Neural gut-to-brain communication for postprandial control of satiation and glucose metabolism

Introduction

The brain plays an essential role in coordinating feeding behavior and systemic glucose metabolism. For example, the brain determines the timing, composition, and quantity of nutrients that are consumed. On the other hand, it coordinates glucoregulatory mechanisms in peripheral organs through defined neural and hormonal communication pathways (Myers & Olson 2012, Ruud et al. 2017). Thereby, glucose metabolism in the whole body is centrally orchestrated according to nutrient availability. The GI tract constitutes a major information source for the brain regarding available nutrients. Given the GI tract’s ability to rapidly digest and absorb nutrients (Goodman 2010), information about the consumed food must be promptly relayed to the brain. This is crucial in order to limit excessive food consumption and to prime the body for postprandial changes in nutrient availability. Hence, the brain requires detailed information about the amount of consumed food, which GI tract organs it passes through at a given time, and which and how many nutrients it contains. This spatial and quantitative information must be transmitted in a rapid and precise manner so that the brain can timely adjust behavior and adequately adapt blood glucose levels.

Peripheral sensory neurons (PSN) that innervate GI tract organs represent an important element of this gut-to-brain communication (Fig. 1) (Berthoud et al. 2021, Duca et al. 2021, Wachsmuth et al. 2022). PSN are afferent neurons, some of which innervate GI tract organs with their peripheral terminals, which enable them to sense food-related signals and transform these signals into neural cues. PSN relay the gut-derived cues on downstream circuits in the central nervous system (CNS) via their central terminals (Blackshaw et al. 2007), for further processing and, ultimately, initiating various physiological processes. As previously reviewed and outlined below, early tracing and lesions experiments over the last decades have provided important insights into the anatomy of PSN and their general role in gut-to-brain communication (Wang et al. 2020, Berthoud et al. 2021, Wachsmuth et al. 2022). However, numerous open questions could not be sufficiently addressed in the past because of the functional diversity of PSN, their anatomical location, and the neuronal intermingling of their cell bodies in peripheral ganglia. Recent genetic and biotechnological advances have begun to open new possibilities to address these questions. First, RNA sequencing techniques (Poulin et al. 2016) have allowed cataloging PSN. This has pinpointed molecular markers of defined neuronal subtypes that show distinct innervation patterns of GI tract organs. Second, new neuroscience approaches that enable optical measurements of neuronal activity (Chen et al. 2013, Lin & Schnitzer 2016) have offered new avenues to study how PSN subtypes respond to signals that originate in GI tract organs as well as their recruitment of downstream neurocircuits. Finally, cell-type-specific manipulation techniques (Roth 2016, Atasoy & Sternson 2018) can now be used to probe the functional contribution of molecularly defined PSN subtypes.

View Full Size

Afferent sensory neurons and postprandial control. Left: PSN of vagal and spinal origin densely innervate GI tract organs and present relevant detectors of mechano- and chemosensory stimuli following food consumption. They relay this information to brain pathways, which in turn coordinate food intake and glucoregulatory mechanism. Spinal afferents, whose cell bodies reside in dorsal root ganglia (DRG) project to the spinal cord. The cell bodies of vagal afferents reside in nodose ganglia (NG) and they project to the brainstem (Berthoud et al. 2004). Middle: Schematic cross sections of the stomach, small intestine, and large intestine with peripheral endings from vagal afferents (blue) and spinal afferents (red). Insert in small intestine shows proximity of neuron endings and enteroendocrine cells (orange and green), which secrete various transmitters. Right: Illustration of the relative proportions of mechanical and chemical sensing during food passage. Created with BioRender.com.

Citation: Journal of Endocrinology 258, 3; 10.1530/JOE-22-0320

• Download Figure
• Download figure as PowerPoint slide

In this review, we detail the role of PSN in mediating gut-to-brain communication, with a particular emphasis on recent findings about their anatomical and functional diversity, which have begun to improve our understanding of how they relate to the regulation of satiation and glucose metabolism.

Physiological and anatomical properties of gut-innervating PSN

PSN that innervate GI tract organs are thought to serve two major functions: mechanosensation, for sensing of organ stretch and distension, and chemosensation, for sensing of nutrient amount and composition (Thorens & Larsen 2004). The first step of PSN in relaying these signals from the gut to the brain is their mode of stimulus detection. PSN are pseudounipolar, with one axonal branch extending into the periphery, which associates with peripheral targets. The detection of mechano- and chemosensory stimuli via the peripheral terminals relies on the expression of various receptors, whose stimulation triggers PSN activation. Histological experiments have demonstrated that PSN innervating the stomach, small intestine, and large intestine express a variety of receptors for the detection of stretch, distension, and macronutrients (Blackshaw et al. 2007). Additional electrophysiological studies found that stroking the stomach wall or administrating nutrients directly into the intestine evokes action-potential firing in PSN (Phillips & Powley 2000, Horn 2009), demonstrating their ability to convey relevant stimuli into neural activity. Besides the detection of nutrient-related stimuli, the expression of several other classes of receptors suggests the functional involvement of PSN in additional processes. For example, receptors for the detection of nociceptive temperatures and inflammatory signals are found in the vast majority of GI tract-innervating PSN (Hockley et al. 2019, Kupari et al. 2019), indicating their role in regulating processes that protect the organism against organ damage. Of interest in this context, a recent study uncovered an important role of PSN in the control of sickness behavior by relaying toxin-related signals from GI tract organs (Xie et al. 2022).

PSN synaptically relay neural signals onto second-order neurons in the CNS via their central axons. As with all PSN, gut-innervating subtypes use glutamate as the principal fast-acting neurotransmitter (Berthoud & Neuhuber 2000). In addition, excitatory neuropeptides, including calcitonin gene-related peptide and substance P, are found in PSN that innervate GI tract organs (Green & Dockray 1987, Tan et al. 2008). PSN fibers are categorized into different subtypes according to their myelination, with either myelinated or unmyelinated axons (Bielefeldt et al. 2005). Thus, the propagation of gut-derived signals can occur at different rates, suggesting that there must be a temporal component in the interpretation of different signals by the CNS.

GI-tract-innervating PSN are either inherent to the vagus nerve (vagal afferents), with cell bodies residing within the nodose ganglia (NG), or to spinal nerves (spinal afferents), with cell bodies residing within the dorsal root ganglia (DRG; Fig. 1) (Berthoud et al. 2004). Their central branches penetrate the nucleus of the solitary tractus (NTS) and area postrema (AP) in the brainstem (vagal afferents), and the dorsal horn of the spinal cord (spinal afferents) (Berthoud et al. 2004). One of the hallmarks of PSN is the discrete innervation pattern of the different GI tract organs. The anatomical organization of vagal afferents in this regard has been extensively studied in the past, primarily by using tracer injections into the NG to visualize terminal endings (Berthoud & Powley 1992, Berthoud et al. 1995, 1997, 2004, Berthoud & Neuhuber 2000). Within the GI tract, this has allowed to distinguish between three different types of nerve endings (Fig. 1): i) Intraganglionic laminar endings (IGLE), which innervate the myenteric plexus between the outer longitudinal and inner muscular layer (Berthoud & Powley 1992); ii) intramuscular array endings (IMA) that terminate in muscular layers in an arrow-like manner (Berthoud & Powley 1992); and iii) mucosa-innervating terminal endings of vagal afferents (Berthoud et al. 1995, Berthoud & Neuhuber 2000). Considering the ending types, it is thought that muscular IGLE and IMA endings function as mechanoreceptors, responding to organ stretch and distension. Mucosa-innervating endings are considered to detect chemosensitive signals. The close proximity of these endings to enterocytes positions them ideally to sense absorbed nutrients. Vagal afferents, particularly those innervating the mucosa, also express receptors for hormones and peptides that are secreted by enteroendocrine cells (EEC), including cholecystokinin (CCK), glucagon-like peptide 1 (GLP-1), serotonin (5-HT), and peptide YY (Wang et al. 2020, Zimmerman & Knight 2020). Since EEC secrete these factors upon nutrient detection (Duca et al. 2021), their local action on nearby vagal afferent terminals has been proposed to present an important signaling pathway (Fig. 1). Indeed, recent manipulation studies showed that the reduction of food intake, that was observed following the stimulation of EEC, requires vagal afferents and activation of the corresponding receptors (Bai et al. 2022, Hayashi et al. 2023). In addition to gut hormone-mediated signaling, synapse-like structures between EEC and vagal afferents have recently been found to mediate gut-to-brain communication (Kaelberer et al. 2018). These ‘neuropods’ support the acurate and rapid transmission of gut-derived signal for the regulation of feeding behavior (Kaelberer et al. 2018, Buchanan et al. 2022).

Compared with the vagal afferent system, spinal afferents that innervate GI tract organs have been much less studied. This is, at least in part, due to their extremely complex anatomical organization (Spencer et al. 2014). While the cell bodies of vagal afferents reside in only two NG, there is one pair of DRG for each spinal cord segment. In mice, this comprises 60 DRG, classified into 8 pairs of cervical, 13 pairs of thoracic, 5 pairs of lumbar, and 4 pairs of sacral DRG (Malin et al. 2007). The hierarchical organization of DRG is reflected by their peripheral innervation patterns, where each spinal nerve is known to penetrate specific peripheral targets (Karemaker 2017), most powerfully demonstrated by dermatomes in the skin. This topographic organization is also found within the GI tract. Spinal afferents whose cell bodies reside in thoracic DRG give rise to nerve fibers innervating the upper GI tract (stomach and small intestine), while spinal afferents of more lumbar and sacral DRG innervate distal parts of the small intestine and the large intestine. Based on this anatomical arrangement, two major types of spinal nerves are generally implicated in gut-to-brain communication: i) Thoracolumbar splanchnic nerves, which innervate all organs along the GI tract, and ii) lumbosacral pelvic nerves, whose endings primarily innervate the colon (Berthoud et al. 2004). In addition to the vast number of DRG, their locations close to the vertebral column complicate investigating the precise GI tract innervation patterns of spinal afferents because enhanced surgical procedures for safe and reliable injections are required. Based on single DRG tracings, it is, however, assumed that the relative innervation of GI tract organs increases more distally (Fig. 1) (Spencer et al. 2014, 2016a,b). Tracing studies of spinal afferents also suggest that a variety of different peripheral ending types exist. This has at least been shown for the colon, where spinal afferent endings are found in muscular and mucosal layers, tuned for the detection of mechanical and chemical stimuli based on receptor expression (Spencer et al. 2014, 2016b, Brierley et al. 2018). Interestingly, even though spinal afferents show these innervation and receptor expression patterns, most previous studies focused on their role in transmitting noxious signals and the perception of pain in visceral organs (Spencer et al. 2016b). Yet, more recent studies have begun to explore the role of spinal afferents in regulating food intake and glucose metabolism (De Vadder et al. 2014, Borgmann et al. 2021, Goldstein et al. 2021).

Taken together, the physiological and anatomical properties of vagal and spinal afferents are most consistent with the key role of PSN in sensing food-related signals along the GI tract and acting as fast and precise mediators of gut-to-brain communication. The remarkable diversity of detection systems, along with the different ending types, likely corresponds to the ability of distinct PSN subtypes in precisely mediating defined responses. Despite the assumption that the different afferents are specialized for the detection of distinct signals, the underlying biological function for such differentiation remains incompletely understood. In a physiological setting, both mechano- and chemosensation should be, more or less, initiated at the same time when nutrients reach an organ. This begs the question of how the complex product of mechano- and chemosensory signals that are relayed via vagal and spinal afferents can be integrated by the brain for the fine-tuning of postprandial responses.

Historical approaches for investigating PSN in gut-to-brain communication

The functional contribution of PSN had long been interrogated using lesion and ablation approaches (Yox et al. 1991, Norgren & Smith 1994, Walls et al. 1995, Moran et al. 1997, Schwartz et al. 1999). Given their anatomical organization and their ability of sensing organ stretch and nutrients, it was hypothesized that PSN encode anorexigenic signals and facilitate meal termination (Fox 2013). Hence, lesioning PSN should cause larger meal sizes and result, in turn, in increased caloric intake and body weight gain. Indeed, numerous studies that employed lesion and ablation techniques reported changes in feeding behavior (Mordes et al. 1979, Yox et al. 1991, Walls et al. 1995, Moran et al. 1997, Schwartz et al. 1999). However, the observed increases in meal size in deafferentiated animals were largely offset by decreases in meal frequency over time, resulting in little or no changes in total caloric intake and only incremental body weight changes (Berthoud 2008). The main conclusion drawn from these studies was that PSN are rather involved in acutely regulating satiation and less important for the long-term control of energy balance and body weight (Fox 2013).

In addition to adaptations in food intake, adjustments in glucose metabolism are rapidly initiated following food consumption, and lesion and ablation experiments have indicated a role of PSN in controlling these postprandial glucoregulatory effects (Wachsmuth et al. 2022). For example, deafferentiation of PSN impairs adaptations in hepatic glucose production as well as insulin and glucose tolerance in response to nutrient-related signals originating in the gut (Wang et al. 2008, De Vadder et al. 2014). This indicates that PSN have an important role in triggering post-meal counter-regulatory responses to limit alterations in blood glucose levels. Yet, there is no clear consensus that sensory neuron deafferentiation is associated with severe, long-term, alterations in glucose metabolism neither in experimental animal models nor in humans.

To interpret these findings correctly, it is however important to understand the different lesion and ablation approaches and how they ultimately affect gut–brain communication. Vagotomy, where trunks of the vagus nerve are exposed and severed, was one of the first methods to investigate vagal afferents (Schwartz et al. 1999). A clear limitation of this approach is that it falls short in selectively targeting the afferent arm of the vagus nerve, without destroying the axons of vagal efferents (Wang et al. 2020). Hence, any observed effects, or lack thereof, are difficult to attribute to the vagal afferent system, as they could also be due to impaired parasympathetic output from the brain to peripheral organs (Karemaker 2017). To refine this surgical technique and to target vagal afferents selectively, an approach named subdiaphragmatic vagal deafferentation was thereupon developed. Here, the afferent rootlets are cut directly after emerging from the brainstem to destroy only vagal afferents. In addition, the contralateral fibers can be cut beneath the diaphragm (Norgren & Smith 1994, Walls et al. 1995). While this approach achieves a more selective lesion of the afferent arm of the vagus nerve, it falls short to only target GI tract-innervating vagal afferents. It is important to note that vagal afferents, in addition to GI tract organs, innervate other internal organs, including the lung, heart, and pancreas (Prescott & Liberles 2022). An additional limitation of surgical interrogations is the possibility of (partial) nerve regrowth; it has been shown that mice exhibit variable regrowth of vagal afferents within only a few weeks of the vagotomy (Powley et al. 2005). Equally difficult are interpreting results from lesioning experiments of spinal afferents. Of note is that spinal nerve fibers are distributed throughout the whole periphery, including the skin, thereby making surgical removal of only those innervating GI tract organs somewhat impossible.

As an alternative to surgical procedures, PSN can be ablated by targeting the TRPV1 receptor (Yox et al. 1991, Wang et al. 2020). This method relies on the fact that administrating TRPV1 agonists at high concentrations causes Ca²⁺ toxicity and triggers cell death of TRPV1+ PSN subtypes (Suri & Szallasi 2008). Since TRPV1 expression is a molecular feature of afferent neurons, a clear advantage of this approach is that it does not interfere with efferent nerves. However, TRPV1-induced ablation is also not specific to PSN that innervate GI tract organs. Furthermore, potential disparate, or even opposing, functions of PSN cannot be investigated with this ablation or the above-mentioned lesion techniques. An example of this is the proposed orexigenic (food intake promoting) and anorexigenic (satiety promoting) functions of different vagal afferents (De Lartigue 2016). An additional complication of interpreting findings from lesion and ablation studies is that compensatory pathways might adopt gut-to-brain communication and disparage defects over time. Thus, the lack of spatial and temporal specificity clearly limits the interpretation of lesion and ablation studies, highlighting the need for approaches that allow selective, time-locked interrogations of only gut-innervating PSN subtypes.

Deconstructing the molecular heterogeneity of gut-innervating PSN

Based on the physiological properties of PSN and their innervation patterns, it had long been postulated that PSN subtypes can be characterized by the expression of different genes. Since electrophysiological studies showed that vagal afferents differently respond to gut peptides and hormones, including CCK and GLP-1 (Davison & Clarke 1988, Nishizawa et al. 1996, Dockray 2003), expression of the respective G-protein-coupled receptors (GPCR) was postulated to mark functionally defined subtypes. To investigate this, profiling GPCR in vagal afferents using RNA sequencing methods was the first step in defining molecular markers (Chang et al. 2015, Egerod et al. 2018). Analysis of the expressed genes revealed a remarkable variety of receptor expression for gut peptides and hormones, as well as receptors for microbial metabolites, lipid inflammation receptors, and orphan receptors (Chang et al. 2015, Egerod et al. 2018). However, such analysis of bulk expression patterns provides only little insights about the genes expressed by single cells or possible other marker genes, including those expressed in PSN subtypes that innervate only GI tract organs. Recent high-throughput technologies that allow single-cell RNA sequencing (scRNAseq) have enabled critical advances in this area (Tang et al. 2009, Kolodziejczyk et al. 2015, Macosko et al. 2015, Ziegenhain et al. 2017, Chen et al. 2019). Specifically, unbiased scRNAseq of vagal afferents revealed a remarkable heterogeneity of gene expression and identified distinct subtypes that are consistent with mechano- and chemoreceptors (Bai et al. 2019, Kupari et al. 2019). Further, selective assessment of RNA expression in GI tract-innervating vagal afferents has been achieved by injecting retrograde tracers into different organs and subsequent sequencing of labeled neurons (Bai et al. 2019, Zhao et al. 2022).

Together these findings have established that vagal afferents can be classified according to their molecular expression profiles, and that genetically distinct subtypes are tuned to serve specific functions. Importantly, the molecular markers that were identified in these studies have guided the generation and utilization of transgenic mouse lines that express recombinases under the promotor of the genes found in vagal afferents (Chang et al. 2015, Williams et al. 2016, Bai et al. 2019, Prescott et al. 2020, Borgmann et al. 2021, Li et al. 2022). These mouse lines, in combination with recombinase-dependent genetic tools, have allowed accessing and investigating discrete subtypes. As will be discussed in the following section, this has enabled researchers to begin disentangling the anatomical complexity of vagal afferents and probing their functional roles.

Anatomical organization of molecularly defined PSN subtypes

Because the organs along the GI tract fulfill specialized yet complementary functions in the digestion of food and the absorption of nutrients, defining the innervation patterns of PSN subtypes is a key step. Based on bulk RNA and scRNA sequencing experiments, the anatomical characteristics of six molecularly defined vagal afferents have hitherto been identified (Williams et al. 2016, Bai et al. 2019, Borgmann et al. 2021). This has been achieved by expressing genetically encoded fluorophores in transgenic mice. The first one expresses the GLP-1 receptor (GLP-1R) (Williams et al. 2016). Selective projection mapping demonstrated that GLP-1R+ vagal afferents primarily innervate the muscular layers of the stomach, with endings densely penetrating the antrum (Williams et al. 2016). This finding was surprising given that expression of the receptor suggests that these vagal afferents respond to endogenous GLP-1, a hormone that is secreted by EEC that are located mainly in the distal part of the small intestine (Gribble & Reimann 2019). However, further histological analysis of GLP-1R+ vagal afferents showed that some of their endings are located in the intestinal mucosa (Bai et al. 2019, Borgmann et al. 2021).

Analysis of GPCR also identified expression of the orphan receptor GPR65 in a subset (approximately 10%) of vagal afferents (Chang et al. 2015). Importantly, the histological assessment revealed that GPR65+ vagal afferents are distinct from GLP-1R+ ones (Williams et al. 2016, Bai et al. 2019, Borgmann et al. 2021). Consistent with a hypothesized anatomical distinction of the vagal afferent system, tracing experiments of GPR65+ neurons showed that their endings are mostly present in the mucosal layer of the small intestine and do not penetrate gastric muscular layers (Williams et al. 2016, Bai et al. 2019, Borgmann et al. 2021). This anatomical organization indicated that GPR65+ vagal afferents are a subtype for intestinal chemosensation. Interestingly, further tracing studies found that GPR65+ vagal afferent endings also innervate the mucosal layers of the stomach and esophagus (Bai et al. 2019, Borgmann et al. 2021, Zhao et al. 2022), suggesting that they detect nutrient-related signals along the entire upper GI tract.

Sequencing of retrogradely labeled neurons also revealed genetic markers of four other subtypes: Calca, somatostatin (Sst), vasoactive intestinal polypeptide (VIP), and the oxytocin receptor (OxtR) (Bai et al. 2019). Calca+ and Sst+ vagal afferents were found to innervate the stomach mucosa, with Calca+ ones densely penetrating the corpus, while Sst+ endings are mainly located in the antrum (Bai et al. 2019, Zhao et al. 2022). Vagal afferents marked by the expression of VIP were shown to exclusively innervate the mucosa of the small intestine (Bai et al. 2019), indicating that Calca+, Sst+, and VIP+ vagal afferents act as chemosensors. The terminal endings of OxtR+ vagal afferents, on the other hand, were found to primarily innervate the muscular layers of the distal stomach and intestine, with the highest density in the duodenum (Bai et al. 2019). Together, these data highlight that molecularly defined vagal afferents have unique innervation patterns along the GI tract, supporting the idea that they are specialized to sense different stimuli and contribute to the relay of mechano- and chemosensory signals from different organs.

Functional roles of PSN subtypes

To gain functional insights into the aforementioned vagal afferent subtypes, complex imaging and manipulation studies were recently performed (Williams et al. 2016, Bai et al. 2019, Borgmann et al. 2021, Kim et al. 2021, Li et al. 2022). Consistent with the finding that GLP-1R+ vagal afferents densely innervate gastric muscular layers, in vivo Ca²⁺ imaging of their cell bodies in NG showed that they are activated by stomach stretch (Williams et al. 2016). This supports a model in which GLP-1R+ vagal afferents relay pre-absorptive mechanosensory information about the size of an ingested meal for the rapid feedback regulation of satiation and blood glucose levels. Indeed, according to selective chemo- and optogenetic studies, activation of GLP-1R+ vagal afferents reduces food intake, whereas their inhibition ameliorates appetite suppression (Bai et al. 2019, Borgmann et al. 2021). It was further shown that GLP-1R+ vagal afferents contribute to the activity regulation of agouti-related peptide (AgRP) neurons (Bai et al. 2019), a neuronal population that resides within the arcuate nucleus of the hypothalamus and is a key element of the neural pathways that control both hunger and glucose metabolism (Andermann & Lowell 2017, Jais & Bruning 2022). Specifically, optical measurements of AgRP neuron activity demonstrated that chemogenetic activation of GLP-1R+ vagal afferents decreases AgRP neuron activity (Bai et al. 2019), suggesting that these vagal afferents dampen hunger signaling following mechanosensing of food in the stomach. The effect of GLP-1R+ vagal afferents on AgRP neuron activity could also be critical for maintaining glucose homeostasis (Konner et al. 2007, Steculorum et al. 2016). Recent investigations using an intersectional genetic approach support this notion. Selective activation of GLP-1R+ vagal afferents was found to lower blood glucose levels and improve glucose tolerance, while their inhibition increased blood glucose levels during feeding (Borgmann et al. 2021). These findings collectively demonstrate that GLP-1R+ vagal afferents exert an important function in adjusting food intake and maintaining blood glucose levels by predicting the nutritive composition of the consumed food through the signaling of mechanosensory cues from the stomach. However, further research is needed to clarify the specific involvement of AgRP neurons in this vagal afferent-dependent signaling process.

Consistent with the dense innervation of the intestinal mucosa by GPR65+ vagal afferent endings, Ca²⁺ imaging experiments showed that they are activated by the infusion of liquid diet into the small intestine but not by gastric or intestinal stretch (Williams et al. 2016). In addition, these neurons were shown to respond to 5-HT (Williams et al. 2016), which is secreted by intestinal EEC. These findings provide evidence that GPR65+ vagal afferents are involved in sensing and relaying information about nutrient availability in the intestine. Interestingly, acute stimulation of GPR65+ vagal afferents under basal feeding conditions has been consistently found to not to alter food intake (Bai et al. 2019, Borgmann et al. 2021). On the other hand, chemogenetic manipulation experiments revealed that their selective activation increases blood glucose levels and stimulates hepatic glucose production during euglycemic–hyperinsulinemic clamp studies (Borgmann et al. 2021). These data support the notion that GPR65+ vagal afferents are involved in the postprandial regulation of glucose balance; hence, they might contribute to the vagal afferent-dependent lowering of glucose levels that are observed when fat is delivered into the intestine (Wang et al. 2008). That being said, more recent imaging studies found that the activity of GPR65+ vagal afferents is not increased upon the intestinal delivery of solutions that contain only fat or sugar (Li et al. 2022), suggesting that the glucoregulatory action of GPR65+ vagal afferents is more complex as it depends on a combination of macronutrients. Further research is needed to understand the precise mechanisms of how GPR65+ vagal afferents are activated and in turn regulate blood glucose levels.

Intestinal infusion of nutrients was also found to activate VIP+ vagal afferents (Li et al. 2022), whose mucosa innervation is very similar to those described for GPR65+ vagal afferents (Bai et al. 2019). However, unlike GPR65+ subtypes, VIP+ vagal afferents showed clear increases in activity upon intestinal administration of single-component solutions that contained either only fat or only sugar (Li et al. 2022). It is possible that the subtype-specific activity regulation of VIP+ neurons depends on the release of CCK from EEC. VIP+, but not GPR65+, vagal afferents express high levels of the CCK receptor (Bai et al. 2019, Li et al. 2022), and fat and sugar activate EEC that secrete CCK (Duca et al. 2021). Intriguingly, acute stimulation of VIP+ vagal afferents, like GPR65+ neuron activation, was found to not to alter food intake (Bai et al. 2019). This suggests that mucosa-innervating vagal afferents are, in general, not involved in the rapid, postprandial control of satiation. Yet, a recent study revealed that VIP+ vagal afferents are an important mediator for the development of fat and sugar preferences (Li et al. 2022). This nutrient preference-forming function has likely long-lasting consequences on energy balance as it could strongly influence the preferred consumption of food high in sugar and/or fat.

As introduced above, a central integrator of gut-derived signals is AgRP neurons of the hypothalamus. Interestingly, simultaneous monitoring of AgRP neuron activity while chemogenetically activating GPR65+ or VIP+ vagal afferents showed that neither of them affect the activity of AgRP neurons (Bai et al. 2019). On the other hand, activation of OxtR+ vagal afferents, whose peripheral endings were found to primarily innervate the muscular layers of the small intestine, potently inhibits AgRP neuron activity (Bai et al. 2019). This indicates that the relay of mechanical, non-nutritive signals via PSN acts as an important control point for AgRP neuron-regulated (hunger) pathways. Consistent with this idea, activation of OxtR+ vagal afferents led to a profound decrease in food intake (Bai et al. 2019). The high-density innervation of the proximal part of the small intestine by OxtR+ vagal afferents (Bai et al. 2019) suggests that they, like GLP-1R+ ones, are recruited relatively early after food is consumed. However, the activity responses as well as the necessity of OxtR+ vagal afferents in the regulation of postprandial responses have not been directly investigated and should clearly be the subject of future studies.

Another vagal afferent subtype that was recently found to acutely regulate food intake is marked by the expression of Calca (Kim et al. 2021). The peripheral endings of this population primarily penetrate the mucosa of the stomach corpus (Bai et al. 2019). While little data are available on the nutrient-sensing capacity of the stomach, the mucosa-specific innervation indicates that Calca+ vagal afferents mediate chemosensory and not mechanosensory signaling (Bai et al. 2019, Kim et al. 2021). Interestingly, imaging studies found that the gastric delivery of solutions containing sugar and/or fat failed to evoke responses in Calca+ vagal afferents (Li et al. 2022); yet, in line with the notion that Calca+ vagal afferents detect specific chemosensory signals, the selective stimulation of their peripheral endings not only reduced food intake but also conditioned mice to avoid the consumption of a sucrose solution (Kim et al. 2021). These findings indicate that they relay signals from ingested food that could be hazardous to health. This concept may fit well with the strong alterations in food preferences that require PSN signaling. Together with the observation that VIP+ vagal afferents mediate sugar/fat preferences (Tan et al. 2020, Li et al. 2022), these data suggest that PSN have long-lasting regulatory functions in behavior calling into question the interpretation of previous lesion studies about their long-term role.

Although the abovementioned cell-type specific manipulation experiments have provided first functional insights into vagal afferent subtypes, one important limitation that should be considered is that the investigated populations hardly ever innervate only certain areas of GI tract organs. For example, vagal afferents that are marked by GLP-1R, GPR65, OxtR, and VIP expression all show dense innervation of the jejunum, duodenum, and ileum (Bai et al. 2019, Borgmann et al. 2021). Thus, stimulating the whole population, as it has been done, likely fails to mimic the natural activity pattern of these neurons that are triggered during or after the consumption of food, which gradually moves through the GI tract in a chronological manner. Hence, to more accurately determine the functional contribution of PSN in signal transmission from discrete organ parts, approaches that allow region-specific investigations are warranted. Here, recent developments in genetic strategies for targeting neurons based on both projection and genetic features may provide new paths for improving specificity (Fenno et al. 2020).

Given the complexity of the spinal afferent system, research on defined GI tract innervating subtypes suffers from the issues outlined above. Sequencing studies have examined the genetic profile of all spinal afferents originating from lumbar DRG (Usoskin et al. 2015, Zeisel et al. 2018). This has provided important insights about the general molecular diversity and defined groups of neurons according to marker gene expression (Usoskin et al. 2015, Zeisel et al. 2018). Via retrograde tracer injections, a recent study also sequenced spinal afferents in lumbar DRG whose terminal endings innervate the colon (Hockley et al. 2019). Analysis of molecular profiles revealed seven colonic DRG clusters, each characterized by 1 unique marker gene (Hockley et al. 2019). While functional studies examining the acute functions of the identified, or other molecularly defined, GI tract-innervating spinal afferents are currently missing, chemogenetic inhibition of the vast majority was found to increase food intake (Borgmann et al. 2021), suggesting that they are required to rapidly promote satiation. Of interest, a recent study employed a sophisticated retrograde approach to investigate stomach-innervating spinal afferents, revealing their substantial involvement in suppressing food intake upon stomach distension (Zhang et al. 2022). The afferents were also found to relay gastric distension signals that originate in the distal part of the small intestine. The transmission of these signals was facilitated by enteric neurons of the myenteric plexus, whose activity is regulated by endogenous GLP-1 released from EEC (Zhang et al. 2022). These significant findings highlight the emerging role of enteric neurons in facilitating communication between GI tract organs. Consequently, they contribute to the transmission of gut-to-brain signals via PSN, ultimately influencing CNS-mediated behavioral and physiological responses.

In addition, a recent study found that the removal of spinal afferents abolishes the inhibition of AgRP neurons during the intestinal administration of glucose (Goldstein et al. 2021). Notably, glucose infusion into the hepatic portal vein, which bypasses the intestine, was found to be sufficient for inhibiting AgRP neuron activity (Goldstein et al. 2021). These findings indicate that spinal afferents are the primary relay node for the convey of glucose signals that originate in the portal vein. This fits well with prior tracing and lesion studies (Berthoud et al. 2004, Bohland et al. 2014). It is important to emphasize that glucose infusion into the intestine or portal vein potently affects food intake and glucose homeostasis (Tordoff & Friedman 1986, Wang et al. 2008), highlighting the important role of glucose sensing in the GI tract for postprandial control (reviewed inDuca et al. 2021). Additional studies have shown that spinal afferents detect gut microbiota-generated nutrients (De Vadder et al. 2014), sense food-derived peptides via µ-opioid receptors (Duraffourd et al. 2012), and mediate conditioned taste aversion upon 5-HT and substance P release from EEC (Bai et al. 2022). While these findings suggest that the spinal afferent system relays a diverse array of information related to behavioral and metabolic processes, whether there is a site-specific transmission of mechano- and chemosensory signals by distinct spinal afferents requires future investigation.

Organization of vagal afferent inputs in the brainstem

There has been a persistent scientific endeavor to comprehend how gut-derived signals are integrated by the brain and disentangle the specific roles played by PSN in this complex communication. Given the difficulty in investigating spinal afferent subtypes that innervate GI tract organs, there is a major obstacle in elucidating the precise underlying circuits. Yet, we are beginning to understand the complexity of vagal afferent pathways (Fig. 2). The NTS/AP region is the first relay station of vagal afferents, where projections from GI tract-innervating ones converge together with those innervating other internal organs. Early histological studies found that vagal afferent projections display a topographical organization with axons from the stomach and intestine primarily innervating the caudal and medial portions of the NTS (Zhang et al. 1995). Since vagal afferents have an excitatory action onto their postsynaptic targets, the activation of those innervating the stomach/intestine should activate postsynaptic second-order neurons in these areas. Indeed, gastric distension and intragastric nutrient administration increase Fos expression in a spatial pattern that matches the vagal afferent projections (Willing & Berthoud 1997, Donovan et al. 2009). Moreover, opto- or chemogenetic activation of GI tract-innervating vagal afferent subtypes induces Fos expression in the NTS/AP regions their axons innervate (Williams et al. 2016, Bai et al. 2019, Borgmann et al. 2021). These observations substantiate that distinct subtypes act as ‘labeled lines’ for the transmission of mechano- and chemosensory signals from the gut.

View Full Size

The brainstem relays vagally mediated signals. (A) Illustration of internal organs that are innervated by the peripheral endings of vagal afferents. (B) Coronal overview of the medial NTS in the mouse brain with regions responding to visceral inputs from different organs (left) based on recent imaging experiments (Ran et al. 2022, Zhao et al. 2022). Right: Molecular markers of discrete subtypes of NTS neurons that have been shown to contribute to the regulation of food intake and glucose metabolisms (D'Agostino et al. 2016, Roman et al. 2016, Gaykema et al. 2017, Aklan et al. 2020, Chen et al. 2020) and their presumed intermingling. Bottom, parasagittal overview of the NTS and representation of major vagal input from different organs of the GI tract (Ran et al. 2022, Zhao et al. 2022). Color code: light blue, esophagus; green, stomach; yellow, duodenum; red, jejunum; dark green, colon; dark blue, respiratory system; brown, heart. CCK, cholecystokinin; Dbh, dopamine β-hydroxylase; NPY, neuropeptide Y; Pgg, preproglucagon; Th, tyrosine hydroxylase. Created with BioRender.com.

Citation: Journal of Endocrinology 258, 3; 10.1530/JOE-22-0320

• Download Figure
• Download figure as PowerPoint slide

However, a major obstacle in elucidating the precise circuits pertains to the difficulty in defining the identity of the NTS/AP neurons that are the targets of different subtypes. Both the NTS and the AP are remarkably heterogenous and contain phenotypic markers for various neurotransmitters and neuropeptides (Fig. 2A) (Dowsett et al. 2021, Ludwig et al. 2021). Thus, the relay of gut-derived signals, and the implementation of appropriate postprandial responses, is likely propagated by specific sets of downstream neurons. Indeed, discrete NTS/AP neuron subtypes whose activity is regulated by GI tract-innervating vagal afferents (Bai et al. 2019, Borgmann et al. 2021) have been shown to regulate food intake and glucose metabolism (Fig. 2B) (D'Agostino et al. 2016, Roman et al. 2016, Gaykema et al. 2017, Aklan et al. 2020, Chen et al. 2020), furthering the hypothesis of functionally defined pathways.

To provide an improved understanding of how gut-derived signals are integrated at the level of the NTS/AP, a recent study used Ca²⁺ imaging and monitored activity responses in single neurons during organ stimulation (Ran et al. 2022). These experiments in anesthetized mice revealed a remarkably organized topographical map in the NTS for input originating from peripheral organs (Fig. 2A and B). Interestingly, mechano- and chemosensory inputs from the same GI tract organs were found to converge onto the same NTS neurons. Furthermore, investigations on local transmission showed that GABAergic interneurons importantly shape NTS neuron activity and allow efficient stimulus-selective transmission of gut-derived signals (Ran et al. 2022). These data strongly support a model in which defined NTS neurons serve to integrate organ-specific signals and propagate this information in the CNS. However, the logic of this organization with regard to the regulation of satiation and glucose metabolism remains unclear. The NTS/AP region sends projections to numerous brain regions, including the parabrachial nucleus and the paraventricular nucleus of the hypothalamus, which contain discrete neuronal populations that are key for regulating appetite and blood glucose levels (Myers et al. 2021). Going forward, it will be important to determine how gut-derived signals regulate the activity of the NTS/AP output neurons that relay signals to these downstream neurons. Future efforts are also required to identify how the NTS/AP integrates mechano- and chemosensory signals from a given organ that coincide at the same time in freely behaving animals. This would reflect the natural order of events during and after the consumption of food.

Conclusion and outlook

In recent years, there have been notable advancements in deepening our understanding of the neural mechanisms that underly gut-to-brain communication. Molecularly defined vagal afferents have emerged as a core substrate for precisely detecting stimuli along the GI tract and promptly relaying this information to the CNS. However, these breakthroughs have also given rise to several intriguing new questions. For instance, why are some PSN subtypes tuned to detect organ stretch and distension, while others serve as pure chemosensors? Future studies will be crucial in establishing the true function of this differentiation and in clearly defining its role in fine-tuning behavioral and physiological responses.

In addition, future investigations into the identity of the spinal afferents that innervate GI tract organs, whose peripheral terminals penetrate muscular and mucosal layers, will undoubtedly improve our understanding of gut-to-brain communication. Thoracic afferents whose endings penetrate the muscular layers of the stomach likely serve to initiate early feedback responses during or after food consumption. In addition, information about previously digested food products and microorganism-derived molecules conveyed by ileum- and colon-innervating subtypes may provide long-term feedback information about nutrient availability, which could impact postprandial responses. Development in these areas will critically depend on scRNA sequencing of retrogradely labeled neurons and applying recently developed neurobiological approaches to detail the regulatory principles of discrete spinal afferent subtypes.

Another crucial area of research is to determine how disruption of neural gut–brain signaling may contribute to imbalances in energy intake and impact body weight development. Soon after the discovery of PSN in the control of food intake, it was recognized that they display aberrant functions in obese animals. For example, it was observed that vagal afferents in high-fat diet-fed animals exhibit reduced responsiveness to gut distension, as well as to CCK administrations (Covasa & Ritter 1998, Daly et al. 2011). These findings have led to various hypotheses regarding the dysregulation of feeding behavior and glucoregulatory responses. As recent studies have identified that PSN display significant anatomical and molecular heterogeneity, an important question arises: Which specific subtypes exhibit abnormal sensory capacities as well as functions in obesity? Importantly, given the shortcoming of effective treatment options for obesity that are currently available, gaining new insights into the aberrant functions of distinct subtypes holds the potential to develop more targeted pharmacological therapeutics.

Declaration of interest

The authors declare no competing interests.

Funding

This work was supported by funding from the European Research Council under the European Union’s Horizon 2020 research and innovation funding programme (grant agreement 851778; HF), and funding within the Excellence Initiative by German Federal and State Governments (CECAD; HF and LC).