Development and Regeneration of the Vagus Nerve

Abstract

The vagus nerve, with its myriad constituent axon branches and innervation targets, has long been a model of anatomical complexity in the nervous system. The branched architecture of the vagus nerve is now appreciated to be highly organized around the topographic and/or molecular identities of the neurons that innervate each target tissue. However, we are only just beginning to understand the developmental mechanisms by which heterogeneous vagus neuron identity is specified, patterned, and used to guide the axons of particular neurons to particular targets. Here, we summarize our current understanding of the complex topographic and molecular organization of the vagus nerve, the developmental basis of neuron specification and patterned axon guidance that supports this organization, and the regenerative mechanisms that promote, or inhibit, the restoration of vagus nerve organization after nerve damage. Finally, we highlight key unanswered questions in these areas and discuss potential strategies to address these questions.

1). Introduction

For centuries, the vagus nerve has been revered as an anatomical wonder for its structural complexity. It extends a multitude of tendrils that reach from the ear to the colon and touch nearly every organ in between. As the most significant component of the parasympathetic nervous system, the vagus nerve is a major sensor of internal organ state and reflexively controls key bodily functions such as swallowing, coughing, speech, vomiting, respiration, blood pressure, heart rate, digestion, metabolism, sweating, and exocrine, endocrine, and hormone secretion [1,2], as well as being a major mediator of the gut-brain axis [3] and the neuroimmune axis [4]. But from this nerve’s complexity, a paradox emerges. Trace its many tendrils back to their origins, and you’ll find that they emerge from only a few small clusters of neurons. How do these few clusters manage the task of accurately controlling a much larger set of tissues and functions? Recent advances in single-cell transcriptomics have revealed a stunning molecular heterogeneity within these clusters and have begun to unravel the correspondence between a vagus neuron’s molecular identity and its tissue target and functional role, leading to an understanding that the nerve’s multifunctionality arises from the organization of its neurons into topographically and/or molecularly distinct subgroups [5–13]. The next major milestone in vagus biology will be to discover the mechanisms by which the molecular heterogeneity between vagus neurons is translated to promote the organized construction of the nerve’s complex anatomical and functional circuitry, a challenge which lies squarely in the realm of developmental biology. In this review, we explore the current understanding of the genetic and molecular basis of vagus circuit construction in the contexts of embryonic development and regeneration as revealed in fish, avian, rodent, and human models, and discuss opportunities and challenges for the future of vagus neurodevelopment.

2). Vagus anatomical & molecular organization

The precisely organized wiring of neurons into functional circuitry is crucial across the nervous system. This is nowhere more true than in the vagus nerve, whose neurons must integrate into a large number of circuits to support its myriad functions, and whose miswiring can result in disruptions to, and/or aberrant crosstalk between, circuits controlling crucial functions [14,15]. The organization of the vagus nerve can be considered at two scales. On a gross scale, the vagus is a bilateral mixed nerve containing efferent motor and afferent sensory fibers whose cell bodies reside in two brainstem motor nuclei and two peripheral sensory ganglia on each side of the body and extend branches to various tissues (Fig. 1). On a finer scale, there is clear evidence that these nuclei and ganglia are further organized into topographically and/or molecularly distinct subgroups which exhibit differing functional and wiring properties and whose diversity underlies the nerve’s extreme multifunctionality. Here we summarize the gross and fine organization of this nerve, an understanding of which is crucial for considering the mechanisms of its development and regeneration.

Figure 1. Vagus nerve anatomy in mouse and zebrafish.

(A) Mouse and (B) zebrafish vagus motor nuclei, sensory ganglia, and axons (blue), and innervated organs.

2.1). Gross anatomy – vagus nuclei and ganglia and their targets

2.1.1). Gross motor anatomy

The motor component of the vagus nerve is composed of two axially elongated nuclei located in the medulla – the dorsal motor nucleus of the vagus (DMV) and the nucleus ambiguus (nAMB). While all vagus motor neurons are cholinergic, several features distinguish these nuclei. Firstly, they occupy distinct positions within the medulla, with DMV localized dorso-medially and nAMB localized more ventrally within each hemisphere [16]. Secondly, they contain distinct types of efferent fibers – DMV exclusively extends general visceral efferent fibers that provide parasympathetic preganglionic innervation, while nAMB primarily extends special visceral efferent fibers that provide direct muscle innervation, although it also extends some preganglionic fibers to the heart and lungs [1,7]. Thirdly, these nuclei innervate targets in distinct axial regions. DMV extends axon branches to more posterior targets, including the smooth muscles of the esophagus, the heart, the lungs, the stomach, the small and large intestines, the gallbladder, the liver, the pancreas, the kidneys, the spleen, the thymus, and the adrenal gland [1]. nAMB extends axon branches to more anterior targets, including the muscle derivatives of pharyngeal arches (PAs) 4 and 6 (muscles of the soft palate, pharynx, and larynx), as well as to the skeletal muscles of the esophagus and to the heart and lungs [1,7].

2.1.2). Gross Sensory Anatomy

The neurons that contribute the sensory function of the vagus nerve are located in two adjacent peripheral ganglia – the larger nodose (aka inferior) ganglion and the smaller jugular (aka superior) ganglion – situated just outside the jugular foramen. The peripheral targeting of the sensory ganglia generally follows a similar anterior-posterior pattern to that of the motor nuclei, allowing accurate reflexive responses to interoceptive stimuli. The nodose ganglion generally extends axon branches to more posterior targets, including the taste buds of the epiglottis, the esophagus, the distal/lower airways including the lungs, the heart, the stomach, the intestines, the pancreas, the liver, and abdominal blood vessels [1,2]. Nodose neurons are correspondingly diverse, being tuned to detect stretch, temperature, pressure and a variety of nutrients and inflammatory chemicals [9]. The jugular ganglion generally innervates more anterior targets, including the external auditory canal, the proximal/upper airways (pharynx, larynx, trachea, mainstem bronchi) and the heart [1,2]. Its sensory modalities are less diverse, reflecting the largely somatosensory role of these branches of the vagus nerve [9].

2.1.3). Comparative gross anatomy

This section has thus far focused on the anatomy of the tetrapod vagus nerve. However, some of the most detailed studies of vagus nerve development and regeneration have been performed in zebrafish. While vagus anatomy and function are generally well-conserved between zebrafish and tetrapods, there are some key distinctions between these groups (Fig. 1). Most notable are the differences in organization of the motor nuclei and sensory ganglia. While tetrapods have two motor nuclei, zebrafish have one (termed the vagus motor nucleus) [17,18]. However, anatomical and functional analogy to the two tetrapod nuclei is apparent: roughly the anterior half of the zebrafish vagus motor nucleus corresponds to nAMB, providing motor innervation to the muscle derivatives of the posterior pharyngeal arches (PAs4, 5, 6, and 7) as well as to the gills and heart, while roughly the posterior half of the vagus motor nucleus corresponds to DMV, extending a visceral branch that provides preganglionic parasympathetic innervation to visceral organs [5]. While the sensory nodose ganglion is present and, to the best of our knowledge, anatomically and functionally conserved between zebrafish and tetrapods, the jugular ganglion is absent in zebrafish. In its place are several small ganglia, termed the vagal epibranchial ganglia, which are present at the base of PAs 4, 5, and 6 and innervate the derivatives of these structures as well as the heart [19].

2.2). Fine anatomy - topographic and molecular organization

A key question that has long motivated interest in the vagus nerve is how a single nerve can accurately and reliably control so many distinct tissues and functions within the body. This requires a reliable strategy to delegate different roles to different neurons within a nucleus or ganglion. A common organizational solution to this challenge is topography, in which a group of neurons is patterned such that the position of a neuron corresponds to some feature of its target or function. This concept has been clearly demonstrated in many systems, including the primary somatosensory cortex [20] and the visual system [21]. The vagus nerve, especially the motor nuclei, exhibit topographic organization that is constructed around multiple target features:

1. Spatial topography. The subdivision of the motor nuclei and sensory ganglia into anterior-targeting (nAMB and jugular) and posterior-targeting (DMV and nodose) represents a coarse form of such organization, however individual motor nuclei display further spatial topography. This is most clearly demonstrated in the zebrafish vagus motor nucleus, in which single-branch and single-neuron mapping has revealed that the distribution of axons to the five primary vagus axon branches is organized along the anterior-posterior (A-P) axis such that more anterior neurons innervate more anterior targets and more posterior neurons innervate more posterior targets [5,22]. This topography appears to be at least partially conserved – in the rat nAMB, for example, neurons innervating PA4 derivatives via the superior laryngeal nerve are more anterior to those innervating PA6 derivatives via the recurrent laryngeal nerve [12]. Spatial distinctions between nAMB neurons innervating the larynx, pharynx, trachea, and esophagus have also been reported in mammals [6,23]. Furthermore, neurons innervating different sub-diaphragmatic organs exhibit viscerotopy along the mediolateral axis of the rat DMV, with a stark left-right asymmetry reflecting internal organ asymmetry [24,25].

2. Functional topography. This, too, is represented in the subdivision of the motor component into muscle-innervating (nAMB) and preganglionic (DMV) nuclei. Additional, finer-scale examples of functional topography have also been reported in mice. For instance, two topographically distinct groups of DMV neurons, both of which innervate the glandular stomach, were found to target distinct subtypes of enteric neurons that regulate gastric contraction and relaxation, respectively [8]. Likewise, two topographically distinct groups of heart-innervating neurons in the nAMB mediate distinct reflexive behaviors - the dive reflex and the baroreflex – respectively [7].

Surprisingly, the sensory ganglia exhibit a clear absence of topography; in the nodose, in particular, neurons innervating individual visceral organs show a scattered “salt-and pepper” distribution [11,13]. A lack of ganglionic topography, however, does not indicate a lack of organization as two other forms of organization appear in the vagus sensory system. First, retrograde labeling and functional imaging studies have revealed that spatially distinct regions of the nucleus of the solitary tract, a brainstem region that receives afferent vagal signals, are responsive to sensory signals from distinct peripheral organs [13,26,27]. Thus, vagus sensory and peripheral targets appear to maintain some viscerotopy despite being separated by a disorganized intermediary (the vagus ganglia). Second, it has been shown in multiple systems, including the olfactory nerve [28,29], that molecular differences between neurons can direct them to distinct targets even in the absence of spatial organization. In recent years, the rise of single-cell RNA sequencing, which is ideal for the identification of molecularly but not necessarily spatially distinct groups of cells, has revealed molecular codes that distinguish vagus sensory neurons analogous to the olfactory receptor codes that distinguish olfactory neurons [6–11]. Although spatially intermingled within the nodose ganglion, vagus sensory neurons innervating distinct organs, tissue layers within organs, and different sensory stimuli within a tissue layer can all be readily distinguished at the transcriptional level [9–11]. Many more molecular subgroups have been identified in both vagus motor nuclei and sensory ganglia but have not yet been anatomically or functionally characterized [6–11].

The diverse forms of topographic and molecular organization observed in the vagus give some clues as to the developmental mechanisms that may underly its patterning. Topography often results from the patterning of neuronal identity via extrinsic developmental signals, such as morphogens, which are inherently spatial. Non-topographic molecular patterning could indicate either that spatial signals exist but are disconnected from a neuron's final position, for instance that identity is spatially determined prior to neuronal migration but jumbled during the migration process, or that spatial signals do not exist, for instance that identity is stochastically determined, or that neurons agnostically select a target followed by identity determination via target-derived signals. The known and potential roles of these mechanisms will be further discussed in the next section.

3). Vagus nerve development

A great deal of recent progress has been made in dissecting the anatomical and molecular organization of the vagus nerve, largely via the examination of adult animals. But how is such a well-organized structure built during animal development? Development, in its most extreme distillation, is a process of cell-fate specification and morphogenesis. In this section, we examine these two developmental concepts as they relate to the vagus nerve: How do embryonic signals specify the various cell identities within the vagus, and how do morphogenetic programs, in the form of axon guidance, translate aspects of cell identity into the organized circuit architecture of the adult vagus nerve?

3.1). Vagus motor neuron specification and assembly into nuclei

Vagus motor neurons are specified in the ventral region of the posterior hindbrain neuroepithelium within rhombomere 8 via a confluence of dorsal-ventral and anterior-posterior embryonic signals. Motor neuron progenitor fate is specified ventrally by a floor plate- and notochord-derived Sonic Hedgehog gradient acting through the Gli family of zinc-finger transcription factors to drive coincident expression of the homeodomain genes Pax6 and Nkx6.1 [30–32]. Downstream activation of the key factors Isl1, Phox2a/2b, and Tbx20 defines the transition to cranial motor neuron identity, and co-expression of Isl1, Phox2a, and the proneural gene Ngn2 is sufficient to transform embryonic stem cells into a cranial motor neuron fate in vitro [33].

Within the presumptive hindbrain, opposing posterior and anterior gradients of Retinoic Acid and FGF8, respectively, establish domains of Hox gene expression to regulate both segmental hindbrain organization and motor neuron diversification along the anterior-posterior axis [34,35]. For instance, Hoxa2 expression in rhombomere 2 (r2) plays a key role in specifying trigeminal motor neuron identity, while Hoxa1/b1 expression in rhombomere 4 is needed to correctly specify facial motor neuron identity [35–38]. The Hox code responsible for specifying motor neurons as vagal has not been explicitly examined, though Hox4 genes, which have an anterior limit at the r6/r7 boundary, likely play an important role [39,40]. Hox5 genes, which have an anterior limit within r8, also appear to play a key role in vagus motor identity – in zebrafish, the Hox5 expression boundary sits right in the middle of the vagus motor nucleus, effectively separating the nucleus into an anterior Hox5-negative domain and a posterior Hox5-positive domain, and ectopic expression of Hoxb5a in anterior vagus motor neurons transforms them to a posterior vagus motor identity as defined by axon targeting to the heart or viscera [5]. This phenomenon is likely conserved, as Hox5 genes have a similar expression boundary in the avian DMV [40], although their function has not been examined in this context.

Once specified, vagus motor neuron precursors delaminate from the neuroepithelium and migrate dorsally to the site of nucleus formation [16,41]. This process is poorly understood, but migration relies on the fucosyltranferase Gmds [41] and nucleus coalescence likely relies on cadherin-mediated adhesion, as has been reported in facial and abducens motor nuclei [42]. Neurons of the nAMB and DMV are closely related – they initially migrate together to form a single nucleus which soon thereafter separates along the dorsal-ventral axis via an unknown mechanism to form two discrete nuclei [16,43].

3.2). Vagus sensory neuron specification and assembly into ganglia

The two sensory ganglia, nodose and jugular, are derived from distinct sources within the embryo – epibranchial placodes and neural crest, respectively, [44] – which are initially specified in adjacent regions at the neural plate border by an interplay of Fgf, Wnt, and BMP signaling [45,46]. After neural tube closure, the placodal ectoderm becomes situated dorsal to the pharyngeal pouches. Several distinct placodes are specified along the anterior-posterior axis via signaling from adjacent body regions; FGF signaling from adjacent cephalic mesoderm plays a key role in specifying the epibranchial placode [19], which is subsequently induced to undergo neurogenesis by Bmp7 signaling from adjacent pharyngeal endoderm [47,48]. This is followed by the delamination and dorsal migration of neuronal precursors from the most posterior epibranchial placode, termed the nodose placode, to generate the nodose ganglion [46].

Meanwhile, cranial neural crest cells are specified in the dorsal hindbrain and migrate in streams into the head periphery, carrying their Hox code with them [49,50]. Jugular neurons come not from the so-called vagal neural crest, which contributes to the enteric nervous system, but rather from posterior cephalic neural crest [49,50]. Jugular sensory precursors migrate ventrally and differentiate as jugular sensory neurons after reaching their destination [51]. Due to their distinct origins, jugular and nodose neurons can be molecularly distinguished by expression of the homeodomain transcription factor Phox2b (specific to nodose) and the zinc-finger transcription factor Prdm12 (specific to jugular) [9].

The developmental origins of the zebrafish vagal epibranchial ganglia, which subserve the PA-innervating functions of the tetrapod jugular ganglion, are less clear. They are generally considered to be primarily derived from epibranchial placodes [19,48,52], but recent evidence suggests that they may receive more substantial contributions from neural crest than previously appreciated [53]. In any case, this represents a substantial difference from the exclusively neural crest origins of the jugular ganglion, the evolutionary basis of which is intriguing but unknown.

3.3). Vagus axon guidance

The spatial and molecular patterning of neuronal identity serves to ensure that neurons have the necessary information to navigate their growing axons to the correct targets in order to establish organized circuit connectivity. Axon guidance also requires molecular distinctions in the environments through which the axons grow, which serve as navigational cues that can be differentially interpreted via molecular distinctions within the axons themselves. Here, we review what is known of how vagus axons extend into the body and how they interpret environmental signals to select the appropriate innervation targets.

3.3.1). Axon guidance signals

Target-derived attractive and repulsive cues commonly guide growing axons. Several major classes of axon guidance signals have been implicated in vagus axon guidance to various tissues. The clearest data on how axon guidance signals support the topographic organization of the vagus nerve comes from studies of zebrafish vagus motor innervation of PA-derived muscles [5,22]. Topographic axon guidance is generally considered to be regulated by an underlying spatial pattern of cell identity that matches a spatial pattern in the target field [54]. Paradoxically, zebrafish vagus motor axon topography is not regulated by innate spatial identity factors. In fact, these neurons are all guided by a common chemoattractant-receptor pair – Hgf and Met, respectively – and are broadly promiscuous regarding which innervation targets they can select. Specificity in axon target selection instead results from an unusual temporal mechanism in which an anterior-to-posterior wave of innervation competence (in the form of Met expression) passes across the vagus motor nucleus with a timing that matches an anterior-to-posterior wave of Hgf expression in the forming pharyngeal arches [22]. Consequently, manipulating either the timing of axon outgrowth or the timing of Met expression is sufficient to alter a neuron’s target selection [5,22]. This mechanism is likely conserved between zebrafish and mammals, as the expression of Met in vagus motor neurons [55,56], the role of the Met ligand Hgf as a cranial motor axon chemoattractant [57], and the anterior-posterior sequence of PA development [58] are all conserved features.

In mammals, the Netrin-DCC signaling pathway has been implicated in the guidance of nodose sensory axons to the gut wall. Enteric neurons and the outer mesenchyme and mucosa of the foregut, which are targeted by nodose sensory axons, express the guidance molecule Netrin-1, which attracts the axons of nodose neurons that express the netrin receptor DCC [59–61]. Additionally, Laminin-111, a component of the extracellular matrix, is present in the regions of the gut where nodose axons terminate; its presence can switch Netrin from a nodose axon attractant to repellent in vitro, suggesting that it may ensure the appropriate termination of nodose axon growth upon reaching their targets [62]. Thus, Netrin-DCC has been proposed to act as a dual-purpose signal, first as a long-range cue to attract vagus sensory axons to the foregut, and then as a short-range cue to keep them there. Netrin-1 and Laminin-111 also have intriguing spatiotemporal localization patterns in developing laryngeal muscles, suggesting they may also play a key role in organizing laryngeal motor innervation by nAMB neurons [63], although their function in this context has not been examined. Repulsive Slit-Robo signaling may also contribute to vagus axon guidance in the gut. Some regions of the gut express the repulsive guidance molecules Slit1, Slit2, and Slit3, and nodose neurons express the slit receptors Robo1 and Robo2 and can be repelled by Slit-expressing cells in vitro, although the relevance of this interaction in vivo remains to be determined [64]. Determining how many, and which, nodose neurons express DCC and Robo will be an important next step in refining our understanding of how these guidance molecules contribute to nerve organization.

Finally, Semaphorin signaling has been implicated in vagus axon guidance in mammals, although its role is less clearly defined. Neuropilin-1 and Plexin-A4, co-receptors for repulsive class III Semaphorins, are expressed in vagus sensory and motor neurons [65,66]. Exposure of cultured vagus sensory neurons to the Semaphorin 3A ligand causes growth cone collapse [67], and vagus axons appear to overgrow and target a broader area in Neuropilin-1, Plexin-A4, and Semaphorin 3A mutant mice [65,68–70]. Neuropilin-1 and Plexin-A4 also regulate vagus axon fasciculation at least partially independently of Semaphorin [69–71]; the extent to which the fasciculation defect causes the guidance defect is unclear.

3.3.2). Axon-axon interactions

In cases where several axons grow to the same target, interactions between them are often important for guidance. The most well-known such phenomenon is the pioneer-follower model, in which the first “pioneer” axon(s) to extend play a pathfinding role, while later extending “follower” axons fasciculate with the pioneer and follow it to the target [72]. In the case of motor axon guidance to the PAs, there is no evidence for such interactions; rather, each axon appears to independently select its target based on the timing of Met expression and axon outgrowth [5,22]. However, in other contexts, axon-axon interactions, particularly between sensory and motor axons, which are intermingled within vagus fascicles, are required for guidance. In zebrafish, the extension of vagus motor axons to PA targets precedes, and is independent of, sensory axon growth; however, vagus sensory central processes require motor axons to reach the brain [73], possibly in a N-Cadherin-dependent manner [74]. The relationship between motor and sensory axons is reversed in the innervation of the mouse gut, where vagus peripheral sensory processes extend earlier and provide a substrate for motor axon growth [75]. Because these studies examined distinct vagus branches in zebrafish and mice, it is unclear whether different leader-follower relationships are species-specific or branch-specific. It is also unclear whether the pioneering axons provide target-specific information to followers, and therefore represent an axon sorting signal, or whether they provide a generic growth substrate, therefore still requiring the follower axons to interpret target-derived guidance signals to reach the correct targets.

3.4). Conclusions and future directions

While we have learned a great deal about the specification and axon guidance signals that support the developmental patterning of the vagus nerve, our current understanding of these processes does not fully account for the breadth or specificity of vagus nerve organization, indicating that there remains much to learn about the mechanisms that allow a given vagus neuron to find the correct target. Because we know that there is, in many cases, a correspondence between a neuron’s spatial and/or molecular identity and its target and/or function, and because logic dictates that axons’ abilities to discriminate between targets likely requires molecular distinctions between those axons, it is reasonable to assume that specificity in axon guidance relies on the specification of a diversity of vagus neuron subtypes fated to innervate particular targets and carry out specific functions. It is also tempting to assume that the extent of molecular heterogeneity discovered in recent single-cell sequencing experiments supports this notion. But because this data is derived from adult neurons, there is no assurance that any differential gene expression identified in these studies is present during, and therefore relevant to, the period during which vagus neurons are finding their targets. This is particularly important in light of the evidence that, while molecular heterogeneity often precedes, and informs, axon guidance and circuit construction [28,76,77], it is also true that the selection of a particular target can cause subsequent changes in a neuron’s molecular identity [78,79]. Thus, a major step forward in this field will result from efforts to characterize the emergence of molecular heterogeneity of vagus neurons during embryo development. Such data both will reveal candidate factors that may regulate the specification and guidance of neurons into organized circuitry and will provide the knowledge needed to develop genetic tools to visualize and manipulate specific subgroups of vagus neurons in order to better define vagus nerve organization and rigorously test the roles of these candidate factors in its development.

4). Vagus nerve regeneration

Regeneration constitutes the regrowth and repatterning of damaged tissues after injury or disease. In the context of nerve regeneration, injured neurons must survive the insult, reestablish a growth cone, and regrow axons and/or dendrites to the correct targets to ensure properly organized circuit connectivity. While this process is in many ways similar to development, there are also critical differences. In contrast to the stereotypy of development, regeneration must be sufficiently flexible to proceed from a diversity of starting points, based on the nature and extent of the damage, which all converge on the proper pattern. Additionally, major contextual differences exist between the embryonic and regenerative environments through which axons must grow, including the maturity of surrounding tissues, the presence or absence of growth and patterning signals, the maturity of the neurons themselves and their ability to respond to such signals, as well as, in many cases, the presence of a wound, with its corresponding disruptions in tissue/ECM structure, inflammation, and scarring, through which a regenerating nerve must navigate. Because of these distinctions, and the challenges they present to the prospect of promoting developmental processes far from the idealized environment of the embryo, scientists have long wondered about the extent to which nerves are even capable of regenerating. In many organisms, including humans, peripheral nerves can regenerate in some contexts [80], a testament to the robustness of living systems. But this regenerative capacity has its limits, and often fails due to defects in both the growth and guidance of damaged axons. While many questions remain, research has begun to reveal the signals that impede regeneration in the cases in which it fails, the mechanisms that promote regeneration in the cases in which it succeeds, and their similarities and differences to the mechanisms that promote nerve development, as well as interventions that can improve regenerative outcomes.

4.1). Regenerative capability of the vagus nerve

The history of research into vagus nerve regeneration dates to the very birth of the field of neuroregeneration [81,82]. In the first known study of its kind, in 1776 William Cruikshank, armed with the knowledge that bilateral, but not unilateral, vagotomy (severing of the vagus nerve in the neck) was lethal, surgically removed a small piece of the vagus nerve from one side of the neck of a dog, allowed the dog eight days to recover, and then severed the vagus nerve on the other side [83]. The dog died, but not nearly as quickly as normally occurs after bilateral severing, and an autopsy revealed evidence that new growth had bridged the gap left during the first surgery, thus spawning the theory that injured nerves could regenerate. Subsequent work showed that a dog could survive second-side vagus severing if the recovery period after the first sever was extended, providing the first concrete evidence for functional regeneration [84].

Many studies over the ensuing centuries have expanded our understanding of the vagus nerve’s regenerative capacity. These include studies that have observed regenerated fibers in the heart [82], the esophagus and gut [85–88], the larynx [82,89], and the liver [87,90], as well as functional recovery of the ability to effect cardiac inhibition [82,91], insulin secretion [85], gastric secretion [86], and gastric motility [86,92] upon stimulation of the severed and regenerated nerve.

Despite these successes, the above studies also reveal a clear limit to vagus regeneration. Evidence of regeneration was almost always observed in a minority of the animals tested and was, even in these animals, notably incomplete. There are also reports disputing, for instance, the recovery of cardiac fibers [93–95]. Such variability is likely due in part to the nature of the injury, including severity and proximity to the target, both factors that have been broadly implicated in regenerative success [80]. Regenerative outcomes may also be more fundamentally linked to the identities of the targets and/or the neurons, as it has also regeneration of visceral sensory fibers is reported to be more successful than that of motor fibers [87,88], perhaps due to sensory neurons’ greater ability to survive after axotomy [96–99]. Finally, even in cases with clear evidence of regeneration, the innervation pattern is often disorganized. These include inappropriate growth trajectories and ectopic localization of afferent terminal subtypes to incorrect tissues in the gut [87,88], as well as the nonspecific reinnervation of the muscles of the larynx after recurrent laryngeal nerve injury, which can lead to erroneous innervation of antagonistic muscles, synkinesis (a condition in which voluntary contraction of one muscle causes involuntary contraction of another), and vocal fold paralysis [14,100]. In all, evidence indicates that vagus nerve regeneration in mammals is rare, incomplete and non-specific, lending a need to understand the mechanisms that promote and impair this process, and how to make it a more successful one.

4.2). Axon guidance and the restoration of nerve organization

Beyond the gross ability of the vagus nerve to reinnervate some of its targets, some progress has been made in understanding the patterning of this reinnervation (i.e., whether a tissue is being reinnervated by the appropriate neurons) and the signals involved in regenerative axon guidance.

4.2.1). Mammalian regeneration

Neurotrophic factors appear to be involved in regeneration of the recurrent laryngeal nerve. In the rat, three laryngeal muscles express the neurotrophic factor glial-derived neurotrophic factor (GDNF) at distinct timepoints post-injury. nAMB neurons concurrently express GDNF receptors [101], and the timing of GDNF expression in each muscle corresponds with the timing of its reinnervation [89]. It was initially assumed that GDNF acts to promote target-specific reinnervation in this context, although this theory is rebutted by both the fact that target-specific innervation generally fails and by further complicating results. For instance, if GDNF function is blocked via antibody injection during regeneration, the reinnervation pattern is altered but vocal fold functional recovery improves relative to wild-type regeneration, suggesting that endogenous GDNF expression may actually impair reinnervation patterning [102].

Altering the endogenous expression or function of other factors, via gain- and loss-off-unction, has been similarly shown to improve regenerative outcomes. Inhibition of the Nerve growth factor (NGF) receptor Tropomyosin receptor kinase A (TrkA) at the injury site also alters the innervation pattern and improves vocal fold function [103]; increasing the expression of Vascular endothelial growth factor (Vegf) [104] or Insulin-like growth factor-1 (Igf-1) [105] also improves functional recovery, although the impact on innervation patterning is not clear. Finally, Netrin-1 and DCC are expressed in laryngeal muscles and nAMB, respectively, in a manner similar to GDNF and its receptors in response to injury. Also, as with GDNF, the endogenous Netrin-1 expression pattern is not capable of guiding proper reinnervation, and injection of Netrin-1 at the injury site changes the reinnervation pattern, although the impacts on function are not clear [101,106]. Why several factors would exhibit specific stereotyped post-injury expression patterns that at best fail to guide appropriately patterned reinnervation, and at worst promote improper innervation, is a mystery.

Exogenous administration of Netrin-1 has also been reported to enhances vagus motor reinnervation of the mouse liver following hepatectomy [90]. No endogenous role for this factor has been described for vagus axon guidance to the liver, although these findings warrant further examination.

4.2.2). Zebrafish regeneration

In zebrafish, which in general possess a greater neuroregenerative capability than mammals, vagus nerve regeneration is more successful. Vagus motor axons severed during the larval period not only reliably regrow and restore target function, but they also almost perfectly reestablish their original spatial topography, even after severe injury [107]. Thus, this model provides a valuable opportunity to elucidate the mechanisms of successful regeneration and to ultimately apply these insights to improve the regenerative process in mammals. Interestingly, target-specific regeneration relies on neither the Hgf-Met guidance signal nor the temporal signaling dynamics (described in section 3) that are responsible for establishing topography in the embryo, indicating that nerves can successfully utilize almost completely distinct patterning mechanisms during development and regeneration. This mechanistic switch corresponds with a switch in vagus motor neuron identity from a state of promiscuity regarding which targets they can innervate to a state in which neurons are determined to innervate a particular target [107]. The nature and causes of this switch are unknown, but they correlate with the completion of axon targeting, suggesting that a combination of retrograde cues, positional cues, and neuronal maturation may “lock-in” neuronal identity. Further, while the signals guiding target-specific axon regeneration have not yet been identified, reinnervation of a target requires that it had been previously innervated, suggesting that some component of the nerve branch may be the source of such cues [107].

4.3). Therapeutic interventions to improve regenerative outcomes

A major goal in regenerative medicine is to establish interventions that can improve functional outcomes after nerve injury. The motor component of the recurrent laryngeal nerve, due to its well-characterized disorganization after reinnervation (described above) and its surgical accessibility, has become a major model for therapeutic intervention. A surgical approach, termed selective reinnervation, aims to restore laryngeal function without promoting vagus regeneration per se. Variations of this technique involve a complex and challenging surgery in which the distal stumps of the injured nerve that innervate different muscles are anastomosed to other cranial nerve roots – specifically the phrenic nerve and the thyrohyoid branch of the hypoglossal nerve – to restore vocal fold function without synkinesis [108–111]. Others involve the introduction of ECM scaffolds to support axon regrowth [112,113] or, more promisingly, ECM scaffolds containing growth & guidance factors such as GDNF [114] or a TrkA inhibitor [103] to support target-specific regrowth; and gene therapy [115], including viral strategies to induce expression of neurotrophic factors including GDNF [116], Vegf [104], or Igf-1 [105]. While some success has been obtained with these approaches, a robust and reliable method to promote organized, functional regeneration remains elusive. A continued interplay between therapy development and basic research into the regenerative signaling environment is imperative to continue moving this field forward.

4.4). Conclusions and future directions

Despite the long history of vagus nerve regeneration, our understanding of the mechanistic details of this process is in its infancy. What is clear is that this is a tenuous and error-prone process in mammals. It is unclear why it should be the case that, although we know vagus neurons at this stage exhibit a high degree of molecular heterogeneity which corresponds with target identity, and in some cases corresponds with molecular heterogeneity between targets, such heterogeneity is insufficient to guide appropriate repatterning. Efforts to address this question will, in large part, mirror those of development, as they will rely on the precise ability to better understand the molecular identities of neurons and their targets, to manipulate relevant aspects of those identities, and to examine effects on nerve regeneration. As in studies of development, it will be prudent to use caution when applying single-cell identity data gleaned from uninjured adults, as the process of injury may induce key gene expression changes in both neurons and targets; understanding such injury-responsive changes will be of great value. In contrast to development, where endogenous patterns of gene expression and signaling can generally be assumed to be conducive to correct patterning, during vagus regeneration it appears that some injury-induced signals impair regeneration, and it will be important to understand both the signals that promote regeneration and those that inhibit it. This complicating factor also highlights the value of studying nerve regeneration in an animal such as the zebrafish, where signals that prevent regeneration are less prevalent, to better understand the complexities of the mammalian post-injury signaling environment and how to improve it. We have already begun to dissect the post-injury signaling environment and, in several cases, manipulate it to improve regenerative outcomes. In the end, continuing to understand and improve vagus regeneration will come down to sorting out the signaling environment into the good (what’s helping), the bad (what’s missing), and the ugly (what’s disrupting)?

5). Conclusions and future directions

Even after centuries of scientific attention, the vagus nerve maintains its mystique. Now, armed with a modern toolkit that allows the ability to molecularly define, genetically manipulate, and observe in vivo neurons of distinct identities at exquisite resolution, we finally stand ready to dissect the molecular and developmental basis of the vagus nerve’s complex organization and multifunctionality. Such efforts are well underway, having revealed and begun to connect the dots between patterns of molecular and functional heterogeneity in neurons and their targets. Continued work to expand, deepen, and refine our understanding of these patterns across developmental and regenerative stages, coupled with a growing capacity to exert genetic control with greater spatial, temporal, and cell type-specific precision, will allow us to dissect how diversity in the patterning, identity, and signaling of neurons and their targets organizes the many diverse branches and functions of the vagus nerve.