Expression of Hand2 is sufficient for neurogenesis and cell type–specific gene expression in the enteric nervous system



The basic helix-loop-helix DNA binding protein Hand2 is expressed in neural crest–derived precursors of enteric neurons and has been shown to affect both neurogenesis and neurotransmitter specification of noradrenergic sympathetic ganglion neurons. In the current study, our goal was to determine whether Hand2 affects neurogenesis and/or expression of vasoactive intestinal polypeptide and choline acetyltransferase in developing enteric neurons. Gain-of-function of Hand2 in HNK-1+ immmunoselected precursor cells resulted in increased neurogenesis. The number of neurons expressing vasoactive intestinal polypeptide increased in response to Hand2 overexpression although choline acetyltransferase was not affected. Targeted deletion of Hand2 in neural crest cells resulted in loss of all neurons expressing vasoactive intestinal polypeptide along the length of the gastrointestinal tract, patterning defects in the myenteric plexus of the stomach, and altered number and morphology of neurons expressing TH. Our data demonstrate that expression of Hand2 is sufficient and necessary for neurogenesis and expression of a subset of cell type-specific markers in the developing enteric nervous system. Developmental Dynamics 236:93–105, 2007. © 2006 Wiley-Liss, Inc.


The enteric nervous system is derived from neural crest cells that migrate from vagal and sacral levels of the neural axis to populate the anlage of the developing gastrointestinal tract (reviewed in Young et al., 2003; Newgreen and Young, 2002; Gershon and Ratcliffe, 2004). Neural crest–derived progenitor cells differentiate into a variety of neurons exemplified by diversity in their morphology, connectivity, and neurotransmitter. Each of these phenotypic characteristics not only describes aspects of neuronal morphology but defines neuronal function as well. Mechanisms responsible for generation of neuronal diversity thus become tools for understanding neuronal development.

It is well established that there is a core network of DNA binding proteins required for the development of autonomic neurons (Southard-Smith et al., 1998; Lo et al., 1999; Pattyn et al., 1999; Howard et al., 2000; Coppola et al., 2005; reviewed in Bertrand et al., 2002; Howard, 2005). The development of sympathetic, parasympathetic, and enteric neurons depends upon expression of the homeodomain DNA binding protein Phox2b and the basic helix-loop-helix (bHLH) DNA binding protein Mash1. Development of each branch of the autonomic nervous system depends further upon recruitment of additional DNA binding proteins that occurs with both spatial and temporal specificity. In the enteric nervous system, two interacting networks of regulatory molecules, one necessary to support migration and/or neurogenesis and the other to support cell type-specific gene expression, are beginning to be elucidated. Expression of Ret and/or endothelin receptor B, which is required for survival and cell fate determination in neural crest–derived precursors of enteric neurons, can be regulated by Phox2b, Sox10, or Pax3 (Pattyn et al., 1999; Bondurand et al., 2000; Lang et al., 2000; Lang and Epstein, 2003; Cantrell et al., 2004; Elworthy et al., 2005; Stanchina et al., 2006). Interestingly, loss-of-function of these (and other) genes, results in loss of enteric neurons in a spatially distinct manner.

Mash1 is required for the development of neurons in the proximal bowel (Guillemot et al., 1993) that express serotonin (Blaugrund et al., 1996). Deletion of Pax3 (homozygous) results in loss of enteric neurons in the small and large intestine (Lang et al., 2000; Lang and Epstein, 2003; Engleka et al., 2005) while homozygous loss of Sox10 is correlated with an absence of enteric neurons along the length of the gastrointestinal tract; interestingly, heterozygous loss of Sox10 is related to aganglionosis in the distal bowel (Pingault et al., 1998; Herbarth et al., 1998; Southard-Smith et al., 1998; Lang et al., 2000). It is difficult to separate effects on neural crest cell migration and neuronal differentiation in response to expression of Sox10 and Pax3; it is likely that these genes, in concert with additional regulatory molecules, function in multiple aspects of migration and specification. To date, Mash1 is the only DNA binding protein expressed in the ENS shown to function in both neurogenesis and cell subtype-specific gene expression (Blaugrund et al., 1996). Because of the spatial restriction of Mash1 expression as well as the great diversity of neurotransmitters and neuropeptides expressed in enteric neurons, identification of additional regulatory molecules that support development of this diversity becomes important. The neuronal expression of the bHLH DNA binding protein Hand2 (dHand) is restricted to sympathetic and enteric ganglia (Cserjesi et al., 1995; Howard et al., 1999; Wu and Howard, 2002) suggesting a potential role in the development of these ganglia.

Hand2 serves critical functions impacting development of the autonomic nervous system (reviewed in Howard, 2005; Sarkar and Howard, 2006). Development of sympathetic ganglion neurons is dependent upon Hand2 (Howard, unpublished data); expression of Hand2 is sufficient to support both neurogenesis and expression of the catecholaminergic marker gene TH and the noradrenergic cell type-specific marker DBH in vitro (Howard et al., 1999; Liu et al., 2005) and in vivo (Howard et al., 2000). In concert, Hand2 and Phox2a appear to couple neurogenesis and expression of noradrenergic marker genes.

Transcript encoding Hand2 is expressed throughout the developing gastrointestinal tract (Wu and Howard, 2002) and early becomes confined to cells located within the submucosal and myenteric plexi. In the ENS of chicken embryos, it appears that Hand2 is expressed in neurons but not support cells (Wu and Howard, 2002). Interestingly, expression of Hand2 is not extinguished with increasing developmental age, suggesting an ongoing function for Hand2. Since Hand2 is expressed early in neural crest–derived precursors of enteric neurons in all anatomical domains of the developing ENS, we were interested to determine whether, in a manner similar to what we have shown for sympathetic chain ganglion neurons, Hand2 supports neurogenesis as well as specification of neurotransmitter identity. Our data suggest that Hand2 supports neurogenesis and affects neurotransmitter choice in a subset of neural precursor cells derived from the ENS.


Hand2 Supports Neurogenesis in Gut-Derived Precursor Cells

As a mechanism to determine whether expression of Hand2 is sufficient to support neurogenesis in gut-derived neural crest-derived cells, we overexpressed Hand2 in neural crest–derived precursor cells immunoselected from embryonic day (ED) 5/6 chick gut (Fig. 1). Hand2 was overexpressed in HNK-1+ selected and residual cells (HNK-1) using retrovirus-mediated gene transfer as previously described (Howard et al., 1999, 2000). Selected and residual cells were assessed for neurogenesis based on TUJ1 immunoreactivity (IR) and cells expressing Hand2 were identified based on expression of the viral coat proteins, p27 or p19. We used p27 IR or P19 IR as an index of retroviral infection and thus as a marker for cells expressing Hand2.

Figure 1.

Expression of Hand2 supports neurogenesis. Neural crest–derived cells were immuno-purified from ED5/6 chicken embryo gut using the HNK-1 antibody. Retrovirus-mediated gene transfer was used to overexpress Hand2 and the number of cells expressing the neuron marker, neuron specific β-tubulin (TUJ1 IR), was determined on culture day 7. The total cell population was visualized based on incorporation of DAPI. The number of cells expressing TUJI IR and DAPI was counted in a minimum of 20 microscope fields per dish. The data are expressed as ratios in order to: (1) quantify the proportion of total cells that differentiate as neurons (TUJ1/DAPI); (2) determine the number of neurons that also express the retroviral marker p27 (indicator of Hand2 overexpression) as a function of the total cells (TUJ1+p27/DAPI); (3) determine the number of neurons that also overexpress Hand2 (TUJ1+p27/TUJ1); and (4) determine the infection efficiency of RCAS-Hand2 (p27/DAPI). The data suggest that gain-of-function of Hand2 (sense RCAS-Hand2) significantly increased the numbers of cells expressing the pan-neuronal marker neuron-specific β-tubulin (TUJ1). Blockade of Hand2 expression (antisense) reduced the number of cells expressing TUJ1 IR. Detection of cells overexpressing Hand2 relied upon expression of the viral coat protein p27 (see Experimental Procedures section for detail). Infectivity of both sense and antisense RCAS-Hand2 was equivalent, suggesting effects on neurogenesis were specific. Data shown are from one experiment with triplicate cultures for each condition. Studies employing sense RCAS-Hand2 were repeated 6 times in triplicate. Studies employing antisense RCAS-Hand2 were repeated 3 times in triplicate.

Neurogenesis was significantly (P < 0.001) increased in cells expressing Hand2. On average, 9.6 ± 3% of control cells (range 5.5–13.0%) differentiated as neurons (TUJ1/DAPI) compared to 22.8 ± 3% of cells (range 19.4–26.7%) expressing Hand2. Interestingly, although expression of Hand2 consistently supported an increase in neurogenesis, expression of the retrovirus declined over time with an average 14% of neurons (range 12.3–16.5%) expressing both TUJ1 and p27 after 7 days. These data suggested that blockade of Hand2 expression should result in an inhibition of neurogenesis. Indeed, if HNK-1+ cells were treated with a retrovirus expressing antisense Hand2, neurogenesis was significantly (P < 0.05) reduced. In cultures treated with antisense Hand2, 9.0 ± 0.9% of cells expressed TUJ1 compared to 22.8% of cells in cultures exposed to sense RCAS-Hand2. In the cells infected with sense Hand2, 13.9 ± 1.6% of cells expressed both TUJ1 and p27 while in cells infected with antisense Hand2 1.4 ± 0.5% of cells expressed both TUJ1 and p27. Analysis of the differentiated neurons demonstrated that 64 ± 3% of the total neuron population expressed Hand2 (TUJ1+p27/TUJ1). This value is significantly (P < 0.001) higher than that in neurons expressing antisense Hand2 (15 ± 2%). The percentage of cells expressing Hand2 or antisense Hand2 was comparable (p27/DAPI) suggesting equivalent infectivity between cultures as well as between retroviral constructs. This bolsters the conclusion that expression of Hand2 is sufficient to support neurogenesis in HNK-1+ selected neural crest–derived cells from the gastrointestinal tract.

We tested the response to expression of Hand2 in residual cells (HNK-1). Expression of Hand2 did not support neurogenesis in the residual cell population. We were unable to detect cells expressing TUJ1 in either control or in Hand2 infected cells (not shown). This is an interesting result because ectopic expression of Hand2 in precursor cells resident along developing peripheral nerves supports their differentiation as neurons (Howard et al., 2000); this suggests that HNK-1 cells in the gastrointestinal tract do not have properties of neural precursor cells.

HAND2 Is Involved in Expression of VIP

In precursors of noradrenergic sympathetic ganglion neurons (Howard et al., 2000) and cholinergic ciliary ganglion neurons (Muller and Rohrer, 2002), expression of Hand2 is sufficient to support expression of the catecholaminergic and noradrenergic marker genes TH and DBH, respectively, as well as pan-neuronal marker proteins neurofilament and SCG10 (Howard et al., 2000). Because Hand2 appears necessary and sufficient to support cell type–specific gene expression in sympathetic noradrenergic neurons, we were interested to determine whether Hand2 serves a similar role in developing enteric neurons. To ask whether Hand2 has a global function in neurogenesis in ENS-derived precursor cells or supports neurogenesis in a cell subtype–specific manner, we chose choline acetyltransferase (ChAT) and vasoactive intestinal polypeptide (VIP) as our first two cell type–specific markers to assess. VIP is expressed in a subset of interneurons and descending inhibitory neurons and as such has a restricted expression pattern compared to ChAT, which is expressed in afferent neurons, interneurons, and excitatory muscle motoneurons (Schemann, 2005).

Overexpression of Hand2 in HNK-1+ immunoselected cells resulted in a significant (P < 0.001) increase in the number of cells expressing VIP IR (Fig. 2A,B). In non-infected cells, 5.0 ± 0.02% of cells expressed VIP; this value increased to 8.5 ± 0.7% in cells overexpressing Hand2. For cells infected with retrovirus expressing sense Hand2, 4.2 ± 0.1% of cells expressed both VIP and p19. Based on the total population of cells expressing VIP IR, 37.9 ± 6% also expressed Hand2 (VIP+p19/VIP). Interestingly, Hand2 had little effect on development of immunoreactivity to choline acetyltransferase (ChAT). In control cells, 6.0 ± 0.6% expressed ChAT IR (Fig. 2C). This value is not significantly different from 7.7 ± 1.2% of cells expressing both Hand2 and ChAT suggesting that expression of ChAT may be regulated independent of Hand2.

Figure 2.

Overexpression of Hand2 affects development of VIP but not ChAT. Hand2 was overexpressed in HNK-1+ immunoselected cells from ED4/5 chicken embryo gut using retrovirus-mediated gene transfer. The rationale for expressing the combined data as ratios is described in the legend for Figure 1. A: The proportion of cells expressing VIP was increased by overexpressing Hand2 (VIP/DAPI). As suggested, the incidence of cells expressing VIP IR was significantly increased in cells also expressing Hand2 (VIP+p19/VIP). For these studies, we used expression of the viral coat protein p19 (AMV3c2 antibody) to identify cells expressing Hand2 in response to retrovirus-mediated gene transfer; this allowed us to identify cells also expressing VIP in co-immunolabeling studies with a rabbit anti-VIP antibody, which detects this antigen efficiently in avian cells. Blockade of Hand2 expression (antisense Hand2) significantly reduced the number of cells expressing VIP IR induced by Hand2 (VIP+p19/VIP) supporting the conclusion that endogenous Hand2 influences the expression of VIP. B: Gut-derived neural crest–derived cells (HNK-1+ immunoselected) grown under control growth conditions differentiate into neurons expressing VIP IR (red) at a lower density than in HNK-1+ immunoselected cells overexpressing Hand2 (RCAS-Hand2). The total population of cells was identified using DAPI (blue). Interestingly, expression of Hand2 did not affect a change in the number of cells expressing ChAT. C: Neither the proportion of total cells expressing ChAT (ChAT/DAPI) nor the number of cells expressing both ChAT and Hand2 (ChAT+p19/DAPI) was affected by overexpression of Hand2. Although the efficiency of infection of antisense RCAS-Hand2 was higher than sense RCAS-Hand2 (p19/DAPI), there was no apparent effect of either gene product on ChAT expression (ChAT+p19/ChAT). Data shown are from one experiment with triplicate cultures. Studies examining expression of VIP were repeated three times in triplicate. Expression of ChAT was assessed in two experiments in triplicate cultures.

For neural crest–derived noradrenergic sympathetic ganglion neurons, we found that Hand2 induced expression of DBH (Howard et al., 2000) and interacted with Phox2a to regulate transcription of DBH (Xu et al., 2003) suggesting that Hand2 might function at the VIP promoter to regulate its transcription. We tested this notion using transient transfection of P19 embryonal teratocarcinoma cells with a VIP-luciferase reporter construct (kindly provided by Lee Eiden) in the presence and absence of Hand2 (Fig. 3). Activity of the luciferase reporter significantly (P < 0.01) increased 2.2- ± 0.2-fold over control in the presence of Hand2. This suggests that Hand2 may regulate expression of VIP by direct regulation of transcription.

Figure 3.

Hand2 increases VIP-luc reporter expression in P19 cells. P19 cells were transiently transfected, using our standard protocol (Xu et al., 2003), with a VIP5.2-luc reporter construct in the absence (VIP) or presence of Hand2 (VIP+Hand2). For each transfection, pRL-tk was included as an internal control for transfection efficiency. Cells were harvested after 48 hr and luciferase assays performed on cell lysates. Hand2 significantly (P < 0.01) enhanced basal VIP-reporter activity. The data are presented as the mean ± S.E.M. (five independent transfections each in triplicate) of normalized Firefly/Renilla luciferase units.

Disruption of Hand2 Expression Affects Patterning, Neurogenesis, and Expression of VIP

Because deletion of Hand2 is embryonic lethal at E10.5 (Srivastava et al., 1997), due to defects in cardiac and vascular development (Yamagishi et al., 2000), we generated a conditional allele of Hand2 (Fig. 4A). To generate a conditional Hand2-null allele, we introduced loxP sites flanking exon 1 by homologous recombination in ES cells. The targeted allele was transmitted through the germline in chimeric male mice. Heterozygous Hand2loxpneop mice were mated to Sycp1-Cre mice to remove the neomycin-resistance cassette in the male germline. Deletion of the neomycin cassette was confirmed by RT-PCR and sequence analysis. Hand2loxp/loxp mice are phenotypically normal and fertile. To target selective deletion of Hand2 in neural crest–derived cells, we crossed Hand2loxp/loxp mice with mice carrying Cre recombinase under transcriptional control of the Wnt-1 promoter (Danielian et al., 1998; Jiang et al., 2000; Brewer et al., 2004). We confirmed loss of Hand2 expression in the Hand2loxp/loxp;Wnt1-Cre (Hand2 conditional null) mice by in situ hybridization using a dig-labeled Hand2 cRNA probe (Fig 4B).

Figure 4.

Conditional Disruption of Hand2 affects ENS development. A: A floxed allele of Hand2 was fabricated by inserting loxP sites flanking exon1 in the 5′ UTR of Hand2. The flowchart shows the Hand2 genomic locus and the targeting vector. The 5′ and 3′ targeting arms are shown in dark blue and green, respectively. Germline-mediated excision of the neomycin resistance cassette was achieved by crossing Hand2ploxpneop animals with Sycp1-Cre mice. Excision of the neomycin resistance cassette was confirmed by sequencing. Selective deletion of Hand2 was achieved by crossing Hand2loxp/loxp mice with Wnt1-Cre transgenic mice (Jackson Laboratories, Bar Harbor, ME). It is well established that the Wnt1-Cre transgene allows floxed genes to be targeted in the neural crest (Danielian et al., 1998; Jiang et al., 2000; Brewer et al., 2004). B: Efficacy of Hand2 deletion was demonstrated using in situ hybridization in gut sections from E14 littermate embryos. In Hand2wt/wt stomach (Stm), small intestine (Sm Int), and colon (Col), mRNA encoding Hand2 is localized in myenteric ganglia. There are no cells evident in the tissue derived from Hand2loxp/loxp;Wnt-1-Cre stomach, small intestine, or colon expressing transcript encoding Hand2. Sites of Hand2 expression were visualized in the stomach using NBT/BCIP, which gives a red signal, and BM Purple in the small intestine and colon, which gives a blue signal.

Morphological analysis of the gastrointestinal tract of E14 or E16 Hand2 conditional null embryos revealed no gross abnormality. We predicted that loss of Hand2 expression would affect development of enteric neurons. To assess neuronal development, we examined TUJ1 IR (green) in whole-mount preparations of tissue obtained from the entire length of E14 GI tract (stomach antrum shown in Fig. 5). Using confocal imaging, we were able to demonstrate a striking effect of Hand2 deletion on the overall patterning of the ENS. When viewed at low magnification (Fig. 5A compared to C), the higher density of neurons in wild-type embryos compared to Hand2 conditional null embryos is clearly seen (compare Fig. 5A,B with C,D). The region of the stomach shown is comparable to Figure 7C and G. In addition, the ganglia are disorganized. The disorganization of ganglia is readily apparent when the pattern of cells expressing TH IR (red) is compared between normal animals (Fig. 6A–C) and Hand2 conditional null embryos (Fig. 6G–I). Not only is the density of TH+ cells decreased in the Hand2 conditional null embryos but the size of the cells is markedly enlarged when compared to wild-type embryos (Fig. 6E,F compared to 6K,L). To gain further insight into the apparent patterning defect resulting from the loss of Hand2 expression, we used acetylcholinesterase histochemistry to visualize patterning of the neuronal plexus in E16 stomach (Fig. 7). A comparison of the acetylcholinesterase staining pattern in Hand2 conditional null embryos (Hand2loxp/loxp;Wnt1-Cre) with that of Hand2loxp/+ embryos shows patterning defects along the oral to anal length of the stomach. At the oral end (Fig. 7A,B, compared to E,F), there is a decrease in the density of fibers as well as an apparent decrease in the density of ganglia. Although the density of fibers increases toward the pyloric region (Fig. 7D compared to H) in both wildtype and conditional null samples, the decreased development of ganglia and fibers in the Hand2 null tissue is visible. In the pyloric region of the Hand2 conditional null stomach, the patterning of the ganglia appears more normal than in other parts of the stomach although the density of both fibers and ganglia remains less than in wildtype stomach. This pattern is also seen in specimens labeled with the TUJ1 antibody where the patterning of the pyloric region of the stomach appears more normal in the Hand2 conditional null tissue compared to wildtype stomach (Fig. 6D–F compared to J–L).

Figure 5.

Conditional deletion of HAND2 affects patterning of enteric ganglia and neuron number. Whole-mount sections of stomach were dissected from E14 Hand2wt/wt (A,B) and Hand2loxp/loxp;Wnt-1-Cre (C,D) embryos and immunolabeled with TUJ1 antibody to identify neurons. Patterning defects seen in the stomach body suggest a decrease in density of fibers and relative disorganization of the plexus (A,C). The white boxes outline equivalent regions toward the antrum in wildtype (B) and Hand2 conditional null (D) stomach, shown at higher magnification, demonstrating the more normal overall pattern of the neuronal plexus in the pyloric region, but with many fewer neurons compared to wildtype. Scale bar = 250 μm (A,C), 50 μm (B,D).

Figure 6.

Conditional deletion of Hand2 affects expression of TH IR. Pieces of E14 stomach were immunolabled (in whole-mount) with TUJ1 (green) to identify neurons and anti-TH (red) to identify cells expressing tyrosine hydroxylase. In Hand2wt/wt (A–F) stomach, TH IR is spread throughout the ganglionic plexus. The cell bodies are small and display dense processes. The pattern of both TUJ1 IR and TH IR is affected in Hand2loxp/loxp;Wnt-1-Cre (G–L) embryos. Although the number of TH-expressing neurons is decreased, the neurons are large with small stubby processes. Scale bar = 250 μm (A–C and G–I) and 50 μm (D–F and J–L). For images shown in A–C and G–I, the magnification is 10×. For images shown in D–F and J–L, the magnification is 65× (6.5× zoom of 10× image).

Figure 7.

Conditional deletion of Hand2 affects myenteric plexus patterning in the stomach. Acetylcholinesterase histochemistry was used to demonstrate the pattern of the neuronal plexus in the stomach of wildtype (top) and Hand2 conditional null E16 embryos (bottom). The stomach was removed from the embryos, fixed, bisected, and stained. Incubation in methyl salicylate was used to clear the tissue post-staining. Regions shown in A–D from Hand2wt/wt stomach and E–H from Hand2loxploxp;Wnt1-Cre stomach are high magnification (20×) images from the regions shown by the blue arrows in the low magnification images (4×). Loss of Hand2 results in less densely packed ganglia and decreased fiber growth.

We considered the possibility that precursor cells might be arrested at this developmental stage in the absence of Hand2. We tested this idea by examining the expression pattern of neuron specific β-tubulin (TUJ1), to identify neurons and neuronal precursor cells, and HuC/D that identifies post-mitotic neurons in E14 proximal bowel. A very interesting phenomenon was uncovered by comparing the expression patterns of TUJ1 (red) and HuC/D (green) in wildtype and Hand2 conditional null embryos (Figs. 8 and Fig. 9, respectively). In wild-type embryos (Fig. 8), we observed essentially complete overlap of TUJ1 and HuC/D IR. Examination of z-stack-images that have been projected through the z-x axis (Fig. 8D,L), or the z-y plane (Fig. 8 H) show coincident expression of TUJ1 and HuC/D as well as the compactness of the neurons and their processes. Interestingly, in Hand2 conditional null samples (Fig. 9), it becomes evident that TUJ1 and HuC/D IR are to a large degree non-overlapping. Based on the density of cells expressing TUJ1 IR and/or HuC/D IR viewed at low magnification (10×, Fig. 9A–C), not only is the number of neurons reduced when compared to equivalent sections obtained from wildtype embryos (Fig. 8A–C) but the depth of ganglia appears decreased. When viewed edge-on in a three-dimensional volume reconstruction (Fig. 8D compared to Fig. 9D), the apparent depth of the plexus is reduced in the Hand2 conditional null proximal bowel. There are many areas where co-localization of TUJ1 and HuC/D is essentially lacking (Fig. 9G,H,K,L) although there are some cells that do express both TUJ1 and HuC/D IR (Fig. 9C,D,G,L; yellow cells).

Figure 8.

TUJ1 and HuC/D are co-localized in E14 Hand2wt/wt embryonic proximal bowel myenteric plexus. Frozen sections (16 μm) were immunolabeled with TUJ1 (red) and anti-HuC/D (green). Confocal microscopy was used to acquire optical sections at 0.5-μm steps through the tissue sample. The images are shown as projections of the entire stack in (A–C; E–G; I–K). Three-dimensional volume reconstructions were compiled using the Leica LAS software package. The reconstructions were rotated 90° or 45° (D, L, respectively) in the z-y plane or z-x plane (H) in order to visualize the edge-on view of the stack; this allowed us to examine the thickness of the neuron plexus as well as directly observe sites of coincident labeling. Merged projection images are shown in C, G, and K. The majority of neurons co-localize HuC/D and TUJ1. For the images shown in A–D, a 20× objective was used. For images E–H, a random region was chosen, the 20× image was optically zoomed 3.6× and digitally acquired. For a second random region, the 20× images (I–L) were optically zoomed 8× and digitally acquired.

Figure 9.

TUJ1 and HuC/D are not co-localized in the ENS of E14 Hand2loxp/loxp;Wnt-1-Cre embryonic proximal bowel. The samples were treated and imaged as described in the legend to Figure 8. The embryo shown is a littermate to the wildtype embryo shown in Figure 8. Examination of the staining pattern and density of immunoreactivity for TUJ1 and HuC/D as visualized in the projections (A–C; E–G; I–K) demonstrates an overall decrease in neurogenesis in the ENS of embryos where Hand2 expression has been deleted (compare A–C in Figs. 8. and 9). One striking characteristic of the merged image shown in C is the limited number of cells expressing both TUJ1 and HuC/D (yellow). Two regions were randomly chosen to examine further the overlap of TUJ1 and HuC/D and are shown in E–H and I–L. From the 3-dimensional reconstructions, it appears that the depth of the myenteric plexus is decreased in the Hand2 conditional null ENS (compare Figs. 8D to 9D). Very striking is the lack of coincident labeling for TUJ1 and HuC/D in the majority of cells. Images were acquired as described in the legend to Figure 8.

Results of our gain-of-function study suggest that deletion of Hand2 in the developing enteric nervous system should affect expression of VIP. Indeed, examination of VIP IR in stomach indicated that neurons expressing VIP are essentially absent (Fig. 10A,C compared to B,D) in the ENS of Hand2loxp/loxp;Wnt1-Cre embryos. In the pyloric region of the stomach, comparison of the pattern of TUJI IR in wildtype (Fig. 10C) and Hand2loxp/loxp;Wnt1-Cre (Fig. 10D) tissue shows a fairly normal plexus (comparable to Fig. 8H) but lacking in VIP IR. The abnormality we observe in neuron plexus patterning as well as the loss of VIP-expressing neurons suggests a critical function of Hand2 in both neurogenesis and cell subtype–specific neuropeptide expression in the developing ENS.

Figure 10.

VIP-expressing neurons are absent in the ENS of Hand2loxp/loxp;Wnt-1-Cre embryos. The stomach was removed from E16 Hand2wt/wt (A) and Hand2loxp/loxp;Wnt-1-Cre (B) littermate embryos and immunostained in whole-mount to localize TUJ1 IR and VIP IR. Examination of VIP IR alone (A,B) showed that VIPergic neurons develop in wildtype animals but are absent from Hand2loxp/loxp;Wnt-1-Cre stomach (10× magnification). We confirmed that neurons had developed in the Hand2loxp/loxp;Wnt-1-Cre stomach by demonstrating co-immunolabeling of TUJ1 and VIP (80× magnification) in the wildtype embryo (C) but only TUJ1 IR in the Hand2 conditional null tissue (D). Scale bars = 250 μm for A,B, 25 μm for C,D. The images shown in A,B were acquired at 10× magnification and the images shown in C,D were acquired using a 40× oil immersion objective with an additional 2× optical zoom.


The mechanisms underlying generation of phenotypic diversity in the ENS remain poorly understood. For the developing nervous system in general, a common theme is becoming apparent where the interplay between homeodomain DNA binding proteins and bHLH DNA binding proteins is responsible for the generation of specific neuron sub-types. There is an emerging understanding of the transcription factor code responsible for generating neuron subtype specificity in the spinal cord (Lee and Pfaff, 2001; Caspary and Anderson, 2003; Cheng et al., 2005; Glasgow et al., 2005) and sympathetic branch of the autonomic nervous system (Guillemot et al., 1993; Groves et al., 1995; Morin et al., 1997; Lo et al., 1999; Pattyn et al., 1999; Stanke et al., 1999; Howard et al., 2000; Brunet and Pattyn; 2002; Goridis and Rohrer, 2002; Capolla et al., 2005) but such a code for the ENS has yet to be elucidated. In the current study, our goals were to: (1) determine whether expression of the bHLH DNA binding protein Hand2 is sufficient and/or necessary for neural crest–derived precursor cells resident in the developing ENS to differentiate as neurons, and (2) ask if expression of Hand2 has an impact on expression of VIP and/or ChAT. Our results suggest that expression of Hand2 in neural crest–derived precursors of enteric neurons differentially affects cell fate determination, neurogenesis, and patterning of the myenteric plexus.

Impact of Hand2 Expression on Neurogenesis

Gain-of-function of Hand2 has demonstrated its sufficiency both in vitro (Howard et al., 1999; Morikawa et al., 2005) and in vivo (Howard et al., 2000) to support expression of the pan neuronal marker genes SCG10, neurofilament and neuron-specific β-tubulin. An analysis of the expression pattern of transcripts encoding Hand2 suggested an early but sustained function in development of the ENS (Wu and Howard, 2002). In chicken embryos, Hand2 is expressed throughout development in all anatomical domains of the GI tract in both submucosal and myenteric ganglia. These results prompted us to posit that once neural crest–derived cells reach the wall of the GI tract, expression of Hand2 would support their differentiation as neurons. This notion was supported by demonstrating a significant increase in the number of neurons differentiating in vitro in response to overexpression of Hand2. Our data suggest that ectopic expression of Hand2 in neural crest–derived cells will induce their differentiation as neurons regardless of their specified fate in the enteric anlage. Interestingly, neurogenesis did not occur in non-neural crest–derived cells ectopically expressing Hand2. This supports the idea that this gene product is likely not serving a proneural function in the ENS and confirms our previous conclusion that Hand2 is not a proneural gene in the autonomic nervous system (Howard, 2005).

The ability of antisense Hand2 to inhibit neurogenesis in gut-derived neural precursor cells raises the possibility that Hand2 is both sufficient and necessary for neurogenesis in the ENS. This conclusion is not borne out by the phenotype of our Wnt1-Cre Hand2 conditional mutant mice. Our expectation was that deletion of Hand2 in neural crest–derived cells populating the gut would obviate their differentiation as neurons. Our results suggest a more complicated scenario. Using the TUJ1 antibody to mark neurons, we found an effect on the overall patterning of the myenteric plexus in the stomach. The differences in patterning were clearly demonstrated along the oral to anal axis using acetylcholinesterase histochemistry to visualize the neural plexus in the stomach. Our results suggest that there is less uniformity to the patterning of the ganglia, alterations in fiber density, and a decrease in the neuron population although not a complete absence of TUJ1+ cells. This result is open to interpretation. It is possible that expression of Hand2 may be sufficient but not necessary for enteric neural crest–derived cells to differentiate as neurons. This conclusion is difficult to reconcile with our gain-of-function data, data indicating that Hand2 is absent in the GI tract of Phox2b knock-out animals (Christos Goridis, personal communication) and that blockade of Hand2 expression in tissue culture results in a substantial loss of neurons. It is formally possible that expression of Hand2 is required for the differentiation of a subset of neurons and that these cells are lost in the mutant. Our data support this conclusion (discussed in greater detail below). It is also possible that neuron-specific β-tubulin is not expressed exclusively in neurons. Indeed, it appears likely that some neural precursor cells can be labeled with the TUJ1 antibody (Michael D. Gershon, personal communication). The non-overlap we observe comparing TUJ1 IR and HuC/D IR suggests that in the absence of Hand2, neural precursor cells may become arrested at this stage of their development and that these cells express TUJ1 IR. Inasmuch as there is a large disparity between the overall expression of neuron-specific β-tubulin and the number of cells expressing HuC/D in Hand2 conditional null embryos, we propose: (1) that Hand2 function is necessary for the development of a subset of ENS neurons, and (2) that one function of Hand2 in neurogenesis involves the transition from a precursor to a differentiated neuron. It is also possible that Hand2 is not efficiently deleted at early stages of ENS development although our in situ hybridization results do not support this conclusion. Additional analysis will be required to support or refute these possibilities.

Expression of Hand2 Is Necessary for the Development of VIPergic Neurons

Our data suggest that expression of Hand2 is sufficient and necessary for the development of neurons expressing VIP in the ENS. Gain-of-function of Hand2 in ENS-derived precursor cells supports expression of VIP while having no effect on the expression of ChAT. This is an intriguing result because we chose these two molecules with the intent of being able to distinguish between an effect on neurotransmitter/neuropeptide specification and cell type–specific effects of Hand2. While ChAT is expressed in the majority of neurons, VIP is expressed in interneurons and descending inhibitory neurons, allowing us to distinguish a global function of Hand2 in cell-type specific marker expression versus a more defined role in specifying subsets of phenotypic characteristics. A role for Hand2 in specifying only a subset of neurotransmitter characteristics is supported by the loss of VIP-expressing neurons in Hand2 conditional null embryos without the complete loss of neurons. Although we have not done a comprehensive examination of neurotransmitter/neuropeptide expression patterns in Hand2 conditional null embryos, the fact that some neurons do develop in the absence of Hand2 suggests that acquisition of cell type–specific characteristics will depend upon additional factors.

Our results suggest that Hand2 may couple specification and differentiation of VIPergic neurons to regulation of VIP transcription. This is an intriguing possibility and fits into the established function of Hand2 in transcriptional regulation of DBH (Xu et al., 2003; Rychlik et al., 2003). It may be the case that the sustained role of Hand2 in the nervous system is to regulate the function of genes whose expression requires Hand2 early for neurotransmitter specification. Since it appears that Hand2 does not have a global function in regulating neurogenesis in the ENS, identification of additional regulatory molecules supporting the expression of neurotransmitter/neuropeptide phenotypic characteristics remains an important area of investigation.

Patterning Defects in Hand2 Conditional Null ENS

Our results suggest that one function of Hand2 is associated with patterning of the neuronal plexi in the ENS. This effect could be a result of the decreased neurogenesis we observe, an increase in the development of glial cells, or rather an effect on migration of cells along the developing enteric anlage. Although we have shown an effect on neurogenesis along with an apparent increase in expression of brain-specific fatty acid binding protein (data not shown) suggesting an increase in gliogenesis as a consequence of loss of Hand2 function, we favor a model where Hand2 affects the migration of neural crest–derived cells. Although the mechanisms are just beginning to be elucidated, there is evidence that neuronal differentiation and migration are coupled by the activity of the neurogenic bHLH DNA binding protein, neurogenin 2 (Ngn2), in the central nervous system (Ge et al., 2006). Inasmuch as Hand2 is a neurogenic bHLH DNA binding protein that we have shown affects neurogenesis and patterning of the ENS, it is tempting to speculate that Hand2 couples these developmental events in a manner similar to Ngn2.

Two factors shown to be important for regulating the activity of Ngn2 in neurogenesis and migration are DNA binding and phosphorylation, respectively (Ge et. al., 2006). Interestingly, we have previously shown that DNA binding and phosphorylation regulate different aspects of Hand2 function (Liu et al., 2005). DNA binding by Hand2 is not required for its NE cell type–specific function at the dopamine-β-hydroxylase promoter (Xu et al., 2003) but it is required for neurogenesis (Howard, unpublished data) and proper limb patterning (Firulli et al., 2003; Liu et al., 2005). The profound effect we observe in patterning of the neuronal plexus in the stomach of Hand2 conditional null embryos suggests that similar to Ngn2, Hand2 may influence migration of neuronal precursor cells. While it is possible that the patterning defects we observe result from an overall reduction in the number of enteric neurons, our data suggest a mechanism whereby Hand2 might couple neurogenesis and migration of neural crest–derived cells along the developing wall of the GI tract. We posit that Hand2 affects expression of migration-related genes in neural crest–derived cells as they enter the gut wall as an early aspect of its function in neurogenesis and cell type–specific gene expression. These data support the growing body of literature suggesting that neurogenic bHLH DNA binding proteins influence multiple processes that are reflected in positive and negative regulation of downstream genes in transcription factor networks that are functioning in parallel.


Tissue Culture

Gut-derived neural crest–derived cells were purified from ED5/6 chick embryo gut by immunoselection using the HNK-1 antibody. Virus-free embryonated eggs were purchased from SPAFAS (Design Learned, Inc. Norwich, CT). The gastrointestinal tract was removed in its entirety from 20 embryos per experimental condition. The tissue was minced and incubated in 0.5% collagenase A (Roche Applied Bioscience, Indianapolis, IN), prepared in phosphate buffered saline (PBS), at room temperature for 2 × 15 min. Cells were released by gentle trituration through a fire-polished Pasteur pipette and collected by centrifugation at 160g for 5 min at 25°C. The pellets were resuspended in high glucose DMEM (1 ml) containing 5% 11-day chick embryo extract (CEE) (Howard and Bronner-Fraser, 1985), 1.0% fetal bovine serum (FBS) (Invitrogen, Carlsbad, CA), and HNK-1 (20 μg/ml) antibody (Sigma, St. Louis, MI) at 4°C on a rotating shaker for 1 hr at room temperature. Cells decorated with the HNK-1 antibody were purified, following incubation in rat anti-mouse IgM conjugated to magnetic microbeads (Miltenyi Biotec Inc., Auburn, CA) at 12°C for 2 hr, by passing through a magnetic field (SuperMacsTMII Separator, Miltenyi, Biotec Inc., Auburn, CA). The HNK-1 selected and negative cells were plated on 35-mm tissue culture plates coated with human fibronectin (24 μg/dish) (Chemicon International Inc., Temecula, CA) at a density of 5–10 × 103 cells/dish. Growth medium (DMEM, 10% FBS, and 5% CEE) was changed every other day. The unselected residual cells were plated in a like manner. The efficiency of the immunoselection and purity of the cultures was tested by counting the numbers of cells expressing the HNK-1 antigen at 20 hr post-plating. We routinely found that approximately 80 to 89% of cells expressed the HNK-1 antigen in the immunoselected cultures while less than 1% expressed this antigen in the residual cells. This degree of purity is moderately higher than previously published (Pomeranz et al., 1993; Chalazonitis et al., 1994). The majority of cells lose expression of the HNK-1 antigen by culture day 4.


Retrovirus-mediated gene delivery was used to introduce Hand2 (RCAS-Hand2) into HNK-1+ and residual cells as previously described (Howard et al., 1999, 2000). Briefly, 1× 106 IU of sense RCAS-Hand2 or antisense RCAS-Hand2 was added to cells at the time of plating; an additional 1× 106 IU was added after 12 hr of culture. The virus generated by RCAS-Hand2 is replication-competent so that cells dividing throughout the culture period are susceptible to infection and thus expression of Hand2. Uninfected cells served as controls. Virus titer was determined in DF-1 chicken embryo fibroblasts (CEF), a continuous cell line lacking endogenous expression of ASLV-like sequences. CEF cells were seeded at 50% confluence and infected with serial dilutions of viral stock. The number of infected cells was counted after 48 hr, at which time the cells had reached confluence. Identification of infected cells was based on expression of the viral coat protein p19 (AMV3c2 antibody, Developmental Studies Hybridoma Bank) or p27 (anti-p27, Charles River Laboratories, Franklin, CT). The mouse anti-p19 antibody was used in co-immunolabeling studies with rabbit anti-VIP antibodies. The differentiation of immunoselected and residual cells into neurons was determined on culture day 7 based on expression of neuron-specific beta tubulin (TUJ1). Cell type–specific expression of neurotransmitters/neuropeptides was determined based on immunoreactivity to vasoactive intestinal polypeptide (VIP) or choline acetyltransferase (ChAT) as described below.


To identify neurons (TUJ1; 1:2000, Covance, Richmond, CA; HuC/D 1:30, Molecular Probes, Carlsbad, CA) and cells expressing ChAT (antibody kindly provided by Miles Epstein) or VIP (1:100, RDI, Flanders, NJ), we performed immunocytochemistry according to our established procedures (Howard et al., 1999; Wu and Howard, 2002; Liu et al., 2005). Briefly, tissue culture plates are placed on ice and cells washed in ice-cold PBS for 3 × 5 min. Cells are fixed for 20 min in fresh 4% paraformaldehyde made in PBS. Following washing 4 × 5 min in PBS, cells are permeabilized in blocking buffer containing 0.3% Triton X-100 and 10% horse serum in PBS for 30 min at room temperature. Cells are then incubated in primary antibody diluted in a buffer containing 0.3% Triton X-100 and 4% horse serum overnight at 4°C. Cells are then washed 3 × 10 min in PBS containing 0.3% Triton X-100 to remove unbound antibody and incubated in directly coupled species-specific secondary antibody for 3 hr at room temperature (Kirkegaard and Perry, Gaithersburg, MD) to visualize sites of antibody binding. Mouse tissue sections and whole-mount preparations of gut were immunostained using the following modifications. The blocking step was increased to an overnight incubation and primary antibody was applied in a buffer containing 0.1M Tris, pH.7.5, 0.3% Triton X-100, and 1.5% NaCl. Mouse tissues were washed in the same buffer and incubated in donkey anti-rabbit IgG coupled to Cy3 (Jackson ImmunoResearch, West Grove, PA) or Goat anti-mouse IgG conjugated to FITC (Kirkegaard and Perry) for VIP IR and TUJ1 IR, respectively. Anti-HuC/D was directly coupled to Alexa 488. For avian tissue, cells were washed in PBS buffer containing 0.3% Triton X-100 and 4% horse serum and incubated in goat anti-rabbit IgG conjugated to TRITC (Kirkegaard and Perry) for VIP IR and anti-rabbit IgG conjugated to FITC (Kirkegaard and PerryMD) for ChAT IR and TUJ1 IR.

Visualization of enteric ganglia in whole-mount preparations of stomach was achieved using acetylcholinesterase histochemistry. The stomach was fixed overnight in fresh paraformaldehyde made in PBS, extensively washed in PBS, and then rinsed in water 3 × 10 min. The tissue was then incubated for 2 hr at 37°C with shaking in a solution (25 ml) containing 6.25 mg acetylthiocholine idodide, 0.19N NaOAc, 0.01N glacial acetic acid, 25 mM Na citrate, 15 mM CuSO4, 2.5 mM K3Fe(CN)6, and 0.125 ml OMPA solution (6.84 mg OMPA/2 ml water). To stop the reaction, the tissue was rinsed 3 × 10 min in water and dehydrated. The tissue was then cleared in methyl salicylate at room temperature for 1 hr, incubated in 100% ETOH, cleared in xylenes, and mounted. Tissue was photographed using an Olympus AX70 microscope fitted with a Spot camera.

Confocal Microscopy

Images were acquired using a Leica TCS SP5 broadband confocal microscope system coupled to a DMI 6000CS inverted microscope equipped with multiple continuous wave lasers and a Chameleon XR tunable pulsed IR laser. Confocal z-series were obtained using either a 10× objective (n.a. = 0.4) or 40× immersion oil objective (n.a. = 1.25). FITC and CY3 fluorescence was excited using 488 nm or 561 nm laser lines, respectively. Acquisition of emitted light was optimized using a tunable SP detector.

Gene Targeting

We used our Hand2 bacterial artificial chromosome (BAC) clone to generate the Hand2ploxp targeting vector. A 12.8-kb fragment extending from −5.2 (SpeI-fill in restriction site) to +7.6 kb (Xho1-fill in restriction site) was digested with BglII and Kpn1 to yield a 2.6-kb fragment (−824 to +1802), which was ligated into pBS246 to introduce loxP sites flanking exon 1. A 4.4 kb 5′ UTR (Xba1 to EcoRV) fragment was cloned upstream of an exon 1 containing fragment to generate a 9.4-kb upstream targeting arm. To generate the 3′ targeting arm, a 5.8-kb Kpn1 to Xho1 (+1802 to +7623) fragment was ligated into a pPNTlox2 targeting vector (pPNTHAND2UP). To generate the final targeting construct, a Spe1 to Kpn1 upstream arm was ligated into pPNTHAND2UP 5′ to pGK-NEO to introduce the third loxP site. We confirmed reliability of the targeting vector by restriction analysis and DNA sequencing. The complete HAND2NEO-loxP (pCKO) targeting vector was linearized with SalI, electroporated into 129/SVIMJ ES cells, and screened by G418 and FIAU. We screened 282 resistant colonies by Southern blotting of RcaI and VspI digested genomic DNA using probes from the 3′ and 5′ flanking region, respectively, as indicted (Fig. 4). Correct targeting was confirmed by Southern analysis of VspI digested genomic DNA using a PCR amplified 5′ probe ranging from −5.8 to 5.2 kb, and from −6.8 to 5.2. Two correct clones (clones 245 and 246) were expanded and injected into C57BL/6 blastocysts and implanted into pseudopregnent mice. Twelve chimeric males were identified. Seven males with germline transmission have been confirmed. The neomycin resistance cassette was removed in the male germline by crossing heterozygous Hand2ploxpneop mice with Sycp1-Cre mice (provided by Douglas Pittman); excision of the neomycin resistance cassette was confirmed by sequence analysis. Mice homozygous for the Hand2ploxp allele are fertile and viable indicating that loxP sites in the 5′UTR do not adversely affect expression from the targeted locus. Animal care, breeding, and experimental protocols were approved by the Medical University of Ohio Animal Care and Use Committee (IACUC) prior to initiating this work.

In Situ Hybridization

To identify cells expressing mRNA encoding Hand2, non-radioactive in situ hybridization was performed, according to our published procedures (Wu and Howard, 2002; Ruest et al., 2004). Briefly, E14 Hand2plox/+ and Hand2loxp/loxp;Wnt1-Cre embryos were fixed in 4% paraformaldehyde, washed in PBS, embedded in OCT compound, and sectioned at 20 μm. Tissue sections were hybridized at 65°C with a digoxigenin-labeled Hand2 riboprobe. Sites of hybridization were localized following color development using NBT/BCIP or BM Purple AP (Roche Diagnostics Corp., Indianapolis, IN), mounted and photographed.

Transient Transfection

P19 cells were maintained in alpha minimal essential medium supplemented with 2.5% fetal bovine serum and 7.5% bovine calf serum (Invitrogen Life Technologies, Carlsbad, CA). One day prior to transfection, cells were plated in 12-well dishes at a density of 1 × 105 cells per well. Transfection assays were set with VIP5.2-luciferase reporter plasmid (generously provided by Lee Eiden) using FuGENE6 (Roche Diagnostics Corp., Indianapolis, IN) according to our previously published procedures (Xu et al., 2003). A total of 0.5 μg/well DNA was used; 150 ng/well of pcDNA3.1-Hand2 expression plasmid, 192 ng/well of VIP5.2-luc, and the balance was empty pcDNA3.1 vector DNA (Invitrogen Life Technologies, Carlsbad, CA). Cells were harvested after 48 hr and luciferase assays performed on cell lysates using a Dual Luciferase assay according to the manufacture's instructions (Promega Corp., Madison, WI). Each well included pRL-tk (8 ng/well) as an internal control for transfection efficiency. The luciferase assay results were transformed and are reported as values normalized to 1 according to the ratio of Firefly to Renilla reporter activity in control transfections. Data are presented as the mean ± S.E.M. of normalized values.


Data are presented as the mean ± S.E.M unless otherwise stated. Statistical significance was determined using an ANOVA and Bonferroni post hoc test unless otherwise stated.


The authors thank Janet Lambert for expert technical assistance. Interesting discussions with Dr. Michael Gershon and Dr. Heather Young are greatly appreciated. This work was supported by the National Institutes of Health: NIDDK067064, NS040644 (to M.J.H.), DE14181, DE14765 (to D.E.C.) and the American Heart Association 0555176B (to D.E.C.). All confocal images were acquired using the resources of the Kaptur-Rogowski Multiphoton Suite located in the Advanced Microscopy and Imaging Center at the Medical University of Ohio.