Key points

What and how much we choose to eat are influenced by a variety of factors. These include the palatability or taste of particular foods, what we have learned about specific foods through experience, social and cultural influences on what foods and what amounts of food are appropriate to consume, the relative availability and the cost of specific foods, and an interacting system of physiologic controls that serve to both maintain adequate nutrition and limit intake to maximize our use of consumed nutrients. The recent obesity epidemic makes it clear that environmental influences can have a tremendous effect on overall energy balance. Obesity rates began to increase in the United States in the 1970s and this can all be attributed to changes in the food environment. However, the changing food environment interacts with a set of physiologic controls that are important in the meal-to-meal controls of eating.

In this review, we concentrate on the roles of 3 interacting physiologic and neural systems important in feeding control ( Fig. 1 ). These are systems that mediate (1) signals related to metabolic state and nutrient availability, (2) signals that arise during a meal that serve to end that meal and maintain as state of satiety, and (3) affective signals related to taste and nutritional consequences that serve to reinforce aspects of eating. We will also identify how these systems interact in the defense of overall energy balance.

Overall physiologic controls of eating behavior.

Nutrient availability signaling

Studies of rodent genetic obesity models had long suggested the importance of circulating factors in overall body weight control. Having identified 2 different mutations in mice that led to obesity, led to parabiosis experiments involving 2 strains of obese (obese [ob/ob] and diabetic [db/db]) and normal mice in which the blood supply between 2 mice in a parabiotic pair was shared. The results led to the conclusion that ob/ob mice lacked a circulating satiety factor that, in its absence, results in greatly increased food intake and obesity, whereas the db/db mouse produced the factor but lacked the ability to appropriately respond to that factor. Twenty years later, Friedman and colleagues cloned the ob gene and named the protein that it produced “leptin” from the Greek “leptos” meaning thin, because this was a factor that helped maintain a normal body weight. Shortly thereafter, the leptin receptor protein was identified as the product of the db gene. Leptin is produced primarily in white fat and circulating leptin levels correlate positively with the fat mass, increasing in circulation as animals or humans become obese. Thus, leptin serves as a signal of the available stored energy.

The study of leptin’s actions has illuminated many of the brain circuits that contribute critically to the control of energy balance and provided a basis for understanding earlier lesion work demonstrating a role for hypothalamic nuclei in energy balance. Leptin receptors are expressed throughout the brain with a particularly high expression within hypothalamic nuclei and other brain regions with identified roles in energy balance. Interactions of leptin with its receptors within these hypothalamic nuclei result in the activation or inactivation of hypothalamic pathways containing various peptides that when administered into the brain either stimulate or stop eating.

A major hypothalamic site of leptin’s actions is the arcuate nucleus. The arcuate contains 2 distinct neuronal populations that express leptin receptors. The first are neurons that express the prepropeptide proopiomelanocortin (POMC). POMC is processed into multiple opioid and melanocortin peptides including the anorexigenic peptide α-melanocyte stimulating hormone. Central administration of α-melanocyte stimulating hormone or synthetic melanocortin agonists potently inhibits food intake. Leptin activates POMC neurons, resulting in both increased POMC expression and α-melanocyte stimulating hormone release at terminal sites. Arcuate nucleus POMC expression decreases with food deprivation and increases with overfeeding, suggesting a regulatory role for this peptide in overall feeding control. Important roles for melanocortin signaling in energy balance have been demonstrated in experiments examining the effects of POMC and melanocortin-3 or melanocortin-4 receptor knockouts. Furthermore, genetic mutations in various aspects of the melanocortin signaling pathway have been identified as monogenic causes of human obesity.

Leptin also interacts with arcuate neurons that express the orexigenic peptides, neuropeptide Y (NPY) and the endogenous melanocortin antagonist agouti-related peptide (AgRP). Leptin inhibits neuronal activity in these cells, reducing NPY and AgRP release and downregulates the expression of these peptides. When leptin levels are low, in times of nutrient depletion or food restriction, the leptin inhibitory tone on NPY/AgRP neurons is diminished, activity in these neurons is increased, and the orexigenic peptides NPY and AgRP are released. Lesions of these NPY/AgRP–containing neurons in adulthood results in rapid starvation.

The feeding stimulatory actions of both NPY and AgRP have been well-documented. Intracerebroventricular or direct hypothalamic injection of NPY potently stimulates feeding and repeated or chronic NPY administration results in obesity. Cell bodies of neurons expressing NPY are found in multiple hypothalamic nuclei, including the arcuate and dorsomedial hypothalamic nuclei. Chronic treatment with NPY or viral-induced NPY overexpression can result in obesity.

AgRP is an endogenous melanocortin antagonist whose expression is limited to the NPY/AgRP–expressing neurons within the arcuate nucleus. AgRP expression is upregulated in response to fasting. AgRP or synthetic melanocortin antagonists increase food intake when administered into the brain and their effects are long lasting. GABAergic signaling is an additional important output of NPY/AgRP expressing neurons in their interactions with arcuate POMC neurons exerting an inhibitory tone on anorexigenic signaling and on neurons in the midbrain parabrachial nucleus.

The hypothalamic paraventricular nucleus and the perifornical area of the lateral hypothalamus are important projection sites for arcuate POMC and NPY/AgRP neurons. The paraventricular nucleus contains neuronal populations that mainly express anorexigenic peptides and thus the outputs from this nucleus serve to limit food intake. Leptin and/or melanocortins activate paraventricular nucleus neurons containing corticotrophin releasing factor, oxytocin, and gastrin-releasing peptide and each of these peptides reduce food intake when centrally administered.

The perifornical region of the lateral hypothalamus contains neurons expressing the orexigenic peptides orexin and melanin concentrating hormone (MCH). Preproorexin expression is increased in response to deprivation and decreased in response to leptin administration and central orexin administration increases food intake. Furthermore, administration of an orexin 1 receptor antagonist inhibits eating, suggesting a role for endogenous orexin in food intake control. MCH expressing cells are located similarly in the perifornical region of the lateral hypothalamus, although they represent a distinct neuronal population. MCH expression is increased in response to fasting and is decreased by leptin administration. MCH administration increases food intake in a dose-related fashion and genetic overexpression of MCH results in obesity.

Although leptin is the adiposity signal that has received the most attention, insulin also acts in the hypothalamus as an adiposity signal. Insulin levels increase with increased adiposity, insulin is transported from the circulation into the brain, and insulin receptors are localized to the hypothalamus with a high concentration in the arcuate nucleus. Central insulin administration inhibits food intake and has been shown to modulate activity in the leptin responsive arcuate circuit, decreasing NPY messenger RNA expression and increasing activity in POMC neurons.

Arcuate neurons that respond to leptin and insulin have also been proposed to be responsive to alterations in the local concentrations of nutrients and in this way serve as sensors for both short- and long-term nutrient states. For example, arcuate POMC neurons are activated and NPY/AgRP expressing neurons can be either activated or inhibited by increasing glucose concentration. However, the role of these glucose-induced alterations in electrophysiologic activity in the control of eating is uncertain as brain glucose concentrations do not necessarily reflect changes in circulating glucose or increase in response to meals. Hypothalamic neurons are also responsive to changes in the local concentration of fatty acids and intraventricular administration of a long chain fatty acid has been shown to reduce food intake. These data have been interpreted to suggest a role for brain fatty acid concentrations as signaling nutrient availability. Finally, local hypothalamic administration of some amino acids has been shown to decrease food intake. However, whether such a mechanism is involved in signaling circulating protein availability has yet to be demonstrated.

Although the hypothalamus has been a primary focus of the study of anorexigenic and orexigenic neuropeptide signaling, the neural pathways regulating energy balance are clearly distributed to multiple brain sites. For example, leptin receptors are expressed in the nucleus of the solitary tract (NTS) in the dorsal hindbrain. Local leptin administration at this site reduces food intake and downregulation of NTS leptin receptors attenuates the ability of leptin to reduce food intake. Data such as these strongly support the view that the adiposity controls of food intake are distributed rather than simply localized to the hypothalamic arcuate nucleus.