Genetic screen for mutations affecting development and function of the enteric nervous system

Authors


Abstract

An intact enteric nervous system is required for normal gastrointestinal tract function. Several human conditions result from decreased innervation by enteric neurons; however, the genetic basis of enteric nervous system development and function is incompletely understood. In an effort to increase our understanding of the mechanisms underlying enteric nervous system development, we screened mutagenized zebrafish for changes in the number or distribution of enteric neurons. We also established a motility assay and rescreened mutants to learn whether enteric neuron number is correlated with gastrointestinal motility in zebrafish. We describe mutations isolated in our screen that affect enteric neurons specifically, as well as mutations that affect other neural crest derivatives or have pleiotropic effects. We show a correlation between the severity of enteric neuron loss and gastrointestinal motility defects. This screen provides biological tools that serve as the basis for future mechanistic studies. Developmental Dynamics 236:118–127, 2007. © 2006 Wiley-Liss, Inc.

INTRODUCTION

The enteric nervous system (ENS) is composed of gastrointestinal (GI) tract resident sensory neurons, interneurons, and motoneurons that direct GI movement by controlling smooth muscle, as well as affecting local blood vessels and mucosal secretion. In mammals and avians, ENS neurons typically form enteric ganglia that are distributed along the entire length of the GI tract. Although the ENS has intrinsic reflex activity, it is also influenced by inputs from the central nervous system (CNS; for review, see Wood et al.,1999; Costa et al.,2000; Bornstein et al.,2004; Furness,2006). The importance of the ENS to human health is highlighted by several syndromes affecting the number and distribution of enteric ganglia (Newgreen and Young,2002a,b). One of these conditions, known as Hirschsprung's disease is characterized by megacolon resulting from a decrease or complete absence of enteric ganglia in the distal colon and associated lack of GI motility (Angrist et al.,1996; Robertson et al.,1997; Stewart and von Allmen,2003; Gershon and Ratcliffe,2004). Another condition, known as Waardenburg-Shah syndrome is also characterized by megacolon and decreased GI tract motility, but in addition has accompanying pigmentation defects (Shah et al.,1981; Ambani,1983; Baynash et al.,1994; Southard-Smith et al.,1998).

The ENS is derived from neural crest and is considered part of the peripheral nervous system (PNS). Neural crest is generated by cells located at the confluence of neural and non-neural ectoderm (Le Douarin and Kalcheim,1999) that segregate from the neural tube and migrate throughout the body, generating a diverse set of derivatives, including PNS neurons and glia, chromatophores, and craniofacial cartilages (Hall,1999; Le Douarin and Kalcheim,1999; Le Douarin and Dupin,2003; Le Douarin et al.,2004). There have been several recent reviews describing the molecular mechanisms underlying neural crest induction, migration, and cell fate specification (Huang and Saint-Jeannet,2004; Le Douarin et al.,2004; Meulemans and Bronner-Fraser,2004; Young et al.,2004; Howard,2005; Taneyhill and Bronner-Fraser,2005), as well as several reviews of the molecular mechanisms underlying formation of the ENS (Young et al.,2000; Newgreen and Young,2002a,b; Hagl et al.,2003; Gershon and Ratcliffe,2004; Burns,2005; Grundy and Schemann,2005).

To date, only a handful of genes encoding proteins that affect mammalian ENS development have been identified (Newgreen and Young,2002a,b; Gershon and Ratcliffe,2004). Interestingly, mutations in some of these genes are pleiotropic, affecting not only ENS neurons, but a variety of other neural crest derivatives as well (Yanagisawa and Masaki,1989; Guillemot et al.,1993; Baynash et al.,1994; Southard-Smith et al.,1998). Genes affecting ENS development in mammals have been uncovered in several different ways. Some of them, for example Piebald lethal (Endothelin-B receptor) and Lethal spotting (Endothelin 3), were first uncovered as spontaneous mouse mutations (Dunn,1920; Phillips,1959; Lane,1966) and later confirmed in gene targeting experiments in mouse (Yanagisawa and Masaki,1989; Baynash et al.,1994; Hosoda et al.,1994). Other genes, for example Mash1, were uncovered in mouse studies in which particular genes were specifically targeted for disruption (Guillemot et al.,1993). More recently, several genes and proteins expressed in the ENS, and thus potentially important in ENS development, have been uncovered in gene and protein profiling studies (Iwashita et al.,2003; Marvin-Guy et al.,2005; Heanue and Pachnis,2006).

Although some genes have been linked to human disease pathologies, they only account for a proportion of the human syndromes that affect GI motility (Newgreen and Young,2002a,b; Gershon and Ratcliffe,2004). Moreover, because only a fraction of the genes involved in GI tract development and function have been isolated, we still do not have a comprehensive picture of the molecular mechanisms involved in ENS specification, migration of ENS precursors into the GI tract and along the intestinal epithelium, and establishment of a cohesive neuronal network that coordinates GI movement (Newgreen and Young,2002a,b; Gershon and Ratcliffe,2004). Our lack of a mechanistic understanding of all these processes provides an important motivation for uncovering additional genes involved in GI tract development and function.

Zebrafish has many advantages as a model system for studying GI tract and ENS development. The general features of GI tract development and function have been described and are similar to those of mammalian models (Wallace and Pack,2003; Wallace et al.,2005; Ng et al.,2005). Several genes involved in zebrafish ENS development have been identified by homology with genes involved in mammalian ENS development, for example phox2b, genes encoding glial cell line-derived neurotrophic factor (GDNF) family ligands, receptors, and coreceptors, and nos1 (Bisgrove et al.,1997; Shepherd et al.,2003; Elworthy et al.,2005; Pietsch et al.,2006). sox10, a gene implicated in Hirschprung's disease and Waardenburg-Shah syndrome, was initially isolated from a genetic screen for zebrafish mutants lacking pigmentation, and subsequently discovered to have many additional phenotypes, including absence of enteric neurons (Kelsh and Eisen,2000; Dutton et al.,2001). Given the morphological and molecular similarities between the zebrafish and mammalian GI tracts (Wallace and Pack,2003; Wallace et al.,2005; Ng et al.,2005), the most compelling reason for using zebrafish to study ENS development and function is the ability to carry out forward genetic screens. Such screens are still the best way to find new genes involved in a process of interest (Grunwald and Eisen,2002). That ENS development is amenable to forward genetics has been beautifully demonstrated by the recent identification of trap100 as a new gene that affects enteric neuron development (Pietsch et al.,2006).

We undertook a similar forward genetic approach in zebrafish to uncover additional genes important in development and function of the enteric nervous system. Our strategy was based on phenotypes of known mutations that affect enteric neurons. Most of these mutations alter the number and/or distribution of enteric neurons within the GI tract (Lane and Liu,1984; Yanagisawa and Masaki,1989; Guillemot et al.,1993; Baynash et al.,1994; Southard-Smith et al.,1998; Kelsh and Eisen,2000; Dutton et al.,2001; Garcia-Barcelo et al.,2003; Hagl et al.,2003). Thus, similar to the screen performed by Pietsch and colleagues (Pietsch et al.,2006), we screened fixed zebrafish larvae for mutations that affect the number or distribution of enteric neurons using an antibody to HuC/D that recognizes these cells (Marusich et al.,1994; Henion et al.,1996).

Because known mutations affecting enteric neurons in humans and other mammalian models have motility defects, we characterized the development of GI motility in living zebrafish larvae and established an assay to determine whether an altered number or distribution of enteric neurons resulted in changes in GI motility. Here, we present an overview of our screen, a brief description of the phenotypes we isolated, and an initial characterization of the relationship between the number and distribution of enteric neurons and GI motility.

RESULTS

Classification of Mutations

We screened gynogenetic diploid progeny of 746 generation 1 (G1) females and identified 13 mutations that fell into two main classes (Table 1): (1) mutations that were specific to enteric neurons and (2) mutations that were pleiotropic, affecting both enteric neurons and other cell types. One of these pleiotropic mutations, b1037, apparently affects only a subset of neural crest derivatives. In contrast, all of the other mutations in this class affected non–neural crest-derived cell types as well as enteric neurons. Based on their phenotypic differences, these mutations may represent 13 different genes. However, future complementation analyses will be necessary to confirm whether this is the case.

Table 1. Mutations Affecting Enteric Neurons
ClassAllele gene name (symbol)Enteric neuron phenotypeOther phenotypesViability
  1. N/A, data not available; dpf, days postfertilization; DRG, dorsal root ganglia.

Enteric neuron    
 b1006Fewer Embryonic lethal
 b1088 gutwrencher (gwr)Fewer N/A
 b1112Fewer N/A
 b1115Fewer N/A
Pleiotropic    
 b871 gutless wonder (glw)FewerGut lumen formation delayed, narrower neural tube, somites chevron shaped but muscle appears disorganized, smaller head, ectopic melanophoresLethal 5 dpf
 b875Fewer (anteriorly restricted)Delayed melanophore migration, edema 4 dpfLethal 7 dpf
 b896FewerNarrower neural tube, smaller head, DRG mislocalizedEmbryonic lethal
 b953FewerNarrower neural tube, jaw malformedEmbryonic lethal
 b984Fewer (anteriorly restricted)Rounded somites, smaller head, spindly melanophoresEmbryonic lethal
 b1035FewerSmaller head, gut appears flattenedEmbryonic lethal
 b1037FewerFewer melanophoresEmbryonic lethal
 b1039Fewer (anteriorly restricted)Shortened fins, DRG mislocalized, edema 4dpfEmbryonic lethal
 b1074FewerEnlarged head, shortened embryosEmbryonic lethal

All 13 mutations have a decrease in the number of enteric neurons in the mid- and posterior intestine, as defined by Ng et al. (2005; see also Fig. 1C). Of interest, we did not isolate any mutations that completely lacked enteric neurons in both of these regions. We did not screen for mutations affecting enteric neurons in the intestinal bulb or the esophagus. Within the class of mutations specific to enteric neurons (Table 1), all four had fewer enteric neurons than in wild-types, but these were distributed along the length of the mid- and posterior intestine. The pleiotropic mutants (Table 1) fell into two subclasses. Six mutations have fewer enteric neurons, but these neurons are distributed along the entire length of the mid- and posterior intestine. In contrast, three mutations have fewer enteric neurons, but these are restricted to the anterior mid-intestine; the posterior intestine entirely lacks enteric neurons.

Figure 1.

gutwrencher and gutless wonder mutants have fewer enteric neurons than wild-types. A: Wild-type 4 days postfertilization (dpf) zebrafish embryo. Boxed area illustrates mid- and posterior intestine. B,D: Enteric neurons labeled with anti-HuC/D antibody. B: Enteric neurons (arrowheads) in the mid-intestine of a 4 dpf wild-type larva and dorsal root ganglia (DRGs; some marked with asterisks) show HuC/D immunoreactivity. C: Differential interference contrast (DIC) image of B showing the location of the mid- and posterior intestine (after Ng et al.,2005). D:gutwrencher mutant larva showing an overall decrease in the number of enteric neurons; the DRGs appear normal. E: DIC image of D. F–K: Four dpf larvae in transverse section. F–H: Sections of a wild-type larva. F: DIC image showing that ventral stripe melanophores are restricted to the region ventral of the somites. G: Fluorescent image showing HuC/D immunoreactivity of enteric neurons. H: Merged images of F and G. I–K: gutless wonder mutant larva. I: DIC image showing ventral stripe melanophores ectopically located within the somites. J: Fluorescent image showing HuC/D immunoreactivity; these mutants have fewer enteric neurons than wild-types. K: Merged images of I and J.

gutwrencher (gwr) represents an example of a gene that appears to affect only enteric neurons (Fig. 1). gutwrencher mutants have fewer enteric neurons than wild-types, but the neurons that are present populate the entire mid- and posterior intestine. To learn the extent of the differences in the number of enteric neurons between wild-types and gutwrencher mutants, we labeled fixed larvae with an antibody to HuC/D, sectioned the stained larvae, and counted HuC/D-positive enteric neurons in every other section of the mid-intestine. We found that gwr mutants had 1.77 ± 0.89 enteric neurons per two sections (n = 15 larvae) compared with 6.46 ± 0.67 enteric neurons per two sections in wild-type larvae (n = 12, standard t-test, P = 7.56 × 10−14; see also Fig. 3D). To learn whether mutations in the gwr gene affect all enteric neurons or only specific subsets, we stained fixed larvae with an antibody to serotonin (5HT; see Pietsch et al.,2006), sectioned them, and counted 5HT-positive enteric neurons in the mid-intestine, using the same counting protocol as for HuC/D-positive enteric neurons. Wild-type larvae had 1.09 ± 0.18 5HT-positive enteric neurons per two sections (n = 12 larvae), whereas gwr mutant larvae had 0.16 ± 0.09 5HT-positive enteric neurons per two sections (n = 4 larvae, P = 2.06 × 10−6). This represents a 3.5-fold decrease in the number of HuC/D-positive enteric neurons, but a 6.2-fold decrease in the number of 5HT-positive enteric neurons, suggesting that the gutwrencher mutation has a more severe effect on 5HT-positive enteric neurons in the posterior intestine than on posterior intestine enteric neurons in general.

Figure 3.

gutwrencher mutants have fewer and colourless mutants lack enteric neurons. A: Enteric neurons in the mid-intestine of a 5.5 days postfertilization (dpf) wild-type larva revealed by HuC/D staining. B: The mid-intestine of a 5.5 dpf gutwrencher mutant showing an overall decrease in enteric neurons. C: A 5.5 dpf colourless (sox10) mutant larva showing complete absence of enteric neurons; the out of focus green spot is tissue autofluorescence iridophore. The larvae shown here were used to make the supplemental movies and still frames in Figure 4, and then fixed and processed for HuC/D immunoreactivity. D: Graphical representation of the number of HuC/D-positive neurons in wild-type, gwr mutant, and b875 mutant larvae. Each dot represents a single larva, and the horizontal lines represent the average for each genotype. HuC/D-positive neurons were counted in every other section, as described in the Experimental Procedures section. Like b875 mutants, b984 and b1039 mutants lack enteric neurons in the posterior intestine (see Table 1). We did not count HuC/D-positive enteric neurons in sections of b984 and b1039 mutants; however, visual inspection of whole-mount HuC/D staining revealed that enteric neurons were essentially absent.

Figure 4.

Gastrointestinal (GI) motility correlates with the number of enteric neurons. Still frames taken from movies of 5.5 days postfertilization (dpf) wild-type, colourless mutant, and gutwrencher mutant larvae. Arrowheads denote sites of contraction. A–C: Rostral-to-caudal contraction waves in wild-type; the arrowhead moves caudally with the wavefront (Supplementary Movie 6). D–F:cls mutants lack obvious rostral-to-caudal contraction waves (note that the arrowhead marks the same spot in each frame; Supplementary Movie 5) but have small spontaneous focal contractions similar to 3.5 dpf wild-types (see Fig. 2A–C and Supplementary Movie 1). G–I: In gwr mutants, there is a lack of coordinated wave progression and many regions of the gut contract simultaneously (Supplementary Movie 7). J: Schematic showing the correlation between the number of enteric neurons and GI motility. Hatched arrows show spontaneous, focal contractions. Long arrows show directions of contraction waves. Green dots represent enteric neurons.

b875 represents an example of a gene that affects both enteric neurons and at least one other neural crest derivative, melanophores. In contrast to gwr, b875 mutants have enteric neurons in the anterior mid-intestine, but lack enteric neurons in the posterior mid- and posterior intestine. To learn the extent of the differences in the number of enteric neurons between wild-types and b875 mutants, we labeled fixed larvae with HuC/D antibody, sectioned the stained larvae, and counted HuC/D-positive enteric neurons in every other section of the mid-intestine. We found that b875 mutants had 0.12 ± 0.18 enteric neurons per two sections (n = 22 larvae, standard t-test used to compare these mutants with wild-types as described above, P = 8.32 × 10−12; see also Fig. 3D). As for gwr mutants, we asked whether all enteric neurons or only specific subsets were affected in b875 mutants by counting 5HT-positive enteric neurons in the mid-intestine. b875 mutants had 0.00 ± 0.00 5HT-positive enteric neurons (n = 8 larvae, P = 2.64 × 10−9 compared with wild-types as above). Thus, as in gwr mutants, b875 mutants have a more significant decrease in 5HT-positive enteric neurons than in HuC/D-positive neurons.

gutless wonder (glw) represents an example of a gene that affects both enteric neurons and melanophores (Fig. 1). gutless wonder mutants have enteric neurons distributed along the mid- and posterior intestine. However, there appears to be a progressive rostral-to-caudal decrease in the number of enteric neurons, unlike wild-types, in which the number of enteric neurons appears to be more uniform along the mid- and posterior intestine. gutless wonder mutants also have ectopically located melanophores (Fig. 1F,I). In wild-type larvae, ventral pigment stripe melanophores form a coherent stripe just dorsal of the intestine. In contrast, in gutless wonder mutants, this stripe is much less coherent and some melanophores are situated much more dorsally than in wild-types. The melanophore phenotype varies along the rostrocaudal axis of gutless wonder mutants. Thus, at some rostrocaudal levels of the intestine, the ventral pigment stripe of gutless wonder mutants appears wild-type, whereas at other levels it appears mutant. However, this variability is not progressive from mid- to posterior intestine; thus, the regions of ectopic melanophores are not correlated with the decrease in enteric neurons (data not shown).

GI Motility Is Correlated With the Number of Enteric Neurons

Enteric neurons direct and coordinate contractile movements of the GI tract smooth muscle; thus, a decrease in the number of enteric neurons is associated with decreased motility in humans and in mammalian models. To determine whether alterations in the number of enteric neurons resulted in changes in motility of the mid-intestinal tract in zebrafish, we first characterized the development of mid-intestine motility in living wild-type larvae (see also Holmberg et al.,2003; Bates et al.,2006). We used real-time video microscopy to watch the movements of a three-somite-length segment of the mid-intestine in larvae at 3.5, 4.5, 5.5, and 6.5 days postfertilization (dpf; Fig. 2). We then rescreened some of the enteric neuron mutations using this motility assay at either 5.5 or 6.5 dpf to learn whether decreasing the number of enteric neurons altered GI motility. Some of our mutations are embryonic lethal (Table 1); thus, we were unable to assess whether the decreased number of enteric neurons altered the motility of their mid-intestines because these mutants did not survive to an appropriate stage to assay.

Figure 2.

Development of motility in wild-type larvae. Still frames taken from movies of live wild-type larval gastrointestinal (GI) tracts (see Supplementary Movies 1–4). A–C: Focal contractions (asterisks) in 3.5 days postfertilization (dpf) larva at 0, 4, and 8 sec to illustrate random locations. D–F: Caudal-to-rostral contraction wave of intestinal bulb of 4.5 dpf larva. The long arrow shows the direction of contraction; the short arrow shows the advancing wavefront. G–I: Caudal-to-rostral contraction wave originating at the anus of a 4.5 dpf larva. The long arrow shows the direction of contraction; short arrows in G and H show advancing wavefronts; the short arrow in I shows the beginning of a new contraction wave. J–L: Rostral-to-caudal contraction wave in mid-intestine of 5.5 dpf larva. The long arrow shows the direction of the contraction; short arrows show advancing wavefronts. M: Schematic representations showing locations of each set of still figures and the movies they were taken from. N: Schematic representation of GI motility development. Red dots represent origins of contractions. Arrows show directions of contraction waves; dashed arrows represent focal contractions without apparent directionality. ib, intestinal bulb; a, anus.

Development of GI Motility.

At 3.5 dpf the mid-intestine of wild-type larvae exhibited spontaneous, focal contractions along its entire length (Fig. 2A–C; Supplementary Movie 1, which is available at http://www.interscience.wiley.com/jpages/1058-8388/suppmat). One day later, at 4.5 dpf, in addition to these focal contractions, the larval mid-intestine had contractile waves that traveled either rostral-to-caudal or caudal-to-rostral, depending on where along the intestine the wave was initiated. Waves zthat appeared to originate at the caudal end of the intestinal bulb traveled in both directions, caudal-to-rostral through the intestinal bulb (Fig. 2D–F; Supplementary Movie 2) or rostral-to-caudal toward the anus (data not shown). The caudal-to-rostral waves extended at least to the rostral end of the intestinal bulb; we did not investigate whether they went further rostrally. Waves that appeared to originate at the anus traveled caudal-to-rostral, and only extended three to four somite lengths (Fig. 2G–I; Supplementary Movie 3).

By 5.5–6.5 dpf, the rostral-to-caudal waves that originated at the caudal end of the intestinal bulb appeared more coordinated and moved with regular periodicity along the entire length of the mid-intestine (Fig. 2J–L; Supplementary Movie 4). These waves appeared peristaltic, with alternating areas of intestinal wall relaxation and contraction (see also Holmberg et al.,2003; Bates et al.,2006), and had a regular periodicity (one wave every 42 ± 2 sec; n = 37 larvae). Because the regular waves of motility in the mid-intestine are easily visualized between 5.5 and 6.5 dpf, we rescreened mutations by observing GI motility in this region.

Motility Changes in Enteric Neuron Mutants.

To understand whether the number or distribution of enteric neurons correlates with GI motility, we initially examined colourless (cls; sox10) mutants that completely lack enteric neurons (Kelsh and Eisen,2000; Dutton et al.,2001; Fig. 3). We reasoned that, if enteric neurons are required for GI motility in zebrafish, GI motility should be absent from cls mutants. We found that, at 5.5 dpf, cls mutants have spontaneous, small, focal contractions (Fig. 4; Supplementary Movie 5) in the mid-intestine, similar to those seen in 3.5 dpf wild-types (Fig. 2A–C; Supplementary Movie 1). However, cls mutants lack the coordinated rostral-to-caudal GI tract motility waves exhibited by 5.5 dpf wild-types (Fig. 4; Supplementary Movie 6). This finding suggests that enteric neurons are required for the coordination of the rostral-to-caudal contractile waves that travel along the intestine.

Having characterized GI motility in cls mutants that entirely lack enteric neurons, we next investigated whether mutants that had some enteric neurons also had motility defects. To address this question, we examined larvae with mutations in the gwr gene. These larvae have fewer enteric neurons than wild-types, but these neurons are distributed along the entire length of the GI tract (Figs. 1D,E, 3). We found that, at 5.5 dpf, gwr mutants have rostral-to-caudal waves of contractions that originate at the caudal end of the intestinal bulb (Fig. 4; Supplementary Movie 7). However, several discrete regions of the gut contract simultaneously in gwr mutants; thus, the contraction waves are less coordinated than those of wild-types. To ensure that the motility changes we observed were correlated with the mutant phenotype, after imaging, all larvae were processed for HuC/D immunohistochemistry (five mutant larvae, seven wild-type larvae; Fig. 3). Taken together, these results provide evidence of a correlation between the number of enteric neurons and the coordination of contractile waves that travel rostral-to-caudal along the intestine of zebrafish larvae and, thus, suggest that coordinated contractile waves require a minimum number of enteric neurons.

DISCUSSION

We used a simple screen that allowed us to identify mutations affecting the number or distribution of enteric neurons in the mid- and posterior intestine, as well as linking the number of enteric neurons to GI tract function. Some of the mutations we isolated, for example gutless wonder, affect both enteric neurons and other neural crest derivatives. The genes defined by these mutations are likely to act early in neural crest development. Alternatively, they may act non–cell-autonomously in other cell types that interact with a variety of neural crest derivatives, as has been described for the trap100 gene that is required in mesendoderm for proper development of several neural crest derivatives (Pietsch et al.,2006). gutless wonder mutants appear to have a defect or delay in mid-intestinal morphogenesis. One intriguing possibility is that, like trap100, gutless wonder is also required in mesendoderm to regulate ENS development; this hypothesis can be tested in future experiments. We also isolated mutations that appear to affect enteric neurons specifically, for example, gutwrencher. The genes defined by these mutations are likely to act later in neural crest development, and some of them may act specifically within enteric neurons.

Our mutations fall into two classes in the severity of enteric neuron loss. One class, for example b875, has phenotypes resembling Hirschsprung's disease and Waardenburg-Shah syndrome, with regions of the mid- and posterior intestine devoid of enteric neurons (Shah et al.,1981; Ambani,1983; Baynash et al.,1994; Angrist et al.,1996; Robertson et al.,1997; Southard-Smith et al.,1998; Stewart and von Allmen,2003; Gershon and Ratcliffe,2004). The other class, for example gutwrencher, has neurons along the entire length of the GI tract, but there are considerably fewer neurons than normal. We have begun to characterize the subtypes of enteric neurons affected by the gutwrencher and b875 mutations. Of interest, in both cases, the number of 5HT-positive neurons was decreased much more significantly than the number of HuC/D-positive neurons. One possible interpretation is that both mutations have a primary effect on this enteric neuron subset. However, this possibility seems unlikely, given the differences in the phenotypes of the two mutations: gwr mutants have a decrease in the number of enteric neurons along the entire length of the mid- and posterior intestine, whereas b875 mutants have enteric neurons in the anterior of this same region, but essentially lack enteric neurons posteriorly. Other possibilities are that, in the enteric nervous system, a decrease in the total number of enteric neurons simply leads to a decrease in all subtypes, or that development of 5HT immunoreactivity is particularly sensitive to a variety of perturbations. Future characterization of the gwr and b875 genes should reveal where they function, and how these functions are required during the development of 5HT-positive enteric neurons.

Our mutations should help to elucidate genes involved in human GI disorders. For example, mutations such as b875 may serve as new models to understand the etiology of Hirschsprung's disease and Waardenburg-Shah syndrome, and mutations such as gutwrencher may serve as models for understanding other types of GI disorders in which there is no obvious aganglionosis. It is clear that further characterization of all of our mutations, including mapping and cloning of the genes identified by the mutations as well as further characterization of the subsets of enteric neurons they affect, will be required to understand how they fit into what is currently known about enteric nervous system development and function.

Our studies provide evidence that, in zebrafish, as in other vertebrates, enteric neurons are required for normal GI motility (Olsson and Holmgren,2001; Furness,2006). Several factors need to be considered to understand the link between enteric neuron number and GI motility in more detail. First, nothing is currently known about enteric glia in zebrafish. Glia are thought to contribute to ENS function (Bush et al.,1998; Cornet et al.,2001; Cabarrocas et al.,2003; Savidge et al.,2004; Ruhl,2005); therefore, it will be important to learn whether our mutants also lack enteric glia. Second, the circuitry of the zebrafish enteric nervous system has not been mapped out, and the contributions of spontaneous smooth muscle contractions and interstitial cells of Cajal (ICC) to GI motility have not been studied. ICCs are a mesodermally derived pacemaker population that regulates some aspects of GI motility (Cajal,1911; Lecoin et al.,1996; Young et al.,1996; Ward and Sanders,2001). Thus, it will be important to identify these cells in zebrafish to understand how they contribute to the motility we have observed. One intriguing possibility is that they are responsible for the small, focal contractions seen early in development and in sox10 mutants, which essentially entirely lack enteric neurons. This possibility can be tested in future experiments; however, it will require identification and characterization of zebrafish ICCs. Third, although we and others (Holmberg et al.,2003,2004; Bates et al.,2006) have begun to characterize motility of the zebrafish GI tract, thus far all of the assays have been conducted on fasting larvae. Most of our understanding of GI tract motility comes from mammalian models in which the GI tract is dissected out and observations and physiological recordings made after the introduction of a bolus of material (Smith et al.,2003). It will be important to establish such assays in zebrafish to correlate what we can learn from our mutants with what has been established using other models.

EXPERIMENTAL PROCEDURES

Generation of Mutations

N-ethyl-N-nitrosourea (ENU) treatment was used to induce point mutations in G0 wild-type males of the *AB or Tübingen strains following the procedures outlined in Henion et al. (1996), with the following modifications: ENU concentration was standardized spectraphotometrically and males were lightly anesthetized in tricaine methane sulfonate (MS222) during exposure to ENU and for 1–2 hr after removal of ENU (Trevarrow,2005). Mutagenized males were outcrossed to wild-type *AB or Tübingen females to produce a G1 generation. G1 females were screened by examining their G2 gynogenetic half-tetrad diploid progeny produced by the “early pressure” (EP) method in which eggs were activated with ultraviolet-irradiated sperm and pressure applied to prevent the second meiotic division (Streisinger et al.,1986; Grunwald and Streisinger,1992; Henion et al.,1996; Beattie et al.,1999). Gynogenetic diploid G2 larvae were screened morphologically and by immunohistochemistry. G1 females found to carry potentially interesting mutations were outcrossed to males of the same genetic background (*AB or Tübingen). Mutations were recovered by pair-wise incrosses of the G2 generation; all mutations we report here appear to be recessive. Embryos and larvae were raised at 28.5°C according to standard zebrafish husbandry (Westerfield,2000) and staged by hours postfertilization at 28.5°C (hpf) or days postfertilization at 28.5°C (dpf).

Immunohistochemistry

Embryos and larvae were raised at 28.5°C for 4 days. At 2, 3, and 4 dpf, embryos and larvae were examined for morphological changes in chromatophore number, pattern, and morphology before being processed for whole-mount immunohistochemistry with the anti HuC/D monoclonal antibody 16A11 (Marusich et al.,1994). Whole-mount immunohistochemistry was performed as follows: larvae were anesthetized in 100 mg/L MS222 (Sigma) before being fixed for 2 hr at room temperature in 1% formaldehyde. After washing several times in phosphate buffered saline (PBS), larvae were blocked in 1% dimethyl sulfoxide, 0.5% saponin, 1% bovine serum albumin, 2% goat serum in PBS. After blocking, larvae were incubated overnight at 4°C in the same blocking solution with the addition of HuC/D antibody. After washing with PBS, larvae were incubated with anti-mouse antibodies conjugated to Alexa488 or Alexa568 (Invitrogen) for 4 hr at room temperature. This modified protocol was used to minimize HuC/D labeling in the spinal cord, enabling visualization of dorsal root ganglia. Larvae were screened using a fluorescence stereomicroscope.

Enteric Neuron Counts

We counted enteric neurons in two representative mutants, b1088 (gutwrencher) which has fewer enteric neurons distributed along the length of the posterior intestine and b875 which lacks nearly all enteric neurons in the posterior intestine, and compared these counts with that of wild-types. In these experiments, wild-type larvae were generated from crosses of wild-type adults; we did not examine wild-type siblings of the mutants. Larvae were stained by whole-mount immunohistochemistry, embedded in agar:agarose:sucrose (Jensen et al.,2001), and sectioned transversely at 12–16 μm, and the number of labeled neurons was counted. Labeled enteric neurons were counted in 4 dpf larvae along a five-somite length in the mid-intestine. To ensure that individual neurons were not counted twice, neurons were counted in alternate sections; thus, the number of enteric neurons is always presented as the number per two sections. Therefore, the counts obtained in this study may be an underrepresentation of actual numbers.

Motility Assay

Larvae were anesthetized with MS222 [100 mg/L in embryo medium (EM)] and mounted left-side facing up in 5% methylcellulose on a microslide. Live images were captured on a Y/C video camera mounted on a Zeiss Axioplan microscope equipped with differential interference contrast optics and a 20× objective. Images were digitized with a Canopus analog-to-digital converter and data captured, stored, and analyzed on a MacIntosh computer through imovie and Adobe premier software. Movies characterizing movement of wild-type GI tracts were taken at 3.5, 4.5, 5.5, or 6.5 dpf at various rostrocaudal positions along the intestine. Movies of the mid-intestine of mutants were taken at 5.5 or 6.5 dpf at approximately three to six somite lengths rostral to the anus (also known as the vent). Approximately 6 min of real-time video was captured for each time point. In every case, wild-type siblings were imaged and included as controls to account for any possible variation in developmental timing of clutches. Periodicity was measured as time between contractile waves as they passed a fixed point within the image frame. Larvae were fixed and processed for HuC/D immunoreactivity after imaging to confirm genotypes.

Acknowledgements

We thank Chuck Kimmel, John Postlethwait, and Jim Weston for comments on the manuscript, Robert Kelsh, Macie Walker, and Paul Henion for performing some complementation analyses, Poh Kheng Loi for help with histological sectioning, Jerry Gleason for help with video microscopy, Amanda Lewis and the staff of the University of Oregon Zebrafish Facility for animal husbandry, and Eisen lab members for encouragement and support.

Ancillary