Development of the Autonomic Nervous System: Clinical Implications

Abstract

Investigations of the cellular and molecular mechanisms that mediate the development of the autonomic nervous system have identified critical genes and signaling pathways that, when disrupted, cause disorders of the autonomic nervous system. This review summarizes our current understanding of how the autonomic nervous system emerges from the organized spatial and temporal patterning of precursor cell migration, proliferation, communication, and differentiation, and discusses potential clinical implications for developmental disorders of the autonomic nervous system, including familial dysautonomia, Hirschsprung disease, Rett syndrome, and congenital central hypoventilation syndrome.

The autonomic nervous system (ANS) is essential for the regulation of homeostasis; when ANS circuits malfunction, the consequences cause considerable morbidity and can be fatal. Autonomic disorders are marked by dysfunction of the cardiovascular, respiratory, and/or digestive systems, and include congenital central hypoventilation syndrome, familial dysautonomia, Rett syndrome, and Hirschsprung disease. Understanding the etiological and pathophysiological mechanisms underlying ANS disorders is necessary for designing therapeutic strategies for intervention. To this end, it has proven useful to investigate the cellular and molecular interactions that drive the emergence of a normally constructed and functioning ANS. Key events that constitute normal ANS development depend on the precise spatial and temporal regulation of neurogenesis, gliogenesis, migration of precursor cells to their appropriate locations, axon guidance to specific targets, target innervation, synapse formation, and, finally, a careful pruning to fine tune and consolidate neural circuits. All of these critical developmental events are orchestrated by the coordinated expression of specific cell surface molecules, transcription factors, secreted growth factors, and extracellular matrix molecules. Should any of those events or cellular signaling pathways go awry, disorders of the nervous system will emerge. Studies on the developing nervous system have identified numerous essential genes and signaling pathways that can be potential therapeutic targets in ANS disease. The goal of this review is to discuss the key developmental events that must occur for the normal formation of the ANS. Where known, this article will also consider the clinical implications of dysregulation of particular genetic pathways and/or cell biological events on ANS development and function. This article will focus on the development of the peripheral components of the ANS and not include the central components of the ANS.

The Neural Crest Role in Autonomic Nervous System Development

First identified over 150 years ago by His, the neural crest is arguably the most intriguing cell population to arise during vertebrate evolution. Their identity is specified during neurulation,^1,2 as the neural folds fuse to form the neural tube. Neural crest cells then undergo an epithelial to mesenchymal transition, delaminate from the neural tube, and migrate throughout the embryo (Fig. 1). Neural crest cells give rise to the entire trunk peripheral nervous system (PNS), both neurons and glia, including all the sympathetic and parasympathetic ganglia, adrenal medulla, carotid body, enteric nervous system (ENS), and the entire chain of dorsal root (sensory) ganglia, in addition to melanocytes.³ In the cervical region, the neural crest also gives rise to much of the cranial-facial skeleton, connective tissue, and cartilage, in addition to the cardiac outflow tract. Failures in normal neural crest cell development cause neurocristopathies, which include DiGeorge syndrome and neuroblastoma.⁴

Fig. 1.

Cross-section of chick embryo (embryonic day 2.75) that has been immunolabeled with an antibody to neural crest cells (HNK-1; green). Neural crest cells are observed migrating ventrally to form the (1) dorsal root ganglia, (2) vertebral chain of sympathetic ganglia, (3) chromaffin cells in the adrenal medulla, and (4) enteric nervous system. (Photograph taken by Dr. Lynn George.)

Neural crest cells delaminate along the entire neuraxis except for the most anterior region. Those that delaminate from the head region are referred to as cranial crest, those that will migrate into the heart are referred to as cardiac crest, and finally those that will comprise the trunk PNS derive from the trunk neural crest. A modification of this classification scheme subdivides neural crest cells into cranial, vagal, and trunk neural crest cells, with the latter further characterized according to the distinct regions from which they emerge along the neuraxis: cervical, thoracolumbar, and sacral. In the trunk, neural crest cells delaminate over several hours such that three temporal cohorts of cells delaminate, giving rise to distinct derivatives: the earliest cells to delaminate migrate straight ventrally between the neural tube and somites in close proximity to developing blood vessels and give rise to the sympathetic and adrenal lineages; the second cohort migrates ventrolaterally through the anterior half of the sclerotome to give rise to the neurons and glia of the dorsal root ganglia (DRG) and to a later wave of sympathetic and adrenal cells; the last cohort to delaminate migrates dorsal-laterally and gives rise to the body’s melanocytes.^5–8

How one cell population can generate such a wide variety of cellular derivatives has been the subject of extensive study. Fundamental questions that have been investigated include whether neural crest cells are a homogenous, multipotent population when they delaminate, with their cell fate determined stochastically by the particular environmental cues they encounter along their migratory path or at their destination, or, alternatively, are neural crest cells a heterogenous population, with their cell fates already predetermined before or as they delaminate, and hence migrate along their cell-specific appropriate routes and stop at their prescribed destinations.^8,9 Furthermore, how plastic is their identity and to what extent do environmental cues instruct and determine whether a neural crest cell will differentiate into, for example, a Schwann cell versus a sympathetic neuron? Elegant embryonic manipulations and transplantation studies, fate mapping and lineage tracing studies, and high-resolution 4D live time lapse imaging^10–15 have revealed that both mechanisms occur—some neural crest cells are prespecified, while the majority are multipotent and can give rise to a wide array of derivatives. Moreover, within a developmental time window, certain transplanted cells from different axial levels can transdifferentiate into the cell type of their new, host-site axial level, illustrating that at least some neural crest cells exhibit considerable plasticity.

One of the most surprising and important recent discoveries in neural crest cell biology is the identification of the specialized neural crest stem-like cell called the “Schwann cell precursor (SCP).”^16–18 Using genetic lineage tracers, SCPs were identified as neural crest derivatives that closely associate with and travel along preganglionic sympathetic and parasympathetic axons.^19–23 Once these developing nerves penetrate into their target region, SCPs detach from the nerve, colonize their target, proliferate, and differentiate into parasympathetic neurons, enteric neurons, and/or adrenal chromaffin cells.^17,20–22 Depending on the tissue type, SCPs can also differentiate into both myelinating and nonmyelinating Schwann cells, endoneural fibroblasts, and melanocytes in addition to tooth pulp cells and odontoblasts.^{19,20,24–27}

Molecules mediating neural crest cell development:

Some of the molecules mediating neural crest cell migration are attractive and several are repulsive, and it is their combination that can direct and channel the spatial and temporal pattern of neural crest migration. For example, trunk neural crest cells are repulsed from entering the foregut, while vagal neural crest cells are not.^28,29 Neural crest cells migrating ventrally in the trunk are repulsed by specific proteins expressed on the posterior sclerotome, and instead migrate along the permissive anterior sclerotome.^30,31 Delamination is mediated by members of the Wingless-related integration site (Wnt) and bone morphogenetic protein (BMP) secreted signaling protein families.^32,33 The transcription factor FoxD3 specifies the identity of neural crest cells that migrate on the ventral pathway, but not for those neural crest cells that migrate dorsolaterally.³⁴ The Sox family of SRY-related HMG-box transcription factors plays key roles: all migrating neural crest cells express Sox10, and its deletion causes a failure in neural crest derivatives.^35,36 The neural crest cells that stop to form DRG neurons and glia express basic helix–loop–helix transcription factors called “neurogenins,” while neural crest cells destined to form sympathoadrenal cells instead express Ascl1/Mash.^37,38

The fate of neural crest cells that have migrated ventrally beyond the site of DRG formation is determined by a gene regulatory network triggered by cues associated with the ventrally positioned dorsal aorta. Factors secreted from the dorsal aorta program neural crest cells into a sympathoadrenal fate by activating a gene regulatory networkof transcription factors.³⁹ Key regulators include BMP-4 and −7, which are members of the transforming growth factor beta (TGF-β) superfamily of growth factors that are expressed by smooth muscle cells that line the dorsal aorta^40–42 and turn on expression of the transcription factors Phox2b, Ascl1, Hand2, and GATA.^43–46 Deletion of the Phox2b gene in mice leads to a failure in the development of all autonomic ganglia.⁴³

Epibranchial Placodes Contribution to the Development of Visceral Sensory Neurons

In addition to the critical role of the neural crest in generating sensory neurons and autonomic neurons and glia, the second source of precursor cells for the developing ANS are specific ectodermal thickenings in the head called “placodes.” These epibranchial placodes give rise to the visceral sensory neurons and glia of the geniculate (cranial ganglion VII), petrosal (cranial ganglion IX), and nodose (cranial ganglion X) ganglia. Interestingly, some cranial ganglia are entirely derived from placodal cells, while the majority are of mixed placodal and neural crest origin.^47,48 Much like neural crest cell, placodal cells delaminate from the columnar ectodermal epithelium, migrate, and then aggregate to form or contribute to the cranial ganglia.⁴⁹ The vagus and glossopharyngeal nerves monitor oxygen, carbon dioxide, and pH in the carotid sinus and aortic arch, information which is then conveyed to visceral motor neurons in the brainstem to trigger appropriate homeostatic responses. While the visceral sensory neuron cell bodies reside within the ganglion, they extend two processes—one that innervates peripheral targets and one that conveys that afferent information to the second-order neurons in the brainstem, specifically to the nucleus tractus solitarius (NTS) and nucleus ambiguous in the rostral medulla. The vagus nerve innervates such key targets as the carotid body, heart, blood vessels, lungs, spleen, liver, pancreas, and gastrointestinal tract, and is the major component of the parasympathetic nervous system.⁵⁰ One of the most exciting recent clinical findings regarding the role of the vagus nerve in disorders of the ANS is that abnormally misfolded α-synuclein appears to be retrogradely transported from the gastrointestinal system to the CNS along the vagus nerve to cause Parkinson disease.⁵¹

As with the neural crest, a stereotyped set of genes is expressed that regulates the differentiation of nodose neurons. And, like the neural crest derivatives, the nodose ganglion is composed of a heterogeneous cell population—nodose neurons not only innervate a wide diversity of peripheral targets, but they also differentiate into such distinct cell types as baroreceptors, chemoreceptors, nociceptors, and mechanoreceptors. The transcription factor Phox2b is absolutely essential for the differentiation and survival of the vast majority of visceral sensory neurons derived from the epibranchial plac-odes; targeted deletion of the Phox2b gene from mice results in the absence of over 90% of neurons in the nodose and petrosal ganglia.⁴⁵ Several neurotrophic factors are essential for the survival of the full complement of nodose and petrosal neurons, including the target-derived factor, brain-derived neurotrophic factor (BDNF), NT-4, and glial-derived neurotrophic factor (GDNF).^52–57 Whether the requirement for so many neurotrophic factors corresponds to the needs of different classes of sensory neurons and/or innervation of distinct target tissues is not fully resolved. Baroreceptors are located in both the petrosal and nodose ganglia, and are mechanoreceptors that innervate the carotid sinus and aortic arch (respectively), and convey changes in arterial blood pressure. Baroreceptors express the BDNF and NT-4 receptor, TrkB, and targeted deletion of both BDNF and NT-4 causes the death of over 90% of nodose and petrosal neurons.^53–55 The subset of nodose neurons that are dopaminergic innervates the glomus cells of the carotid body and are completely dependent on BDNF/TrkB signaling for survival.^53,58 TrkB is also expressed in the brain stem in the pre-Bötzinger complex, the site that regulates respiratory rhythm generation. Because of the combined dependency of baroreceptor, chemoreceptor, and brainstem respiratory circuitry on BDNF signaling, mice that are null for BDNF have severe respiratory impairment with deficits in breathing control.⁵⁹

The molecular mechanism used by vagal axons to navigate toward and innervate their wide array of target tissues is not fully understood, but there is evidence for the adhesion molecules Slit/Robo, netrin/DCC, and laminins in influencing the paths forged by developing vagal axons.^60,61 Recent single-cell RNA sequencing of nodose ganglion neurons has generated an “atlas” of differential gene expression patterns in clusters of nodose ganglion cells that should help us understand how different subclasses of nodose neurons establish and maintain their distinct identities.^62–64

Development of Parasympathetic Neurons

Two characteristic features of parasympathetic ganglia, as opposed to sympathetic and sensory ganglia, are that they are located deep within the body in close proximity to or within their target organs, and they develop later than do the sympathetic and sensory ganglia. While they, like the majority of the PNS, derive from the neural crest, one question is how do they arrive at their distally located targets? Within the past few years, our understanding of this problem has been elucidated by several exciting studies that identified a requirement for preganglionic nerves in the emergence of parasympathetic ganglia. The preganglionic parasympathetic nerves serve as stem cell–like niches for SCPs. Once at the target site, SCPs invade, proliferate, and differentiate into the parasympathetic neurons of the ciliary, pterygopalatine, lingual, submandibular, and otic ganglia.^19,20 If preganglionic parasympathetic nerves are deleted or significantly reduced in size, the corresponding parasympathetic ganglia do not form.^19,20 Once at the target site, under the influence of local extrinsic cues including BMPs, the SCPs turn on the transcription factors Ascl1 and Phox2b^37,39,43 that will induce their parasympathetic genetic identity. The SCPs give rise to both the neurons and glial cells in the parasympathetic ganglia. SCPs express the neuregulin1 receptors, ErbB2 and ErbB3, and deletions in the ligand, neuregulin1, lead to significant reductions in the number of parasympathetic neurons.^19,20 As parasympathetic neurons mature and innervate their targets, they become highly dependent for survival on the growth factors GDNF and neurturin.⁶⁵

Development of Sympathetic Neurons

Sympathetic neurons derive entirely from the neural crest. In the trunk, they are organized in two chains of paravertebral ganglia, which lie just ventrolaterally to the spinal cord (see Fig. 1), and a second prevertebral group of ganglia including the celiac, superior, and inferior ganglia. More rostrally are located the superior cervical ganglion (SCG) and the stellate ganglion that supply the heart and the head. The core of the developing sympathetic chain ganglia is established by an early cohort of neural crest cells that express the chemokine receptor, CXCR4, and migrate ventrally, chemotactically, in response to their ligand, SDF1/CXCL12, which is expressed in the mesenchyme surrounding the dorsal aorta.¹¹ In response to the cues associated with the dorsal aorta, these cells differentiate into sympathetic neuroblasts, and finally sympathetic neurons.^39,66 Upon reaching the dorsal aorta, the sympathetic precursors actively segregate rostrocaudally to form discrete ganglia—a process that is dependent on the cell adhesion molecules N-cadherin and EphrinB/EphB signaling.⁶⁷ As they form ganglia, they turn on expression of sympathetic neuron markers, including tyrosine hydroxylase. At this point, they undergo a second collective migration: filopodia extended from the maturing sympathetic neurons contact their axial-specific preganglionic sympathetic axons, and this contact induces the dorsal migration of the nascent sympathetic ganglia to their final location adjacent to the ventral root of the spinal cord.¹¹ This patterning is mediated by BDNF secreted by the preganglionic sympathetic axons, which activates TrkB receptors on the sympathetic neurons; blockade of BDNF/TrkB binding prevents the dorsal migration of the immature sympathetic ganglion.¹¹ Finally, one of the most unusual characteristics of developing sympathetic neurons is their unique capacity to reenter the cell cycle and divide, even after they have extended axons and express genes that demarcate differentiated neurons.^68–70

The transcription factor Phox2b is essential for differentiation and survival of sympathetic neurons—if deleted, sympathetic precursors undergo apoptosis before they even reach the aorta, and fail to turn on the gene network required for differentiation of sympathetic neurons.⁴³ Loss ofAscl1 inneural crest cells leads to a reduction in the number of sympathetic neurons,⁷¹ and also is required for sympathetic axon outgrowth and guidance. Very intriguing new data have revealed a requirement for the transcription factor hypoxia-inducible factor 1a, Hif1α, during the development of pre- and postganglionic sympathetic neurons.⁷² Hif1 is the master regulator of the cell’s transcriptional response to hypoxia, and its deletion in neurons prevented normal cardiac innervation.

Like all other developing peripheral neurons, embryonic sympathetic neurons are absolutely dependent on neurotrophic factors for their survival, as first illustrated by the work of Levi-Montalcini and Booker in 1960,⁷³ that antibodies against nerve growth factor (NGF) could cause a complete sympathectomy during development. With the identification of the genes for NGF and its homologues BDNF, NT-3, and NT-4 and their respective cognate receptors, TrkA, TrkB, and TrkC in the 1990s, it has been well established that this family is essential for the survival and maintenance of many different neuronal populations in both the ANS and throughout the nervous system.⁷⁴ The neurotrophin “hypothesis” states that these neurotrophins are synthesized by the target tissues and are of limited supply. As axons invade their targets, those axons that successfully bind and endocytose neurotrophins survive, due to activation of a signaling cascade that travels retrogradely back to the nucleus, while those that are unsuccessful in this process die by apoptosis.⁷⁵ Thus, in mice that are null for NGF or its receptor TrkA, virtually all of the sympathetic neurons die before birth.^76,77 NGF is required not only for the survival of sympathetic neurons but also for target innervation.⁷⁸ Prior to their dependency on NGF, neurons in the SCG depend on the growth factor GDNF for survival and target innervation.^57,79 Deletion of the GDNF receptor, Ret, causes a significant reduction in sympathetic neurons in the SCG, without any reduction of neurons in the paravertebral chain of sympathetic ganglia.^65,80 This finding indicates that not all sympathetic neurons share the same growth factor dependencies. Reduction of the Ret coreceptor Gfrα3 in mice also causes a reduction of neurons in the SCG⁸¹; in this case, the critical signaling is via the ligand artemin. Wnt ligands and Frizzled receptors are also required for target innervation by sympathetic axons.^82,83

Development of the Carotid Body

The carotid body is a chemosensory organ that detects hypoxia and hypercapnia, and in response triggers homeostatic respiratory responses.^84–86 It derives entirely from the neural crest,^45,87,88 and is composed of neuroendocrine glomus cells (type I) that express tyrosine hydroxylase and secrete dopamine, and sustentacular cells (type II) which are glial-like. Interestingly, in the adult, sustentacular cells can act as neural stem cells and replace damaged glomus cells in acute and chronic hypoxic conditions.⁸⁹ Recently, a novel population of neural stem cells have been identified that can proliferate and differentiate very rapidly into oxygen-sensitive glomus cells in response to hypoxemia detection by mature glomus cells.⁹⁰ This “fast neurogenesis” is triggered by release of ATP and acetylcholine from mature glomus cells. The transcription factor, Hif2α, is required for normal development of glomus cells; in its absence, glomus cells do not develop and mice lack adaptive physiological responses to hypoxia.⁹¹ Like sympathetic neurons, glomus cell development also depends on Phox2b, Ascl1, and Hand2, but unlike sympathetic neurons, glomus cells do not depend on Ret.^45,46 There are two sources of precursor cells that generate the carotid body: one source is neural crest cells that first form the SCG and then migrate into the carotid body anlagen.^88,92 A second source was discovered recently using genetic lineage tracing, and showed that SCPs located in nearby nerves and/or ganglia also contribute to the glomus cell population, reminiscent of how parasympathetic ganglia form.⁹³

Development of Adrenal Chromaffin Cells

The adrenal medulla contains two types of neuroendocrine chromaffin cells, one that is primarily adrenergic and another that is primarily noradrenergic. When stimulated by preganglionic sympathetic axons, catecholamines are released directly into the blood stream where they regulate the body’s systemic stress response and alter metabolism. Through a series of recent studies, it has become clear that chromaffin cells in the adrenal medulla derive from two distinct neural crest lineages: one source is SCPs that are associated with the preganglionic sympathetic nerve fibers. These SCPs, in association with the preganglionic axons, penetrate into the developing adrenal gland and generate the majority of adrenal chromaffin cells,^22,94 similar to what was observed for the formation of parasympathetic ganglia.^19,20 If those preganglionic neurons were prevented from innervating the adrenal medulla, the number of chromaffin cells was reduced by nearly 80%.²² These preganglionic axons and chromaffin cell precursors require class 3 semaphorin (SEMA3) signaling through neuropilins (NRP) to target the adrenal medulla.⁹⁴

The second neural crest source of adrenal medulla chromaffin cells comes from a subset of neural crest-derived sympathetic precursor cells that are already expressing tyrosine hydroxylase, migrate ventrolaterally beyond the dorsal aorta, and condense centrally within the aggregate of developing adrenal cortical cells (see Fig. 1).^40,41,95 They then begin differentiating into chromaffin cells, with a subset turning on PNMT and expressing adrenalin.^96,97 Like sympathetic neurons, adrenal chromaffin development is dependent on BMP signaling, Phox2b, and Mash1/Ascl.^41,44,46

Neural crest-derived glial cells in the form of sustentacular cells are also present in the adrenal medulla and may act as stem-like cells.⁹⁸ Two peripheral cancers, neuroblastoma and pheochromocytoma, most often originate from chromaffin cells in the adrenal medulla.⁹⁹ Recently, a network of imprinted, regulatory genes have been identified that are highly enriched in developing adrenal chromaffin cells as opposed to sympathetic neuroblasts. These new data have promise to yield interesting insights into not only chromaffin cell development but also how dysregulation of this network can trigger chromaffin cell-derived cancers.⁹⁶

Development of the Enteric Nervous System

A major component of the ANS is the ENS, which is essential for digestion, transport, absorption of nutrients and fluids, and excretion. It also has protective functions by monitoring luminal contents for potential ingested toxins and pathogens. The human ENS is composed of between 400 and 600 million neurons¹⁰⁰ and, if devoid of any extrinsic input, can function independently as the so-called second brain.¹⁰¹ The ENS is organized in two concentric networks of ganglionated plexi, the myenteric (Auerbach) and submucosal plexi that contain both neurons and glia. As with another major neural crest derivative (the DRG), the ENS cell populations are very heterogeneous, with some having motor function—excitatory or inhibitory—while others are interneurons and sensory neurons. Together, these neural networks ensure a coordinated series of muscle contractions for polarized movement of gut contents down the gastrointestinal tract, in addition to organizing secretion, blood flow, fluid and nutrient absorption, and for mediating homeostatic interactions with the intestinal immune and endocrine systems.^102,103

The ENS also receives extensive extrinsic input from the vagus nerve, DRG, and from postganglionic sympathetic nerves. The extrinsic ANS input serves to coordinate gut function and activity with the overall homeostatic state of the organism. Vagal afferents are essential for conveying afferent information concerning luminal metabolites, irritants, and chemosensory stimuli to the brain. The vagal (parasympathetic) efferents serve a key role in regulating digestion, while the sympathetic input modulates fluid transport across the intestinal epithelium, causes vasodilation in the gut, and regulates hormone secretion by the enteroendocrine cells.¹⁰⁴ The molecular and cellular mechanisms that mediate the emergence of a functional ENS that is innervated and synaptically connected to modulatory extrinsic inputs is a challenging problem that remains unresolved.

The ENS is derived from two distinct neural crest lineages. One source of neural crest cells that forms the ENS is vagal (region) neural crest cells that migrate into the foregut and colonize the entire developing GI tract²⁹ (see Fig. 1). Interestingly, while the majority of trunk neural crest cells are repulsed from entering the foregut, vagal neural crest cells are not.²⁸ These neural crest cells then migrate along the entire gut, colonizing it. A second neural crest cell immigration into the developing gut occurs from the sacral region, and these cells migrate into the hindgut.^105,106 The first ENS neurons to form are in the developing myenteric plexus, and beginning in the midgut region, these cells undergo a secondary migration toward the gut lumen to form the submucosal plexus.¹⁰⁷

The other major source of ENS neurons is SCPs that colonize vagus nerve axons as they innervate the esophagus and stomach.²⁰ If the vagus nerve is dramatically reduced in size, then the number of ENS neurons in the esophagus is correspondingly reduced. This coordination is a striking example of developmental convergence from three tissues of distinct origin: neural crest, placode-derived neurons, and CNS neurons. After birth, there is second caudal immigration of SCPs into the GI tract that accompany the developing pelvic and mesentery nerves and contribute up to 20% of neurons in the colon.¹⁷

Molecular mechanisms mediating ENS development:

The molecular and cellular mechanisms that mediate the generation and survival of the ENS have been extensively studied, although numerous intriguing and important questions remain. BMP-2 and −4 are essential for the migration, differentiation, and maturation of enteric ganglia,^108–112 and can also influence the specification of distinct subtypes of ENS neurons,¹¹³ as can the transcription factor Ascl1.¹¹⁴ The correct spatial-temporal patterning of guidance molecules, in particular extracellular matrix molecules, is essential for normal migration of enteric neural crest cells.^115–117 The neural crest cell cohorts that enter the gut via the rostral versus caudal ends exhibit behavioral differences, with the former being much more invasive and expressing higher levels of the Ret receptor than the latter.¹¹⁸ GDNF is the ligand for Ret, and its coreceptor for the ENS is Gfrα1; activation of this cascade is essential for the survival, proliferation, and migration of the enteric neural crest cells.^118–122 Deletion of Ret, Gfrα1, or GDNF in mice results in aganglionosis in the distal colon, the classic hallmark of Hirschsprung disease.^122–129 Distal aganglionosis is the result of insufficient proliferation and survival of ENS progenitor cells. Of clinical interest, colorectal aganglionosis results from a reduction of Ret to one-third of normal levels, indicating that the level of Ret protein dictates the behavior of ENS progenitor cells.^29,129 Like sympathetic neurons, ENS neurons die when deprived of signaling through the tyrosine kinase receptor ErbB3.²¹ The innervation of the gut by the vagal axons is itself dependent on both attractive and repulsive molecular interactions with both the surrounding mesenchymal tissue and the developing enteric neurons.^61,130,131 Another signaling pathway that is required for normal proliferation of ENS progenitor cells is endothelin3/EdnrB^132–134—deletion of the endothelin 3 gene leads to colorectal aganglionosis (discussed later). Once enteric neurons and glia differentiate, subsets of enteric neurons have been shown to also depend on NT3-TrkC signaling via interactions with the BMP-4 signaling pathways for their differentiation and survival.¹³⁵

Neurons in the adult ENS have been shown to die in response to luminal insults—hence, if homeostasis is to be maintained, neurogenesis in the adult should occur. A few studies have provided evidence for regeneration of neurons in the adult ENS: Gershon’s group showed that serotonin, via activation of the serotonin 5-HT4 receptor, can stimulate neurogenesis in the ENS of adult mice.¹³⁶ ENS glial cells can be recruited to generate neurons in response to neuronal injury and loss.^137,138 Most recently, a study identified a population of enteric neuron precursor cells that can generate new neurons in vivo in mice,¹³⁹ and it will be of great interest to determine their future clinical potential.

When Autonomic Nervous System Development Goes Awry: Clinical Implications

Identification of the essential molecular and cellular interactions that mediate normal development of the ANS can inform our understanding of the etiology and pathophysiological processes that underlie and trigger ANS disorders (Table 1).

Table 1.

Genes associated with autonomic nervous system disorders

Disease	Gene(s) associated with the disease	Major nervous system phenotypes	References
Familial dysautonomia	ELPI (IKBKAP)	Developmental loss of sensory, sympathetic, parasympathetic, enteric neurons, and progressive loss of retinal ganglion cells	144−159
Rett syndrome	MeCP2	Autism, motorabnormalities, autonomic dysfunction, in particular respiratory dysfunction due to decreased BDNF	160−162
Hirschsprung disease	RET (50–80% of cases); SEMA3; EDNRB; NRG1	Impaired enteric nervous system development, particularly distal colon agangliogenesis	140−143
Congenital central hypoventilation syndrome	PHOX2B	Impaired autonomic respiratory control due to faulty development of carotid bodies, petrosal ganglia, and/or NTS	163,164

Hirschsprung Disease

Studies on the ENS have identified several signaling pathways that cause Hirschsprung disease, a common genetic enteric neuropathy that affects 1 in 5,000 children and is marked by colorectal agenesis of enteric ganglia.⁹ Genome-wide association studies have linked Hirschsprung disease with several of the key genes that have been identified as vital for development, with about half of cases being associated with mutations in the neurotrophic factor receptor, RET. In fact, recent data indicate that nearly 80% of patients with Hirschsprung disease may harbor mutations in genes within the RET regulatory network,¹⁴⁰ indicating the importance of screening patients for mutations in this gene network. Other mutations associated with Hirschsprung disease include Neuregulin-1, the endothelin 3 receptor, EDNRb, and an axonal guidance molecule SEMA3.¹⁴¹ Intriguingly, the gene responsible for familial dysautonomia, ELP1, has also been linked to Hirschsprung disease in a cohort of 173 Chinese patients,¹⁴² and knock-down of elp1 in zebrafish embryos causes aganglionosis and a reduction in the number of ENS neurons—hallmark features of Hirschsprung disease.¹⁴³

Familial Dysautonomia

Familial dysautonomia or Riley-Day syndrome (OMIM 223900) is a genetic sensory and autonomic neuropathy that results from impaired neural development.¹⁴⁴ Clinical hallmarks of the disease include cardiovascular instability due to dysfunctional baro- and chemoreceptors, decreased pain and temperature sensation, and organ failure including faulty gastrointestinal tract function. Patients typically die by the age of 40 from sudden unexpected death during sleep,¹⁴⁵ most likely due to blunted respiratory and cardiovascular homeostatic activity. Elegant studies have demonstrated that while sympathetic output is not normal in patients with familial dysautonomia, some of the most severe clinical manifestations of the disease are the direct result of failure of afferent input.^146,147 As with Hirschsprung disease, animal models have proven invaluable for the identification of the molecular and cellular developmental mechanisms that are impaired in familial dysautonomia (FD). The disease results from a point mutation in the gene ELP1 (also called IKBKAP), with over 99.5% of patients having a conserved (c.2204 + 6T > C) mutation.^148,149 This mutation results in skipping of exon 20 that varies with tissue type, causing nonsense-mediated decay of the truncated mRNA. Elp1 is the scaffolding subunit of the six-subunit Elongator complex that is essential for modification of wobble uridines in tRNA, and in its absence, protein expression is abnormal.^150,151 Because the neural crest gives rise to the majority of the trunk ANS, the first studies in mice investigated whether the clinical phenotypes of the disease resulted from impaired neural crest cell specification and/or migration. Using mice in which Elp1 was conditionally deleted from the neural crest, the data showed that, in fact, neural crest cells do not depend on Elp1 for their specification and/or migration in the trunk, in that neural crest cells successfully migrated to and colonized all of their normal targets.^152,153 By the end of development, though, mice were born with reduced numbers of sensory, sympathetic, and parasympathetic neurons^152–155 due to reduced proliferation and increased death of mitotically active progenitor cells as a consequence of faulty DNA repair and apoptosis.^150,152 In addition, sensory neurons lacking Elp1 undergo exacerbated programmed cell death, most likely due to impaired retrograde transport of the critical survival factor, NGF.^156–158 Furthermore, by studying the cell biology of developing neurons in FD mouse models, it has become clear that in the absence of Elp1, neurons undergo intracellular stress, marked by elevated p53, increased pJNK, depolarized mitochondria, increased levels of reactive oxygen species, and impaired mitochondrial function.^152,159

Rett Syndrome

Rett syndrome is a developmental disorder that is marked by autism and autonomic dysfunction, including respiratory abnormalities, and results from mutations in the methyl-CpG-binding protein 2 (MeCP2) gene. The cardiorespiratory dysfunction, marked by impaired homeostatic responses, is due to faulty development of the baroreceptors in the nodose ganglion and the brainstem centers in which those neurons synapse.¹⁶⁰ Interestingly, mouse models for Rett syndrome have revealed that the developmental dysfunction of the cardiorespiratory circuitry is due to altered BDNF expression in the absence of MeCP2, and that elevating BDNF levels can be ameliorative,¹⁶¹ as are small molecules that activate the BDNF receptor, TrkB.¹⁶² These findings are hopeful in that there has long been an interest in manipulating neurotrophin levels and/or their Trk receptors for treating neurological disorders.⁷⁴

Congenital Central Hypoventilation Syndrome

Congenital central hypoventilation syndrome (CCHS) or “Ondine’s curse” is due to mutations in the gene encoding the transcription factor PHOX2b and causes faulty autonomic respiratory control with central hypoventilation due to the deficient response of chemoreceptors to hypercapnia and hypoxia. Postmortem analysis has revealed abnormal carotid bodies in two patients,¹⁶³ and the syndrome is thought to be caused by developmental impairment of the carotid bodies, the petrosal ganglion (that innervates the carotid bodies) and/or the NTS in which the visceral sensory afferents synapse in the brainstem. All of these structures require the transcription factor Phox2b for their survival,⁴⁵ and mouse studies have shown that mice lacking one allele of Phox2b have an impaired response to hypoxia and hypercapnia and exhibit dysfunctional respiration that is reminiscent of CCHS. Furthermore, mice that express a Phox2b mutation that is naturally found in patients with CCHS exhibit breathing irregularities, do not respond to increases in carbon dioxide, and die soon after birth due to central apnea.¹⁶⁴

Conclusions and Future Therapeutic Developments

The promise of treating neurodegenerative disease with stem cells is fueling many new approaches for therapeutics. This holds true for the ANS, with a “white paper” being published in 2016 on the potential use of stem cell therapy for treating ENS disease.¹⁶⁵ Depending on the cell type to be replaced in the ANS, stem cells must be purified, expanded, and differentiated into the appropriate cell type, then transplanted into the appropriate ganglion and/or visceral organ, and new functional circuits established. Given that numerous ANS tissues do contain stem cells, and in light of the advances the field has made in identifying the key transcription factors, growth factors, and axonal guidance molecules required for target innervation, these preclinical studies can be conducted now in animal models for various ANS disorders. In addition, with the identification of the genetic pathways that are mutated to cause such disorders as FD, Rett syndrome, Hirschsprung disease, and CCHS, the genetic mutations causing these disorders can be corrected in patient stem cells with gene editing tools such as CRISPR/Cas9, or gene-replacement using adeno-associated viruses. In summary, the time has come to take advantage of the information derived from studying the development of the ANS to curing diseases of the ANS.