The drive to regain is mainly in the brain.

THE EXTREMELY HIGH RATE OF RECIDIVISM in the treatment of human obesity is a well-documented phenomenon (18, 19, 57). The factors that underlie this high failure rate are widely debated, and no consensus has yet been reached. One seemingly insurmountable problem in identifying potential etiologic factors is that such studies become hopelessly mired in the interaction between homeostatic (metabolic) and nonhomeostatic (reward) controls of food intake in humans. For that reason, animal models have provided important new insights into the way in which mammals balance their intake and output of energy under tightly controlled conditions. In this issue of the American Journal of Physiology-Regulatory, Integrative and Comparative Physiology, MacLean et al. (39) present a superbly constructed and beautifully interpreted set of studies on the behavioral and metabolic responses during the development of obesity, chronic food restriction, and relapse after ad libitum access to food. To do this, they assessed energy intake and expenditure in obese rats for up to 16 weeks of restricted feeding (“weight-reduced”) followed by 8 additional weeks of refeeding (“relapsed-obese”). Given the average life span of rats ( 2 yr) vs. humans ( 70 yr), this experiment is comparable to observing reduced obese humans for more than 10 years under strictly controlled conditions. The depressing conclusion is that weight-reduced obese rats had a persistent decrease in their resting energy expenditure and increase in their metabolic efficiency that did not attenuate with continued restriction. Furthermore, when external constraints on food intake were removed, chronically restricted rats inexorably regained lost weight almost exactly back to the level they would have maintained if never restricted. They did this initially by increasing food intake while maintaining their lowered metabolic rate. As refeeding rats regained their projected unrestricted body weights (adipose stores), their rate of intake decreased and their metabolic rate normalized. Because similar responses appear to occur in humans (20, 56), the extrapolated message for obese humans seeking long-term weight loss is to find some way to increase your metabolic rate and control your hunger for life! The MacLean et al. study (39) tells us what happens but not why it happens. What and where is the internal set point that leads animals and humans to doggedly return to their previous body weights when they are either overor underfed for prolonged periods? Clearly, there is no one set point localized in one precise place. The defended body weight and adipose mass can be manipulated by changing dietary content and palatability (27), as well as by conditions that change neural function and the metabolic and hormonal state of the individual (13, 16, 25, 30, 37, 41, 42, 53). In fact, energy homeostasis is controlled by a continuous dialogue between the periphery and brain. Genetic, environmental, and psychosocial inputs are superimposed on this dialogue. The summated inputs are integrated within metabolic sensors in the periphery and brain and lead to changes in the regulated adipose mass (39, 40). Leptin and to a lesser extent insulin are the two currently known hormonal signals that tell the brain how much fat is stored in peripheral depots (60). These hormones converge on select brain areas that contain specialized “metabolic sensing” neurons (33). Such neurons have leptin and insulin receptors but also utilize glucose and other metabolic signals from the periphery as signaling molecules to control the rate of neuronal activity. These neurons summate both tonic and phasic hormonal, metabolic, and other hard-wired neural signals from the periphery and surrounding brain areas to alter membrane potential, firing rate, transmitter release, and gene transcription (33). The integrated outputs from these metabolic sensing neurons is passed on to neuroendocrine and behavioral effector systems involved in the control of energy homeostasis (33). This homeostatic feedback system seems primarily designed to ensure that individuals will consume and store sufficient food during periods of surfeit to allow them to survive times when fuel availability is limited. This suggests that survival is more dependent on the adequate functioning of anabolic than catabolic systems. During prolonged food restriction, adiposederived leptin and adiposity-associated insulin levels fall. Leptin and insulin both exert inhibitory feedback on anabolic hypothalamic peptides like neuropeptide Y (NPY) and positive feedback on catabolic peptides like proopiomelanocortin (POMC), the precursor of the melanocortin-3/-4 receptor agonist -melanocyte-stimulating hormone (25, 30, 52, 60). Central injections of NPY increase food intake and reduce energy expenditure and fat oxidation (2, 3), and melanocortin agonists have the opposite effect (4). When leptin or insulin levels decline during food restriction or when their central signaling is genetically disrupted, NPY expression increases and POMC expression decreases, and animals enter an anabolic state (17, 51, 52, 55, 58, 60). Thus the increased metabolic efficiency of food-restricted rats and their increased food intake and metabolic efficiency during relapse, as demonstrated by MacLean et al. (39, 40) and others (1, 5, 20, 26), are exactly what would be predicted from an examination of their hypothalamic NPY and POMC expression. Furthermore, replacement of either leptin or insulin acts to normalize both the anabolic tilt of central regulatory peptides and the peripheral metabolic rate even if fat stores are not restored (48, 49, 51). While vastly oversimplified with regard to the myriad other peripheral and central regulatory systems involved in this process, the general message is the same: anything that reduces the adipose mass or interferes with central leptin or insulin signaling leads to a net central and peripheral anabolic tone and fosters the development of obesity. The hypothetical set point for the level at which adiposity is regulated is determined by genetic, gender, perinatal, developAddress for reprint requests and other correspondence: B. E. Levin, Neurology Service (127C), VA Medical Center, E. Orange, NJ 07018-1095. (E-mail: [email protected]). Am J Physiol Regul Integr Comp Physiol 287: R1297–R1300, 2004; doi:10.1152/ajpregu.00582.2004.