Balancing on the Crest - evidence for disruption of the enteric ganglia via inappropriate lineage segregation and consequences for gastrointestinal function

Abstract

Normal enteric nervous system (ENS) development relies on numerous factors, including appropriate migration, proliferation, differentiation, and maturation of neural crest (NC) derivatives. Incomplete rostral to caudal migration of enteric neural crest-derived progenitors (ENPs) down the gut is at least partially responsible for the absence of enteric ganglia that is a hallmark feature of Hirschsprung disease (HSCR). The thought that ganglia proximal to aganglionosis are normal has guided surgical procedures for HSCR patients. However, chronic gastrointestinal dysfunction suffered by a subset of patients after surgery as well as studies in HSCR mouse models suggest that aberrant NC segregation and differentiation may be occurring in ganglionated regions of the intestine. Studies in mouse models that possess enteric ganglia throughout the length of the intestine (non-HSCR) have also found that certain genetic alterations affect neural crest lineage balance and interestingly many of these mutants also have functional gastrointestinal (GI) defects. It is possible that many GI disorders can be explained in part by imbalances in NC-derived lineages. Here we review studies evaluating ENS defects in HSCR and non-HSCR mouse models, concluding with clinical implications while highlighting areas requiring further study.

Introduction

The enteric nervous system (ENS) is a complex network of ganglia intrinsic to the intestinal wall that is necessary for normal motility and homeostasis of the gastrointestinal (GI) tract. The ENS originates from vagal, rostral truncal, and sacral neural crest (NC) populations that migrate to the fetal gut in the developing embryo (Burns, 2005). In the mouse, vagal and rostral trunk neural crest cells (NCC) emigrate from the neural tube to the foregut and migrate caudally to populate the entire gut. Sacral NCC delaminate later from the neural tube, enter the hindgut, and migrate rostrally, opposite the vagal NCC, to co-populate the post-umbilical portion of the gut. As these populations migrate from the neural tube and along the gut, cell-autonomous signaling as well as cues from the environment play a role in their lineage divergence (Burns, 2005). NCC contributing to the ENS, appropriately called enteric neural progenitors (ENPs), differentiate from a stem cell-like state to give rise to glia and numerous neuronal types (Heanue and Pachnis, 2011; Pham et al., 1991; Sang and Young, 1996). For successful ENS development, correct timing and dosage of gene expression affecting ENP survival, proliferation, and differentiation is essential. The necessity of balancing these processes is clearly seen in the human functional bowel disorder, Hirschsprung disease (HSCR), which is clinically recognized by the absence of ganglia in a variable portion of the distal intestine.

Research in ENS development has previously focused on HSCR in part because of the dramatic megacolon phenotype that results from aganglionosis of the distal bowel. This spotlight has remained trained on HSCR as a GI disorder due to success in identifying susceptibility genes in patients and the recapitulation of aganglionosis in rodent models bearing mutations in homologous genes (Chakravarti et al., 2006). Concurrent advances in treating HSCR patients through surgical removal of distal aganglionic bowel followed by reattachment of ganglionated proximal intestine to the anus has led to increased survival (Amiel et al., 2008). Yet long term clinical follow-up of HSCR patients has found that surgical intervention does not completely “cure” intestinal symptoms in these individuals who continue to suffer from intestinal disorders, such as enterocolitis and chronic constipation, despite successful surgical removal of the aganglionic gut segment (Amiel et al., 2008; Austin, 2012; Chumpitazi and Nurko, 2011; Kaul et al., 2011). The occurrence of such chronic bowel dysfunction across a variety of surgical procedures (Chumpitazi and Nurko, 2011) and evidence from developmental studies of HSCR mouse models (Paratore et al., 2002; Paratore et al., 2001; Walters et al., 2010) hint that defects in ENS development in the proximal, ganglionated portions of the intestine may be causative. In particular, mechanistic analyses of HSCR mouse models indicate that defects in ENP developmental potential, the ability of NCC to differentiate into a variety of cell types, contribute to HSCR. Moreover, the concept that imbalance of NC-derived cell types within the ENS could lead to GI dysfunction is gaining strength, fueled by evidence from gene-specific mouse mutants that lack aganglionosis yet show deficits in ENS lineages, GI motility and immune response (See Table 1). This review collates multiple studies whose evidence adds weight to this concept and raises future questions for the field to address.

Table 1. Summary of mutant allele effects on Enteric Nervous System structure and function.

Changes in ENS architecture and density apply to ganglionated regions of the bowel for HSCR mouse models. Fields marked as "unknown" indicate data not reported, or mouse model not tested.

Allele	ENS Density	ENS Architecture	NCC Lineage Balance	Effect on GI Function	Inflammatory Response	References
Sox10Dom/+	decrease in ENPs	variable disorganized patterning observed	in vitro data	unknown	unknown	(Kapur, 1999; Southard-Smith et al., 1998; Walters et al., 2010)
Sox10LacZ/+	decrease in ENPs	postnatal unknown	in vitro data; increase in immature neuronal markers in vivo	unknown	unknown	(Paratore et al., 2002; Paratore et al., 2001)
Ednrbs-l/s-l	decreased intensity of AChE fibers	abnormal patterning	unknown	unknown	unknown	(Cantrell et al., 2004)
EdnrBs-l/s-l or EdnrBsl/+	decreased through colon	unknown	unknown	absent or impaired colonic migrating motor complexes	subset susceptible to enterocolitis	(Cantrell et al., 2004; Fujimoto, 1988; Fujimoto et al., 1988; Hosoda et al., 1994)
Ednrbtm1Ywa/+ or Ednrbtm1Ywa/tm1Ywa	unknown	unknown	unknown	unknown	subset susceptible enterocolitis	(Cheng et al., 2009; Hosoda et al., 1994; Zhao et al., 2009)
ET-3−/−	decreased neuronal numbers	unknown	increased NOS+ neurons	absent colonic migrating motor complexes	unknown	(Roberts et al., 2008)
Rettm1Cos/tm1Cos	normal neuronal numbers	decreased neuron size and cholinergic fibers	unknown	reduced contractility; but ENS signaling aberrant	unknown	(Gianino et al., 2003)
Ret DN/+	decreased neuronal numbers	decreased neuronal fiber density	unknown	unknown	unknown	(Jain et al., 2004)
RetC620R/+	decreased neuronal numbers	decreased neuronal fiber density	unknown	unknown	unknown	(Carniti et al., 2006)
Gfra+/−	normal neuronal numbers	decreased neuron size	unknown	reduced contractility; ENS signalling aberrant; delayed GI transit	unknown	(Gianino et al., 2003; Wang et al., 2010)
Gdnf+/−	decreased neuronal numbers	alterations observed, but not significant	reduction in NOS+ and CHAT+ neruons, but proportional to overall neuronal decreases	reduced contractility; ENS signalling aberrant	unknown	(Gianino et al., 2003; Shen et al., 2002; Wang et al., 2010)
Gdnf over expression	neuronal numbers increased	changes in fiber density with increased Gdnf expression	increases in late born neuron subtypes	stronger contractility; increased VIP and Substance P release; accelerated GI transit	unknown	(Wang et al., 2010)
Ascl1−/− (Mash1−/−)	decreased neuronal numbers	wide spacing, erratic arrangement of ganglia	CGRP+ neuronal numbers normal, other subtypes unknown	unknown	unknown	(Blaugrund et al., 1996)
TPH2−/−	decreased neuronal density in ileum	unknown	decreased dopaminergic neurons	slowed total GI transit and colonic emptying; accelerated gastric emptying	unknown	(Li et al., 2011)
D2−/−	unknown	unknown	unknown	faster total GI transit and colonic emptying; absorption affected	unknown	(Li et al., 2006)
Hand2 (Nestin-del)	decreased neuronal numbers	disorganized plexi	decreased CHAT+, NOS+, and calretinin+ neurons and glia	unknown	unknown	(Lei and Howard, 2011)
Hand2+/− or Hand2flox/−	decreased neuronal numbers	disorganized plexi	significant decrease in NOS+ neurons	decreased GI motility	decreased susceptibility to induced inflammation	(D'Autreaux et al., 2011)
BMP over expression	no changes in neuronal density	unknown	region specific increases in dopaminergic, 5-HT+ and TrkC+ neurons; glia increased	unknown	unknown	(Chalazonitis et al., 2004; Chalazonitis et al., 2011; Chalazonitis et al., 2008)
Noggin over expression	increased neuronal numbers	increased density	increased TrkC+, 5-HT+ neurons, decreased GABA+ and CGRP+ neurons	irregular transit; increased stool frequency, weight and water content	increased susceptibility to induced inflammation	(Chalazonitis et al., 2004; Chalazonitis et al., 2011; Chalazonitis et al., 2008; Margolis et al., 2011)
Fgf-2−/−	decreased neuronal numbers	changes in neurite length; larger neurons	decreased calbindin+ neurons	reduced choride-ion secretion; mucosal barrier defects	unknown	(Hagl et al., 2008; Hagl et al., 2012)
NET−/−	decrease in myenteric neuronal numbers	unknown	5-HT+ and calretinin+ neurons reduced	unknown	unknown	(Li et al., 2010)

Evidence from HSCR Mouse Models

Early studies in HSCR mouse models focused on determining the genetic causes of aganglionosis in spontaneously occurring mutants as a means to identify genes that contribute to development of the ENS. Initially, little thought was given to the possibility that ENS defects in ganglionated regions of the bowel might contribute to intestinal dysfunction. However, more recently, abnormalities in proximal ganglionated regions of the intestine have been recognized (Cantrell et al., 2004; Jain et al., 2004; Walters et al., 2010). This discovery led investigators to hypothesize that long term complications in HSCR patients could result from deficiencies in ganglia that had been formed in the proximal intestine (Jain et al., 2004; Walters et al., 2010). As the field has learned more about the developmental processes that contribute to aganglionosis in mouse models, evidence for effects of individual HSCR genes on NCC lineage divergence in the fetal intestine has emerged.

Sox10

Mice with a loss-of-function Sox10^+/− (Sox10^Lacz or Sox10^tm1Weg) (Paratore et al., 2002; Paratore et al., 2001) or a dominant-negative allele of Sox10 (Sox10^Dom/+) (Kapur, 1999; Southard-Smith et al., 1998) both recapitulate the HSCR phenotype of distal bowel aganglionosis. The first hint that perturbations in normal Sox10 levels could produce alterations in development potential came from in vitro studies conducted on neural crest-derived progenitors isolated from dorsal root ganglia of Sox10^−/−, Sox10^+/−, and Sox10^+/+ mice (Paratore et al., 2001). These isolated cells were allowed to grow and then differentiate in culture. The developmental potential of isolated progenitors was scored by the differentiated cell types these cultured cells produced. Sox10 deficient NCC showed altered developmental potential and were deficient in gliogenesis at both high and low plating densities. While this study focused on neural crest-derived cells collected from dorsal root ganglia, a subsequent effort by the same group determined that the temporal appearance of early and late neuronal differentiation markers within the fetal foregut and midgut of Sox10^+/− mutants in vivo were also disrupted (Paratore et al., 2002). Careful analysis revealed that neural progenitor cells expressing early neuronal markers like PGP9.5 were increased in frequency among Sox10^+/− mutants, while more mature neurons marked by expression of neurofilament 160 were reduced. Interestingly, although the NC progenitor pool was decreased in Sox10^+/− mutants, total neuronal numbers were not, suggesting that reduction of Sox10 promotes a neural fate, but may not contribute to later neural differentiation processes. Whether these fetal irregularities persist in the postnatal intestine of Sox10^+/− mutants remains in question.

An independent study assessed developmental potential of ENPs for the dominant-negative allele of Sox10, Sox10^Dom/+, while concurrently investigating the effects of genetic strain background on ENS development (Walters et al., 2010). This analysis found that ENPs from Sox10^Dom/+ mutants also had altered developmental potential in vivo and documented ENS irregularities in fetal and postnatal intestine of heterozygous Sox10^Dom/+ mice. Moreover, the authors reported atypical myofibroblast-like cells with cytoplasmic Phox2b expression in and around myenteric ganglia. While similar cells were also observed in wild-type littermates, the overall number of these unusual cells was increased in Sox10^Dom/+ mutants. Taken together the studies by Paratore et al., (2001, 2002), and Walters et al., (2010), suggest that altered developmental potential of NCC, including enteric cell types, occurs in HSCR individuals carrying disruptions of Sox10. It remains to be determined whether this altered developmental potential ultimately disrupts the balance of cell types in the postnatal enteric ganglia and how such alterations would affect intestinal function.

Ednrb/Edn3

Endothelin receptor B (Ednrb) is expressed in migrating NCC while its preferred ligand, Endothelin 3 (Edn3) is expressed in the gut wall. Several spontaneously occurring and genetically modified rodent models with altered Ednrb or Edn3 expression exist (Baynash et al., 1994; Druckenbrod et al., 2008; Hosoda et al., 1994; Roberts et al., 2008; Rothman and Gershon, 1984). However only a single study has documented alterations in proximal ENS density and patterning among Ednrb^s-l/s-l homozygous mutants (Cantrell et al., 2004). While no analysis of cell types in postnatal intestine of these or other Ednrb mutants has been published to date, Ednrb mutants do exhibit enterocolitis that leads to early death. (Fujimoto, 1988a) first described the presence of enterocolitis in the piebald lethal (Ednrb^s-l/s-l) mouse. Ednrb^s-l/s-l mice have an increase in enterocolitis compared to wild-type littermates and a subset of Ednrb^s-l/s-l mice die presumably due to acute infectious processes as they have no gross megacolon, but exhibit acute splenitis and an increase in intestinal Ig-A secreting plasma cells at time of death (Fujimoto et al., 1988b). Similarly, a subset of Ednrb-null (Ednrb^{Tm1Ywa/Tm1Ywa}, hereafter referred to as Ednrb^−/−) mice die prematurely with signs of widespread bacteremia (Cheng et al., 2010). Interestingly, these inflammatory processes appear mediated, at least in part, by ganglionated portions of the intestine as ~40% of Ednrb^−/− mice that undergo corrective pull-through surgery to remove aganglionic bowel still go on to develop enterocolitis (Zhao et al., 2010). Mice with mutations in Edn3 (Et-3) have not been assessed for enterocolitis, but have been evaluated for altered GI motility and proportions of enteric neurons. ET-3 ^−/− (Edn3^{tm1Ywa/tm1Ywa}) lack colonic migrating motor complexes in the ganglionated portions of their bowel. Additionally, despite having a lower density of neurons in the myenteric plexus, they have twice as many NOS expressing neurons compared to their wild-type littermates (Roberts et al., 2008).

Ret Signaling Pathway

Gene deficits among members of the Ret signaling pathway have been the most extensively studied models for HSCR phenotypes. Ret is a receptor tyrosine kinase expressed on ENPs that has documented roles in proliferation, migration, and survival of these cells. Athough several Ret ligands and co-receptors exist, Ret signaling in the development of the ENS primarily occurs in conjunction with its Gfra1 co-receptor and its preferred ligand, Gdnf (Gianino et al., 2003).

In humans, a single functional mutation in RET is sufficient to cause HSCR. In mice, gene targeting of Ret (Ret^{tm1Cos/tm1Cos}, commonly referred to as Ret^−/−) results in complete loss of the ENS in the small and large intestines (Schuchardt et al., 1994, 1995). However, mice with heterozygous loss (Ret^+/−) appear normal and healthy, with no aganglionosis or loss of neuron numbers in either myenteric or submucosal ganglia (Gianino et al., 2003). Whether the proportions of various neuron subtypes in Ret^+/+ and Ret^+/− are equivalent remains unknown. Interestingly, Ret^+/− mice do show a reduction in neuron size and cholinergic neuronal fiber count in certain regions of the bowel and also exhibit deficits of intestinal contractility (Gianino et al., 2003). The cellular basis of the motility deficits observed in heterozygous Ret^+/− mutants is of interest not only from a mechanistic standpoint, but also because heterozygous deficiencies are more prevalent in patient populations.

Intermediate phenotypes in Ret mutant mice that exhibit the distal aganglionosis present in human HSCR patients have been produced. These include the Ret^DN/+ (Ret^tm3.1Jmi/+) mouse that harbors a dominant-negative mutation (Jain et al., 2004) and the Ret^C620R/+ (Ret^tm1Cti/+) mouse that carries a gain-of-function point mutation (Carniti et al., 2006). NC lineage segregation, processes that generate normal proportions of neurons and glia or neuronal subtypes in development, has not been evaluated in either of these two models. However, the ENS in the proximal bowel of the Ret^DN/+ mouse contains fewer neurons and exhibits decreased neuronal fiber density. These heterozygous models provide an ideal opportunity to investigate roles for Ret in NC lineage segregation.

Gfra1 (Gdnf receptor alpha-1) is a GPI-linked co-receptor that preferentially binds Gdnf (glial derived neurotrophic factor) with Ret (Jing et al., 1996; Treanor et al., 1996). Deficits in ENS development in Gfra1 mutants tend to recapitulate those found in Ret mutants. This evidence supports the theory that Ret signaling through Gfra1 and Gdnf, not other co-receptor:ligand pairs, is largely acting in early ENS development. Similar to complete loss of Ret, ablation of Gfra1 (Gfra^{tm3Jmi/tm3Jmi}, commonly called Gfra^−/−) leads to a nearly complete absence of neurons in the small intestine and colon (Cacalano et al., 1998; Enomoto et al., 1998). And, as is seen for Ret heterozygous mutants, animals lacking a single copy of Gfra1 possess normal numbers of neurons in both myenteric and submucosal plexi, but significant reductions in neuron cell size are evident in some regions of the intestine and contractility of the intestine is abnormal in these animals (Gianino et al., 2003).

Studies from mouse models with altered Gdnf expression, particularly those that disrupt levels or timing of Gdnf production, provide the best evidence to date that Ret signaling influences ENP lineage divergence. Similar to Ret and Gfra1 knockouts, Gdnf null mice (Gdnf^{tm1Rosl/tm1Rosl} knockout mutants, commonly called Gdnf^−/−) do not develop enteric ganglia caudal to the stomach and exhibit early lethality. However Gdnf^+/− mice, in contrast to Ret^+/− and Gfra-1^+/− mutants, do exhibit hypoganglionosis throughout the intestine (Shen et al., 2002). To investigate the effect of Gdnf expression levels on ENS development, Wang et al (2010) overexpressed Gdnf in the fetal intestine. Overexpression initiating at embryonic day 17 was accomplished using transgene constructs that conferred expression in enteric glia or the muscle cells, while postnatal, global Gdnf expression was accomplished through daily peritoneal injections starting at birth. Both modes of Gdnf overexpression resulted in increased neuronal proliferation in submucosal neurons of the small intestine as well as neurons of both plexi in the colon. While myenteric neurons in the small intestine showed no significant increases in proliferation, the timing of GDNF overexpression affected submucosal neuronal subtypes. Early born neuron subtypes that were established before the onset of GDNF overexpression, such as serotonergic neurons, and ChAT expressing neurons, were not significantly affected. However, neurons that appear early but also continue to be born in the postnatal period, such as NOS+ neurons, were significantly increased in number in both Gdnf overexpression models. In addition, gross intestinal contractility changes were found in mice with Gfap-driven Gdnf overexpression. While the altered contractility may indirectly correlate with increased numbers of NOS neurons, the mechanistic basis remains to be determined as electrical properties of specific neuronal subtypes were not evaluated and multiple neuronal subtypes were not examined. It will be interesting to determine if overexpression of Gdnf before embryonic day 17 alters ratios and functionality of earlier born neuronal subtypes or if postnatal reduction of Gdnf levels cause reductions in late born neuron types.

It is worth noting that while Ret signaling through the Gfra1:Gdnf (co-receptor:ligand) combination plays an early and major role in ENS development, signaling through interactions with Gfra2:Nrtn plays a later role. This is especially relevant for the maintenance and proliferation of neurons in the submucosal plexus of the small intestine and both plexi of the colon (Gianino et al., 2003). The possibility that Gfra2:Nrtn are able to compensate for a loss of Gdnf in late stages of development expression or whether this axis selectively affects NC subtypes is yet to be determined. While heterozygous Ret^+/−, Gfra-1^+/−, and Gdnf ^+/− mouse mutants do not exhibit absence of enteric ganglia in the distal intestine that is characteristic of HSCR, all of these models share loss of intestinal contractility and exhibit reduced release of neurotransmitters in vitro (Gianino et al., 2003). Understanding the causes of altered GI motility is crucial for developing strategies to treat patients with functional GI deficits that comprise a substantial proportion of the population. The realization that heterozygous deficiency of Ret, Gfra1, and Gdnf alter the composition and functionality of the ENS likely would not have come about without studies to better understand the rare disorder HSCR. These results nicely illustrate how studies of rare disorders can inform common problems.

Evidence from Non-HSCR Models

Although mouse mutants have been used to model the obvious aganglionosis that occurs in human HSCR, studies in non-HSCR mouse models provide the most substantial evidence that disruption of NC-derived lineages can occur in the ENS and that these imbalances can alter GI function. Understanding the etiology of GI dysmotility is relevant to a large fraction of the population as millions of people suffer from intestinal motility disorders. Surprisingly, the etiologies of chronic constipation, pseudo-obstruction, Crohn's disease, ulcerative colitis, and other GI disorders that appear to have genetic components remain relatively obscure (Karban et al., 2004; Ostwani et al., 2010; Van Limbergen et al., 2009; Villani et al., 2010). Given the evidence detailed below, it is likely that ENS defects, including abnormal NC lineage segregation, contribute significantly to many GI disorders. Discoveries in non-HSCR mouse models should aid in teasing apart the etiologies of these disorders.

Ascl1 (formerly Mash-1)

Ascl1 (Mash-1) is a basic helix-loop-helix transcription factor most extensively studied for its role in neurogenesis in the central and sympathetic nervous systems (Bertrand et al., 2002). However, Ascl1 also largely influences development in a subpopulation of ENPs. Blaugrund et al., (1996) determined that Ascl1-dependent enteric progenitors give rise to a subset of transiently catecholaminergic cells which differentiate into serotonergic neurons. Transiently catecholaminergic cells initially express catecholaminergic gene products, such as tyrosine hydroxylase, but lose catecholominergic gene expression upon differentiation. Ascl1 null embryos (Ascl1^{tm1And/tm1And}, commonly referred to as Ascl1^−/−) have severely reduced neuronal serotonin (5-HT) expression and exhibit gross ENS deficits, such as a reduction in overall neuronal number and more widely spaced and erratically arranged ganglia. This study only evaluated one other neuronal subtype, the late born CGRP (calcitonin gene related peptide) expressing neurons, which appeared unaffected in Ascl1 null mutants. Recent studies described below have found that perturbations in neuronal serotonin expression alter neuronal subtype proportions and development, thus it would be interesting to evaluate other neuronal subtypes in the Ascl1^−/− mutants.

Serotonin (5-HT)

Although serotonergic neurons only comprise 1–2% of the myenteric plexus in rodents (Furness, 2006), they are one of the earliest types of neurons that exit the cell cycle in the developing ENS (Pham et al., 1991) and they also affect the development of later born neurons, GI motility (Li et al., 2011), and GI epithelial growth (Gross et al., 2012). The exact role 5-HT plays in GI development and function has been somewhat obscured by the fact that two 5-HT expressing cell types exist in the gut — enterochromaffin cells within the gut epithelium and neurons within the myenteric and submucosal plexi. The recent, fortuitous discovery that 5-HT expressing epithelial cells rely on the enzyme Tph1 (tryptophan hydroxylase 1) for 5-HT synthesis while enteric neurons rely on the enzyme Tph2 (tryptophan hydroxylase 2) (Gershon and Tack, 2007; Neal et al., 2009) has permitted a more thorough investigation of 5-HT in ENS development. Li et al., (2011) conducted a study utilizing Tph1, Tph2, and Tph1/2 knockout (KO) mice to determine the effects of 5-HT depletion (epithelial vs neuronal vs total gut respectively) on enteric development and GI motility. These studies identified perturbations of both ENS development and GI motility in the Tph2 KO (neuronal 5-HT depleted) mice. In contrast such perturbations were not seen in Tph1 single knockout or exacerbated in the Tph1/2 double knockouts. In comparison to wild-type ENS development, Tph2 KO mice had a significantly lower neuronal density in the ileum and a significant decrease in the late born dopaminergic neurons. Complementary gain-of-function experiments using ENPs isolated from Slc6a^−/− (serotonin transporter) mutants, as well as treatment of ENP cultures with 5-HT both confirmed and extended the finding that 5-HT impacts development of dopaminergic neurons in the fetal intestine. Functional deficits due to loss of neuronal 5-HT were also documented in Tph2 KO mice including significantly faster gastric emptying but overall slower total GI transit and colonic emptying. A schematic illustrating how disruption of one enteric neuronal population, like 5-HT+ neurons, might impact cell types within the ganglia is diagrammed in Figure 1. Experiments to investigate the effect of neuronal 5-HT overexpression on ENS development and GI motility have yet to be conducted.

Figure 1.

Schematic diagram illustrating potential mechanisms by which alterations in gene expression or function could lead to multiple ENS defects. At left a typical normal myenteric ganglia is shown consisting of multiple neuronal subtypes (ChAT+ green, 5-HT+ dark blue, and GABA+ peach). Enteric glia normally reside within the ganglia, along inter-ganglia fibers, and within the underlying muscle (grey ovals/stellate cell shapes). Changes in neuron and glia numbers and imbalance of specific neuronal subtypes (such as the decrease in CHAT+ neurons seen in the inset, fewer green neuronal cell bodies) have already been described in some mouse mutants. These alterations — as well as more subtle changes — could disrupt the electrical properties of neurons and glia thus interfering with signaling between ENS components and other cell types like those of the immune system. Such changes could ultimately result in GI dysfunction due to deficits in motility and inappropriate immune response.

Dopamine Receptor 2 (Drd2)

Do alterations of ENS dopaminergic signaling significantly alter GI function? Li et al., (2006) conducted a study in which dopaminergic signaling was disrupted through dopamine (Drd2) receptor knockouts. Five dopaminergic receptors (Drd1a-5) are expressed in the gut, but Drd2 is restricted to enteric neurons. Drd2^−/− (allele name Drd2^{tm1Schm/tm1Schm}) knockout mice had faster total GI transit times as well as faster colonic motility when compared to wild-type littermates. Absorption of food was affected as Drd₂^−/− mice were runted although they ate more food and drank more water than wild-type siblings. Other neurons and other cell types receiving signals from dopaminergic neurons in the bowel may be able to compensate for changes in dopaminergic neuron number or function by altering DA receptor expression. However, this possibility remains to be investigated in detail.

Hand2

Hand2 (heart- and neural crest derivatives-expressed protein 2) is a basic helix-loop-helix DNA binding protein that plays a role in neurogenesis and proliferation (Rohrer, 2011). Several studies in recent years have implicated Hand2 in lineage divergence among ENS progenitors (D'Autreaux et al., 2011; D'Autreaux et al., 2007; Hendershot et al., 2007; Lei and Howard, 2011). Collectively, these efforts demonstrate how loss of a single gene can alter proportions of specific neuronal classes in the ENS with a significant impact on intestinal motility.

Initially, Hendershot et al (2007) ablated Hand2 expression in NC derivatives through crosses of a conditional floxed allele (Hand2^{tm1.1MajH:Wnt1-Cre}) with a Wnt1 promoter driven-Cre transgene (Danielian et al., 1997). This strategic approach removed Hand2 from all NCC with the result that organization and connectivity of enteric ganglia was disturbed accompanied by an overall reduction in neuron numbers. Complete absence of VIP expressing neurons from the fetal intestines of these mutants, while ChAT+ neurons were maintained, was the first indication that loss of Hand2 could have a specific effect on a discrete neuron lineage. Subsequent analyses by Lei et al (2011) depleted Hand2 from a specific subset of progenitors during ENS development using a Nestin-Cre driver (allele Hand2^{tm1.1MajH:Nestin-Cre}). Through this refined ablation of Hand2, a Nestin-independent population of ENPs that accounted for approximately 15% of total neural precursors was revealed. In Hand2 depleted mutants, this population was able to proliferate and compensate for a decrease in nestin-dependent neurons. However, this compensatory effect was unique to specific subsets of neurons in Hand2 mutants. For example, Nestin-independent precursors were able to proliferate and compensate for the loss in nestin-dependent ChAT neurons, but could not compensate for the loss of NOS or calretinin expressing neurons. Additionally, Nestin-independent precursors appeared skewed towards a neuronal fate. In wild-type animals and Nestin-driven Hand2 depleted mice, greater than 90% of glia were Nestin-dependent. Glia were reduced overall in Nestin-driven Hand2 depleted animals and nestin-independent precursors did not compensate for this loss of glia. In addition to skewed ratios of NC derivatives, nestin-driven Hand2 depleted mice had an overall disorganized plexus and typically died by postnatal day 20. While the cause of death was uncertain, these animals exhibited massive distension of the entire GI tract, and it is certainly plausible that GI dysfunction contributed to their premature death.

D'Autreaux et al., (2007) also observed that NC-specific depletion of Hand2 disrupted terminal differentiation of enteric neurons. Moreover, this group later demonstrated that Hand2 depletion in the ENS impacted not only GI motility, but inflammation as well (D'Autreaux et al., 2011). They observed disruption of enteric plexus organization as well as a significant decrease in NOS expressing neurons in Hand2 deficient mice (Hand2^+/− and Hand2^flox/−), but they also found that neuronal subtype specification is Hand2 dosage specific by analyzing ratios of NOS+ neurons in various combinations of Hand2 targeted alleles. In contrast to the Lei et al., (2011) study, D'Autreaux et al., (2011), found glia numbers were not significantly reduced in the myenteric plexus of Hand2 haplo-insufficient mutants. This finding could either reflect the capacity of enteric glia to differentiate despite low levels of Hand2, temporal effects of Hand2, and/or the effect of differential Hand2 expression levels in distinct NCC subpopulations. Effects of transcription factor levels and timing have previously been shown to play a role in cell lineage segregation in the pancreas (Collombat et al., 2003; Hang and Stein, 2011; Liu et al., 2011; Pan and Wright, 2011). And, differential expression transcription factor levels may be functional in the ENS as well since this has been documented in ENPs at stages concurrent with lineage segregation (Corpening et al., 2008). Regardless of the underlying mechanism, the D’Autreaux and Lei studies differ since Nestin-Hand2 depleted mutants die near weaning with gross bowel distention, but Hand2^+/− mice are fertile and have a normal lifespan. However, Hand2^+/− heterozygotes do not have normal intestinal function, as D'Autreaux et al., (2011) documented these mutants have a significant increase in total GI transit time and markedly slower colonic motility compared to wild-type littermates. Surprisingly, despite a reduction in overall neuronal density and decrease in GI motility, Hand2^+/− mice are protected against TNBS induced inflammation when compared to wild-type littermates (Margolis et al., 2011). This finding was unexpected as increased gut motility is thought to protect against microbial overgrowth and infection (Powell, 1995). Margolis et al., (2011) concluded that an overall decrease in neuronal density is protective against GI inflammation despite the decrease in motility seen in these mutants. Indeed, neurons have been implicated in GI inflammation (Lakhan and Kirchgessner, 2010). However, it is not yet known which enteric neuron subtypes contribute to inflammatory processes, and future studies elucidating roles for distinct neuronal subtypes and enteric glia in initiating and propagating inflammation will be essential to understanding how the ENS mediates enterocolitis.

Bone Morphogenic Proteins (BMPs)

BMPs orchestrate several developmental programs in a dosage and time specific manner during the course of development including neurogenesis in the developing gut. ENPs encounter BMP signaling throughout development as the notochord and somite mature (Faure et al., 2002). In addition, it is known that BMP 2, 4, and 7 are expressed in the developing fetal intestine (Chalazonitis et al., 2004; Chalazonitis et al., 2008).

The influence of BMPs agonists and antagonists on ENP lineage segregation has been studied in vitro and in vivo for multiple enteric lineages. In vitro, isolated rat ENPs treated with BMP2 and BMP4 levels above those normally encountered in vivo appear to specify TrkC expressing neurons prematurely and cause an overall increase the number of TrkC expressing neurons (Chalazonitis et al., 2004). Mouse models with neuronal BMP4 overexpression show no changes in ENS neuronal density in vivo; however, neuronal subtype proportions are skewed with significant increases in dopaminergic neurons, serotonergic neurons, and TrkC expressing neurons in specific gut regions (Chalazonitis et al., 2008). Additionally, BMP4 overexpression causes glial numbers to increase markedly at the expense of neurons (Chalazonitis et al., 2011).

Antagonsim of BMP signaling in neurons through NSE-driven noggin overexpression corroborated findings in regards to BMPs role in ENS development. Noggin overexpression in neurons increases overall ENS neuronal density which is consistent with BMPs’ opposing role in promoting neural differentiation at the expense of proliferation (Chen and Panchision, 2007). Noggin overexpression increased the proportion of serotonergic neurons (Chalazonitis et al., 2008), but led to decreases in TrkC and GABA and CGRP expressing neurons (Chalazonitis et al., 2004) (Chalazonitis et al., 2008) as well as the glia to neuron ratio (Chalazonitis et al., 2011). Importantly, these mice also exhibited irregular bowel transit (Chalazonitis et al., 2008) as well as increased sensitivity to chemically induced inflammation (Margolis et al., 2011). It is clear that decreased BMP signaling tips the balance to favor more neurons and glia as well as certain neuronal subtypes; however, we can only speculate now as to how these imbalances are exactly influencing GI function. Also, it is imperative to mention that these studies (Chalazonitis et al., 2004; Chalazonitis et al., 2011; Chalazonitis et al., 2008) and others (Chen and Panchision, 2007) note the dosage and timing of BMP expression affects the timing and differentiation capacity of cell types.

Norepinephrine Transporter (NET)

Although the ENS is not known to house any intrinsic norepinephrine producing neurons, developing transiently catecholaminergic neural precursors express norepinephrine transporter (NET). Transiently catecholaminergic subtypes include, but are not limited to, 5-HT expressing neurons and some (but not all) Calretinin and NOS expressing subtypes (Blaugrund et al., 1996; Li et al., 2010). To determine if NET is essential for the specification of these transiently catecholaminergic neuronal subtypes, (Li et al., 2010) evaluated transiently catecholaminergic subtypes in the myenteric plexus of NET null mice (Slc6a^2tm1Mca common name NET^−/−). NET^−/− mice have an overall decrease in myenteric neuron numbers, with 5-HT and Calretinin+ subsets significantly reduced in number. In NET^−/− mutants NOS+ neurons were observed to be slightly reduced in number, but the reduction was not significant. These results indicate that NET is essential for certain transiently catecholaminergic subtypes. However, it is unknown whether both the transiently catecholaminergic NOS and non-transiently catecholaminergic NOS populations of neurons were largely unaffected by the NET deficiency or if one of these populations was able to proliferate and compensate for an undocumented loss in the other population.

Fgf2 and Sprouty2

Mutations in Sprouty2, an FGF2 antagonist, have been associated with intestinal neuronal dysplasia (IND) in humans (Borghini et al., 2009) and Fgf-2^−/− (Fgf2^{tm1Zllr/tm1Zllr}) mice similarly exhibit abnormally large ganglia that contain fewer, but larger, neurons than wild-type mice (Hagl et al., 2008). (Hagl et al., 2012) classified enteric neurons in Fgf-2^−/− mice based on morphology (Hanani and Reichenbach, 1994) and found a significant decrease in calbindin expressing Dogiel Type 2 neurons. Neurons may be classified by their shape, function, neurotransmitters, electrical properties, and/or gene expression patterns. Most of the studies mentioned previously used gene expression detected by immunohistochemistry to classify neurons. Because the studies by Hagl et al (2012) assessed morphology and calbindin expression as characteristic features, the authors hypothesized that the neuronal population lost was sensory in nature. Additionally, Fgf-2^−/− mice exhibit alterations in chloride secretion and translocation of bacteria across the mucosal barrier. How the ENS modulates permeability to bacterial penetration has not been determined (Hagl et al., 2008).

Transgenic Models

Recent development of fluorescent transgenic mouse models (Corpening et al., 2008; Druckenbrod and Epstein, 2005; Enomoto et al., 2001) has permitted refined analysis of ENS development that were not previously possible by immunohistochemical studies of mutant strains. In particular, it has long been accepted that ENPs either enter the foregut mesenchyme proximally and migrate down its length in a rostral to caudal fashion (vagal ENPs) or they enter the gut at the distal end and migrate caudal to rostral (sacral ENPs). These processes left unexplained how subsets of HSCR patients exhibit 'skip segment' aganglionosis where small region(s) of the colon contain ganglia (O'Donnell & Puri, 2010) instead of continuous aganglionosis that is typically seen in the distal colon. How skip segment aganglionosis might arise given known mechanisms remained a quandary although several groups suggested that extramural ENPs could account for the ganglionated 'skip segments' seen in some HSCR cases (Coventry et al., 1994; O'Donnell and Puri, 2010). Synthesis of information from multiple transgenic models now suggests these skip regions occur because normal migration of ENPs across the mesentery between the small intestine and the colon occurs, but subsequent filling of the proximal colon by ENPs that migrate from the cecum fails. Mesenteric ENPs were initially reported by (Druckenbrod and Epstein, 2005). However, only recently has mechanistic data emerged that can explain the role of these mesenteric ENPs in skip segment aganglionosis. Recent analysis by (Nishiyama et al., 2012) using elegant live cell imaging has unequivocally shown that ENPs migrate across gut mesentery to populate the majority of the colon before cells migrating down the length of the gut tube arrive (Nishiyama et al., 2012). The integration of these analyses is just one more example of how studying fundamental processes in normal development can serve to elaborate the etiology of disease.

Finding Balance in the ENS

Post-surgery, chronic constipation and/or enterocolitis afflicts many HSCR patients, and some Ednrb mouse models suffer from enterocolitis as well despite surgical removal of the aganglionic segment (Zhao et al., 2010). Because the distal aganglionic region has been removed in these patients and mouse models, the effect of and role played by ganglionated proximal intestine in long-term GI function has to be considered. To date evidence from two distinct sources both suggest the balance of cell types within enteric ganglia that derive from NC during lineage segregation can significantly alter the ENS. Studies from mouse models that mimic the diagnostic feature of HSCR, aganglionosis, are in their infancy, but initial data is consistent with the possibility of disrupted lineage segregation in some mutants. Likewise analysis of enteric development in mouse mutants that do not exhibit overt aganglionosis (non-HSCR) also indicates that any one of several gene deficits can disrupt the normal processes of ENS ontogeny and skew the resulting profile of neurons and glia away from normal composites. It should be emphasized that while several non-HSCR mouse mutants exhibit alterations in enteric neural crest lineages, how these imbalances directly or indirectly cause GI dysfunction remains to be determined. Many studies evaluating ENS architecture and composition do not assess GI function and those preliminary studies that do understandably focus on gross assessments, such as overall changes in GI transit time or colonic motility. A need for further histological, physiological, and electrophysiological studies to determine exactly how certain populations and imbalances in these populations affect GI function would be valuable for better understanding the etiology of intestinal dysmotility.

As the ENS field moves forward, we must consider several factors. How much does one have to tilt the balance of neural crest derivatives for GI dysmotility to occur? The answer to this question is paramount. Some variability must exist for normal bowel function to occur as inbred mice strains have different proportions of neuronal subtypes (Neal et al., 2009), yet still exhibit normal bowel function. It may be that only certain environmental stressors, such as infection, cause certain imbalances to manifest as disease. Reduced penetrance seen in some human diseases could be due to a slight tip of the balance by genetic modifiers, causing the balance to wax and wane between disease and unaffected amongst family members. Other functional bowel disorders in humans appear to have a genetic component, but the genetic etiologies and risk factors remain unknown. Small perturbations that do not lead to overt aganglionosis, but cause imbalances of cell types within the ENS, could explain disorders such as chronic constipation despite the presence of detectable enteric ganglia throughout the intestine as well as residual deficits seen in HSCR patients that suffer GI dysfunction long after surgical intervention.

Additionally, recent studies have unearthed previously unrecognized ENP subpopulations that appear to be distinct in their developmental capacity. It was formerly believed that ENPs were homogenous with regards to expression of Nestin, but Lie et al (2011) identified a previously unknown ENP nestin-independent population. A study by Mundell and colleagues (Mundell et al., 2012) revealed a distinct ENP population that does not express Ednrb and appears to have substantial capacity to compensate for loss of other enteric progenitors during ENS development. The ability to fate-map, isolate, and manipulate these and other subpopulations as the ENS forms will be important. In the future, such efforts will provide insight into how and when subpopulations are specified, how distinct subpopulations influence lineage diversification, and define the ability of distinct subpopulations to compensate for disturbances in other enteric cell types that might otherwise derail GI motility and immune response.

Highlights.

• ►Neurons and glia of the enteric nervous system derive from neural crest
• ►Mouse models that alter enteric ganglia are reviewed
• ►Ganglia architecture and composition require normal neural crest differentiation
• ►Deficits in enteric ganglia are related to clinical gastrointestinal disorders