Vagal Interoceptive Modulation of Motivated Behavior

Abstract

In addition to regulating the ingestion and digestion of food, sensory feedback from gut to brain modifies emotional state and motivated behavior by subconsciously shaping cognitive and affective responses to events that bias behavioral choice. This focused review highlights evidence that gut-derived signals impact motivated behavior by engaging vagal afferents and central neural circuits that generally serve to limit or terminate goal-directed approach behaviors, and to initiate or maintain behavioral avoidance.

Interoception and Biofeedback to the Brain

The viscera and brain are engaged in constant conversation. A key facilitator of this conversation is the gut-brain axis, comprising top-down neural pathways through which the central nervous system (CNS) controls gastrointestinal (GI) digestive functions, and bottom-up pathways through which the CNS receives moment-to-moment interoceptive feedback about visceral state (208). In humans and other mammals, the luminal surface of the GI tract is more than 100 times larger than the surface area of the skin, and represents the principal body surface interface for host-environment interactions. Indeed, the scope and complexity of gut-brain signaling pathways far exceed those related to the skin and other organ systems (155). The impact of GI processes on higher-level executive functions, affective state, and motivated behavior has received ample experimental attention within the fields of physiology, neuroscience, and psychology (21, 50), including a growing interest in gut microbial influences on brain and behavior (36, 88, 140, 151, 156, 171). In addition to fine-tuning ingestive and digestive processes, GI sensory feedback to the brain provides the neural basis for so-called “gut feelings,” which shape the subconscious emotional responses of humans and other animals to the daily threats and opportunities that drive motivated behavior (85, 155). Such threats and opportunities can originate internally (e.g., gut pathogens or nutrients) or within the environment (e.g., predator or food cues), and elicit a constellation of behavioral, physiological, and affective responses that are organized by complex CNS circuits. These responses can be innate or conditioned through learning, and typically are accompanied by endocrine and autonomic adjustments that alter GI and other visceral functions and interoceptive feedback to the brain.

The powerful influence of gut-brain signaling on behavioral state permits animals to anticipate or flexibly respond to contingencies by biasing the emotional and cognitive processes that guide motivated behavior. William James and Carl Lange were the first to formally propose that emotional feelings represent the perceptual consequences of sensory feedback from the body, providing the conceptual basis for the highly influential James-Lange theory of emotion (136). In James' words, “If our hypothesis be true, it makes us realize more deeply than ever how much our mental life is knit up with our corporeal frame, in the strictest sense of the term” (115). The core of this theory persists today amid mounting evidence that interoceptive feedback about GI and other visceral functions strongly impacts the emotional and cognitive processes that shape motivated behavior (37, 48, 50, 53, 56, 155). Accordingly, the body's physical state is proposed to serve as the basis for mood, affect, and other cognitive components of motivated behavior in both reductionist and complex systems accounts of emotion. The pivotal role of interoceptive feedback on motivated behavior arises in large part from the vital homeostatic control functions of spinal and hindbrain visceral circuits and their interconnections with higher limbic and cortical regions, offering a means through which goal-directed approach and avoidance behaviors are generated in accordance with emergent physiological and cognitive challenges and anticipated outcomes (208). Although many moment-to-moment decisions and choices appear to be under voluntary control, a constant undercurrent of subconscious influence over motivated behavior arises from within the body (273, 277). This article will review key aspects of interoceptive signaling pathways from the GI tract to the CNS, and describe how approach and avoidance behaviors are impacted by vagal sensory feedback from the gut in experimental rodent models. We also highlight evidence pointing to specific neurons in the caudal brain stem that transmit vagal sensory signals from the gut to brain regions that organize motivated behaviors. We propose that vagally mediated recruitment of these central neural pathways serves to limit or terminate positively reinforced behaviors, while increasing avoidance behaviors.

Interoceptive Signaling from Gut to Brain

Sensory signaling from the GI tract to the CNS is critical not only for regulating food seeking, intake, and digestion, but also for biasing emotional state and prioritizing motivated behaviors (25, 56, 267). Gut sensory information is encoded by intrinsic and extrinsic primary afferent neurons, by immune cells, and by enteroendocrine cells within the GI tissue (155) (FIGURE 1). GI endocrine and immune cells release gut hormones, cytokines, and other signaling factors that act peripherally on visceral sensory nerves (73, 123, 176, 194, 215), or may directly access the CNS at circumventricular brain regions lacking a blood-brain vascular barrier (79, 113), via active transport across the barrier (18) or through passive diffusion in cases of barrier disruption, as occurs in inflammation, neurodegenerative disease, or autoimmune disorders (17, 109, 114, 163).

FIGURE 1.

Encoding sensory signals in the gut

Gut signals are carried by vagal and spinal primary afferent neurons projecting into the hindbrain and spinal cord, respectively. Mechanical stimuli (stretch, pressure, distortion) can activate spinal, vagal, and intrinsic primary afferent neurons (IPANs) directly, without intermediary enteroendocrine (EE) or enterochromaffin (EC) cells. Although no synaptic connections have been found between IPANs and extrinsic afferents, the latter form networks around myenteric ganglia neurons that receive synaptic input from IPANs. Serotonin (5-HT), peptides, cytokines, and other molecules produced by gut endocrine and immune cells in response to microbial signals, luminal nutrients, mechanical stimuli, and other factors can signal to the brain through the circulation, through EE and EC cells, and also via receptors expressed by primary vagal and spinal afferent neurons. Figure is adapted from Ref. 155, with permission.

Circulating Factors

An abundance of pharmacological evidence supports the role of pancreatic-, adipose tissue-, and GI-derived signaling factors in shaping motivated behavior by accessing the brain via the systemic circulation. The principle sources of caloric/nutritional feedback to the brain are hormones such as leptin, insulin, cholecystokinin (CCK), glucagon-like peptide 1 (GLP-1), peptide YY (PYY), and other regulatory molecules signaling current and future availability of nutrients and stored energy (22, 46, 73, 194). At least one gut hormone (ghrelin) is released from gastric tissue when the stomach is empty, signaling an absence of nutrient absorption that facilitates appetitive food-seeking behavior (60, 131, 179). Leptin, CCK, GLP-1, ghrelin, and other peripherally derived signaling factors are released into the circulation and appear to affect neural activity within a variety of CNS regions, including those involved in motivation and approach behavior (179, 232, 245, 277). Although it is not yet clear whether or how CNS circuit activity is modified by these circulating factors under physiological (as opposed to experimental) conditions, food intake and the motivation to work for food is reduced after intraparenchymal administration of “anorexigenic” factors such as leptin, amylin, GLP-1, or insulin within specific CNS nuclei implicated in the appetitive and consummatory phases of ingestion (discussed further, below) (81, 109, 142, 162, 245, 266, 277), whereas central administration of ghrelin increases palatable food intake and reinforcement (179, 233). Stronger evidence for physiologically relevant, direct central actions comes from studies in which central receptors are disabled to interrupt the actions of endogenous (as opposed to administered) factors. For example, operant responding for palatable food is increased in rodents after virally mediated knockdown of leptin receptor expression within the midbrain ventral tegmental area (VTA) (61), whereas administration of a ghrelin receptor antagonist into the VTA or other brain regions not only suppresses food intake and reward but also interferes with the reinforcing properties of drugs such as cocaine, alcohol, and amphetamine (1, 69). Thus circulating GI-related factors that signal metabolic states of caloric surfeit or deficiency may directly access CNS circuits to influence reward circuit activity and approach behaviors in general, thereby linking motivated behavior to current physiological state (110, 277).

Extrinsic Neural Pathways

Despite the existence of hormonal signaling pathways from the GI tract to the CNS, the linkage between gut functions and emotional/cognitive processes is afforded primarily through extrinsic afferent neural signaling pathways to the spinal cord and caudal brain stem. It is important to note that neural activity within these afferent pathways is both sensitized and supplemented by gut microbial, immune, and endocrine signaling (97, 123, 124, 140, 157, 176, 219), generating an impressive degree of plasticity and versatility in the quality and magnitude of neural signals reaching the CNS (73, 74).

Neural signals from the gut (FIGURE 1) reach the CNS via spinal and vagal primary sensory neurons whose pseudo-unipolar cell bodies occupy the dorsal root or nodose ganglia, respectively, and whose central axon terminals synapse within the spinal cord or the caudal nucleus tractus solitarius (cNTS) in the lower brain stem (FIGURE 2, A AND B) (4, 5, 99, 164, 230, 237, 259). Spinal afferents from the GI tract enter lamina I of the dorsal horn and converge centrally with vagal sensory pathways by virtue of collateralized spinal inputs to the cNTS, the pontine parabrachial nucleus (PBN), and other brain stem regions (160). Despite this convergence, spinal and vagal sensory signals from the GI tract are qualitatively different, as revealed by direct electrophysiological recordings of afferent traffic en route to the CNS. Spinal afferent neurons innervating the upper GI tract are specialized to detect and protect against tissue damage, and are critical for nociceptive signaling (103, 155). On the other hand, GI vagal afferents appear to lack the properties of true nociceptors; instead, activation of vagal sensory inputs to the CNS has been implicated in antinociception through engagement of central opioid signaling (72, 270). Spinal pain signaling pathways from gut to brain clearly impact emotional state and motivated behavior, and peripheral factors that chronically sensitize and activate these pathways likely contribute to the high comorbidity of GI disorders with dysphoric motivational states of anxiety and depression, as thoroughly reviewed elsewhere (83, 143, 155). The remainder of this review will focus not on spinal afferents or pain signaling from the GI tract but instead on interoceptive signals conveyed from the gut to the brain via vagal afferents that terminate within the cNTS.

FIGURE 2.

cNTS neural projections to key CNS regions that shape motivated behavior

The peripheral axons of GI vagal afferent neurons, whose cell bodies occupy the nodose ganglia, ramify throughout the upper GI tract (A) and are specialized to detect distension, smooth muscle tone, locally released gut hormones, and other signaling factors, including microbial products and immune mediators (see FIGURE 1). The central axons of vagal afferents enter the dorsal region of the caudal medulla, terminating densely within the cNTS (B). Of the many chemically distinct neural populations residing within the cNTS, three are responsive to interoceptive vagal sensory input and also project to higher brain regions: GLP-1 neurons, NA neurons of the A2 group, and CCK-expressing neurons. These interoceptive-responsive cNTS neurons modulate motivated behavior via direct projections to CNS regions implicated in approach (blue) and avoidance behaviors (red). Inputs to approach behavior circuits (blue): GLP-1 and A2 NA neurons directly innervate the MDS, targeting both the VTA (D) and NAc (G). cNTS projections to the LHA (E) and other central sites (e.g., medial hypothalamus, thalamus, lateral septum) provide additional routes for ascending interoceptive neural pathways to influence brain regions involved in behavioral reinforcement and reward. Inputs to avoidance behavior circuits (red): GLP-1 and A2 neurons also form direct and indirect projections to the BST (F) and CeA (E). The ventrolateral (vl) BST receives particularly dense input from medullary NA neurons that co-express PrRP. GLP1, A2, and CCK-expressing cNTS neurons provide dense input to the pontine PBN (C), which densely innervates both the CeA and BST. cNTS, caudal nucleus tractus solitarius; GLP-1, glucagon-like peptide-1; NA, noradrenergic; CCK, cholecystokinin; MDS, mesolimbic dopamine system; VTA, ventral tegmental area; NAc, nucleus accumbens; LHA, lateral hypothalamic area; BST, bed nucleus of the stria terminalis; CeA, central nucleus of the amygdala; PrRP, prolactin-releasing peptide.

Vagal Sensory Pathways from Gut to Brain

Vagal sensory afferents monitoring pre- and post-absorptive sites within the GI tract comprise the large majority of fibers within the vagus nerve (23). Vagal afferents are tonically active, providing a steady stream of glutamatergic input to postsynaptic cNTS neurons that is tuned on a moment-to-moment basis by gut distension, smooth muscle tone, locally released gut peptides, and immune mediators. The peripheral axons of some vagal afferents ramify within the mesentery and in intramuscular arrays throughout the smooth muscle layers of the esophagus, the gastric fundus and antrum, the pylorus, and the small intestine (26, 27, 189, 195, 224), where they are poised to detect muscle tension and contraction. Some vagal sensory axons terminate in close proximity to interstitial cells of Cajal within the smooth muscle layers (190) or within the myenteric and submucosal plexuses of the enteric nervous system, where they interact with several classes of intrinsic primary afferent neurons (155). Other vagal afferent fibers permeate the mucosal lining to terminate within a single epithelial-cell width of the luminal gut surface along the basal lamina of the epithelium (23, 24, 191), where their firing activity is regulated in a receptor-mediated manner by signaling factors released from mucosal immune cells, enteroendocrine cells, and other intermediate cells in the lamina propria (194). It is important to note that, although individual vagal afferents can be classified as mechano- or chemoreceptive, the majority appear to be polymodal and respond to a wide range of both mechanical and chemical stimuli (97). Vagal sensory neurons express a diverse array of receptors specialized for detecting locally released and circulating gut hormones, cytokines, and microbial products that vary according to diet, feeding status, and stress exposure (73, 74, 97, 123, 176, 194, 215), significantly expanding the potential for physiological plasticity of vagal sensory signaling from gut to brain. In this manner, vagal afferent neuronal plasticity effectively fine-tunes the major neural route through which information about the gut environment and ingested nutrients reaches the CNS to influence behavior.

The central axons of GI vagal sensory neurons enter the dorsal region of the caudal brain stem via multiple rootlets that converge in the solitary tract and terminate within the cNTS (FIGURE 2, A AND B) (4, 120). The cNTS is considered the “visceral” NTS, distinct from the more rostral “gustatory” NTS (141), and is a key component of the dorsal vagal complex (DVC), which also includes the area postrema (AP) and dorsal motor nucleus of the vagus. The DVC is a critical central node for interoceptive feedback from body to brain (100, 101, 200, 202, 205, 273, 275). Vagal sensory inputs to the cNTS are glutamatergic (181, 193, 241), and their activation generates tightly synced, large-amplitude excitatory postsynaptic receptor-mediated currents in cNTS neurons to provide high-fidelity transmission of sensory nerve activity (9). Glutamatergic vagal afferents also express a large, heterogeneous array of neuropeptides (97) whose synaptic signaling and/or neuromodulatory properties within the cNTS are largely unknown. In addition to afferent neural inputs, the AP and a significant portion of the subjacent cNTS contain fenestrated capillaries (51, 119, 229), providing local parenchymal access to blood-borne factors arising from the gut (e.g., toxins, cytokines, hormones, and transported glucose and other nutrients) that can affect local neural activity (31, 111, 187, 194, 199, 269). Thus cNTS neurons receive direct and relayed synaptic input from vagal GI and other sensory neurons surveying the entire visceral landscape, and also are privy to circulating factors that provide additional interoceptive feedback signals.

Since vagal sensory neurons express receptors for CCK, ghrelin, GLP-1, and many other gut hormones, receptor-mediated mechanisms can efficiently modify vagal afferent signaling to the cNTS in response to hormone stimulation within peripheral tissues or via the circulation (22, 59, 73, 97). Gut hormones released in response to intestinal nutrients sensitize (i.e., lower the activation threshold) of vagal afferents that deliver distension-related negative feedback signals to the cNTS, which generate satiety and the cessation of feeding-related approach and consummatory behavior (78, 154, 166, 224, 234, 268). The firing activity of distension-sensitive vagal afferents also is increased by factors such as circulating leptin (182, 183). Leptin levels increase or decrease on a slow time scale in proportion to adipose tissue mass, but also increase acutely after release from the gastric mucosa in response to mechanical stimulation (13, 38). Thus factors that increase GI vagal afferent signaling to the cNTS can be envisioned primarily as activating a vagal sensory “brake” that attenuates feeding-related approach and consummatory behavior. To extend the analogy, increased ghrelin signaling from an empty stomach reduces the firing activity of GI vagal afferents (59), effectively releasing the vagal afferent brake and facilitating ingestive approach and consummatory behavior.

In the remainder of this focused review, we will highlight evidence that gut-derived signals influence motivated behavior by altering GI vagal afferent activity and/or the activity of cNTS neurons, which innervate a central neural network implicated in emotionality and goal-directed behavior. Research findings in rats and mice indicate that sensory feedback from the gut attenuates not only food intake but a range of other positively reinforced approach and consummatory behaviors, while also increasing avoidance behaviors.

Motivated Behaviors and Vagally Mediated Influences

An animal's survival depends critically on its ability to obtain the resources necessary for health and well-being, and to avoid harmful and potentially life-threatening situations. The adaptive behavioral processes that facilitate these critical abilities are called motivated behaviors because they energize, activate, and direct behavior toward a motive or goal. Motivated (i.e., goal-directed) behaviors are closely linked to emotional/hedonic state, and can be broadly divided into two categories: approach and avoidance. Motivated approach and avoidance behaviors can be driven by innate factors (e.g., the rewarding properties of sweet taste and nutrient absorption, or the aversive properties of hunger, pain, and nausea), or by learned associations (e.g., cue-potentiated feeding, drug self-administration, or conditioned avoidance of flavors previously paired with gastric malaise). The complex CNS circuits that control initiation, maintenance, and termination of motivated behaviors are highly conserved across species, guiding humans and other animals toward stimuli and situations that may satisfy critical homeostatic needs, and away from those that may threaten homeostasis. These complex, multifaceted neural systems include key circuit nodes located at every level of the neuraxis, with each node receiving interoceptive feedback signals relayed directly or indirectly from the cNTS (FIGURE 2).

Approach Behaviors

Goal-directed approach behaviors are guided by the positive reinforcing properties of current or anticipated appetitive stimuli and associated cues (e.g., food, sexual partners, pain relief, social interaction) that generate hedonic pleasure and/or diminish a negative hedonic state. The reinforcing properties that motivate goal-directed appetitive behaviors such as feeding, sexual activity, and drug self-administration largely depend on recruitment of the mesolimbic dopamine system (MDS), comprising dopaminergic (DA) projections from the midbrain VTA to the ventral striatum [i.e., nucleus accumbens (NAc)] (FIGURE 2, D AND G) and other corticolimbic regions that access the motor systems that generate goal-directed behavior (211). In experimental animals and in humans, increased DA signaling within the NAc follows exposure to innate stimuli or learned cues with emotional significance and incentive salience, and DA transients are closely linked to reinforcement and reward-based learning (105, 177, 221, 238). For example, extracellular DA levels within the NAc increase in response to the rewarding taste of sucrose or after intestinal glucose transport, but fall in response to the aversive taste of quinine or after systemic administration of the nauseogenic agent lithium chloride, perhaps “marking” these innately rewarding or aversive stimuli as worthy of attention and learning about associated cues (35, 86, 212, 226). Interestingly, extracellular DA levels within the rat striatum closely reflect the caloric density and nutritive features of GI contents (64, 65, 226), suggesting moment-to-moment “tuning” of MDS reward circuits based on postabsorptive nutrient signaling from the GI tract. Indeed, intestinal transport/absorption of glucose, protein, fat, and other nutrients is sufficient to increase DA signaling within the MDS, which is necessary for the reinforcement that initiates appetitive food seeking and maintains consumption, as well as for feeding-related Pavlovian and instrumental learning (58, 211, 226, 236).

Ascending neural projections from the cNTS reach the VTA and NAc directly (200, 242, 260) and also indirectly via relays within the PBN (47, 89) and lateral hypothalamic area (LHA) (276). LHA inputs to the VTA heavily influence DA signaling within the MDS (12, 108, 122, 192, 248), and the LHA receives substantial synaptic inputs from the cNTS (116, 210, 242, 272) that include inputs originating from gastric vagal afferents (210). Classic studies demonstrated that LHA stimulation elicits a robust feeding response that can be blocked not only by antagonism of DA receptors (184) but also by subdiaphragmatic vagal nerve transection (14, 188), evidence that vagal sensory and/or motor loops are a critical component of the LHA-MDS-mediated effect on feeding-related approach behavior. Thus, by influencing neurophysiological signaling within the MDS and “extra-MDS” circuits, GI sensory signals relayed through the cNTS can modulate the motivational properties of environmental stimuli and thereby shape approach behavior.

Despite this evidence, results from other experiments support neither the necessity nor the sufficiency of GI vagal afferent signaling in driving CNS reward circuits that facilitate nutrient-associated positive reinforcement. A refined surgical approach called “subdiaphragmatic afferent vagotomy” (SDA) (172), in which half of the vagal motor innervation of the GI tract is preserved despite complete interruption of GI vagal afferent feedback to the CNS, has been used to more selectively examine the role of vagal afferents in GI-based nutrient reward signaling and reinforcement. Results from SDA experiments in mice and rats argue against any necessary role for vagal afferent signaling in the primary reinforcing properties of intestinal nutrients (227). Although the inability of SDA to eliminate flavor conditioning reinforced by intestinal glucose, lipid, and other nutrients challenges the idea that vagal afferents are required, it must be acknowledged that half of the GI vagal afferent cell bodies and their central axonal inputs to the cNTS persist after SDA, despite transection of the subdiaphragmatic portion of their peripheral axons that physically contact GI tissue. Thus leptin, insulin, GLP-1, and other factors released into the circulation in response to intestinal nutrients may be detected by semi-intact vagal afferents and conveyed to the cNTS in a manner that is sufficient for nutrient reward.

The available evidence indicates that central neural pathways for vagally mediated influences on MDS reward systems appear to be utilized primarily to deliver negative feedback signals from the GI tract that reduce the rewarding properties of nutrients and other reinforcing stimuli, in part by inducing competing motivational states associated with satiety, malaise, and nausea (see below; FIGURE 2). Indeed, most gut hormones released by intestinal absorption/transport of glucose, fat, and protein suppress rather than stimulate feeding (277), similar to other natural and experimental stimuli that activate vagal afferent inputs to the cNTS and reduce food intake and other motivated approach behaviors (201, 204). Importantly, vagally mediated, feeding-related cNTS stimulation appears to promote normal meal-induced satiety, whereas threat-induced stimulation of the cNTS (e.g., due to excessive GI distension, inflammation, or toxin exposure) recruits higher numbers or different populations of cNTS neurons that contribute to aversion, malaise, and avoidance behavior (134, 147, 213, 214).

Avoidance Behaviors

Avoidance behaviors are geared toward reducing exposure to negative, dysphoric emotional states such as fear, anxiety, pain, and nausea. In contrast to positively reinforced approach behaviors, avoidance behaviors are often characterized by behavioral suppression or withdrawal coupled with defensive posturing and increased vigilance/arousal, during which attention is focused on the aversive stimulus. Indeed, “anxiety-like behavior” in rodents is characterized by behavioral suppression and avoidance (e.g., reduced food intake, social interaction, locomotion, and exploration) accompanied by increased sympathetic activity, muscle tone, and startle responses (139, 178). Avoidance behaviors are generally guided by the negative reinforcing properties of successfully reducing interaction with aversive stimuli and associated cues. Similar to positive reinforcement, DA signaling within the MDS is critical to maintain and strengthen avoidance behaviors that prevent or minimize aversive experiences or outcomes (174).

The coordinated physiological and behavioral signs of avoidance behavior depend on the output of widely distributed neural circuits, including the bed nucleus of the stria terminalis (BST) and the central nucleus of the amygdala (CeA), often referred to as the “central extended amygdala” (29, 43, 62, 66, 257, 263). Research in rodents supports the view that the CeA is critical for eliciting short-lasting, stimulus-specific fear responses, including conditioned fear and avoidance behavior, whereas the BST is necessary for more sustained and generalized anxiety-like responses to innate or conditioned threat stimuli and contexts (256, 258). The highly interconnected CeA-BST receive robust direct and relayed interoceptive input from GI-related regions of the cNTS and PBN that are important for conditioned avoidance behavior (29, 39, 42, 222, 223), together with direct input from the cortex, hippocampus, and other amygdalar subnuclei that process stimuli associated with environmental threats. Projections from the CeA and BST target widespread areas of the CNS that coordinate the physiological and behavioral components of fear, anxiety, and other avoidance behaviors (75, 218, 240, 252, 258). Thus GI sensory feedback to the brain can initiate and modify avoidance behaviors associated with nausea, fear, and anxiety by modulating neural activity within the CeA-BST (30, 39, 42).

Studies in rodents have implicated vagal afferent signaling to the cNTS and higher brain regions in innate anxiety, learned fear, and the effect of food intake and physiological arousal on emotional learning and memory (80, 125, 158, 165, 175, 216, 264). In rats, interruption of vagal afferent signaling from the GI tract to the CNS via SDA reduces anxiety-like behavior, attenuates conditioned fear extinction, promotes an imbalance of noradrenergic and GABAergic systems within the prefrontal cortex and NAc (127), and impacts cognitive flexibility (128), clearly supporting a role for vagally mediated conveyance of “gut feelings” to the brain. The vagus nerve also is proposed as the most important neural pathway for communication between gut microbes and the brain (28, 84, 140, 144), and vagal nerve integrity is necessary for anxiolysis and certain other CNS-mediated effects induced by intestinal probiotics in rodents (33, 88, 180). Considered together, the anxiolytic effects of SDA and the necessity of an intact vagus for the anxiolytic effects of GI probiotic treatment suggest that vagal afferent signaling normally promotes an avoidance-dominated motivational state that can be initiated by gut inflammatory signaling and can evolve into full-blown “sickness behavior,” as discussed below.

Behavioral Reorganization During Inflammation and Illness

The importance of interoceptive feedback signals for assigning priority to motivated behavior is particularly apparent during times of visceral challenge related to GI infection, toxin exposure, and inflammation. Such challenges elicit a myriad of defensive and protective physiological and behavioral responses associated with a motivational state of avoidance; when these responses escalate and converge, they are collectively referred to as “sickness behavior.” Sickness behavior in humans and other animals is characterized by self-reported or observed symptoms that can include lethargy, dysphoria, nausea, anxiety, social isolation, malaise, anhedonia, loss of appetite, sleepiness, hyperalgesia, reduced grooming (self-care), and loss of libido. These symptoms and responses represent an adaptive reorganization of motivated behavioral priorities that generally support survival and recovery after exposure to toxins and pathogens, including microbial pathogens in the gut. Bacterial translocation from gut lumen to mucosa can be triggered not only by injury but also by stress and high-fat diet, both of which promote local release of proinflammatory cytokines from gut immune cells (6, 88, 171, 249, 274). In this regard, the critical role of gut microbial populations in emotionality is supported by evidence that transplantation of gut microbes harvested from high-fat-diet-fed donor mice into chow-fed recipient mice promotes increased anxiety-like behaviors and other cognitive alterations in the recipients (36). Importantly, recipient mice in that study also displayed disrupted intestinal barrier function, leading to increased circulating endotoxin and signs of central neuroinflammation.

GI bacterial translocation can be mimicked experimentally by systemic administration of endotoxin, which stimulates cytokine production and reliably reduces food intake, willingness to work for food, sexual behavior, and social interaction in rats and mice (34, 49, 185, 271). Instead, avoidance behaviors predominate, as evidenced by increased alertness but decreased locomotion and exploration (32, 144, 170). Whereas systemic cytokines can influence the CNS via numerous routes, vagal afferents express receptors for proinflammatory cytokines and are activated in response to peripheral endotoxin or cytokine administration (118, 133, 146). Subdiaphragmatic vagotomy attenuates many of the behavioral and physiological consequences of systemic endotoxin or cytokine administration, including activation of the neuroendocrine stress response, hyperthermia, conditioned taste aversion, reduced willingness to work for food, and decreased social interaction (32, 34, 57, 82, 106, 132, 146, 261). Conversely, other studies utilizing total subdiaphragmatic vagotomy or the more refined SDA surgical approach (discussed above) demonstrated little or no effect of vagal interruption on the ability of endotoxin or cytokine treatment to induce sickness behavior (106, 118, 186, 225). It is possible that vagal afferents are critical for signaling immune challenge during early stages of GI bacterial infection, when low levels of cytokines produce local effects on vagal nerve endings but are too “dilute” to signal systemically. Indeed, subdiaphragmatic vagotomy was reported to block hyperthermic responses to low but not high peripheral doses of cytokine (107). Consistent with this idea, gut bacterial infections that are subthreshold to increase systemic proinflammatory cytokines can still promote anxiety-like behavior in rodents (145), concurrent with activation of cFos expression by vagal afferent neurons, cNTS neurons, and neurons within the BST and CeA (96).

Although the experimental evidence cited above was obtained in laboratory rats and mice, the impact of gut-to-brain and vagal afferent signaling in mood, emotion, and motivated behavior also has been demonstrated in humans. For example, an increased prevalence of emotional dysregulation, depression, and mood disorders has been reported in patients with inflammatory bowel diseases—which produce a chronic pro-inflammatory GI state and systemic cytokine imbalances (98, 173) known to affect vagal sensory neurons—suggesting that inflammation-related disruption of normal vagal signaling contributes to clinical mood disorders. Indeed, vagal nerve stimulation (VNS) is an approved and effective treatment for severe clinical depression (40, 95, 102, 150, 217) with efficacy for improving mood and reducing anxiety in clinical populations (40). VNS has been reported to attenuate encoding of affectively negative information, potentially due to engagement of the ventromedial prefrontal cortex, insula, and dorsal pons (40, 50). Given the proposed role of gut vagal afferent signaling in supporting innate anxiety-like avoidance behaviors in rodents (127), it seems paradoxical that VNS is anxiolytic in humans. However, electrical nerve stimulation is clearly quite different than natural recruitment of vagal afferents by GI stimuli, and it is unknown how “psychically effective” levels of VNS actually impact neural signaling within the cNTS and limbic forebrain.

Central Neural Pathways for Vagal Afferent Modulation of Motivated Behavior

As reviewed in the preceding sections, experimental and clinical research strongly supports an important role for vagal afferent signaling in gut-to-brain regulation of motivated behavior and emotional state. On the whole, this evidence suggests that recruitment of vagal afferents contributes to suppression of motivated approach behaviors and augmentation of avoidance behaviors. These contributions must be actualized through the cNTS, whose component neurons innervate CNS approach and avoidance circuits (FIGURE 2). It is important to note that cNTS neurons also receive synaptic input from spinal, brain stem, hypothalamic, limbic forebrain, and cortical regions, and are sensitive (via the area postrema) to blood-borne signals, including those signaling the availability of glucose and the presence of toxins and cytokines. Vagal afferent input from the GI tract modulates cNTS neural responses to these other inputs, providing “interoceptive tuning” of cNTS neurons that innervate central circuits implicated in motivated approach and avoidance behavior. For this reason, it is not surprising that activation of cNTS neurons tends to reduce both ingestive and non-ingestive motivated approach behaviors, while increasing avoidance behaviors.

Chemically Identified cNTS Signaling Pathways That Impact Motivated Approach and Avoidance Behaviors

Projection neurons within the rodent cNTS that carry GI interoceptive signals to higher brain regions include two non-overlapping neural populations [noradrenergic (NA) neurons comprising the A2 cell group, and GLP-1-expressing neurons] (FIGURE 3), plus a third population of CCK-expressing neurons that partially overlaps with the GLP-1 population, at least in mice (213). Together, these neural populations account for the majority of cNTS projections to other CNS regions that shape motivated behavior, including regions that modulate DA signaling within the MDS (11, 68, 110, 159, 169, 196, 200, 202, 232).

FIGURE 3.

Location of A2 and GLP-1 neurons within the hindbrain cNTS

Top: schematic coronal section through the rat caudal medulla illustrating the location of the cNTS (highlighted in blue). The red-lined box indicates the area shown at higher magnification in the photomicrograph on the bottom. In this image, dopamine beta hydroxylase (DBH)-positive NA neurons of the A2 cell group are green, whereas GLP-1-positive neurons are red. AP, area postrema; ts, tractus solitarius; cc, central canal; DMX, dorsal motor nucleus of the vagus. Figure is adapted from Ref. 147, with permission from Frontiers in Neuroscience.

In rats, cFos activation of A2 and GLP-1 neurons is markedly increased after exposure to stimuli that are perceived to threaten homeostasis, including restraint, systemic administration of supraphysiological doses of the gut peptide CCK, exposure to an illuminated elevated platform, exposure to predator odor, systemic administration of endotoxin or lithium chloride, or rapid ingestion of an extremely large meal (134, 147, 149, 168, 201, 204). Consistent with the ability of these “threat” stimuli to recruit cNTS projection neurons, each stimulus also suppresses subsequent food intake and increases avoidance behaviors, including anxiety-like behavior (147, 200). Indeed, as discussed further below, a growing body of evidence implicates signaling from A2, GLP-1, and CCK-positive neural populations to higher brain regions in mediating the powerful influence of interoceptive state on motivated behavior (147, 149, 214).

The A2 cell group.

The A2 cell group comprises NA neurons that are located entirely within the cNTS. Signaling from this NA population is poised to elicit avoidance behavior via receptor activation within numerous midbrain and forebrain regions, including the BST and CeA (16, 169, 202, 278) (FIGURE 2). A2 neurons appear to be critical for activating these limbic forebrain regions in response to vagally mediated interoceptive threats that increase avoidance behavior in rats, including large systemic doses of CCK (203), lithium chloride (207), and endotoxin. Regarding the latter, endotoxin-induced sickness behavior is associated with robust activation of A2 noradrenergic neurons that innervate the BST and CeA (30, 93, 94, 169). Moreover, the integrity of the A2 neurons is critical not only for BST-CeA neural recruitment following endotoxin but also for endotoxin-induced social withdrawal and reductions in exploratory behavior (91, 152).

Exteroceptive threats also recruit CNS circuits with important relays through the cNTS, supporting a role for vagal sensory modulation of exteroceptive stimulus-induced behavioral responses. In rats, immobilization or restraint stress, which elicits robust avoidance behavior, activates the large majority of A2 neurons and substantially increases extracellular levels of NE within the BST and CeA (63), whereas antagonism of NA receptors within the BST-CeA attenuates the ability of restraint stress to suppress social interaction and increase avoidance (44). Furthermore, removal or blockade of NA signaling in the BST blocks the anxiogenic effect of systemic yohimbine (which increases both peripheral and central NA signaling, including signaling from A2 neurons) (167, 278), and also attenuates the aversive properties of precipitated opiate withdrawal (11, 67, 68). In this regard, NA signaling within the BST (which originates primarily from the cNTS) has been proposed to drive the dysphoric, avoidance-dominated state of drug withdrawal by modulating inputs to the VTA that are implicated in anxiety, motivation, and affect (117, 235, 254).

Neurons synthesizing preproglucagon, the protein precursor for GLP-1.

Neurons synthesizing preproglucagon, the protein precursor for GLP-1, reside within the cNTS and subjacent reticular formation. GLP-1 released from intestinal endocrine cells can cross the BBB (45, 121), but its rapid degradation makes it unlikely to reach the brain in sufficient quantities to engage central GLP-1 receptors (112). Instead, the cNTS and medullary reticular formation provide the sole source of synaptic GLP-1 signaling within the brain, at least in rats (104, 137, 161, 204, 245). GLP-1 neurons are located adjacent to A2 neurons within the cNTS, but these are two distinct neural populations (147, 200).

Despite relatively sparse GLP-1 axonal inputs to the rat VTA and NAc (104, 200), GLP-1 signaling plays a significant and physiologically relevant role in suppressing reinforcement and approach behavior toward food, drugs, and other rewards via receptor signaling within the MDS and other brain regions (70, 110, 196–198, 231, 232). For example, selective pharmacological antagonism of GLP-1 receptors (GLP-1R) within the VTA, NAc, or LHA increases food intake (3, 76) and motivation for sucrose reward (243), whereas GLP-1R agonists reduce food palatability and motivation (70). Moreover, central administration of GLP-1R agonist attenuates the ability of cocaine to elicit phasic DA signaling within the NAc (87), whereas genetic knockdown of GLP-1R expression within the VTA increases cocaine self-administration (220).

Beyond reducing the reinforcing properties of otherwise rewarding stimuli, GLP-1 signaling also is implicated in innate and conditioned avoidance behavior (135, 209, 228, 244, 251). In rats, central antagonism of GLP-1R is anxiolytic (149), whereas GLP-1R agonists are anxiogenic (7, 126). GLP-1 neurons within the cNTS express cytokine receptors and display increased calcium permeability in response to cytokines (8), suggesting that GI or systemic inflammation may sensitize GLP-1 neurons to shift the animal toward behavioral states of avoidance. Together, these studies provide compelling evidence that central GLP-1 signaling pathways from the cNTS play a constitutive role in limiting approach and supporting avoidance behaviors.

The neuropeptide CCK.

The neuropeptide CCK is co-expressed by a subset of GLP-1 neurons in mice (90), whereas other CCK neurons appear to lack GLP-1 expression, and vice-versa. Similar to A2 neurons, CCK-expressing neurons within the cNTS are responsive to nutritional state in mice (213), and their activation reduces appetite (52). Interestingly, separate populations of cNTS CCK neurons innervate the PBN and paraventricular nucleus of the hypothalamus (PVH) (52, 214), and optogenetic activation of the CCK pathway from cNTS to PBN inhibits food intake and produces behavioral signs of aversion, including conditioned taste avoidance, whereas activation of the CCK pathway to the PVH inhibits food intake but has reinforcing effects in fasted mice (214). The reinforcing effect of CCK signaling from the cNTS to the PVH may reflect the satisfaction of satiety, which “puts the brake” on motivated approach behavior related to further food seeking and intake (52). Conversely, CCK inputs to the PBN may be recruited by excessive or pathological stimulation of vagal GI or other inputs to the cNTS that increase avoidance behavior (213, 214).

Interoceptive tuning of A2 and GLP-1 signaling from the cNTS.

The accumulated evidence reviewed here suggests that both interoceptive and exteroceptive signals can redirect motivated behavior from approach to avoidance via recruitment of cNTS neurons that directly or indirectly suppress activity within MDS-related reward circuits and increase activity within BST/CeA-related avoidance circuits. Motivated behaviors are profoundly affected by physiological (i.e., interoceptive) state (246), as exemplified by the strong influence of food restriction/deprivation on drug self-administration and other motivated approach and avoidance behaviors in rats (20, 41, 149). Recently, we discovered that overnight food deprivation eliminates or significantly reduces the ability of interoceptive and exteroceptive stressors to activate cNTS A2 and GLP-1 neurons in rats (148, 149), offering a potential mechanism for interoceptive tuning of motivated behavior. Overnight fasting also reduces anxiety-like behavior and attenuates BST neural activation in rats after acute stress (149). Thus overnight fasting-induced alterations in sensory signaling from the GI tract, likely due to altered nodose sensory signaling to the cNTS, may desensitize A2 and GLP-1 neurons to stimuli that normally activate these neurons in the fed state, thereby attenuating anxiety and other avoidance behaviors. The anxiolytic properties of caloric deficit may increase approach behaviors such as exploration and foraging, even in potentially dangerous environments, to ensure survival by replenishing energy stores (2, 54, 55, 149).

Conclusions and Future Directions

More than 130 years ago, William James and Carl Lange placed interoceptive signaling at the center of emotional regulation and motivated behavior. In this review, we have highlighted research demonstrating that neural, hormonal, cytokine, and microbial signals originating in the gut strongly influence goal-directed approach and avoidance behaviors, primarily by reducing approach and increasing avoidance. GI vagal afferents are well-positioned to carry gut-derived signals to the brain via synaptic inputs to the cNTS, with A2, GLP-1-, and CCK-positive neurons implicated in signaling to pontine, midbrain, hypothalamic, and forebrain regions that shape approach and avoidance behaviors.

Although substantial evidence supports a role for vagal afferents in the gut-brain connection, many questions remain. For example, although diet, stress, and other factors clearly interact to alter gut microbial populations, and although the gut microbiome can powerfully influence motivated behavior, focused research approaches are needed to determine the underlying mechanisms and precise role of vagal afferents in carrying microbial-related signals to the CNS. What is the role of vagal sensory signaling in the ability of probiotics or dietary changes to impact centrally mediated stress responses and affective behavior? What are the specific effects of microbial signals on plasticity of gene expression by vagal sensory neurons within the nodose ganglia, and on neurons within the cNTS that carry gut-derived signals to higher brain regions? Furthermore, given the powerful impact of early life events on the developmental assembly and life-long function of central visceral sensory and motor circuits, stress responsiveness, and emotionality (15, 88, 129, 130, 206, 208, 255, 262), it is critical to determine how pre- and postnatal stress exposure, maternal diet during pregnancy and lactation, offspring diet from infancy through adolescence, and other environmental factors affect gene expression and signal transmission in vagal sensory pathways from gut to brain.

It also is unclear whether gut-brain signaling pathways that impact motivated behavior differ substantially in male and female animals, including humans. The vast majority of the animal research reviewed in this article was conducted using male rats and mice, typically to avoid the potential “complications” of estrus cyclicity and/or the added time and expense necessary to conduct sufficiently powered experiments that include both sexes (19). However, gonadal hormone status clearly affects GI vagal afferent signaling to the cNTS and behavioral responses to GI signals in rodents (10, 77, 153). Future research on gut-brain signaling pathways must incorporate female subjects, especially considering the important role of vagal afferents in innate anxiety-like behavior in rodents (33, 127). In this regard, many anorexia nervosa sufferers report that their anxiety is alleviated by voluntary fasting/caloric restriction (which alters vagal afferent signaling to the cNTS), and many experience a resurgence of anxiety when food is reintroduced therapeutically (250). Anxiety is implicated in etiological and maintenance models of anorexia nervosa, bulimia nervosa, and other clinical eating disorders that occur predominantly in females (138, 253). For animal research to provide insights into these and other anxiety-related disorders that affect food intake and other motivated behaviors in humans, the research must include well-designed, adequately powered comparative analyses of both sexes.

To reveal the impact of GI vagal afferent signaling pathways on motivated behavior, it also is imperative that future studies utilize the most precise methodologies as they continue to evolve. The SDA technique (172) has been the “gold standard” for eliminating GI vagal afferent inputs to the brain, while preserving GI vagal motor functionality. However, SDA is technically challenging, disrupts half of the vagal motor signaling to the gut, and does not selectively target phenotypically unique subsets of GI vagal afferents. A promising new technique involves bilateral, toxin-induced destruction of GI vagal afferent cell bodies that express CCK receptors (71), representing a new, highly selective approach that preserves vagal motor innervation of the GI tract. Another laboratory is using powerful Cre-based technologies to identify and characterize the functional role of genetically specified subsets of vagal afferent neurons (265), opening up new opportunities for manipulating vagal afferents via optogenetics or DREADDs (designer receptors exclusively activated by designer drugs), approaches that already have been used to dissect the role of specific subtypes of cNTS neurons and their downstream targets in regulating motivated behavior (e.g., Refs. 52, 92, 214, 247). Considering that both the A2 and GLP-1 neural populations co-express glutamate and other neuropeptides (239, 279), selective opto- or chemostimulation techniques offer an advantage over the use of pharmacological approaches that target only a subset of neurotransmitter receptors, although these stimulation techniques are unlikely to closely model natural stimulation parameters. Nevertheless, these and other promising developments offer novel opportunities for animal researchers to revisit the role of GI vagal afferent signaling in the physiological and emotional components of motivated behavior. Such studies can only improve our relatively limited understanding of how interoceptive gut-to-brain signaling pathways affect human physiology, emotion, and behavior under conditions of health and disease.