able association between meal size and the Signals That Regulate Food time lag before initiation of a subsequent meal suggests that factors determining meal onset are coupled to those terminating the Intake and Energy meal (8). However, if confronted with pe- riodic food-associated stimuli, variable food Stephen C. Woods,* Randy J. Seeley, Daniel Porte Jr., availability, changing social situations, or Michael W. Schwartz novel stimuli, animals readily modify their eating schedule while maintaining long- Feeding behavior is critical for survival. In addition to providing all of the body’s ma- term energy homeostasis (5). Likewise, if cronutrients (carbohydrates, lipids, and proteins) and most micronutrients (minerals and physical constraints are placed on meal size vitamins), feeding behavior is a fundamental aspect of energy homeostasis, the process or the number of available meals each day, by which body fuel stored in the form of is held constant over long animals readily modify their meal pattern so intervals. For this process to occur, the amount of energy consumed must match that sufficient are consumed to precisely the amount of energy expended. This review focuses on the molecular signals maintain stores (5). Thus, neither the that modulate food intake while integrating the body’s immediate and long-term energy timing nor the size of meals is fixed, and needs. animals can accommodate a wide array of schedules to maintain energy balance. Be- cause of this flexibility, controls must exist that determine meal size once eating has For the past 50 years, two types of model energy needs, nor is meal termination tied begun, to ensure that total intake is regu- have dominated the study of food intake. to the replenishment of depleted substrates. lated. Consistent with this, a sizable litera- The conceptually simpler “depletion-reple- Rather, meal onset can occur for many rea- ture has documented the existence of meal- tion” models propose that some parameter sons, including habits and learned associa- generated signals, or “satiety factors,” that of immediately available energy be con- tions, opportunity, social factors, and time accumulate during eating and ultimately stantly monitored, with declining amounts of day (5). Similarly, meal termination can contribute to meal termination (and hence triggering meal onset. Thus, a meal is initi- be influenced by many extrinsic factors, as determine meal size) (Fig. 1). The ability of ated when available energy (for example, well as by signals generated by the con- these factors to impact meal size is modu- blood glucose or lipid availability or total sumption of food (5), including signals gen- lated (at least indirectly) by the size of the energy derived from these fuels) falls to a erated in proportion to fat mass. Hence, fat mass. threshold value and is terminated when animals consume meals when their lifestyle Compelling evidence that satiety factors substrate levels are sufficiently replenished. and the environment permit, and energy exist came in the early 1970s, when it was In principle, these models can account for regulation occurs through modulation of found that administration of the gut peptide both meal onset and meal termination. A the amount of food eaten at each meal to cholecystokinin (CCK) to rats before the well-known example is Mayer’s glucostatic maintain energy stores. The continuous but time of food availability caused a dose-de- hypothesis (1), which postulates that small variable needs of specific tissues are met by pendent decrease in meal size (9). Since declines in glucose concentrations or utili- utilization of recently ingested calories dur- then, hundreds of animal and human stud- zation trigger meal initiation. Both the liver ing and immediately after meals and by ies have documented the generalizability of (2) and the brain (1, 3) have been hypoth- drawing on stored energy at other times. this phenomenon (10). Key conclusions esized to monitor and respond to changes of The depletion of energy stored in the form from this literature are as follows. immediately available energy in the control of adipose tissue, therefore, increases food 1) CCK is but one of several peptides of food intake. However, although key pa- consumption, and this increase in consump- secreted from the gut during meals that, rameters related to energy depletion and tion occurs primarily by increasing meal when administered exogenously, reduce repletion correlate well with energy intake, size. When food availability or energy stores meal size. Other potential satiety peptides they correlate poorly with energy expendi- are severely depleted, however, animals also include members of the bombesin family ture. Depletion-repletion models, therefore, initiate more frequent meals to survive (5, (bombesin, gastrin-releasing peptide, and do not explain the matching of energy in- 6). A large and rapidly growing literature neuromedin B) (11) and glucagon (12). take with expenditure that results in the supports the hypothesis that food intake is 2) Blocking the action of endogenous long-term stability of fat stores. controlled within a lipostatic system for satiety factors with specific antagonists or The second type of model links homeostasis. purified antibodies increases meal size (10, intake to the amount of stored energy (fat 13), implying that meal size is normally mass) in the body. This “lipostatic model,” Regulation of Meal limited by these factors. originally articulated by Kennedy (4), posits Number and Size 3) Satiety peptides combine with other that signals proportional to the size of fat signals to influence meal size. For example, stores become integrated with other regula- The average number of meals per day varies when low-dose CCK-8 (a synthetic oc- tors of food intake. Thus, the onset of eat- widely among and within animal species. tapeptide of CCK) is coupled with mild ing is not necessarily tied to immediate When the daily light-dark cycle is fixed and gastric distension, meal size is reduced syn- when other constraints are controlled (for ergistically (14). S. C. Woods and R. J. Seeley are in the Department of Psychiatry, University of Cincinnati Medical Center, Post example, when there is ample food, low 4) At doses that elicit modest reductions Office Box 670559, Cincinnati, OH 45267–0559, stress, and no predators or social competi- of meal size, satiety factors do not produce USA. D. Porte Jr. and M. W. Schwartz are in the Depart- tors), species-specific meal patterns become nausea or distress in animals (10, 15). ment of Medicine, University of Washington and Veterans Administration Puget Sound Health Care System, Seat- apparent. Nocturnal laboratory rodents eat When administered small doses of satiety tle, WA 98108, USA. most food during the dark, with the largest factors, humans report feeling sated earlier *To whom correspondence should be addressed. E-mail: meals occurring near the time that lights go in a meal without other untoward symp- [email protected] on or off (7). Evidence of a weak but reli- toms (16).

1378 SCIENCE ⅐ VOL. 280 ⅐ 29 MAY 1998 ⅐ www.sciencemag.org REGULATION OF BODY WEIGHT:ARTICLES 5) Satiety peptides signal the brain hypothalamic melanocortin system. Hor- eries that the ob mutation resides in the through peripheral nerves (for example, va- mones that are regulated by adipose tissue gene encoding (23), a hormone se- gal afferent fibers) as well as through recep- ( and leptin) inhibit central anabolic creted from adipocytes, that the db muta- tors within the brain itself (17). This meal- pathways and stimulate central catabolic tion resides in the leptin receptor gene (24), related information is transmitted initially pathways (Fig. 2). and that leptin administration reverses obe- to the nucleus of the solitary tract, a brain- Parabiosis studies performed by Coleman sity in ob/ob but not in db/db mice (25). stem area that integrates afferent signals 30 years ago (22) suggested the existence of Because direct administration of leptin into arriving from the tongue (gustation) and hormones that regulate food intake in in- the CNS potently reduces food intake and gastrointestinal system (18). Afferent neu- verse proportion to fat mass. Specifically, because leptin receptors are expressed in ronal information then passes anteriorly genetically obese ob/ob mice were hypothe- hypothalamic areas important in the con- through the brainstem to the sized to lack such a hormone, and geneti- trol of food intake (26), the brain is thought and other forebrain areas. Importantly, cally obese db/db mice were proposed to be to be a primary target for leptin’s anorexic CCK is effective at reducing meal size in insensitive to the same hormone. These effect. Leptin appears to be transported into chronic decerebrate animals in which all hypotheses were confirmed by the discov- the CNS by a saturable receptor-mediated connections between the lower brainstem and the forebrain are severed (19). The necessary neuronal circuitry for this action Fig. 1. The role of satiety of satiety factors is therefore contained signals in the control of food intake. Once eating within the lower brainstem. has begun, food inter- 6) Although satiety peptides can alter acts with receptors on the size of individual meals, their repeated the tongue, the orophar- administration does not alter body weight. ynx, the stomach, and For example, when CCK-8 is automatical- the duodenum, as well ly administered to rats at the start of each as in the liver and other spontaneous meal, the size of each meal is organs. The detection, reduced, but the animals compensate by processing, and absorp- initiating more meals and thereby main- tion of food generate tain body weight (20). Hence, satiety fac- “satiety” signals that pro- vide negative feedback tors can potently affect food intake over Satiety to the CNS, and these the course of individual meals but by signals signals accumulate and themselves have limited influence on ad- interact to bring a meal iposity. It is this property that, when cou- to an end. The signals pled with the success of energy homeosta- reach the brain through sis over long intervals, implies the exis- visceral afferent nerve fi- tence of other signals, presumably propor- bers and through the tional to the size of the adipose mass. Such blood. long-term signals are not satiety signals per se but act over longer spans of time to suppress food intake by interacting with meal-related stimuli. It is through this in- teraction between long-term adiposity sig- nals and meal-related satiety signals that the control of food intake is integrated into the homeostasis of fat stores. Fig. 2. The general cir- Long-Term Regulation of cuitry underlying the reg- Negative Energy balance Positive Energy Balance ulation of body weight. Energy homeostasis is Low insulin and leptin High insulin and leptin Energy homeostasis is accomplished through achieved when anabolic and catabolic influences a highly integrated and redundant neurohu- ÐÐ+ + moral system that minimizes the impact of are in balance over long intervals. The hormones Catabolic Anabolic short-term fluctuations in energy balance on leptin and insulin are se- Anabolic Catabolic fat mass. Critical elements of this control creted in direct propor- system are hormones secreted in proportion tion to the size of the to body adiposity, including leptin and insu- adipose mass. During lin, and the central nervous system (CNS) states of negative energy Increased Decreased targets upon which they act (21). Candidate balance, the adipose food intake and food intake and CNS targets must exert potent unidirection- mass contracts, and less weight gain al effects on energy balance in response to leptin and insulin are se- changes in body fat. They include those that creted and reach the brain. As a result, anabolic pathways are disinhibited and catabolic pathways are suppressed, a stimulate food intake and promote weight condition that favors increased food intake and energy storage. Conversely, during states of positive gain (anabolic pathways), such as the hypo- energy balance, the adipose mass expands, leptin and insulin concentrations both increase, and the thalamic Y (NPY) axis, and resulting output from the brain favors reduced food intake and a reduction of the size of the adipose those that reduce food intake and promote mass. These key negative feedback circuits help ensure stability of the size of the adipose mass over (catabolic pathways), such as the time.

www.sciencemag.org ⅐ SCIENCE ⅐ VOL. 280 ⅐ 29 MAY 1998 1379 process, and the efficiency with which cir- (43). Thus, leptin and insulin induce a decline in body fat stores. This response culating leptin enters the brain is reduced spectrum of responses that leads to loss of occurs during fasting as well as in uncon- when plasma concentrations are elevated body fat stores. trolled insulin-deficiency diabetes mellitus, (27). Leptin receptors in brain capillary Except for leptin-deficient obese mice, and it arises through increased NPY gene endothelial cells (28) may mediate leptin’s most obese mammals have elevated plasma expression in ARC neurons and increased transport from blood to brain, and the ob- concentrations of leptin and insulin (30, NPY release into the PVN (50). NPY ac- servation that leptin concentrations in hu- 34, 44), and they appear to be resistant to tivity in this pathway is increased in other man cerebrospinal fluid correlate directly leptin-induced . Thus, it remains conditions associated with weight loss, such with plasma concentrations is consistent to be determined whether human as caloric restriction, lactation, and intense with its entry to the CNS from the plasma can be successfully treated with continuous (51), and this response is mediated, (29). leptin administration. Systemic insulin ad- at least in part, by reduced negative feed- Leptin and insulin share many properties ministration is not a viable option for in- back from insulin and leptin (21). NPY is as adiposity signals. Although insulin is se- ducing weight loss because of its peripheral overexpressed in the ARC of leptin-defi- creted from pancreatic beta cells rather effects that enhance fat storage and reduce cient ob/ob mice and leptin-resistant db/db than adipocytes, its circulating concentra- blood glucose concentrations. Moreover, at mice (52), and this response is attenuated tions are proportional to adiposity (30). least some forms of obesity are associated by leptin administration in ob/ob (but not Insulin also enters the CNS by a receptor- with resistance to insulin’s effects in the db/db) mice (53). In normal rats, leptin mediated, saturable transport process across brain. Thus, genetically obese Zucker rats administration also blunts the effect of fast- brain capillary endothelial cells (31), and (fa/fa, with a mutation of the leptin recep- ing to increase hypothalamic NPY messen- insulin receptors are located in the same tor gene) do not reduce their food intake or ger RNA (mRNA) levels (26). Similarly, key hypothalamic areas as leptin receptors body weight when given insulin intracere- central insulin administration attenuates (32). Finally, like leptin, insulin reduces broventricularly (45), suggesting that cen- the increase in hypothalamic NPY mRNA food intake and body weight in a dose- tral leptin activity may be necessary for levels that is associated with both fasting dependent manner when administered di- insulin signaling to occur. The nature of the and insulin-deficiency diabetes (54). Com- rectly into the CNS, and neither hormone interaction between leptin and insulin in bined with evidence that receptors for lep- produces symptoms of malaise (33). The the control of food intake, however, re- tin and insulin are concentrated in the secretion of both leptin (34) and insulin quires further study. ARC (26, 32), these results suggest that the (30) is influenced by the overall amount of hypothalamic NPY system is normally in- fat stores as well as by short-term changes in Central Effector Pathways hibited by negative feedback provided by energy balance (35), although insulin secre- both insulin and leptin. Weight loss lowers tion is stimulated acutely in response to The hypothalamus contains multiple neu- the concentration of these hormones, an meals, whereas leptin secretion is not. The ronal systems important in the regulation of effect that in turn activates the NPY sys- mechanisms governing leptin synthesis and energy homeostasis. For some systems (an- tem, facilitating the recovery of lost weight. secretion remain to be fully elucidated, but abolic), stimulation results in a net increase The finding that mice genetically deficient insulin appears to play a key role (36). In of energy intake and storage, and for others in NPY have apparently normal food intake contrast to its immediate effect on circulat- (catabolic), stimulation results in a net de- and body weight (55) suggests that other ing glucose, however, insulin’s effect on crease of energy intake and storage (Fig. 2). systems can compensate for NPY’s normal circulating leptin concentrations is delayed NPY is a that is widely activities in energy homeostasis. The ame- for several hours (37). expressed throughout the brain. In the hy- lioration of the obesity and hyperglycemia One remarkable aspect of the catabolic pothalamus, a well-defined pathway that is in ob/ob mice deficient in NPY, however, response to leptin administration is that the implicated in NPY’s effects on energy ho- demonstrates the potential contribution of weight loss appears to be due entirely to loss meostasis originates in the arcuate nucleus unchecked NPY signaling in the syndrome of fat (38). In fact, in some studies of nor- (ARC). Axons project from NPY cell bod- that results from reduced leptin signaling mal, lean animals, continuous leptin ad- ies in the ARC to the paraventricular nu- (55). ministration can virtually eliminate detect- cleus (PVN) (46), a major integration site Glucocorticoid (GC) hormones secreted able body adipose stores because of a rela- for inputs related to energy homeostasis. by the adrenal cortex are also implicated in tive increase of metabolic rate coupled with Central NPY administration promotes a energy homeostasis by effects on NPY. Ad- reduced energy intake (38). Under the in- state of positive energy balance and in- renalectomy attenuates the effect of fasting fluence of exogenous leptin, metabolic rate creased fat storage, with the most sensitive to increase both food intake and hypotha- remains normal or elevated despite progres- injection site being the PVN and adjacent lamic NPY gene expression, and these im- sive weight loss (39). In contrast, metabolic perifornical area (47), where NPY receptors pairments are reversed by GC administra- rate falls sharply in animals with compara- (both Y1 and Y5) are abundant (48). NPY tion (56). Moreover, GC deficiency en- ble weight loss due to caloric restriction, an injection into this brain area also reduces hances the ability of insulin and leptin to effect associated with reduced activity of SNS outflow to brown adipose tissue (49), promote anorexia and weight loss, and this the sympathetic nervous system (SNS) thereby lowering energy expenditure while effect is also reversed by GC administration (40). Because leptin increases SNS outflow simultaneously increasing the expression of (57). Taken together, these findings suggest (41), increased sympathetic activity may enzymes involved in lipogenesis in white that GCs are endogenous antagonists of mediate its action on metabolic rate. Cen- adipose tissue (49). Thus, central NPY ad- leptin and insulin in the control of energy tral insulin administration also reduces ministration increases energy intake, de- homeostasis. weight to a greater extent than can be creases energy expenditure, and increases NPY is not unique in its ability to in- accounted for by reduced caloric intake lipogenesis. Repeated NPY administration crease food intake and body energy stores (42). Furthermore, when animals infused into the PVN produces obesity within a (Table 1). Central administration of other with insulin centrally are given a choice, matter of days (47). hypothalamic [melanin-con- they reduce their intake of dietary fat while The ARC-PVN NPY pathway is activat- centrating hormone (MCH) and the re- sparing carbohydrate and protein reserves ed in response to signals associated with a cently described orexins A and B (also

1380 SCIENCE ⅐ VOL. 280 ⅐ 29 MAY 1998 ⅐ www.sciencemag.org REGULATION OF BODY WEIGHT:ARTICLES identified as “hypocretins 1 and 2”)] also (64). Leptin’s effect on energy homeosta- leptin and insulin is corticotropin-releasing stimulates food intake (58). As with NPY, sis, therefore, appears to involve, at least hormone (CRH), which is synthesized in expression of these peptides increases in in part, the activation of the hypothalamic PVN neurons (67). Central administration response to fasting (58), suggesting that melanocortin pathway. From this perspec- of CRH (or its recently described relative they may also play an important role in tive, it is not surprising that impairment of urocortin) reduces food intake and body energy homeostasis. melanocortin receptor signaling can cause weight, and endogenous CRH may be in- Of particular interest among central cat- obesity. volved in stress and illness (68). Hypotha- abolic systems are the melanocortins, pep- Evidence that melanocortins play a lamic CRH gene expression is increased by tides cleaved from the proopiomelanocortin critical role in energy homeostasis derives leptin administration (26) and inhibited by (POMC) precursor polypeptide. In the from the observation that genetic defi- GCs. Overproduction of CRH is implicated mammalian forebrain, POMC gene expres- ciency of the MC4 receptor in mice results in the anorexia associated with adrenal in- sion is limited to ARC neurons that project in hyperphagia and obesity (65). Ectopic sufficiency (21, 69), and reduced CRH sig- to areas that participate in energy ho- production of agouti, an endogenous an- naling may contribute to the actions of GC meostasis [such as the PVN (59)]. These tagonist of MC receptors that is normally hormones to promote weight gain and obe- brain areas also express melanocortin (MC) only expressed in skin, also produces an sity (21, 69). receptors (specifically, MC3 and MC4 re- obesity phenotype. Production of agouti in Leptin and insulin act, in part, by in- ceptors), and agonists of these receptors the brain of “yellow obese,” or “agouti,” fluencing the efficacy of meal-generated elicit anorexia, whereas antagonists have (Ay) mice antagonizes brain MC4 recep- satiety peptides. For example, the effect of the opposite effect (60). The endogenous tors and thereby results in obesity, whereas CCK to reduce meal size is potentiated by melanocortin implicated most strongly in production of agouti in skin antagonizes coadministration of either insulin or lep- the control of food intake and body weight melanocyte MC1 receptors and results in tin (70). In this way, the size of the fat is ␣-melanocyte–stimulating hormone (␣- yellow coat color (65). The agouti-related stores can influence daily feeding behavior MSH), which binds with high affinity to protein (AGRP), another product of ARC by modulating sensitivity of the animal to MC3 and MC4 receptors (61). neurons, shares sequence homology with signals generated by eating per se. An Because the CNS melanocortin system agouti and is an antagonist of MC3 and individual who has reduced exerts effects opposite to those of NPY, it MC4 receptors (66). Transgenic overex- leptin and insulin concentrations is there- was anticipated that expression of POMC pression of AGRP also produces an obesity fore less sensitive to single-meal satiety in the ARC would be regulated in a man- syndrome (66). signals; hence, larger meals are consumed ner opposite to that of NPY, and indeed Another hypothalamic catabolic neu- when conditions permit. Likewise, an an- fasting has been found to reduce POMC ropeptide that contributes to energy ho- imal that has recently overeaten and mRNA levels in the ARC (62). This re- meostasis and that is regulated in part by gained excess weight will be more sensi- sponse is likely to be a consequence of reduced leptin signaling, as the level of POMC mRNA is also reduced in the ARC of ob/ob mice and leptin administration to these animals reverses this defect (62). Because leptin receptors are expressed on ARC POMC neurons (63), melanocortin neurons appear to be a target of leptin action. Consistent with this hypothesis, the ability of centrally administered leptin to lower food intake and to activate PVN neurons (as measured by induction of c- Fos expression) is blocked by pretreatment with a melanocortin receptor antagonist

Hypothalamus Table 1. Candidate signaling molecules involved in energy homeostasis in the CNS.

Catabolic Anabolic Adiposity signals CRH* NPY* ␣-MSH* AGRP* CCK MCH Bombesin orexins A and B (ϭ Brainstem hypocretins 1 and 2) Satiety Somatostatin galanin signals Thyrotropin-releasing ␤-endorphin hormone Fig. 3. Integration of feeding-related signals from adipose tissue, the gut, and the brain. The circulating Calcitonin-gene– dynorphin adipose signals leptin and insulin penetrate the blood brain barrier and stimulate receptors on neurons related peptide in the hypothalamus. Satiety signals generated by ingested food enter the caudal brainstem (either as Neurotensin norepinephrine hormones that act locally on neurons within the brainstem itself or through visceral afferent signals Glucagon-like growth hormone– originating in the mouth or gut), where they influence reflexes related to the acceptance or rejection of peptide–1 releasing hormone food. Satiety information is also relayed anteriorly to the hypothalamus, where it is integrated with Serotonin cognitive information and adiposity signals. Increased activity of adiposity signals enhances the ability of *These molecules are particularly important in the regu- satiety signals to terminate a meal. The integrated information is then relayed back to the brainstem to lation of adiposity. areas controlling food intake and energy expenditure.

www.sciencemag.org ⅐ SCIENCE ⅐ VOL. 280 ⅐ 29 MAY 1998 1381 tive to meal-generated signals and tend to Langhans, ibid. 55, 497 (1996). Dorsa, Trends Neurosci. 11, 107 (1988). 3. L. A. Campfield and F. J. Smith, Int. J. Obes. 14,15 33. M. Chavez, R. J. Seeley, S. C. Woods, Behav. Neu- eat smaller meals over time. Two points (1990); P. Even and S. Nicolaidis, Brain Res. Bull. 15, rosci. 109, 547 (1995); T. E. Thiele et al., Am. J. are noteworthy. First, the location of in- 429 (1985). Physiol. 272, R726 (1997); S. C. Woods, E. C. Lot- tegration of adiposity and satiety signals is 4. G. C. Kennedy, Proc. R. Soc. London B Biol. Sci. ter, L. D. McKay, D. Porte Jr., Nature 282, 503 not known, nor is it clear how satiety 140, 579 (1953). (1979); S. C. Woods et al., Neurosci. Biobehav. Rev. 5. S. C. Woods and J. H. Strubbe, Psychon. Bull. Rev. 20, 139 (1996). signals interact with the hypothalamic an- 1, 141 (1994). 34. R. V. Considine et al., N. Engl. J. Med. 334, 292 abolic and catabolic systems described 6. J. H. Strubbe and J. Gorissen, Physiol. Behav. 25, (1996); M. Rosenbaum et al., J. Clin. Endocrinol. above. The observation that decerebrate 775 (1980); S. C. Woods, Psychol. Rev. 98, 488 Metab. 81, 3424 (1996). (1991). 35. G. Boden, X. Chen, M. Mozzoli, I. Ryan, J. Clin. animals respond to satiety signals but do 7. H. R. Kissileff, Physiol. Behav. 5, 163 (1970); J. Le Endocrinol. Metab. 81, 3419 (1996). not regulate the size of their adipose mass Magnen, (Cambridge Univ. Press, London, 36. W. M. Mueller et al., Endocrinology 139, 551 (1998). (19) suggests that the forebrain is key to 1985). 37. S. Dagogo-Jack, C. Fanelli, D. Paramore, J. Broth- 8. J. Le Magnen and S. Tallon, J. Physiol. 58, 323 ers, M. Landt, Diabetes 45, 695 (1996); J. W. Kolac- the integration process. This concept is (1966). zynski et al., ibid., p. 699. compatible with the role of the hypothal- 9. J. Gibbs, R. C. Young, G. P. Smith, J. Comp. Physiol. 38. G. Chen et al., Proc. Natl. Acad. Sci. U.S.A. 93, amus as a major integration site for adi- Psychol. 84, 488 (1973). 14795 (1996). 10. G. P. Smith and J. Gibbs, in Multiple Cholecystokinin 39. N. Levin, C. Nelson, A. Gurney, R. Vandlen, F. De- posity signals (Fig. 3). Second, the time Receptors in the CNS, C. T. Dourish, S. J. Cooper, Sauvage, ibid., p. 1726. constant with which adiposity signals in- S. D. Iversen, L. L. Iversen, Eds. (Oxford Univ. Press, 40. T. Sakaguchi, K. Arase, J. S. Fisler, G. A. Bray, fluence food intake is much longer than London, 1992), pp. 166–182. Am. J. Physiol. 255, R284 (1988). 11. J. Gibbs et al., Nature 282, 208 (1979); E. E. Laden- 41. W. Haynes, S. Walsh, A. Mark, W. Sivitz, J. Clin. the span of one or a few meals. The best heim, K. E. Wirth, T. H. Moran, Pharmacol. Biochem. Invest. 100, 270 (1997). estimates in humans and other mammals Behav. 54, 705 (1996); L. J. Stein and S. C. Woods, 42. M. W. Schwartz et al., Endocrinology 130, 3608 suggest that whereas social and other con- Peptides 2, 833 (1983). (1992). straints constantly influence how much is 12. N. Geary, Neurosci. Biobehav. Rev. 14, 323 (1990). 43. M. Chavez, C. A. Riedy, G. van Dijk, S. C. Woods, 13. R. D. Reidelberger and M. F. O’Rourke, Am. J. Am. J. Physiol. 271, R727 (1996). eaten in individual meals, regulation of Physiol. 257, R1512 (1989). 44. R. C. Frederich et al., Nature Med. 1, 1311 (1995); adiposity is integrated only over intervals 14. G. J. Schwartz, P. R. McHugh, T. H. Moran, ibid. M. Maffei et al., ibid., p. 1155; M. K. Sinha et al., of several days (71). 265, R872 (1993). J. Clin. Invest. 97, 1344 (1996). 15. J. Antin, J. Gibbs, J. Holt, R. C. Young, G. P. Smith, 45. H. Ikeda et al., Appetite 7, 381 (1986). J. Comp. Physiol. Psychol. 89, 784 (1975); J. Holt, J. 46. T. L. O’Donohue et al., Peptides 6, 755 (1985). Summary Antin, J. Gibbs, R. C. Young, G. P. Smith, Physiol. 47. B. G. Stanley, S. E. Kyrkouli, S. Lampert, S. F. Lei- Behav. 12, 497 (1974); P. J. Kulkosky, L. Gray, J. bowitz, ibid. 7, 1189 (1986). Gibbs, G. P. Smith, Peptides 2, 61 (1981). 48. C. Gerald et al., Nature 382, 168 (1996). Like other homeostatic systems, weight 16. N. E. Muurahainen, H. R. Kissileff, A. J. Derogatis, 49. C. J. Billington, J. E. Briggs, M. Grace, A. S. Levine, regulation is notable for its highly inte- F. X. Pi-Sunyer, Physiol. Behav. 44, 645 (1988); N. E. Am. J. Physiol. 260, R321 (1991); G. Bray, Am. J. grated and redundant nature. It is there- Muurahainen, H. R. Kissileff, F. X. Pi-Sunyer, Am. J. Clin. Nutr. 55, 265 (1992). Physiol. 264, R350 (1993); F. X. Pi-Sunyer, H. R. 50. A. Sahu, P. S. Kalra, S. P. Kalra, Peptides 9,83 fore not surprising that multiple adiposity Kissileff, J. Thornton, G. P. Smith, Physiol. Behav. (1988); J. D. White and M. Kershaw, Mol. Cell. Neu- signals exist and that many CNS pathways 29, 627 (1982). rosci. 1, 41 (1989). participate in the response to these sig- 17. D. P. Figlewicz, A. J. Sipols, D. Porte Jr., S. C. 51. D. E. Lewis et al., Am. J. Physiol. 264, E279 (1993); nals. The response to weight loss resulting Woods, Brain Res. Bull. 17, 535 (1986); D. P. Fi- M. S. Smith, Endocrinology 133, 1258 (1993). glewicz, A. J. Sipols, D. Porte Jr., S. C. Woods, R. A. 52. S. C. Chua et al., Mol. Brain Res. 11, 291 (1991); from inadequate caloric intake is a case in Liddle, Am. J. Physiol. 256, R1313 (1989); T. H. J. P. H. Wilding et al., Endocrinology 132, 1939 point. Insulin and leptin concentrations Moran, L. Shnayden, A. M. Hostetler, P. R. McHugh, (1993). decrease and GC concentrations increase, ibid. 255, R1059 (1988); G. P. Smith, C. Jerome, R. 53. M. W. Schwartz et al., Diabetes 45, 531 (1996); T. W. Norgren, ibid. 249, R638 (1985). Stephens et al., Nature 377, 530 (1995). a combination that activates pathways 18. S. Ritter, T. T. Dinh, M. I. Friedman, Brain Res. 646, 54. M. W. Schwartz, D. P. Figlewicz, D. G. Baskin, S. C. that stimulate appetite and promote 53 (1994); J. B. Travers, S. P. Travers, R. Norgren, Woods, D. Porte Jr., Endocr. Rev. 13, 387 (1992); weight gain while simultaneously inhibit- Annu. Rev. Neurosci. 10, 595 (1987). A. J. Sipols, D. G. Baskin, M. W. Schwartz, Diabetes 19. H. J. Grill and G. P. Smith, Am. J. Physiol. 254, R853 44, 147 (1995). ing pathways that have the opposite ef- (1988); R. J. Seeley, H. J. Grill, J. M. Kaplan, Behav. 55. J. C. Erickson, K. E. Klegg, R. D. Palmiter, Nature fect. This combined effect maximizes the Neurosci. 108, 347 (1994). 381, 415 (1996); J. C. Erickson, G. Hollopeter, R. D. homeostatic response to weight loss and, 20. D. B. West, D. Fey, S. C. Woods, Am. J. Physiol. Palmiter, Science 274, 1704 (1996). 246, R776 (1984). 56. P. K. Green, C. W. Wilkinson, S. C. Woods, Endo- consequently, the efficiency with which 21. J. S. Flier and E. Marantos-Flier, Cell 92, 437 (1998); crinology 130, 269 (1992); P. Ponsalle, L. S. Srivas- depleted fuel stores are replenished. Al- M. W. Schwartz and R. J. Seeley, N. Engl. J. Med. tava, R. M. Uht, J. D. White, J. Neuroendocrinol. 4, though such a robust system for defending 336, 1802 (1997). 585 (1993). fat stores may have conferred a survival 22. D. L. Coleman, Diabetologia 14, 141 (1978). 57. M. Chavez et al., Physiol. Behav. 62, 631 (1997); 23. Y. Zhang et al., Nature 372, 425 (1994). K. E. Zakrzewska, I. Cusin, A. Sainsbury, F. Rohner- advantage during human evolution, it now 24. S. C. Chua et al., Science 271, 994 (1996); L. Gwo- Jeanrenaud, B. Jeanrenaud, Diabetes 46, 717 poses a formidable challenge for the treat- Hwa et al., Nature 379, 632 (1996). (1997). ment of obesity. Individuals who suffer 25. M. A. Pelleymounter et al., Science 269, 540 (1995); 58. L. DeLecea et al., Proc. Natl. Acad. Sci. U.S.A. 95, J. L. Halaas et al., ibid., p. 543; L. A. Campfield, F. J. 322 (1998); D. Qu et al., Nature 380, 243 (1996); T. from a regulatory defect that results in the Smith, Y. Guisez, R. Devos, P. Burn, ibid., p. 546. Sakurai et al., Cell 92, 573 (1998). defense of an elevated amount of body fat 26. J. G. Mercer et al., FEBS Lett. 387, 1113 (1996); 59. J. Z. Kiss, M. D. Cassell, M. Palkovits, Brain Res. appear to resist changes in energy stores M. W. Schwartz, R. J. Seeley, L. A. Campfield, P. 324, 91 (1984). Burn, D. G. Baskin, J. Clin. Invest. 98, 1101 (1996); 60. W. Fan, B. Boston, R. Kesterson, V. Hruby, R. Cone, with the vigor characteristic of lean indi- R. J. Seeley et al., Horm. Metab. Res. 28, 664 Nature 385, 165 (1997); K. Mountjoy, M. Mortrud, M. viduals (72). Finding a “magic bullet” for (1996). Low, R. Simerly, R. Cone, Mol. Endocrinol. 8, 1298 the treatment of obesity may therefore be 27. W. A. Banks, A. J. Kastin, W. Huang, J. B. Jaspan, (1994). L. M. Maness, Peptides 17, 305 (1996); J. F. Caro et 61. H. B. Schioth, R. Muceniece, M. Larsson, J. E. S. unrealistic. Rather, interventions directed al., Lancet 348, 159 (1996). Wikberg, J. Endocrinol. 155, 73 (1997). at multiple targets in the energy ho- 28. P. L. Golden, T. J. Maccagnan, W. M. Pardridge, 62. M. W. Schwartz et al., Diabetes 46, 2119 (1997); meostasis system might be necessary to J. Clin. Invest. 99, 14 (1997). J. E. Thornton, C. C. Cheung, D. K. Clifton, R. A. achieve and maintain weight loss. 29. M. W. Schwartz, E. Peskind, M. Raskind, E. J. Steiner, Endocrinology 138, 5063 (1997). Boyko, D. Porte Jr., Nature Med. 2, 589 (1996). 63. C. C. Cheung, D. K. Clifton, R. A. Steiner, Endocri- 30. J. D. Bagdade, E. L. Bierman, D. Porte Jr., J. Clin. nology 138, 4489 (1997). REFERENCES AND NOTES Invest. 46, 1549 (1967); K. S. Polonsky, B. D. Given, 64. R. J. Seeley et al., Nature 390, 349 (1997). ______E. Van Cauter, ibid. 81, 442 (1988). 65. R. D. Cone, D. Lu, S. Koppula, D. Inge, Recent Prog. 1. J. Mayer and D. W. Thomas, Science 156, 328 31. G. Baura et al., ibid. 92, 1824 (1993); D. Wu, J. Yang, Horm. Res. 51, 287 (1996); D. Huszar et al., Cell 88, (1967). W. M. Pardridge, ibid. 100, 1804 (1997). 131 (1997). 2. M. I. Friedman, Proc. Nutr. Soc. 56, 41 (1997); W. 32. D. G. Baskin, B. J. Wilcox, D. P. Figlewicz, D. M. 66. M. Graham et al., Nature Genet. 17, 273 (1997);

1382 SCIENCE ⅐ VOL. 280 ⅐ 29 MAY 1998 ⅐ www.sciencemag.org REGULATION OF BODY WEIGHT:ARTICLES

M. M. Ollmann et al., Science 278, 135 (1997). S. C. Woods, Am. J. Physiol. 250, R856 (1986); C. A. 73. Supported by grants from the NIH (DK 17844, DK 67. M. F. Dallman, Trends Endocrinol. Metab. 4,62 Matson, M. F. Wiater, J. L. Kuijper, D. S. Weigle, 54080, DK 54890, DK 12829, DK 52989, and NS (1993). Peptides 18, 1275 (1997). 32273), the Diabetes Endocrinology Research Cen- 68. N. Rothwell, Neurosci. Biobehav. Rev. 14, 263 71. J. M. De Castro, Neurosci. Biobehav. Rev. 20, 1119 ter and Clinical Nutrition Research Unit of the Univer- (1990); M. Spina et al., Science 273, 1561 (1996). (1996); J. Nutr. 128, 61 (1998). sity of Washington, and the Merit Review Program of 69. G. A. Bray, J. Fisler, D. A. York, Front. Neuroendo- 72. E. J. Drenick and D. Johnson, Int. J. Obes. 2, 123 the Department of Veterans Affairs and funds from crinol. 11, 128 (1990). (1978); D. D. Stallone and A. J. Stunkard, Ann. Be- the Division of Metabolic Diseases, Hoffmann-La 70. D. P. Figlewicz, L. J. Stein, D. B. West, D. Porte Jr., hav. Med. 13, 220 (1991). Roche, Nutley, NJ.

metabolic fitness as a measure of success, Strategies and Potential health professionals can shift the patient’s focus from unrealistic, culturally imposed Molecular Targets for Obesity goals (for example, dress size or belt size), to the more appropriate and achievable goal of Treatment better health (7). Classes of Anti-Obesity Drugs L. Arthur Campfield,* Franc¸ oise J. Smith, Paul Burn Anti-obesity drugs can be classified accord- Obesity is an increasingly prevalent and important health problem. Although treatment ing to their primary mechanism of action is available, the long-term maintenance of medically significant weight loss (5 to 10 on energy balance. When daily energy in- percent of initial body weight) is rare. Since 1995 there has been an explosion of research take matches daily energy expenditure, focused on the regulation of energy balance and fat mass. Characterization of obesity- body weight remains constant. If intake associated gene products has revealed new biochemical pathways and molecular tar- exceeds expenditure, then a state of positive gets for pharmacological intervention that will likely lead to new treatments. Ideally, these energy balance is achieved and body weight treatments will be viewed as adjuncts to behavioral and lifestyle changes aimed at will increase. Conversely, if energy expen- maintenance of weight loss and improved health. diture exceeds intake, then a state of nega- tive energy balance is achieved and body weight will decrease. The goal of all anti- obesity drugs is to induce and maintain a Obesity is an increasingly prevalent, costly, Assessing the Efficacy of state of negative energy balance until the and important health problem throughout Obesity Treatments desired weight loss is achieved (4, 5, 9–11). the world (1, 2). In the United States, the There are four general classes of anti- prevalence of obesity in adults is now 32%, Traditionally, the efficacy of a new obesity obesity drugs. (i) Inhibitors of energy (food) and the prevalence in children has risen by treatment is assessed by its effect on body intake (or appetite suppressants) reduce 40% over the last 16 years. Similar trends weight. By this criterion, a treatment is hunger perception, increase the feeling of are being seen worldwide (1). considered successful if it (i) prevents fur- fullness, and reduce food intake by acting Obesity is a particularly challenging ther weight gain, (ii) inducesa5to10% on brain mechanisms. As a result, these medical condition to treat because of weight loss from the initial body weight, drugs facilitate compliance with caloric re- its complex etiology. Body weight repre- and (iii) allows long-term maintenance of striction. (ii) Inhibitors of fat absorption sents the integration of many biological the weight loss once it is achieved (1, 7). reduce energy intake through a peripheral, and environmental components. The en- Recently, an alternative, medically gastrointestinal mechanism of action and vironmental components (3) can be mod- based outcome measure for obesity treat- do not alter brain chemistry. (iii) Enhancers ulated through behavioral changes such ment has been advocated by scientists and of energy expenditure act through periph- as healthy eating and physical activ- physicians (7). Rather than focusing pri- eral mechanisms to increase thermogenesis ity, whereas the biological components marily on body weight, body fat, or the body without requiring planned increases in are much more difficult to address. Chang- mass index (BMI ϭ weight/height2), this physical activity. (iv) Stimulators of fat mo- es in body weight are resisted by very measure, called “metabolic fitness,” tracks bilization act peripherally to reduce fat mass robust physiologic mechanisms that we the metabolic health of obese individuals. or decrease triglyceride synthesis or both are only beginning to understand (4–6). Metabolic fitness is defined as the absence without requiring planned increases in However, the recent explosion of research of biochemical risk factors associated with physical activity or decreases in food intake. on the altered biochemical pathways obesity, such as elevated fasting concentra- Importantly, the beneficial actions of all caused by single gene mutations in ani- tions of cholesterol, triglycerides, glucose, four drug classes can be easily overcome by mal models of obesity has dramatically or insulin; impaired glucose tolerance; or increased intake of food (especially calori- expanded our knowledge base of these elevated blood pressure. In this school of cally dense food items) or decreased volun- physiologic mechanisms (6). As a result, thought, weight loss is viewed not as a goal tary physical activity. efforts to develop innovative anti-obesity but as a modality to improve health (7). The major drugs used to treat obesity are drugs have intensified. Here, we discuss Many studies have shown that during peri- shown in Table 1. Currently, the only drugs some of the potential drug targets that ods of weight loss there is a uniform im- approved for use are a small set of centrally have emerged from this “new science” of provement in the profile of risk factors (1). acting appetite suppressants that reduce obesity. Interestingly, reductions in the biochemical food intake by modulating the concentra- risk factors may not always be dependent on tions of monoamine (se- The authors are in the Department of Metabolic Diseases, Hoffmann–La Roche Incorporated, 340 Kingsland Street, weight loss. For example, insulin sensitivity rotonin and norepinephrine or norepineph- Nutley, NJ 07110, USA. and cholesterol levels can be improved by rine alone) in the brain. This modulation *To whom correspondence should be addressed. E-mail: physical activity in the absence of weight can occur at the level of neurotransmitter [email protected] loss (1, 3, 8). The hope is that by using release or re-uptake or both. The identifi-

www.sciencemag.org ⅐ SCIENCE ⅐ VOL. 280 ⅐ 29 MAY 1998 1383