Learning of food preferences: mechanisms and implications for obesity & metabolic diseases

Abstract

Omnivores, including rodents and humans, compose their diets from a wide variety of potential foods. Beyond the guidance of a few basic orosensory biases such as attraction to sweet and avoidance of bitter, they have limited innate dietary knowledge and must learn to prefer foods based on their flavors and postoral effects. This review focuses on postoral nutrient sensing and signaling as an essential part of the reward system that shapes preferences for the associated flavors of foods. We discuss the extensive array of sensors in the gastrointestinal system and the vagal pathways conveying information about ingested nutrients to the brain. Earlier studies of vagal contributions were limited by nonselective methods that could not easily distinguish the contributions of subsets of vagal afferents. Recent advances in technique have generated substantial new details on sugar- and fat-responsive signaling pathways. We explain methods for conditioning flavor preferences and their use in evaluating gut-brain communication. The SGLT1 intestinal sugar sensor is important in sugar conditioning; the critical sensors for fat are less certain, though GPR40 and 120 fatty acid sensors have been implicated. Ongoing work points to particular vagal pathways to brain reward areas. An implication for obesity treatment is that bariatric surgery may alter vagal function.

Fig. 1. Schematic diagram showing the main flow of information during the task of choosing food.

(1) Before ingestion, available foods with their environmental context are perceived through visual, olfactory, and taste cues that may recall memories from previous encounters. (2) Food items found safe and providing positive nutritional signals are selected/preferred over other available foods and ingested. (3) Selection is thereby modulated by the overall nutritional state monitored by the master metabolic sensor in the basomedial hypothalamus. (4) Once accepted and ingested, the chosen food elicits a large number of temporally contingent signals from interaction with components of the alimentary canal, including enteroendocrine cells and neuropod cells. (5) Select signals in the circulation or via primary afferents are used by the brain to initially sustain ingestion (appetition), and later stop ingestion (satiation). They are also used to update existing memories of the selected food, or form new memories. The three general functional brain areas indicated and the specific brain structures included do not necessarily represent the exact neural pathways and systems and rather serve heuristic models. Abbreviations: Acb nucleus accumbens, BA bile acids, IC insular cortex, OFC orbitofrontal cortex, PFC prefrontal cortex, VTA ventral tegmental area (mesolimbic dopamine system).

Fig. 2. Nutrient signaling in the gastrointestinal tract and its communication pathways to other organs and the brain.

The volume and osmotic effects of ingested foods interact with the muscular wall of the alimentary canal and can activate vagal stretch (Intramuscular arrays, IMAs) and tension receptors (Intraganglionic laminar endings, IGLEs). Entry of carbohydrates, proteins, and fats into enterocytes is facilitated by specific transporters localized to the brush border apical membrane. Sugars, amino acids, and lipids are then diffusing into the mucosal lamina propria. Enteroendocrine cells (ECs) represent about 1% of all intestinal epithelial cells that can synthesize and release one or more gut hormones. Once transported into these ECs, carbohydrates, amino acids, and lipids differentially engage intracellular signaling pathways eventually leading to membrane depolarization, increased calcium concentrations, and the release of hormone-containing vesicles into the lamina propria. Some ECs with specialized extensions into the lamina propria (neuropod cells) can release the neurotransmitter glutamate onto vagal afferent nerve terminals bearing glutamate receptors. In addition, sugars and nonnutritive sweeteners are detected by the sweet receptor T1R2/3 and can trigger the synaptic release of ATP in neuropod cells acting on P2R on vagal afferent terminals. Nutrients and hormones in the lamina propria then have access to the bloodstream, mucosal vagal nerve endings, and the lymph system. Nutrients and hormones taken up into the bloodstream (either directly or after transport through the lymphatic system) can interact with vagal sensors in the portal vein or liver and eventually with sensors in all other organs and specific areas of the brain. Crosstalk between different ECs and between ECs and common enterocytes, as well as crosstalk between ECs and the enteric nervous system (ENS) are not shown for simplicity. Also note that innervation of the gut, portal hepatic vein, and liver by dorsal root afferents (DRG), which can also mediate signals to the brain are not shown. Abbreviations: Molecular transduction mechanisms: GLUT2 glucose transporter-2, GLUT5 glucose transporter-5, SGLT1 sodium-glucose transporter-1, T1R2/3 sweet taste receptor, T1R1/3 umami taste receptor, T2R bitter taste receptor, PEPT1 peptide transporter-1, α-Gust α-gustducin, PLC phospholipase C, TRPM5 transient receptor potential cation channel subfamily M member 5, IP₃ inositol triphosphate, ASBT apical sodium-dependent bile acid transporter, GPBAR1 G protein-coupled bile acid receptor 1. Hormones and enzymes: GLP-1 Glucagon-like peptide-1, PYY peptide YY, GIP Gastric inhibitory peptide, CCK cholecystokinin, 5-HT serotonin, GOAT ghrelin-O-acetyl transferase, DPPIV dipeptidyl peptidase-4, FGF19 fibroblast growth factor 19/15, Apo A-IV apolipoprotein-4. Receptors on vagal afferents: GLP1R GLP-1 receptor, Y2R PYY-2 receptor, GIPR gastric inhibitory peptide receptor, CCK1R cholecystokinin-1 receptor, 5-HT3R serotonin-3 receptor, GHSR growth hormone secretagogue receptor, GLUR glutamate receptor, P2R purinoreceptor. Brain: PBN parabrachial nucleus, AP area postrema, NTS nucleus tractus solitarius, SC spinal cord.

Fig. 3. Nutrient-conditioned flavor preferences.

A Naïve mice given two-bottle access to “isosweet” nutritive sugars (glucose or sucrose) and nonnutritive sweeteners (sucralose, AceK) take 24 h or more to develop a preference for the sugar. Once trained, a sugar preference is expressed in less than 2 min [65, 68, 136]. B Naïve mice given one-bottle access (1 h/day) to a CS+ flavored saccharin solution paired with IG 16% glucose infusion increase their licking response within 10 min in the first test session (CS+1) compared to prior sessions with a CS- flavor paired with IG water (CS-0). In subsequent one-bottle CS+ sessions licking is increased from the very first min. In two-bottle tests all mice licked more for the CS+ than CS−; 80% CS+ preference. Because mice were not infused in 2-bottle tests they licked much more than in one-bottle tests [94].

Fig. 4. Proposed gut–brain pathways mediating postoral sugar and fat appetition in mice.

(1) SGLT1-mediated glucose transport across the brush border membrane leads to enterocyte depolarization and the release of glutamate from neuropod cells reaching into the lamina propria. The synaptically released glutamate excites glutamate receptors located on sensory nerve terminals originating from unknown vagal afferent neuron populations in both the left and right nodose ganglia and projecting through both left and right cervical vagus [68]. (2) Glucose activates via SGLT1 a selective population of vagal afferent neurons and in turn a selective population of proenkephalin-expressing neurons in the left and right NTS [65]. (3) Glucose metabolism can influence brain reward circuitries by an unknown metabolic sensor and pathway [96]. (4) After absorption and reaching the hepatic-portal vein and liver, glucose activates the mesolimbic dopamine system by acting in an unknown fashion on sensory terminals of vagal afferent fibers passing through the common hepatic branch associated with the left cervical vagus [83]. (Note that these authors speculate that the postoral sucrose may act on neuropod cells or hepatic-portal sensors, admit that there must be pathways in addition to the hepatic vagus; and their outcome behavior is operant sugar seeking) (5) The presence of intestinal glucose is signaled in an SGLT1-dependent fashion via dorsal root afferent neurons passing through the celiac ganglia to inhibit hypothalamic AgRP neurons [103]. (Note that there was no preference testing in this study). Inhibiting AgRP neurons conditions flavor preferences [137]. (6) Fatty acids (FA) derived from dietary fat acting in part on intestinal GPR40 and GPR120 sensors signal brain reward circuits via undefined pathways to condition CS+ flavor preferences and promote fat-seeking behavior [112]. (7) Dietary fat acting on unspecified intestinal sensors activate brain reward systems via CCK-sensitive vagal afferent fibers passing through the right nodose ganglion to condition relative preferences for dilute or concentrated fat emulsions and promote operant fat-seeking behavior [84]. (8) Dietary fat acting on unspecified intestinal sensors via vagal afferent neurons to inhibit hypothalamic AgRP neurons [103]. Note that studies in rats indicate that the upper small intestine is partially innervated by vagal fibers traveling in all the anterior and posterior celiac, the anterior and posterior gastric, as well as the gastroduodenal branch dividing from the common hepatic branch [28].