzyme sites are underlined within the oligo sequences. The AatII PCR product was subcloned into AatII-digested pBOS 5′Phox2b-H2BCFPSacI to yield pBOS 5′Phox2b-H2BCFPSacI3′Sox10. Finally SpeI was used to release the FRT-Tet-FRT fragment from another shuttle vector. This fragment contained the FRT-flanked tetracycline resistance gene originating from pFRT-Tet129 and was ligated into SpeI-digested pBOS-5′Phox2b-H2BCFPSacI to yield the final Phox2b-H2BCFP targeting construct.

The 5′Phox2b-H2BCFP FRT-tet-FRT 3′Phox2b targeting sequence was linearized with AscI sites outside the actual homology arms. The targeting vector was electroporated into EL250 E. coli (Lee et al.,2001) previously transformed with the Phox2b 95M11BAC. Successful targeted homologous recombination resulting in the modified Phox2b-H2BCFP BAC was first identified by screening for tetracycline resistance (7.5 μg/ml) and subsequently confirmed by PCR. Removal of the tetracycline cassette was achieved by L-arabinose induction of flp recombinase present in the EL250 E. coli strain (Lee et al.,2001). Restoration of tetracycline sensitivity was confirmed by replica plating.

Generation of Phox2b-H2BCFP BAC Transgenic Lines

Phox2b-H2BCFP BAC DNA was prepared by CsCl banding and subsequent dialysis into microinjection buffer (10 mM Tris, pH 7.5; 0.1 mM EDTA, 10 mM NaCl, 30 mM spermine, and 70 mM spermidine). The modified BAC was injected into fertilized mouse eggs obtained from the mating of (C57BL/6 × SJL)F₁ females with (C57BL/6 × SJL)F₁ males by the Transgenic Animal Model Core at the University of Michigan using standard protocols (Camper et al.,1995). Transgene copy numbers were determined in offspring by using a semi-quantitative PCR assay (data not shown) in comparison to dilution standards containing 1, 2, 4, and 8 copies of each BAC/genome equivalent.

Genotyping

Genomic DNA was isolated from tail snips by proteinase K digestion and phenol-chloroform extraction. BAC transgenes were detected by PCR amplification of three independent sequences. Primers targeting the Histone2B portion of the construct were utilized to amplify the “internal” BAC sequence: 5′-CTGGTCGAGCTCGACGGCGACGTA-3′ and 5′-AGTCGCGGCCGCTTTACTTG-3′. The ends of the BAC construct were also detected by PCR with primers that primed from the vector backbone into the BAC genomic insert at both the SP6 (5′-GCCGTCGACATTTAGGTG-3′ and 5′-CAGAATCCAAAAGCAACAAG-3′) and T7 arms (5′-AGCCGCTAATACGACTCACTA-3′ and 5′-AGTGTTGCTTTCTTTGAGTGG-3′). STR markers used to screen for integrity of the Phox2b BAC 95M11 insert were identified by custom software (STRFinder, J.R. Smith, unpublished data). These repeats were evaluated for differences in repeat length or internal variants by polyacrylamide gel electrophoresis or single-strand conformational polymorphism analysis as previously described (Cantrell et al.,2004) using routine PCR conditions. Primer sequences and allele information are provided in Supplementary Table 1.

Harvesting of Enteric Gut Muscle Strips

The small intestine was dissected from adult mice (10–13 weeks old) and cut into smaller pieces, making sure to keep track of duodenum versus colon. Gut pieces were fixed for 30 min in ice-cold neutral buffered formalin (NBF) with 0.5% Triton X-100. After fixation, the inner and outer enteric muscle layers, together with the myenteric plexus, were stripped away intact from the submucosa of the gut.

Immunohistochemistry on Enteric Muscle Strips

After fixation as described above, muscle strips were washed in cold 1×PBS with 0.1% TritonX-100. Tissue was blocked at room temperature in 1×PBS containing 0.1% TritonX-100, 5% heat-inactivated normal donkey serum (Jackson Immuno Research), and 1% bovine serum albumin. After blocking, primary antibodies were diluted into block as follows: Phox2b (Pattyn et al.,1997, 1:750); Hu (Fairman et al.,1995, 1:800); Sox10 (Santa Cruz, 1:25); Cy3-conjugated GFAP (Sigma, 1:800); S-100β (Dako, 1:500). Tissue was incubated with primary antibodies overnight at room temperature, followed by washes in 1×PBS with 0.1% TritonX-100. Cy3-conjugated secondary antibodies (Jackson Immuno Research) were diluted 1:1,000 in block, and the tissues were incubated for 1 hr at 37°C with occasional agitation. Tissues were then washed in 1×PBS containing 0.1% TritonX-100 and flat-mounted in Aqua Poly/Mount (PolySciences, Inc).

Embryonic Dissections

Staged mouse embryos were obtained from timed matings with designation of the plug morning as 0.5dpc. Embryos were screened for CFP transgene expression using an Olympus BX41 fluorescent microscope with a CFP filter. GI tracts were subdissected from CFP-positive embryos and fixed in ice-cold NBF. Fixed guts were flat mounted on microscope slides in Aqua Poly/Mount. Other tissue subdissections were performed on 14.5dpc embryos, and individual organs were fixed before transferring to 1×PBS for imaging. All Phox2b-H2BCFP tissues were imaged within 48 hr as CFP signal fading occurred within a few days of isolation.

Flat-Mounts of 10.5dpc Hindbrains

Phox2b-H2BCFP and B6D2.Phox2b^tm1Jbr/+ embryos were dissected at 10.5dpc into 1X PBS. CFP+ embryos from Phox2b-H2BCFP BAC transgenic litters were screened for CFP expression. All embryos harvested from B6D2. Phox2b^tm1Jbr/+females were processed to detect β-galactosidase (β-gal) staining and visualize Phox2b^tm1Jbr/+ reporter expression. The hindbrain, from the developing eye to the level of the otic vesicle, was dissected for imaging. This tissue was fixed in ice-cold NBF for 30 min and then washed three times in PBS. Phox2b-H2BCFP embryonic hindbrains were mounted and imaged immediately. Phox2b^tm1Jbr/+ hindbrains were stained for β-galactosidase activity by routine methods (Deal et al.,2006). For mounting, hindbrain tissue was positioned on glass slides in Aqua Poly/Mount with the brachial arches downward so the hindbrain was in view. The neural tube was filleted open with a 30-gauge needle so that the ventricles could be splayed flat laterally and flattened by coverslipping.

Fluorescent Microscopy

Confocal microscopy of 10.5dpc embryos and enteric muscle strips was performed on a Zeiss Scanning Microscope LSM510 using a CFP band-pass filter to visualize the fluorescent reporter present in this transgene, as well as rhodamine (Cy3, Texas Red, LP560) and Cy5 (LP650) long pass filters to visualize endogenous Phox2b protein and other glial markers stained with antibodies in gut muscle strips. Images of sub-dissected 14.5dpc organs were captured on a Zeiss Stereo Lumar V12 fluorescence stereomicroscope equipped with a Q-Imaging Retiga 4000R digital camera and software. Additional images were obtained on an Olympus BX41 equipped with a DP70 camera and software.

Analysis of Signal Intensity

Stacks of images acquired using the LSM510 confocal microscope were converted to single extended focus projections in order to determine CFP-related fluorescence intensity values. MetaMorph image analysis software was used to determine CFP fluorescence levels within cells in the migration wavefront of 9.5dpc, 10.5dpc, and 13.5dpc Phox2b-H2BCFP fetal guts. An 8-bit pixel intensity scale of 0–256 arbitrary intensity units was used to determine the threshold value for CFP fluorescence compared to non-fluorescent areas of the gut lumen and adjacent tissue. Cells of interest were outlined using the region of interest (ROI) tool and average intensity values of each ROI were determined. An average background value based on two independent regions within each gut sample was calculated and subtracted from the average CFP fluorescence values to obtain the CFP intensity value for each ROI.

Acknowledgements

We thank Alexandre Pattyn and Jean-François Brunet for generously sharing the Phox2b-LacZ knock-in line (Phox2b^tm1Jbr/+) and the Phox2b antibody used in these studies; Dr. Neil Copeland for the EL250 E. coli strain and BAC recombineering methods; Jeffrey R. Smith and Kevin Bradley for assistance in identification of STR primers within the Phox2b BAC interval; Dr. Dave Piston for providing the Cerulean coding region construct; Elizabeth M. Stein for technical support in mapping STR markers within the Phox2b-H2BCFP transgenic lines; Dr. Miles Epstein for generously providing the Hu antibody; Drs. Thomas Saunders and Maggie Van Keuren for outstanding support in the generation of BAC transgenics in the University of Michigan Transgenic Animal Model Core. We gratefully acknowledge and thank Dr. Sam Wells and the support staff of the Cell Imaging Shared Resource Core for advice and assistance in confocal imaging. The Cell Imaging Shared Resource Core is supported by NIH CA68485, DK20593, DK58404, HD15052, DK59637, EY08126. This work was also supported by research funds provided to EMS² from NIH DK064251 and the March of Dimes FY06-390.