Within the last 10 to 15 years, a number of discoveries have revised the way in which scientists view the role of the brain in the control of food intake (1). One aspect of the brain's influence is often characterized as the control of energy homeostasis. This term accounts for a number of factors arising from experimental studies on molecules and food consumption but seems to stop well short of explaining how brain processes articulate the variety of patterns of human feeding. It should be kept in mind that eating is 100% behavior, and this activity links the internal world of molecules and physiological processes with the external world of physical and cultural systems. It is not always clear the extent to which human eating patterns are a function of physiological or environmental pressure; this is, of course, the subject of extensive experimental study and debate. Because much of the current scientific activity on neural control of feeding is driven by the need to understand (and deal with) the causes of obesity, it will be necessary, at some stage, to reconcile the effects of the physiological mechanism believed to be responsible for eating control in the obese with the actual patterns of eating displayed (eating phenotypes) by obese people. Ultimately, the mechanisms and the behavioral phenotypes must match up. Initially, it is useful to consider which components of energy homeostasis are codified in specific molecular processes and neural pathways and to describe how the integration of diverse signaling systems (the codification) is translated into the expression of behavior and the accompanying subjective sensations.
BLUNDELL, JOHN E. Perspective on the central control of appetite.
Tonic and Episodic Processes
Current thinking about the brain's role in feeding is often traced to the concept of homeostasis proposed initially by Claude Bernard. This energy homeostasis is managed by two sets of signals. One set arises from tissue stores, especially adipose tissue, and reflects what is often referred to as long-term signaling, although an alternative term is tonic signaling (2). The chemical signals include leptin, insulin, and certain cytokines and possibly amylin, visfatin, and adiponectin. This signaling system reflects the metabolic state of adipose tissue and, it is argued, acts as a driver for the feeding component of energy homeostasis. Episodic signals arise largely from the gastrointestinal (GI)1 tract and are generated periodically by the act of eating. These signals exhibit a rhythmicity synchronized with episodes of eating, and the chemical components include cholecystokinin, glucagon-like peptide (GLP) 1, oxyntomodulin, ghrelin, peptide YY (PYY), and possibly other peptide hormones released from cells in and around the GI tract (3).
The integration of the tonic and episodic signals reflects the brain's recognition of the current dynamic state of energy stores and the oscillating flux of nutrients derived from eating and detected by episodic signaling. This integration is instantiated in a set of neural pathways and receptors that extend from the nucleus tractus solitarius and area postrema in the hindbrain, through to the discrete hypothalamic nuclear in the basal forebrain (4, 5). A representation of this system for energy homeostasis is depicted in Figure 1.
The episodic signals rise and fall in harmony with the pattern of eating. The largest group of these peptides provides signals for satiation (meal termination) and satiety (postmeal inhibition of eating). For example, cholecystokinin is released in response to food ingestion (especially fat and protein) and influences meal size through an action to terminate eating. More recently, it has been proposed that oxyntomodulin also acts as a terminator of eating. Additionally, GLP-1, which rises after eating, plays a role in inhibiting eating in the postprandial period; GLP-1 could, therefore, be considered a satiety peptide. Recently, considerable interest has been shown in peptide PYY 3–36, whose action may be mediated by the ileal brake phenomenon. The gastric peptide ghrelin is a meal-related signal with an appetite-inducing effect because it is increased before meals (when concentrations in the blood increase) and decreases after meals (6). Indeed, one common feature of the action of these peptides is to adjust the rate of gastric emptying, and this may play a critical role in the effect on appetite control. This means that the ultimate effect on the brain is mediated by afferent stimulation to the hindbrain.
However, it also appears that some of these potent chemical signals have a direct action on receptors in the brain. An important site is the arcuate nucleus, which contains two sets of neurons, one responsible for orexigenic and the other anorexigenic activity. Figure 1 indicates how these episodic signals can influence the orexigenic drive embodied in neuropeptide Y (NPY) and agouti-related peptide neurons. Substantial evidence indicates that activation of NPY can produce a potent increase in food intake. In close proximity to the NPY/agouti-related peptide neurons in the arcuate nucleus are neurons producing melanocortins. These neurons suppress feeding activity by action at specific melanocortin receptors (melanocortin 3 and melanocortin 4 receptors) on downstream neurons. This second stage activity is mediated by neuronal pathways comprising orexins and melanin-concentrating hormone in the lateral hypothalamic area. Consequently, at this level, the activation and inhibition of eating are embodied in distinct molecules with distinct neuronal connections. Both ghrelin and peptide PYY (3-36) are believed to exert at least part of their action on feeding by an interaction with NPY orexigenic neurons. For PYY (3-36), the action is thought to be mediated by the Y2 autoreceptor.
Integration between the episodic and tonic signals is also achieved at this level because the NPY and melanocortin neurons are influenced by both leptin and insulin that signal the state of energy-related metabolic activity. The system is said to be leptin dependent. Leptin, providing a signal from adipose tissue, inhibits NPY neurons and stimulates the melanocortin system. In principle, therefore, high levels of leptin, signaling large adipose stores, contribute to energy homeostasis by suppressing appetite. In practice, because obesity is characterized by the phenomenon of leptin resistance, this appetite-suppressing action may be quite weak.
The Paradox of Hunger and Obesity
One output of this central regulation of energy homeostasis is the tuning of sensations of hunger and fullness that accompany eating, which, some researchers claim, are responsible for the pattern of eating itself. This means that the feeling of hunger carries a strong likelihood that eating will occur, whereas a strong feeling of fullness would inhibit eating and preserve the intensity and duration of postprandial satiety. All of the episodic signals noted above produce marked changes in hunger and/or fullness when the peptides are experimentally manipulated or when they vary naturally according to the pattern of feeding. Because these peptides may exert their action by NPY neurons (which also interact with the melanocortins), it can be assumed that the change in hunger is mediated by molecular mechanisms in the arcuate nucleus. Therefore, the sensation of hunger, which obviously has cortical representation, is driven by the arcuate nucleus under the modulating influence of episodic signals coming directly from the GI tract peptides or by the hindbrain afferent pathways.
It follows that tonic signaling from adipose tissue by leptin levels will also influence the expression of hunger. Indeed, it is known that after weight loss, an increased intensity of hunger is inversely related to the blood level of leptin (7). In addition, the administration of leptin (in cases of lipidodystrophy) can suppress the feeling of hunger (8). However, there seems to be a paradox because in cases of obesity, where adipose tissue stores are large and leptin levels high, obese people have strong feelings of hunger. This can be demonstrated through ratings of hunger intensity and also by high scores on psychometric traits that reflect chronic hunger (9). This raises the question of why people carrying huge amounts of stored energy should feel hunger and experience the driving need to eat. Episodic signals clearly play a role, and it is likely that an empty gut, in which the satiety signals will be low, or a short period of fasting, in which the satiation signal will be low, will impact on the baseline orexigenic activity in the arcuate nucleus. Therefore, activating signals arising from an empty gut, or a short fast can overcome tonic signals arising from adipose stores and generate strong and compelling urges to eat. This activation could arise from the increased expression of endocannabinoid receptors or melanin-concentrating hormone-1 receptors known to take place in the gut after food deprivation. The existence of leptin resistance will only lead to a further weakening of tonic control over hunger. Insulin also has a role to play because insulin resistance is known to dampen the impact of satiety signals, thereby preventing an appropriate level of caloric compensation (10).
These mechanisms may help to explain why obese people, despite very large energy stores, still feel hungry and experience strong urges to eat. One strategic possibility for limiting food intake in obesity is to replicate the signaling effects of a full gut or the effects of a recent meal. Of course, this approach is well recognized and is being actively exploited by pharmaceutical companies developing anti-obesity drugs.
Homeostatic and Hedonic Systems
Considering dietary synergies in appetite control, it is important to recognize the existence of two influential systems. Alongside the tonic and episodic aspects of the system for energy homeostasis, there exists a brain system that mediates the role that sensory pleasure plays in eating; namely, the hedonic system. It is currently a matter of debate whether the weight gain that arises from a loss of effective appetite control is due to disturbances in the homeostatic pathways or to inappropriate sensitization of the hedonic network (11). A further question is whether or not the systems operate independently or whether they interact (12).
A reasonable proposal is that an exaggerated response to the sensory features of foods constitutes a risk factor that could undermine effective appetite control. Consequently, physiological satiation or satiety signals could be overwhelmed by the potency of the hedonic response to foods. In parallel with the pathways of the energy homeostatic system, the hedonic system involves glutamate, opioids, benzodiazepines, γ-aminobutyric acid, endocannabinoids, and dopamine pathways. The mesolimbic system is widely believed to provide the structural elements that codify the hedonic aspects of appetite. In turn, the nucleus accumbens appears to be a focus of hedonic processing. For example, opioid agonists induce feeding when injected into the nucleus accumbens, whereas opioid antagonists inhibit feeding. Because these effects are more pronounced when the food is palatable, it is argued that arcuate nucleus pathways modulate eating by adjusting the hedonic response to food. Some confirmation of the role of mesolimbic dopamine pathways in hedonic processing is provided by positron emissiontomography scans, indicating that an increase in dopamine release is correlated with the degree of experienced pleasure. A further study has demonstrated that the concentration of dopamine D2 receptors varies in obese people in relation to BMI.
A further step in understanding how the brain controls appetite is to consider the degree of separation or integration of the homeostatic and hedonic systems. Putting the question on the level of human experience, we can ask whether hunger and pleasure interact. Although experimental evidence indicates that the two systems can be separately manipulated and, therefore, operate with a degree of independence, the segregation is only partial. At the structural level, adiposity-related signals such as leptin can influence the expression of a behavior elicited through stimulation of a central reward pathway. In addition, endocannabinoid receptors, regarded as mediators of food related pleasure, are widely found in close relationship with neurons in the arcuate and lateral hypothalamic zones that are implicated in the physiological signaling of energy homeostasis (13).
At an experimental level, there is good evidence that palatable food can distort the normal compensatory responses to meals in human volunteers and, therefore, weaken satiety. In other words, strong hedonic signaling can overcome physiological satiety signaling. It has been cogently argued that palatable food can disrupt appetite regulation (14). These data indicate significant capacity for an interaction between homeostatic and non-homeostatic (hedonic) processes in the central control of appetite. Together, the systems will modulate not only the choice of foods consumed but also the temporal and quantitative expression of appetite control. Further light will be shed on this issue when it becomes possible to confirm the molecular instantiation of the twin attributes of liking and wanting in neural pathways (15).
Human appetite control is expressed through integrated sequences of behavior accompanied by oscillating states (e.g., hunger, fullness, urges to eat) and more stable traits (e.g., disinhibition and hunger). In part, these markers reflect the operations of the homeostatic and hedonic processes that have a molecular and structural representation in the brain. The balance between tonic and episodic signaling (and the transformation of these signals within the brain) and the balance between homeostatic and hedonic processes ultimately determine the willingness or reluctance of people to eat or not eat. Within the prevailing obesigenic environment, the precision of appetite control is undermined, and the neural integration favors overconsumption leading to weight gain. This gain in adipose tissue appears to further deregulate the control of appetite through the actions of both leptin and insulin resistance and through a favoring of hedonic over homeostatic mechanisms. Therefore, the control over appetite becomes progressively less accurate and sensitive as obesity develops. This makes it critically important to understand the neural processes that are involved in the appetite control system.
Nonstandard abbreviations: GI, gastrointestinal; GLP, glucagon-like peptide; PYY, peptide YY; NPY, neuropeptide Y.