Molecular Cell Types as Functional Units of the Efferent Vagus Nerve

Abstract

The vagus nerve vitally connects the brain and body to coordinate digestive, cardiorespiratory, and immune functions. Its efferent neurons, which project their axons from the brainstem to the viscera, are thought to comprise “functional units” - neuron populations dedicated to the control of specific vagal reflexes or organ functions. Previous research indicates that these functional units differ from one another anatomically, neurochemically, and physiologically but have yet to define their identity in an experimentally tractable way. However, recent work with genetic technology and single-cell genomics suggests that genetically distinct subtypes of neurons may be the functional units of the efferent vagus. Here we review how these approaches are revealing the organizational principles of the efferent vagus in unprecedented detail.

1. Introduction

The vagus nerve, a cranial nerve named for its “wandering” path from brainstem to bowel, is a vital conduit for brain-body communication. The vagus nerve connects the central nervous system to the digestive, cardiorespiratory, and immune systems and coordinates physiological processes to support homeostasis. This is necessary for survival since significant damage to the vagus nerve (e.g., a cervical vagotomy) can be fatal across many mammalian and avian species [1–3]. Among the tens of thousands of axons constituting the vagus nerve in mammals [4], only a minority signal from brain to viscera. These descending axons comprise the vagal efferent system and are the focus of this review. Most vagal axons transmit sensory information from viscera to the brain and are the subject of other recent reviews – e.g., [5, 6]. Also, since vagal efferent anatomy and function have been extensively reviewed elsewhere, e.g., [6–15], this review focuses instead on recent insights into the molecular organization of the efferent vagus gained from single-cell genomics and genetic technology.

2. Functional Anatomy of the Efferent Vagus

The efferent fibers of the vagus nerve are divided into two functional classes, general visceral efferents and special visceral efferents. Vagal general visceral efferents are parasympathetic, preganglionic axons which control smooth muscle and cardiac muscle activity and glandular secretion in the digestive and cardiorespiratory systems. In contrast, vagal special visceral efferents fibers are motor neuron axons that control striated muscles of the upper airways and esophagus. In mammals, the cell bodies which give rise to the vagal general visceral efferents and special visceral efferents reside largely in two nuclei of the medulla. The larger of these two nuclei, the dorsal motor nucleus of the vagus (DMV), contains around 13,800 vagal efferent neurons in humans [16] and about onetenth as many in mice [17]. DMV efferent neurons are the parasympathetic preganglionic neurons for the heart, esophageal smooth muscle, and abdominal organs including the stomach, pancreas, gall bladder, and intestines [9] (Figure 1). The smaller vagal efferent nucleus, the nucleus ambiguus (nAmb), contains only 587 neurons in mice [17]. Unlike the DMV, the nAmb is a mixed nucleus, containing both the motor neurons for striated muscles of the upper airways and esophagus (vagal special visceral efferents) as well as parasympathetic preganglionic neurons for the heart, trachea, and lungs [18–24] (Figure 2). Together the DMV and the nAmb contain nearly all the cell bodies of vagal efferent neurons, with the remaining ones scattered sparsely across the intermediate zone between DMV and nAmb. The intermediate zone contains vagal efferent neurons that innervate the heart in many species – e.g., references [25–27]. While less is known about this vagal efferent region, some reports indicate that cardiovagal neurons in the intermediate zone are more like cardiovagal neurons of the DMV than those of the nAmb, in that their activity is not synchronized to respiratory or cardiac cycles [28].

Figure 1 – Model of DMV circuits for digestive, metabolic, and cardiac control.

For simplicity, circuits are shown unilaterally. Figure made with BioRender.

Figure 2 – Model of nAmb circuits for cardiorespiratory, upper airway, and esophageal control.

For simplicity, circuits are shown unilaterally. Figure made with BioRender.

Vagal general visceral efferents control many reflexes in the digestive system. Some of these reflexes are anticipatory, such as those prompted by the sight, smell, taste, or even the thought of food, during the cephalic phase of digestion [29]. Cephalic phase stimuli are relayed from higher brain and gustatory regions to vagal efferent neurons in the DMV, which then initiate cephalic phase secretion of gastric acid, pancreatic enzymes, and insulin [29–32]. These reflexes can be conditioned by experience, as famously demonstrated by Pavlov [33], and vary with food palatability [33–35]. Thus, the degree of cephalic phase responses may be matched to the anticipated size of the meal [30]. These responses may have evolved to allow the infrequent consumption of large meals, since vagotomized animals, which are incapable of cephalic phase responses, tend to eat smaller meals more frequently [36].

General visceral efferents in the DMV also mediate gut reflexes during later phases of digestion. Swallowing food causes distension in the esophagus, which signals through a vago-vagal circuit to relax smooth muscle in the lower esophageal sphincter and stomach – the latter, known as the “receptive relaxation reflex,” was described in 1911 by Walter Cannon [15, 37]. DMV neurons reflexively control motility from thoracic esophagus to colon, for instance, slowing gastric emptying when undigested fat is present in the ileum (i.e., the “ileal brake” [38]). Beyond their role in motility, DMV neurons also play important roles in nutrient absorption. Potentially in response to circulating energy signals such as glucose and leptin, DMV neurons alter the rate of gastric emptying into the intestines [39–44], pancreas release of insulin and glucagon [11, 32], and hepatic gluconeogenesis [45]. Through these changes in digestive and metabolic function, the general visceral efferents neurons of the DMV help to maintain energy homeostasis.

In the heart and lungs, other general visceral efferents maintain homeostasis through reflexive changes in cardiorespiratory function. For instance, in the baroreflex, stretch-sensing vagal sensory fibers in the aortic arches detect increases in blood pressure and signal to cardiovagal neurons in the nAmb to decrease blood pressure by slowing heart rate [5, 46]. Another reflex, respiratory sinus arrythmia, involves signaling from inhibitory respiratory neurons in the brainstem to cardiovagal neurons, which decreases heart rate with each breath out [47]. While the physiological value of this particular reflex remains unclear, it may improve ventilatory gas-exchange efficiency and/or reduce cardiac work during deep breathing and exercise [48]. Like the gut-projecting vagal efferents, cardiorespiratory vagal efferents also mediate anticipatory changes. For instance, in the mammalian dive reflex, one of the most robust and conserved of all mammalian autonomic reflexes, cardiovagal neurons slow heart rate from the onset of a voluntary underwater dive, which is thought to conserve oxygen stores in anticipation of a dive [49, 50].

The vagus nerve also contains motor axons which directly synapse on striated muscles of the upper airways and cervical esophagus. These special visceral efferents, or branchiomotor, fibers of the vagus nerve control motor functions of the larynx, pharynx, and cervical esophagus. Interestingly, some of the special visceral efferents fibers appear to collaterally innervate esophageal ganglia [51]. Through their control of upper airway and esophageal muscles, special visceral efferents neurons of the nAmb play important roles in deglutition, phonation, and protecting the airways [52–54].

Research on the efferent vagus has led to a model in which different populations of vagal efferent neurons, or “functional units,” innervate different organs and play different physiological roles [16, 55]. Indeed, some digestive functions of the efferent vagus are antagonistic in nature and so unlikely to be controlled by the same neurons. For instance, different neuron populations appear to have opposite effects on the lower esophageal sphincter [56] and gastric smooth muscle [55, 57] and in driving release of glucagon and insulin [58, 59]. Still, important questions remain unanswered about the identities and organization of the functional units of the efferent vagus.

3. Anatomical Organization of Vagal Efferent Neurons

The earliest evidence for functional units of the efferent vagus came from pioneering studies which used retrograde tracers to label vagal efferent neurons based on which organ they innervate. These studies revealed a viscerotopic organization for vagal efferent neurons in the DMV and nAmb [24, 60–64], where neurons innervating different organs tend to reside in different subregions. Within the rodent DMV, for instance, neurons that project through the gastric branches of the vagus tend to be located medially, whereas those innervating through the coeliac branches are found more laterally [60, 61]. In the nAmb, esophageal and pharyngeal motor neurons occupy distinct subregions [22, 24], as do the motor neurons for the cricothyroid and cricoarytenoid muscles of the larynx [65]. This viscerotopy appears to be conserved across vertebrate species. In mammals, birds, and cartilaginous fish, for instance, vagal efferent neurons in the DMV are organized anatomically according to which branch of the vagus nerve they project through and/or which organ they innervate [61, 66, 67].

There are notable exceptions to the viscerotopic organization of vagal efferent neurons. For example, neurons innervating the same tissue may reside in different anatomical subregions: e.g., neurons at both the rostral and caudal ends of the DMV both innervate the lower esophageal sphincter (LES) in rats [56]. Conversely, neurons may occupy the same subregion but innervate different organs or tissues. For instance, in Guinea pig nAmb, neurons innervating the trachea are distinct from those innervating the esophagus but found within the same subregion [68].

The anatomical distribution of vagal efferent neurons can reflect functional differences too. For instance, stimulating rostral and caudal DMV neurons with L-glutamic acid increases and decreases LES pressure, respectively [51]. In addition, electrically stimulating the medial DMV was shown to be more effective at evoking gastric acid secretion than the lateral DMV, though the opposite is true for pancreas hormone secretion [53]. These results suggest that the functional units controlling LES tone, gastric acid secretion, and pancreas endocrine function may be anatomically distinct.

The functional units of the efferent vagus are also heterogeneous in terms of their cytoarchitecture, neurochemistry, and physiology. Pancreas-projecting DMV neurons can be classified based on their action potential dynamics, response properties, and dendritic structure, and in general differ significantly from gut-projecting DMV neurons [69–71]. One subset (~8%) of pancreas-projecting DMV neurons specifically exhibits a slow afterhyperpolarization which could be modulated by their synaptic inputs [69]. In addition, DMV neuron populations that respond oppositely to gastric stretch and duodenal stretch also differ in their dendritic length and complexity [72], suggesting that the synaptic architecture of DMV neurons may be specialized for different interoceptive modalities. Neurochemically, DMV neurons innervating the gastric fundus are more likely to express neuronal nitric oxide synthase (nNOS) [73] and less likely to express tyrosine hydroxylase (TH) than the caudal DMV neurons that selectively innervate the gastric corpus [74]. In the nAmb, neurons projecting to the upper airways and esophagus extend their dendrites beyond the nAmb borders, have projection-specific patterns of dendritic branching, and tend to express calcitonin gene related peptide (CGRP), in contrast to those innervating thoracic sites [75, 76]. Together, these findings support the idea that vagal general visceral efferents form functionally specialized units that can be differentiated based on anatomical, neurochemical and physiological features. However, while many if not all these features reflect molecular differences between functional units of the efferent vagus, the extent to which the functional units differ molecularly has only recently come into view.

4. Genetic Approaches to Unravel Vagal Efferent Circuits

Over the past decade, the use of genetic technology has shed much light on the functional organization of efferent vagal neurons. The modern neuroscience “toolbox” contains a growing variety of genetically targetable tools for mapping neural circuits and monitoring and manipulating their activity. These tools can be targeted to neuron subtypes with both anatomical and genetic specificity, e.g., packaging the tool gene in a DNA recombinase-dependent viral vector and then injecting the vector into the brain of a mouse expressing the DNA recombinase from the locus of a subtype-specific marker gene [77]. This approach leverages molecular differences between neighboring neurons to investigate their functional differences, by mapping the synaptic circuitry or manipulating and monitoring the activity of a genetically defined population. Given the heterogeneity of vagal neurons, genetically targetable approaches will be critical to identifying the functional units of the efferent vagus.

Optogenetics and chemogenetics have been widely used to control the activity of vagal efferent neurons while measuring physiological and behavioral outcomes (Table 1). A common strategy across these studies is to target vagal efferents generally based on the gene Chat, expressed by most or all vagal efferents [78–81]. Importantly, Chat-based genetic targeting avoids vagal sensory neurons (afferents) in the nodose ganglion [82] (but see also ref. [83]). Thus, in contrast to traditional approaches such as electrical vagus nerve stimulation and vagotomy, the outcomes of manipulating Chat+ vagal neuron activity with optogenetics and chemogenetics can be attributed more confidently to the efferent vagus. However, most studies have applied these approaches generally to vagal efferent neurons, or those within the DMV, and so provide little insight into their functional heterogeneity. Only some have targeted vagal efferents based on their organ innervation or gene expression and are discussed in detail below.

Table 1 –

Summary of studies investigating vagal efferent function with optogenetics or chemogenetics.

Cellular Target	Species	Targeting Approach	Genetic Tool	Expected Effect on Efferent Activity	Physiological or Behavioral Outcome	Citation
All vagal efferents	Mouse	Chat-ChR2-eYFP mouse	Optogenetic, ChR2	↑	↓ kidney damage after ischemia-reperfusion injury	[84]
All vagal efferents	Mouse	Chat-ChR2-eYFP mouse	Optogenetic, ChR2	↑	↑ blood insulin levels ↔ blood glucose	[85]
All vagal efferents in DMV	Mouse	Chat-ChR2-eYFP mouse; fiber optic over DMV	Optogenetic, ChR2	↑	↓ pancreatitis after caerulein injection ↓ heart rate	[86]
All vagal efferents in DMV	Rat	Lentivirus with PRSx8 promoter	Optogenetic, ChIEF	↑	↓ heart damage after ischemia-reperfusion injury	[87]
All vagal efferents in DMV	Rat	Lentivirus with PRSx8 promoter	Chemogenetic, allatostatin receptor	↓	↑ heart damage with ischemia-reperfusion injury	[87]
All vagal efferents in DMV	Rat	Lentivirus with PRSx8 promoter	Chemogenetic, allatostatin receptor	↓	↓ exercise capacity	[88]
All vagal efferents in DMV	Rat	Lentivirus with PRSx8 promoter	Optogenetic, ChIEF	↑	Preserves exercise capacity and left ventricular function after myocardial infarction	[89]
All vagal efferents in DMV	Mouse	Chat-ChR2-eYFP mouse	Optogenetic, ChR2	↑	↓ blood levels of tumor necrosis factor alpha	[90]
All vagal efferents in DMV	Mouse	Cre-dependent (dep.) AAV in Chat-Cre DMV	Chemogenetic, hM3Dq	↑	↑ glucose tolerance ↓ food intake	[91]
All vagal efferents in DMV	Mouse	Cre-dep. AAV in Chat-Cre DMV	Chemogenetic, hM4Di	↓	↓ glucose tolerance ↑ food intake	[91]
All vagal efferents in DMV	Mouse	Cre-dep. AAV in vGAT-Cre NTS	Chemogenetic, hM3Dq	↓*	↑ blood glucose	[92]
All vagal efferents in liver	Mouse	Retrograde Cre-dep. AAV in liver	Optogenetic, JAWS	↓	↑ blood glucose ↑ liver expression of gluconeogenic enzymes	[93]
Liver-proj.	Mouse	Transsynaptic AAV-Cre in liver +	Optogenetic, ChR2	↓*	↑ blood glucose ↑ liver expression of gluconeogenic enzymes	[93]
DMV neurons		Cre-dep. AAV in arcuate hypoth.
Synaptic inputs to DMV	Rat	AAV in substantia nigra, pars compacta; fiber optic over DMV	Optogenetic, eNpHR	↓*	↓ tone in gastric corpus and antrum	[94]
Synaptic inputs to DMV	Mouse	Cre-dep. AAV in Npy-Cre NTS	Optogenetic, ChR2	↑*	↑ gastric motility	[95]
Synaptic inputs to DMV	Mouse	Cre-dep. AAV in Npy-Cre NTS	Optogenetic, ArchT	↓*	↓ gastric motility	[95]
All vagal efferents in heart	Mouse	heart isolated from Chat-ChR2-eYFP mouse	Optogenetic, ChR2	↑	↓ heart rate	[96]
All vagal efferents in nAmb	Mouse	Cre-dep. AAV in Chat-Cre nAmb	Chemogenetic, hM3Dq	↑	↓ heart rate ↓ blood pressure ↑ NREM sleep	[97]
All vagal efferents in nAmb	Mouse	Chat-Cre+/Phox2b-Flp+ neurons; fiber optic over nAmb	Intersectional optogenetic, CaTCh	↑	↓ heart rate ↑ esophageal contractions	[79]
Crhr2+ nAmb neurons	Mouse	Crhr2-Cre+/Chat-Flp+ nAmb neurons; fiber optic over nAmb	Intersectional optogenetic, CaTCh	↑	↔ heart rate ↑ esophageal contractions	[79]
Ghsr+ nAmb neurons	Mouse	Cre-dep. AAV in Ghsr-Cre nAmb	Optogenetic, bReaChES	↑	↓ heart rate ↓ AV conduction ↔ lung resistance	[98]
Calb1+ nAmb neurons	Mouse	Cre-dep. AAV in Calb1-Cre nAmb	Optogenetic, bReaChES	↑	↓ heart rate ↓ AV conduction ↑ lung resistance	[98]

↑, increase; ↓, decrease; ↔, no change; AV, atrioventricular; dep., dependent; proj., projecting; NTS, nucleus tractus solitarius; hypoth., hypothalamus; DMV, dorsal motor nucleus of the vagus; nAmb, nucleus ambiguus;

^*,expected based on manipulation of synaptic inputs to efferent vagus neurons.

Subsets of vagal efferent neurons can be targeted based on the organ they innervate. For instance, Kwon and colleagues combined viral tract tracing and optogenetics to characterize a neuronal circuit connecting POMC neurons of the arcuate hypothalamus to the liver via efferent neurons of the DMV [93]. POMC neurons are activated by feeding, glucose, and leptin, and their activation signals satiety through the melanocortin pathway - recently reviewed in [99]. A subset of DMV neurons expresses the melanocortin-4 receptor (MC4R), a receptor for POMC-derived neuropeptides, the activation of which inhibits DMV neurons [93, 100]. Kwon et al. (2020) found that activating POMC neurons that synapse on liver-projecting DMV neurons increased blood glucose and gluconeogenic enzyme transcription in the liver, without affecting blood insulin or glucagon levels [93]. The hyperglycemic effect was blocked by the pharmacological antagonist, SHU9119, suggesting it requires signaling through melanocortin-4 and/or -3 receptors [93]. Thus, arcuate POMC neurons may drive hepatic gluconeogenesis by inhibiting liver-projecting DMV neurons.

Other studies suggest that MC4R+ DMV neurons innervate the pancreas to control insulin release. One such study used a transgenic mouse in which the promoter of the melanocortin 4 receptor (MC4R) gene, Mc4r, drives expression of green fluorescent protein (GFP) [101]. The authors found that half of DMV neurons that innervate the pancreas are labeled by the Mc4r-GFP reporter [101]. These results are consistent with genetic data linking Mc4r expression to cholinergic control of circulating insulin levels. Specifically, deletion of the Mc4r gene from cholinergic (Chat+) neurons, including essentially all vagal efferent neurons, disinhibits DMV neurons and leads to hyperinsulinemia [100, 102]. In contrast, restoring Mc4r expression to cholinergic or parasympathetic (Phox2b+) neurons on an otherwise Mc4r-null background significantly attenuates hyperinsulinemia [100]. Thus, the DMV may contain two functional subtypes of Mc4r+ neurons: one which innervates the pancreas to drive insulin release, and another which innervates the liver to suppress gluconeogenesis. Given the role of the DMV in cephalic digestion and metabolism [103], these two Mc4r+ DMV neuron subtypes could work complementarily to prevent hyperglycemia after a meal.

Vagal efferent neurons can be genetically targeted to trace their innervation of the viscera. For instance, one study performed anterograde synaptic tracing of vagal efferents expressing the autism-associated gene Met [104]. To label Met+ vagal efferents, the authors used a transgenic reporter mouse line that expresses enhanced GFP (eGFP) downstream of the Met gene promoter, Met^EGFP [104]. The authors found that Met^EGFP labels a subset of vagal efferent neurons in the DMV and nAmb [104]. Anterograde tracing of these neurons in optically cleared, whole embryos showed that they innervate the larynx and esophagus (presumably from nAmb) as well as the stomach (presumably from DMV) [104]. Only some retrogradely-labeled stomach-projecting DMV neurons were MetEGFP+, unlike most cecum- and colon-projecting neurons [104]. These results suggest the Met gene is expressed in multiple vagal efferent circuits, including those controlling motility along the alimentary canal.

The studies summarized above raise important questions about the molecular organization of vagal efferent neurons. Are Mc4r and Met expressed by different populations of vagal efferent neurons, one that decreases blood glucose levels and another that stimulates gut motility, respectively? More generally, how does the molecular identity of vagal efferent neurons relate to their physiological role?

5. Molecular and Functional Taxonomy of Vagal Efferent Neurons

Single-cell RNA-sequencing (scRNA-seq) has provided new insight into the functional organization of the efferent vagus. By profiling and comparing the mRNA content of individual cells, scRNA-seq can systematically identify cell types and states. Comparing expression profiles across cell types reveals genes enriched in each cell type. These marker genes not only raise hypotheses about the potential functions of each cell type but also provide a means to test these hypotheses, through the cell type-targeted use of genetic tools. Combining scRNA-seq with genetic technology can therefore reveal the identity, nature, and function of cell types. Two recently published studies combined single-cell genomics with genetic technology to classify subtypes of vagal efferent neurons in the nAmb [79, 98]. One study, Coverdell et al. (2022), found a genetically distinct vagal circuit for motor control of the esophagus [79]. The authors first used scRNA-seq to profile gene expression in 238 Chat+ neurons from in and around the nAmb of adult mice. Clustering the neurons by their transcriptional similarity revealed three molecularly distinct subtypes, which the authors annotated based on subtype-specific marker genes: Crhr2, Vipr2, and Adcyap1. The authors then used genetically encoded anterograde tracers to compare the axon projections of the Crhr2^nAmb and Vipr2^nAmb neuron subtypes. Their results show that the Crhr2^nAmb neurons innervate the esophagus but not the upper airways, whereas the opposite was true for Vipr2^nAmb neurons. Optogenetically activating cholinergic Crhr2+ fibers in the esophagus caused esophageal contractions which were time-locked to the photo-stimulation. Since the nAmb is the primary source of cholinergic input to the esophagus [105], cholinergic Crhr2+ fibers in the esophagus most likely originated from Crhr2^nAmb neurons. Activating all nAmb cholinergic neurons, but not the Crhr2^nAmb or Vipr2^nAmb neuron subtypes, significantly decreased heart rate, suggesting that other nAmb neurons (e.g., Adcyap1^nAmb neurons) control heart rate. Overall, the results of the Coverdell et al. (2022) study suggest that molecularly distinct subtypes of nAmb neurons differentially control the esophagus and heart.

Another scRNA-seq analysis found distinct subtypes of cardiovagal neurons in the nAmb. Veerakumar et al. (2022) used scRNA-seq in neonatal mice to identify two cardiovagal subtypes which differed in their innervation of cardiopulmonary ganglia and ability to induce bronchoconstriction in adults [98]. One of the cardiovagal subtypes innervated only cardiac ganglia (ambiguus cardiovascular neurons, ACV), whereas the other innervated cardiac and pulmonary ganglia (ambiguus cardiopulmonary neurons, ACP). Optogenetically activating either the ACP neurons or ACV neurons caused a similarly rapid and robust reduction in heart rate, whereas bronchoconstriction occurred only with activation of the ACP neurons. Finally, the ACV neurons and ACP neurons were differentially activated by models of the baroreflex and diving reflex. These results suggest that genetically distinct vagal efferent circuits exist for different cardiorespiratory functions.

How do the nAmb neuron subtypes in the Coverdell et al. [79] and Veerakumar et al. [98] studies compare? Both studies used single-cell RNA-seq to molecularly identify nAmb neuron subtypes, though Coverdell et al. focused on the esophagus- and upper airway- (pharynx, larynx) projecting subtypes, while Veerakumar et al. studied the heart- and larynx-projecting subtypes. Still, these two studies appear to describe largely overlapping neuron populations. For instance, the top marker genes for larynx-projecting neurons in the Veerakumar et al. study, Calca and Dlk1, were expressed by upper airway-projecting neurons (Vipr2^nAmb neurons) in the Coverdell et al. study. However, these marker genes were also expressed by the esophagus-projecting neurons (Crhr2^nAmb neurons) in Coverdell et al., suggesting that Calca and Dlk1 may be markers of nAmb motor neurons more generally. In addition, while the Coverdell et al. study did not characterize the organ projections of one neuron subtype, Adcyap1^nAmb neurons, this subtype expressed many molecular markers of heart-projecting neurons identified by the Veerakumar et al. study (Figure 3A). The Adcyap1^nAmb subtype in the Coverdell et al. study may therefore correspond to the heart-projecting neurons of the Veerakumar et al. study. This would agree with previous research in guinea pigs showing immunofluorescence of PACAP, the Adcyap1 gene product, in cholinergic axons innervating the heart [106]. Of note, the marker genes used to target ACP and ACV neurons in the Veerakumar et al. study, Calb1 and Ghsr, respectively, appeared to be differentially expressed among Adcyap1^nAmb neurons in the Coverdell et al. study (Figure 3B). This suggests that the Adcyap1^nAmb subtype may include both the ACP and ACV subtypes of cardiovagal neurons identified by Veerakumar et al.

Figure 3 – Comparison of nAmb neuron subtypes from Veerakumar et al., 2022 and Coverdell et al., 2022.

Veerakumar et al. nAmb neuron subtype marker genes expressed by Coverdell et al. nAmb neuron subtypes (A) and single neurons (dots) of the Adcyap1+ subtype (B). Expression unit in panel B is unique transcripts per cell.

Some key differences are notable between the neuron subtypes identified by Veerakumar et al. and Coverdell et al. Specifically, at least two genes identified as markers of ACV neurons by Veerakumar et al., Ghsr and Htr3b, were more highly expressed in the upper airway-projecting neurons (Vipr2^nAmb neurons) than in the presumptive heart-projecting neurons (Adcyap1^nAmb neurons) in Coverdell et al. (Figure 3A). These discrepancies may be explained by differences in the age of mice used in the two studies. Veerakumar et al. used 2–4 day old mouse pups for single-cell RNA-seq, whereas Coverdell et al. used 28 week old mice. It is possible that expression of Ghsr and Htr3b turns on later in development in upper airway projecting nAmb neurons and so was not detected in the neonatal mice in Veerakumar et al. Future studies could investigate this possibility by using single-cell RNA-seq to characterize the molecular profile of nAmb neurons across developmental stages. In general, more work is needed to understand the molecular development of vagal efferent neurons.

Molecular subtypes of vagal efferent neurons may also play distinct roles in digestion. A study by Tao, Campbell et al. (2020) identified two subtypes of DMV neurons in mice that selectively innervate the glandular stomach but target different populations of enteric neurons [78]. Their scRNA-seq analysis found seven molecularly distinct subtypes of DMV efferent neurons. Tracing the axon projections in a subtype-specific manner showed that two of the DMV subtypes, Cck^DMV and Pdyn^DMV, selectively innervated the glandular stomach. Interestingly, Cck^DMV neurons preferentially contacted cholinergic enteric neurons, characteristic of neurons that control gastric contraction and acid secretion. On the other hand, Pdyn^DMV neurons selectively targeted nitric oxide-expressing enteric neurons, characteristic of neurons that control gastric relaxation. Thus, Cck^DMV neurons and Pdyn^DMV neurons innervate potentially antagonistic populations of enteric neurons in the glandular stomach.

The studies summarized above suggest a molecular logic to the functional organization of the efferent vagus (Figure 4, based on refs. [78, 79, 98]). Specifically, they show that molecular subtypes of vagal efferent neurons differ in their organ innervation and functional capabilities. These findings raise many questions for future investigation: for what vagal functions are these subtypes necessary? Do the subtypes also differ in terms of their afferent circuitry or response to vagal sensory stimuli? What is the nature of the five other molecular subtypes of vagal efferent neurons? Of note, the total number of vagal efferent neurons profiled by scRNA-seq in these studies is less than half the number present in mouse. Thus, additional subtypes of vagal efferent neurons may exist, potentially within or in addition to the subtypes identified so far. Additional scRNA-seq analysis is needed to have a comprehensive “parts list” of vagal efferent neurons and characterize their molecular development.

Figure 4 – Molecular subtypes of DMV and nAmb neurons as functional units of the efferent vagus.

Circles with question marks indicate potentially unidentified neuron subtypes.

6. Anti-Inflammatory Capabilities of the Efferent Vagus

Stimulation of the vagus nerve electrically [107–110], optogenetically [86, 90, 111], or by ultrasound [112, 113] causes an anti-inflammatory effect in rodents and humans [114]. In fact, vagus nerve stimulation is considered a potential treatment for a wide variety of inflammatory conditions and injuries [115], including: sepsis [107]; postoperative ileus [109]; pancreatitis [86, 116]; arthritis [114, 117]; Crohn’s disease [118, 119]; and ischemia and reperfusion injury of the kidney [111, 120, 121] and heart [110, 122, 123]. The immunological effects of the vagus nerve stimulation involves the activation of the cholinergic anti-inflammatory pathway (CAP) [115, 118, 124–127], which refers to the activation of splenic α7-nicotinic acetylcholine receptor (α7nAChR)-positive macrophages by splenic β2-adrenergic receptor-positive CD4+ T cells, rather than a mechanism dependent on acetylcholine release from vagus. Indeed, there is no apparent direct innervation of the spleen by the vagus, with recent studies showing that the effects of vagus nerve stimulation on spleen function rely on the activation of the splenic nerve through two distinct pathways; vagal afferents stimulate splenic sympathetic nerve activity through central sympathetic pathways, whereas vagal efferents (i.e., those thought to originate in the DMV) regulate splenic sympathetic nerve activity by modulating postganglionic neurons in the celiac-superior mesenteric plexus [90, 107, 128]. Notably, studies have challenged whether the vagal efferents control splenic nerve activity [129, 130], while others have found that only some of the branches of the splenic nerve are affected by vagal nerve stimulation [131, 132]. Nevertheless, there is clear evidence that the vagus can regulate immune function through multiple, parallel mechanisms.

Many important questions remain unanswered about the regulation of immune function by the vagus. For instance, which molecular subtypes of DMV neurons can regulate immune function? Are multiple DMV subtypes involved, or perhaps a single subtype that collateralizes to mediate a multi-organ, anti-inflammatory program? Future investigations could target the molecular DMV subtypes with genetic tools (e.g., actuators, tracers) to determine (1) which subtypes are capable of suppressing inflammation, (2) through which organs or tissue these subtype(s) act, (3) which stimuli activate them, and (4) whether their activity is necessary for immune homeostasis. Intriguingly, neurons of the insular cortex, which signal to DMV neurons, retain an anatomically specific memory of inflammation and, when activated, can recapitulate that inflammatory state [133]. Thus, the role of immunomodulatory DMV neurons may be to recall tissue-specific inflammatory states during repeated infections.

7. Conclusions and Future Directions

Recent scRNA-seq studies have raised the possibility that genetically distinct subtypes of vagal efferent neurons form function-specific circuits (Figure 4) [78, 79, 98] and so may correspond to functional units of the efferent vagus. Marker genes identified for each neuron subtype can be used to target genetic tools to that subtype. Leveraging this genetic access, each subtype can be characterized anatomically, physiologically, and functionally and so be integrated with the wealth of previous data on vagal efferent neuron diversity.

Genetic access to vagal efferent neurons will enable previously unanswerable questions to be addressed:

• Does each functional unit integrate different sources of synaptic input? Previous studies have mapped the synaptic inputs to the DMV and nAmb [134] or to organ-specific vagal circuits [27, 135] but not to neuron subtypes with a specific vagal function. Using genetically targetable, monosynaptically restricted rabies virus [136], the synaptic inputs to each vagal efferent neuron subtype can be mapped and compared.

• What are the activity dynamics of each functional unit during digestive and cardiovascular reflexes? Recent reports have demonstrated the feasibility of imaging the activity of genetically defined neurons in the medulla of awake, behaving mice [97, 137, 138]. Applying this approach to a genetically defined vagal neuron subtype in vivo will reveal the stimuli and conditions sensed by that subtype.

• Is each functional unit necessary for a different vagal reflex? Molecularly distinct populations of neurons in and around the nAmb are activated during different cardiac reflexes [98]. Accordingly, it is tempting to speculate that different vagal reflexes may engage distinct molecular subtypes of vagal efferent neurons. Selectively silencing or ablating a subtype of vagal efferent neurons with genetically targeted agents such as tetanus toxin (TetTox), diptheria receptor (DTR), or constitutively active Caspase [139], will reveal if specific subtypes are necessary for each vagal reflex.

Why is it important to define the functional units of the efferent vagus? Given their powerful control of heart function, metabolism, and inflammation, the corresponding circuits of the efferent vagus are attractive therapeutic targets for many diseases. However, electrically stimulating the whole nerve lacks specificity and, since the vagus comprises functionally heterogenous fiber types [140–142], may not optimally activate all vagal efferent fiber types. A more precise approach would target only one functional unit of the efferent vagus – e.g., to suppress inflammation without also decreasing heart rate or contracting the larynx. This could be accomplished pharmacologically, using drugs which selectively act on one functional unit, or even optogenetically. Optogenetic activation of the efferent vagus was recently demonstrated in a large mammal, supporting its translatability to clinical use [143]. However, progress in therapeutically targeting the efferent vagus has been limited in part by a lack of knowledge about how its functional units differ molecularly, including which receptors, neuropeptides, and other druggable signaling molecules they express. Recent studies connecting the molecular and functional subtypes of vagal efferent neurons have revealed subtype-specific expression of many genes encoding potential drug targets [78, 79, 98]. A better understanding of the molecular identity and organization of vagal functional units will reveal new opportunities to harness the therapeutic potential of the wandering nerve.