Brainstem Circuits Regulating Gastric Function

Abstract

Brainstem parasympathetic circuits that modulate digestive functions of the stomach are comprised of afferent vagal fibers, neurons of the nucleus tractus solitarius (NTS), and the efferent fibers originating in the dorsal motor nucleus of the vagus (DMV). A large body of evidence has shown that neuronal communications between the NTS and the DMV are plastic and are regulated by the presence of a variety of neurotransmitters and circulating hormones as well as the presence, or absence, of afferent input to the NTS. These data suggest that descending central nervous system inputs as well as hormonal and afferent feedback resulting from the digestive process can powerfully regulate vago-vagal reflex sensitivity. This paper first reviews the essential “static” organization and function of vago-vagal gastric control neurocircuitry. We then present data on the opioidergic modulation of NTS connections with the DMV as an example of the “gating” of these reflexes, i.e., how neurotransmitters, hormones, and vagal afferent traffic can make an otherwise static autonomic reflex highly plastic.

OVERVIEW

The gastrointestinal (GI) tract possesses an intrinsic nervous plexus that allows the intestine to have a considerable degree of independent neural control. The stomach and the esophagus, however, are almost completely dependent upon extrinsic nervous inputs arising from the central nervous system (CNS) (1–6). CNS control over the smooth and coordinated digestive functions of the stomach is mediated by parasympathetic and sympathetic pathways that either originate in, or are controlled by, neural circuits in the caudal brainstem.

Sympathetic control of the stomach stems from cholinergic preganglionic neurons in the intermediolateral column of the thoracic spinal cord (T6 through T9 divisions), which impinge on postganglionic neurons in the celiac ganglion, of which the catecholaminergic neurons provide the stomach with most of its sympathetic supply. Sympathetic regulation of motility primarily involves inhibitory presynaptic modulation of vagal cholinergic input to postganglionic neurons in the enteric plexus. The magnitude of sympathetic inhibition of motility is directly proportional to the level of background vagal efferent input (7). Recognizing that the GI tract is under the dual control of the sympathetic and parasympathetic nervous systems, we refer the reader to other comprehensive reviews on the role of the sympathetic control of gastric function (7–9). The present review focuses on the functionally dominant parasympathetic control of the stomach via the dorsal motor nucleus of the vagus (DMV) and the means by which this input is modulated, i.e., plasticity of vago-vagal reflexes.

ORGANIZATION OF THE BRAINSTEM VAGAL CIRCUITS CONTROLLING THE STOMACH

Visceral Afferent Inputs to Brainstem Gastric Control Circuits

Vagal afferent fibers carry a large volume of information about the physiological status of the gut directly to brainstem circuits regulating gastric function. The cell bodies of origin for this afferent sensory pathway are contained in the nodose ganglion. Neurons in the nodose ganglia are the cranial equivalent of dorsal root ganglion cells in that they are bipolar and connect the gut directly with nucleus tractus solitarius (NTS) neurons in the brainstem with no intervening synapse (reviewed recently in References 10–13).

There are several types of visceral receptors and afferent fibers (7, 9, 14–31). Independent of their functions or modalities, however, all vagal afferents use glutamate as the primary neurotransmitter to transfer information to the NTS (32–42).

Electrophysiological experiments have shown that glutamate released onto presumptive GI NTS neurons activates both NMDA and non-NMDA receptors (35, 41, 43), as do putative cardiovascular NTS neurons (36, 44, 45). The release of glutamate at this interface between peripheral perception (sensory afferent fibers) and central integration (NTS neurons) is open to modulation by a wide variety of transmitter and hormonal agonists, including glutamate itself (46–51), cholecystokinin (CCK) (52–54), leptin (55, 56), tumor necrosis factor (57), and ATP (58), to name but a few. These data strongly suggest that vago-vagal reflex functions are subject to pre- as well as postsynaptic modulation.

Nucleus Tractus Solitarius

The general hypothesis of a vago-vagal reflex connecting vagal afferent input, brainstem NTS neurons, and vagal efferent projections to the stomach was formulated some years ago and has been reviewed extensively (6, 10, 38, 59–66). The early models presented a framework to explain many of the salient features of brainstem reflex control of the stomach. Simply put, reflex action is initiated by stimulation of sensory vagal afferent pathways, which activate second-order NTS neurons via glutamate action on NMDA and non-NMDA receptors. These NTS neurons can use several different neurotransmitters to control the output from DMV cells, which, in turn, control gastric functions and complete the vago-vagal loop (Figure 1).

Figure 1.

Cytoarchitecture of the dorsal vagal complex. Darkfield photomicrograph of a coronal section of rat brainstem at the level of the area postrema/intermediate level following application of HRP crystals to the subdiaphragmatic vagus. Note the intense labeling of the nucleus tractus solitarius (NTS), dorsal motor nucleus of the vagus (DMV), and nucleus ambiguus (NAmb). An area of the photomicrograph (dotted line) has been expanded into cartoon form to allow a more detailed illustration of the brainstem circuitry of gastrointestinal (GI) vago-vagal reflexes. Note that vagal afferent neurons, whose cell bodies lie in the nodose ganglion, receive sensations from the GI tract. The central terminals of these afferent fibers enter the brainstem via the tractus solitarius and terminate within the NTS, utilizing principally glutamate as their neurotransmitter. These afferent signals are integrated by neurons of the NTS that project to, among other areas, the adjacent DMV using, mainly, glutamate, GABA, or NE as neurotransmitters. Neurons of the DMV are the preganglionic parasympathetic motoneurons that provide the motor output to the GI tract via the efferent vagus, where they release acetylcholine onto their postganglionic target. Postganglionic parasympathetic neurons are either excitatory [cholinergic (ACh)] or inhibitory [nonadrenergic, noncholinergic (NANC)].

Visceral sensory afferents are organized in an overlapping topographic manner within the NTS subnuclei. Terminal fields from the intestine are represented in the subnuclei commissuralis and medialis, the stomach sends its afferent inputs to the subnuclei medialis and gelatinosus, and the esophagus to the subnucleus centralis (66–70).

Although sensory inputs from distinct visceral areas, such as, for example, the aortic branches or gastric branches, do not seem to converge on single NTS neurons (71), the same subnucleus receives sensory information from more than one peripheral organ. Afferent terminals involved in arterial baroreflex circuits as well as terminals involved in vago-vagal gastric reflexes, for example, impinge upon neurons similarly located in the subnucleus medialis of the NTS (17, 38, 67, 70, 72–76). This overlap of the terminal fields from different organs within the NTS makes the anatomical and functional identification of individual neurons difficult.

The subnucleus centralis (cNTS), however, provides a convenient exception in that it seems to receive inputs from vagal afferent fibers originating almost exclusively in the esophagus (17, 38, 66, 67, 77), making it an excellent model for the study of a population of second-order neurons controlling esophageal-mediated reflexes. The cNTS of the rat is located between the tractus solitarius and the lateral third of the DMV at a level of approximately 0.5 to 1.5 mm rostral to the calamus scriptorius (70). Note that although the cNTS projects to the DMV as part of a classic example of a vago-vagal reflex, it also projects to areas that are not related to CNS control of gastric functions. Indeed, the cNTS projects to the nucleus ambiguus (NAmb), the reticular formation, and the ependymal layer of the fourth ventricle, as well as other NTS subnuclei (66). Thus, although cNTS neurons receive sensory information almost exclusively from the esophagus, these neurons control outputs related to gastric, esophageal, cardiovascular, deglutitive, and respiratory functions. Neurons of the cNTS may also be subject to plastic changes in function that translate into dramatic shifts in the reflex responses to afferent input dependent upon, and appropriate to, the feeding status of the animal.

Dorsal Motor Nucleus of the Vagus

The parasympathetic motor supply to the stomach is provided by the efferent vagus nerve originating from neurons located in the NAmb and DMV. The NAmb contains the soma of cells contributing to the vagal control of the esophagus, upper esophageal sphincter, and cardiorespiratory system. We refer the reader to recent reviews dealing with the NAmb (6, 8, 63, 77, 78).

The cell bodies for the great majority of parasympathetic efferent fibers that project to the upper GI tract originate along the whole rostro-caudal extent of the DMV (79–81). The DMV is a paired structure in the dorsal caudal medulla adjacent to the central canal, the majority of whose cells project to neurons in the myenteric plexus or onto interstitial cells of Cajal (ICC) of the upper GI tract (19, 82–85), with the highest density of efferent fibers terminating in the stomach (87).

Retrograde tracing experiments have determined that the DMV is organized in medio-lateral columns spanning its rostro-caudal extent (79–81). Efferent projections originating from soma located in the medial portions of the nucleus form both the dorsal and ventral gastric vagal branches. Neurons in the lateral portions of the DMV send axons to the vagal celiac and accessory celiac branches, whereas scattered neurons in the left DMV provide axons to the hepatic branch. The medio-lateral organization of the columns is not maintained when the target organs of the various vagal branches are considered. In fact, the two gastric branches innervate the stomach, a portion of the proximal duodenum, and some visceral structures such as the pancreas; the celiac branches innervate the GI tract from the duodenum to the transverse colon; and the hepatic branch innervates parts of the stomach, liver, and proximal duodenum (87, 88).

Although the DMV may not be organized in a rigid organotypic manner, evidence suggests that the vagal innervation of the stomach is segregated within the DMV by function. For example, descending vagal pathways responsible for causing gastric contractions versus gastric relaxation [i.e., the nonadrenergic, non-cholinergic (NANC) pathway] appear to be localized in different regions of the DMV. That is, putative NANC-pathway neurons appear to be located in the caudomedial and rostrolateral divisions of the DMV, whereas the gastroexcitatory neurons are located in the more rostral and medial divisions of the DMV (66, 89).

The DMV is comprised of neuronal populations that are nonhomogeneous with respect to morphological features such as soma size and shape, number of dendritic branches, and extent of dendritic arborization (90–96). The significance of these morphological differences is not well understood, although a possible explanation is discussed below.

Huang and colleagues (95) have hypothesized that neuronal DMV subgroups in the human may form functional units innervating specific organs; our studies in the rat suggest a similar organization. Combined retrograde tracing, whole cell patch clamp, and postrecording neuronal reconstruction techniques show a functional and morphological correlation between DMV neurons and the peripheral target organs that they innervate (92). Specifically, in the rat, DMV neurons projecting to the gastric fundus have a smaller soma size and fewer dendritic branches than neurons projecting to the corpus, duodenum, or cecum; neurons that project to the cecum have the largest soma out of all these. Additionally, there is evidence to support a relationship between structure and function of neurons in the rat DMV; e.g., neurons responsive to gastric or intestinal distension can be distinguished into separate morphological groups (96). Furthermore, these different DMV subgroups produce different profiles of extracellular recorded action potentials. This last observation suggests that DMV neurons engaged in different functions may possess different membrane properties.

Indeed, several investigators have reported a large array of unevenly distributed membrane currents in the DMV (92, 94, 97–100). Using the current clamp configuration, Browning et al. (92) showed that gastric-projecting DMV neurons can be easily distinguished from intestinal-projecting DMV neurons by the characteristic shape and duration of the after-hyperpolarization (AHP). That is, intestinal projecting neurons have a much larger and slower AHP, in contrast to the smaller and faster AHP of gastric-projecting neurons. This characteristic implies that gastric DMV neurons are more prone to change membrane potential in response to synaptic inputs than larger DMV neurons that project to other targets. This is of particular relevance, as one of the distinguishing features of DMV neurons is that they maintain a spontaneous, slow (1–2 pulses sec⁻¹) pacemaker-like activity (65, 101, 102). This pacemaker activity implies that afferent inputs altering the membrane potential of these neurons by even a few mV, in either direction, cause dramatic changes to the vagal motor output (97, 103). One immediate implication is that the stomach, at rest, is controlled by a tonic vagal efferent outflow provided by small gastric-projecting DMV cells, the vagal motor output of which is continuously sculpted by impinging inputs. Conversely, larger intestinal DMV neurons need more robust synaptic inputs to achieve the same membrane displacement. This size principle, widely accepted for spinal motoneuron activation and recruitment (104), states that neurons with the smallest cell bodies have the lowest threshold for synaptic activation and, therefore, can be activated by weaker synaptic inputs. The resting membrane potential of gastric smooth muscle is nonuniform, with a gradient from approximately −48 mV in the proximal stomach to −70 mV in the distal stomach (105). The threshold for smooth muscle contraction is approximately −50 mV (106). The myenteric plexus of the stomach essentially serves as a follower of vagal efferent input (107): There is a high degree of fidelity for activation of gastric myenteric neurons by vagal efferent inputs (108). As a result, the proximal stomach is highly sensitive to minute changes in input from either the cholinergic excitatory branch or the NANC inhibitory branch of the DMV, and slight changes in DMV activity in either division translate into dramatic effects on gastric function.

Synaptic Connections Between the Nucleus Tractus Solitarius and the Dorsal Motor Nucleus of the Vagus

The majority of NTS neurons do not possess pacemaker activity; their inputs onto DMV neurons must be driven and are modulated by synaptic activity, either from the afferent vagus, from other CNS areas, or via circulating hormones (10, 41, 53, 57, 65, 109–118). By being subject to a vast array of modulatory activity, the synaptic connections between NTS and DMV by implication play a major role in shaping the vagal efferent output.

There are numerous NTS neuronal phenotypes that can potentially contribute input to the DMV and induce potent effects on vagally mediated gastric function (17, 119–134). Despite the presence of this vast array of neurotransmitters in various NTS subnuclei, electrophysiological data show that the NTS primarily controls the DMV through glutamatergic (37, 61, 101, 135, 136), GABAergic (61, 101, 137), and catecholaminergic (138–140) inputs. Electrical stimulation of the NTS subnucleus commissuralis, for example, evokes noradrenergic α2-mediated inhibitory potentials in approximately 10% of the DMV neurons (140), whereas an α1-mediated current can be evoked by stimulation of the A2 area between the subnuclei medialis and cNTS of the NTS (K.N. Browning & R.A. Travagli, unpublished data). Additionally, stimulation of other portions of the NTS evokes both inhibitory GABA-mediated or/and excitatory glutamate-mediated currents in DMV neurons (61, 101, 103, 141–144). Pharmacological and immunocytochemical experiments have determined that GABA released onto DMV neurons activates GABA-A receptor subtypes (101, 135, 146–148).

Most of the NTS-induced inhibition of the DMV is mediated by GABA, interacting with the GABA-A receptor; conversely, most of the excitation delivered to the DMV by the NTS is mediated by glutamate interacting with both NMDA and non-NMDA receptors (101, 144, 146, 149–152). Recent data from our laboratory show a major functional role of catecholamines (likely originating from the A2 area) in both excitatory and inhibitory control of DMV. Application of glutamate or catecholamine agonists in the dorsal vagal complex (DVC) has profound effects on gastric motility and tone (114, 138, 152–154). Glutamate injected into the medial portion of the DMV causes a brisk gastric excitation, whereas norepinephrine injections cause gastroinhibition, perhaps owing to the simultaneous activation of the NANC inhibitory pathway and inhibition of the cholinergic gastroexcitatory pathway (66, 138, 139).

Microinjections of glutamate and catecholamine antagonists in the same area, however, do not induce noticeable effects on gastric motility and tone unless GABAergic transmission is blocked (37, 135, 138). Conversely, microinjections of GABA antagonists into the DVC induce profound excitatory effects on esophageal motility and on gastric motility and secretion (135, 147, 148). Taken together, these data suggest that the DMV output is restrained by a tonic GABAergic input arising, most likely, from the NTS. Furthermore, they suggest that factors modulating GABAergic inputs from the NTS to the DMV may have a significant impact on vagal reflex control of the stomach. By contrast, glutamatergic and catecholaminergic inputs do not seem to play a major role in setting tonic vagal output to the GI tract but rather seem to be invoked phasically by specific reflexes. Thus, NTS neurons that display some type of spontaneous activity use GABA as their main neurotransmitter. Glutamatergic and adrenergic NTS neurons are probably silent unless activated by afferent input.

Vagal Efferent (Motor) Control

To better appreciate the role of the efferent vagus in controlling GI functions, one has to consider the short-term effects of vagotomy (reviewed recently in References 9 and 155). Acute vagotomy induces an increase in fundic tone and a decrease in antral motility, impairing the reservoir function of the stomach. The conflicting effects of vagotomy suggested to Pavlov that the vagus nerve controlled gastric motility through both inhibitory and excitatory pathways, a concept that has received ample support over the past 100 years.

The vast majority (>95%) of DMV neurons are cholinergic, i.e., choline acetyl-transferase immunoreactive (156). Some DMV neurons also express immunoreactivity for nitric oxide synthase (NOS) or catecholamines [e.g., tyrosine hydroxylase (TH)] (89, 156–159). Although the projections of NOS- (158) or TH- (159) positive DMV neurons target selective areas of the stomach, the physiological significance of this detail is not clear. Interestingly, in both rodent (66, 89) as well as in feline (160) models, these markers for alternate neurotransmitter synthesis (i.e., TH and NOS) tend to coincide with the location of neurons involved in the inhibitory effect of vagal input to the stomach. For example, NOS-containing DMV neurons are found in the extreme caudal and rostrolateral portions of the DMV. Stimulation of these areas evoke gastric relaxation (66, 89, 160). This is far from establishing a connection between NOS- (or TH-) positive neurons and gastroinhibition; the correlation, however, is quite intriguing.

Using an intact vagus-gastric myenteric plexus in vitro preparation, Schemann & Grundy (107) demonstrated that stimulation of vagal preganglionic fibers, likely originating from DMV somata, induces exclusively cholinergic nicotinic potentials. These data indicate that acetylcholine, interacting with nicotinic receptors, is the principal neurotransmitter released from vagal efferent terminals and that preganglionic vagal fibers excite only enteric neurons. Because activation of vagal efferent fibers can induce both excitatory as well as inhibitory effects on gastric smooth muscles (18, 38, 62, 66, 160–180), it is clear that both excitatory and inhibitory postganglionic neuroeffectors are released from enteric neurons in response to excitatory vagal input. Note that the inhibitory vagal action on the stomach is directed toward the control of motility and tone but not of gastric secretion (181–185).

The principal excitatory postganglionic neurotransmitter is acetylcholine acting on muscarinic receptors in gastric smooth muscle, ICC, and parietal cells. Activation of muscarinic receptors depolarize smooth muscle and the ICC to augment the smooth-muscle activity that drives peristalsis and tone (1, 2, 4, 26, 186–190).

Inhibitory postganglionic neurons comprise the NANC path between the nicotinic preganglionic vagal efferent fibers and gastric smooth muscle and ICCs. The two most likely candidates mediating this connection are nitric oxide and vasoactive intestinal polypeptide, although other mediators such as adenosine and/or serotonin (5HT) have also been implicated (84, 162, 164, 165, 169–171, 176, 179, 191–194). Activation of this vagal NANC path produces a profound relaxation of the proximal stomach and depresses motility in the antrum. Together, these excitatory and inhibitory vagal efferent control mechanisms provide the brainstem with the tools to exert a very fine control of gastric motility (10).

Vago-Vagal Reflexes

Afferent input to the NTS is organized in a distinct viscerotopic manner: The intestine is represented in the subnuclei commissuralis and medialis, the stomach in the subnuclei medialis and gelatinosus, and the esophagus in the cNTS (67–69). Elucidation of this viscerotopic organization of afferent information, together with the knowledge that the DMV may be organized in a more functional manner, led to the hypothesis that vago-vagal reflex functions are organized in a sort of functional matrix (82). According to this hypothesis, contributions of vagal afferent input from different regions of the gut lead to different patterns of gut-directed vagal efferent outflow. Accordingly, afferents serving one gut region synapse on a subset of NTS neurons that, in turn, terminate on an appropriate collection of DMV neurons, completing the reflex. Medial NTS neurons (i.e., the cells that receive vagal afferent input from the stomach and intestine) possess long mediolaterally oriented dendrites (~600 μm) that extend across the terminal zones of all gastrointestinal afferent inputs (195). A priori, such an arrangement tends to favor a convergence of vagal afferent input of different modalities and different visceral loci onto select populations of NTS neurons. Selected NTS neurons provide dominant local synaptic control over the appropriate DMV projections to the stomach (61, 82) and determine a specific pattern of gastric control (59, 60, 108). In general, stimulation of the proximal GI (i.e., esophagus, fundic stomach) vagal afferents activates some gastric-projecting vagal fibers while inhibiting others. By contrast, activation of vagal afferents from more distal sites in the antrum and intestine results in inhibition of practically all vagal outflow to the stomach (see reviews in References 10, 61, 62, and 108). In combination with other physiological studies, we see that vagal afferent input integrated by the NTS ultimately evokes gastroinhibition by either the withdrawal of cholinergic tone (i.e., generated by the pacemaker activity of DMV neurons), the activation of the vagal NANC pathway, or a combination of both, as described above.

This model provides the framework for the hardware responsible for coordinating vago-vagal responses. In its static form, one may presume that activation of any given afferent elicits the same hard-wired efferent response. Recent studies have, however, revealed a high degree of plasticity in available responses, such as in the vago-vagal reflex control over the stomach. One agonist signal may “gate” another, and the tonic effects of vagal afferent input may “gate” agonist responses. As mentioned above, DMV neurons are spontaneously active and highly sensitive to inputs from the NTS. This means that influences modulating NTS input to the DMV can have potent effects on the regulation of gastric motility. Although many examples of presynaptic modulation of the NTS are available, we focus below on the regulation of opioidergic presynaptic signaling, as it is perhaps the best-developed example.

Agonist and Afferent “Gating” of Opiate Effects on Connections Between the NTS and DMV

CNS opioid release is proportional to the anticipated quality of a reward (196–198). With respect to feeding, opioid release is associated with the expected quality of food to be consumed. It is clear that activation of central opioidergic mechanisms is associated with the consumption of large amounts of palatable food. In evolutionary terms, rapid consumption of scarce, palatable food was desirable, in that palatability is an excellent predictor of caloric density or metabolic usefulness. There is some evidence suggesting that opiates can act in the dorsal vagal complex to augment feeding as well as to coordinate gastric function in anticipation of large (and in evolution, scarce) palatable meals (199–201). It would be advantageous, under these circumstances, to have a “cephalic-phase” mechanism by which opioids could augment digestive processes in order to assimilate high-quality food as quickly as possible. It would be especially convenient if the effects of opioids to accelerate digestion could be gated or regulated by the presence of other agonists whose release would signal either imminent feeding or that feeding is taking place. In this way, the effects of opioids on digestion would work best while taking food rewards, but not during the taking of other rewards.

Activation of central opioid receptors, abundant also in the DVC, may signal the consumption of palatable food and elicit gastric relaxation (199, 200, 202–208). Enhancement of proximal gastric relaxation during feeding would assist in the assimilation of large, palatable meals. Teleologically speaking, such a mechanism would explain why celebrants at Thanksgiving dinner may complain that they can’t eat another bite of green beans, yet mysteriously “find room” for pie and ice cream. Although this laboratory group did not set out specifically to investigate the hedonic mechanism of gastric relaxation in response to palatable meals, our basic neurophysiological results do suggest that opioid effects within the DVC to provoke gastric relaxation are gated by factors associated with the taking of meals.

When we analyzed the cellular mechanisms of μ-opioid actions in the DVC, our group first showed that endogenous opioid peptides inhibit all of the glutamatergic but none of the GABAergic currents between the NTS and the DMV (141). The selectivity of the effects of opioids on the brainstem circuitry was explained by the presence of μ-opioid receptors on the surface of glutamatergic, but not GABAergic, profiles impinging on DMV neurons (141) (Figure 2). It is possible, however, to induce an inhibitory effect of opioids on approximately 60% of the GABAergic currents by activating the cAMP-PKA pathway either via forskolin or via pretreatment with hormones such as thyrotropin-releasing hormone (TRH) or CCK. TRH and CCK are well-known markers of cephalic-phase digestive functions (10, 209) and the consumption of fatty meals, respectively (210).

Figure 2.

Differential modulation of synaptic transmission. Depiction of the synaptic connections between the NTS and the DMV. (A) Under control conditions, the μ-opioid receptor (μ) is expressed on the nerve terminal of glutamatergic NTS profiles apposing DMV neurons; by contrast, on GABAergic nerve terminals, the μ-opioid receptor is internalized and associated with the Golgi apparatus (upper panel). Opioid agonists (e.g., Enk) can act to inhibit glutamate synaptic transmission, but not GABAergic transmission (lower panel). (B) High-power photomicrographs of the cloned μ-opioid receptor (MOR1)-immunoreactivity (-IR; TRITC filters, red) and γ-glutamyl glutamate-IR (Glu-IR; FITC filters, green; left panel) used as a marker for glutamate nerve terminals, and glutamic acid decarboxylase-IR (GAD-IR; FITC filters, green; right panel) used as a marker for GABAergic nerve terminals in the rat DVC. Note that MOR1-IR is colocalized with glutamate nerve terminals (yellow; arrows) but not with GABA nerve terminals. Scale bar: 15 μm. (C) Representative traces of evoked excitatory postsynaptic currents (eEPSCs) (left panel) and evoked inhibitory postsynaptic currents (eIPSCs) (right panel) in gastric-projecting DMV neurons voltage-clamped at −60 mV and −50 mV, respectively, evoked by electrical stimulation of the NTS. Perfusion with Enk inhibits the eEPSCs but not eIPSCs. Scale bar: 50 pA and 30 ms.

The increase in cAMP-PKA activity in NTS neurons promotes fast (within five minutes) but short-lasting (approximately 60 minutes) trafficking of μ-opioid receptors from intracellular compartments to the outer membrane of the GABAergic terminals. Once on the surface of the synaptic terminal, opioid receptors are available for activation by the μ-agonist, resulting in a decrease in the net charge of the GABAergic current (211). This mechanism may explain, in part, feeding-related changes in the effectiveness of central gastroinhibitory action of opiates. Opioid receptors are normally available for interaction with their endogenous ligand on a limited population of glutamatergic synapses impinging on DMV neurons of the cholinergic (gastroexcitatory) pathway. Inhibition of this glutamatergic input induces a partial cholinergic withdrawal in the stomach and, by consequence, a modest inhibition of gastric motility. Activation of the cAMP-PKA pathway in the NTS, for example by peptides such as TRH or CCK, induces trafficking of μ-opioid receptors on GABA terminals synapsing on DMV neurons that comprise the NANC pathway (211). Opioid agonists, by reducing the GABAergic input to these DMV neurons, increase the vagal output of NANC neurotransmitters to the target organ and, by consequence, further enhance its inhibition. By this means, the effectiveness of opiates to provoke gastroinhibition can be greatly amplified by the action of other agonists.

Agonist modulation of receptor trafficking may also be responsible for the dramatic and very different effects neuropeptide Y (NPY) has on gastric motility in the fed versus fasted state. Briefly, a vagally mediated increase in gastric motility and secretion is observed when NPY is administered centrally to fasted rats (212, 213), but gastric motility is reduced when NPY is administered to fed rats (213, 214). The cellular explanation of this intriguing difference in gastric responses to central administration of NPY is probably similar to those explanations regarding the effects of opioids at the level of the GABAergic synapse between the NTS and the DMV (62, 141, 143, 211, 215). Our group has shown recently that the GABAergic NTS synapse with DMV neurons is subject to receptor trafficking modulation by a number of neurotransmitter and hormone agonists, including TRH, CCK, 5HT, norepinephrine, pancreatic polypeptides, and opiates (103, 211, 216–221).

In another hypothesis, an additional type of short-term plasticity, determined by the state of activation of vagal afferent fibers, also influences receptor availability at the level of the DMV-NTS GABAergic synaptic connection. This, in turn, allows the vago-vagal reflex circuits to adapt their responses to the immediate physiological needs of the animal, i.e., to different phases of the digestive process. As mentioned above, this hypothesis is quite appealing because it suggests that multiple factors regulating the release of glutamate from vagal afferents (e.g., CCK, leptin, tumor necrosis factor, and glutamate itself) may modulate basic reflex functions.

Preliminary data from our group are starting to provide explanations for the cellular and structural mechanisms that underlie the capability of this circuit to adapt its responses to the visceral afferent state of activation. Although this state of activation is reflected in the levels of activity of the cAMP-PKA pathway in the GABAergic NTS nerve terminals impinging on DMV neurons, it is set, principally, by vagal afferent fiber inputs onto NTS neurons and/or terminals. We came to this conclusion by observing that following vagal sensory deafferentation, either by perivagal capsaicin or by selective sensory rhizotomy, opioids reduced the GABAergic currents between the NTS and the DMV, but without the need for increasing the activity of the cAMP-PKA pathway (215). These data imply that vagal sensory afferent inputs chronically dampen the level of activity of the cAMP-PKA pathway at the level of the GABAergic synapses between the NTS and the DMV, making the synapse unavailable for modulation. As mentioned above, glutamate is the main neurotransmitter of vagal afferent fibers impinging on brainstem circuits (32–35, 37–42, 44, 45). Following its release, glutamate activates both ionotropic and metabotropic receptors (mGluRs). All metabotropic glutamate receptor subunits identified to date have been found in the DVC (50). The presence of group II and III mGluRs, both of which play a role in the modulation of brainstem vagal circuits (47–49, 222, 223), is of particular interest, because their activation decreases the levels of cAMP (224). We have recently investigated whether these receptor groups are involved in setting the levels of cAMP-PKA activation in GI brainstem circuits and, by consequence, in determining the availability of the NTS-DMV GABAergic synapse to modulation.

Our data indicate that group II mGluRs, but not group III mGluRs, are involved in this type of modulation of the GI brainstem circuits. In fact, pretreatment with selective mGluR group II antagonists “primes” the μ-opioid receptors on GABAergic NTS-DMV synapses, making it possible for opioids to modulate the inhibitory currents without the need to pharmacologically increase the activity of the cAMP-PKA pathway (K.N. Browning, Z.L. Zheng, & R.A. Travagli, submitted manuscript).

In summary, sensory vagal afferent fibers use group II mGluRs to dampen the activity of the cAMP-PKA pathway in GABAergic NTS nerve terminals. While held in check, the low levels of cAMP-PKA activity keep the μ-opioid receptors in inaccessible intracellular compartments such that enkephalins do not have a modulatory effect on the NTS-DMV GABAergic currents. The use of glutamate both to activate the sensory vagal pathways and at the same time keep the GABAergic NTS neurons in a nonmodulatory state is an apparent conundrum. In fact, it implies that glutamate released by sensory vagal afferent fibers exerts opposite functions via simultaneously acting at ionotropic receptors (which carry information about the visceral organs and prepare the circuit for a sophisticated level of modulation to allow the appropriate response) and at mGluRs (which dampen the brainstem GABAergic circuitry and prevent its modulation by neurotransmitters such as pancreatic polypeptides, opioids, indolamines, and catecholamines).

We have to keep in mind, however, that group II mGluRs may be located perisynaptically (225) and may be activated by the release of glutamate that leaks from vagal afferent terminals on NTS neurons in an action potential–independent manner. If this is the case, then mGluRs and ionotropic glutamate receptors may coexist without interfering with each others’ functions; furthermore, peptides impinging on NTS neurons, such as TRH and CCK, that increase the activity of the cAMP-PKA pathway may overcome the tonic activation of group II mGluRs, increase the levels of cAMP, and induce altered sensitivity of neural circuits to a variety of inputs. The role of glutamate activation of group II mGluR, then, may be to change the synaptic state of the NTS-DMV GABAergic circuit. The widespread use by different neurotransmitters of the cAMP-PKA pathway argues in favor of its utilization in a general manner in the control of vago-vagal circuits. Also, depending on whether the affected GABAergic NTS-DMV synapse controls cholinergic excitatory pathways or NANC inhibitory pathways, a decrease in synaptic transmission may result in an increased or decreased vagal motor output, respectively. The physiological correlate of these experimental conditions is modulation of the NTS-DMV GABAergic synapse by neurotransmitters that are coupled to adenylate cyclase and that generate cAMP. Alterations in sensory inputs from the GI tract, such as those following activation of vago-vagal reflexes or changes in the feeding status, for example, are expected to modify the ability of tonic GABAergic inputs controlling the vagal brainstem circuits to be modulated. For example, the constant perception of ongoing GI tract activity exerts a tonic inhibition of cAMP-PK pathways in the DVC. Activation of vago-vagal reflexes or changes in the state of activation (i.e., from fasted to fed or vice versa) may change the levels of cAMP and influence the ability of circulating hormones or locally released neurotransmitters to modulate inhibitory synaptic transmission between the NTS and the DMV (Figure 3).

Figure 3.

Increasing the levels of cAMP in the brainstem induces receptor trafficking in NTS nerve terminals. (A) In control conditions, i.e., when the levels of cAMP are low either because the tonic release of glutamate from vagal afferent fibers activates group II metabotropic glutamate receptors (left) or because Gαpled receptors are not activated (right), the terminals of GABAergic neurons in the NTS store μ-opioid receptors (μ) in internal compartments associated with the Golgi complex. In this situation, μ-opioid agonists (e.g., Enk) cannot modulate the release of GABA onto DMV neurons. (B) Following increases in cAMP levels within the GABAergic nerve terminal, for example, by (1) activation of a receptor coupled to Gα., TRH or CCK, (2) antagonism of group II metabotropic glutamate receptors, or (3) removal of tonic vagal afferent input, μ-opioid receptors are released from the Golgi apparatus and translocated to the nerve terminal membrane, where opioid agonists can inhibit GABA synaptic transmission between the NTS and the DMV. (C) High-power photomicrographs of the cloned μ-opioid receptor (MOR1)-immunoreactivity (-IR; TRITC filters, red) and glutamic acid decarboxylase-IR (GAD-IR; FITC filters, green) used as a marker for GABAergic nerve terminals in the rat DVC. In control conditions (left panel), note the absence of MOR- and GAD-IR colocalized profiles. Following vagal afferent rhizotomy (right panel), many nerve terminals show MOR- and GAD-IR colocalized profiles (yellow; arrows). Images represent three-dimensional reconstructions from Z-stack image series. Scale bar: 10 μm. (D) Representative traces of eIPSCs evoked in a gastric-projecting DMV neuron voltage-clamped at −50 mV. Perfusion with Enk does not affect the amplitude of IPSCs evoked by electrical stimulation of the NTS. Following five minutes’ perfusion with substances that increase the cAMP levels, however, reapplication of Enk reduces the amplitude of the eIPSCs. Scale bar: 50 ms and 200 pA.

SUMMARY

GI brainstem circuits that are part of vago-vagal reflexes are comprised of sensory afferent fibers whose terminals impinge on NTS neurons. The NTS neurons then project to DMV cells, which in turn provide the preganglionic efferent fibers controlling cholinergic excitatory and NANC inhibitory postganglionic cells. The strategic location outside the blood-brain barrier of portions of brainstem vagal circuits makes them accessible to a multitude of circulating hormones, cytokines, and chemokines that can dramatically alter vago-vagal reflex responsiveness. Add to this the dense and phenotypically diverse descending projections from anterior forebrain structures to the dorsal vagal complex (8), and it becomes clear that the number of mechanisms by which vago-vagal reflex selectivity and sensitivity can be modulated is virtually limitless.

The presynaptic modulation, and hence plasticity, of these circuits allows for a finely tuned control over gastric functions. It is now apparent that these circuits are not static entities devoted to the relay of unprocessed information between the brain and the gut. Indeed, by controlling the levels of activity of the cAMP-PKA pathway in synaptic terminals, circulating molecules, descending neurotransmitter inputs, and primary afferent traffic can act on brainstem circuits to affect the sensitivity of the synapses to modulation. If variations of the cAMP levels are determined, for example by the substances circulating in different situations such as during fasting or following a meal or by differing levels of visceral afferent traffic, then the net effect of activation of a particular circuit can vary according to the cAMP levels within the synapse itself. This implies, then, that gross differences in the performance of gastric control reflexes that occur at different times of the digestive process may result from variations in the cAMP levels at GABAergic NTS-DMV synapses.

In conclusion, we are just starting to understand the cellular mechanisms that underlie the exquisite coordination of digestive processes with ongoing and anticipated changes in behavior as a consequence of the diversity of input. The work to be conducted in the years to come must investigate the circulating factors that can cause rapid changes in brain-gut control through their direct action on brainstem vagal reflex control circuits and the cellular entities involved in this control. A multitude of possible modulatory mechanisms exist within these circuits to guarantee speed, precision, and flexibility in the control of digestive processes.