Unraveling the brain regulation of appetite: lessons from genetics

Transcription

1 F o c u s o n n e u r a l c o n t r o l o f f e e d i n g r e v i e w Unraveling the brain regulation of appetite: lessons from genetics Giles S H Yeo 1 & Lora K Heisler 2 Over the past 20 years, genetic studies have illuminated critical pathways in the hypothalamus and brainstem mediating energy homeostasis, such as the melanocortin, leptin, 5-hydroxytryptamine and brain-derived neurotrophic factor signaling axes. The identification of these pathways necessary for appropriate appetitive responses to energy state has yielded insight into normal homeostatic processes. Although monogenic alterations in each of these axes result in severe obesity, such cases remain rare. The major burden of disease is carried by those with common obesity, which has so far resisted yielding meaningful biological insights. Recent progress into the etiology of common obesity has been made with genome-wide association studies. Such studies now reveal more than 32 different candidate obesity genes, most of which are highly expressed or known to act in the CNS, emphasizing, as in rare monogenic forms of obesity, the role of the brain in predisposition to obesity. The drive to consume food is one of the most primitive of instincts promoting survival. It has been shaped by millions of years of evolution and has provided living creatures with powerful and redundant mechanisms for adapting and responding to times of nutrient scarcity. It is the recent challenge of adaptation to food abundance, coupled with concurrent changes in lifestyle and working practices, which has contributed to obesity becoming one of the greatest public health challenges of the twenty-first century. These changes have occurred only over the past 25 years, against a constant gene pool. Therefore, few could dispute that the obesity epidemic has been driven by an almost seismic shift in lifestyle and environment. However, individuals respond differently to these obesigenic environmental changes, and it is increasingly clear that this variation in response has a potent genetic element. Indeed, studies of body mass index (BMI) correlations of monozygotic, dizygotic, biological and adopted siblings reveal the heritability of fat mass to be between 30% and 70% (refs. 1,2). Consequently, genetic approaches offer a powerful tool for characterizing the molecular and physiological mechanisms of body weight control and allow us to understand how these may become ineffective in the obese state. In the past two decades, insights from human and mouse genetics have shown that the brain is the primary orchestrator of energy intake and body weight. Specifically, monogenic alterations in the melanocortin, leptin, 5-hydroxytryptamine (5-HT; serotonin) and brain-derived neurotrophic factor (BDNF) systems have shown these pathways to be required for normal energy balance and body weight. In this review, we discuss how lessons from genetics, including studies of monogenic disorders and, more recently, polygenic clues arising from genome-wide association studies (GWAS), have highlighted the central role of the brain in the regulation of feeding behavior. 1 University of Cambridge Metabolic Research Laboratories, Institute of Metabolic Science, Addenbrooke s Hospital, Cambridge, UK. 2 Department of Pharmacology, University of Cambridge, Cambridge, UK. Correspondence should be addressed to G.S.H.Y. (gshy2@cam.ac.uk) or L.K.H. (lkh30@medschl.cam.ac.uk). Published online 25 September 2012; doi: /nn.3211 Severe obesity and monogenic syndromes Melanocortin pathway. The melanocortin pathway is a principal nexus through which nutritional cues converge and signal to influence energy balance 3. It is therefore unsurprising that many of the known monogenic and some of the polygenic causes of obesity are linked to alterations in the melanocortin pathway. The CNS melanocortin pathway includes the polypeptide precursor of the endogenous melanocortin receptor agonists, pro-opiomelanocortin (POMC), which is expressed in the arcuate nucleus of the hypothalamus (ARC) and brainstem nucleus of the solitary tract (NTS), and the endogenous melanocortin receptor antagonist/inverse agonist agouti-related protein (AgRP), which occurs exclusively in the ARC 3 (Fig. 1). Brain POMC is post-translationally processed to produce the melanocortin agonists α- and β-melanocyte-stimulating hormone 4,5, and AgRP is released by specific ARC neurons 6, to influence energy balance predominantly through action at melanocortin-4 receptors (MC4Rs) 7. In addition to blocking melanocortin agonist action at melanocortin receptors, AgRP neurons also directly inhibit POMC neurons by releasing GABA 8. ARC POMC and AgRP neurons are responsive to nutritional signals including adipose, gut and pancreatic hormones, brain-derived energy balance associated neurotransmitters and neuropeptides, and dietary nutrients (Fig. 2). POMC and AgRP neurons appear to integrate these signals and, through a number of different effector pathways, a few of which are discussed in detail below, influence energy balance. The AgRP neurons also express neuropeptide Y (NPY) and GABA. Recent reports show that optogenetic 9 or designer receptor 10 activation of AgRP neurons stimulates voracious feeding and weight gain, and this effect occurs in mice in which melanocortin signaling is blocked 9, suggesting that stimulating either NPY and/or GABA release is responsible for the drive to eat. Activation of AgRP neurons stimulates feeding by inhibition of single-minded 1 (Sim1)-expressing target neurons in the paraventricular nucleus of the hypothalamus 11. Conversely, optogenetic activation of POMC neurons inhibits food intake 9. Although traditional targeted gene deletions of mouse Agrp, Npy or Slc32a (encoding the vesicular GABA transporter) has little nature neuroscience VOLUME 15 NUMBER 10 OCTOBER

2 R e v i e w NH 2 POMC COOH mediator of energy intake and MC4R modulation of cholinergic neurons primarily underlies effects on energy expenditure. α >> ACTH, β, γ MC1R Melanogenesis γ 3 -MSH γ 1 -MSH ACTH MC2R Steroidogenesis ACTH β-lph α-msh CLIP γ-lph β-end γ > α, β MC3R AgRP Energy homeostasis, energy partitioning MC4R β-msh β > α >>> γ Energy homeostasis α >> ACTH, β, γ MC5R Sebum production Figure 1 POMC and its receptors. Top: pro-opiomelanocortin is posttranslationally processed to produce several biologically active peptides, including adrenocorticotropin (ACTH) and α-, β- and γ-melanocytestimulating hormone (MSH). Arrows indicate dibasic cleavage sites. Bottom: the actions of the melanocortin peptides are mediated by five receptors (MC1R MC5R). Aside from MC2R, which only binds ACTH, the other receptors are promiscuous for the different peptides. The relative binding efficiency of ACTH and the various MSH peptides for each receptor is indicated. MSHs signal to MC3R and MC4R to mediate energy homeostasis. Additionally, the endogenous antagonist/inverse agonist AgRP competes with the MSHs to bind to both receptors. effect on body weight 12 16, ablation of AgRP neurons in adult mice can result in severe aphagia and ultimately starvation, showing that sudden loss of these neurons has a profound effect on feeding A likely explanation for the disparity of these results is that mice can adapt to the loss of any one of these neurotransmitters if it occurs constitutively. Indeed, ablation of AgRP neurons in neonatal mice results in adult mice that can feed adequately after weaning 19 ; furthermore, adaptation can occur after ablation of AgRP neurons in adult mice if GABA signaling is enhanced by benzodiazepine delivery to the parabrachial nucleus, or if excitatory input to the parabrachial nucleus is suppressed 21,22. But whereas adaptation to loss of signaling by AgRP neurons can occur, adaption to loss of POMC neurons does not occur. POMC and MC4R mutations in both mouse and man reveal that disruption of these genes promotes hyperphagia and obesity. In fact, MC4R mutations are the most common monogenic cause of obesity, responsible for up to 5% of cases of severe childhood obesity, with the severity of receptor dysfunction seen in in vitro assays predicting the amount of food ingested in an ad libitum test meal 29. Thus, this observed genotype phenotype correlation emphasizes that the melanocortin regulatory system is sensitive to quantitative variation in MC4R expression and function. MC4Rs are widely distributed in the CNS, with particularly robust expression in the energy balance associated regions the paraventricular nucleus of the hypothalamus (PVH) and dorsal vagal complex (DVC) 30. Genetic advances have been instrumental in revealing an anatomical divergence in the underlying MC4Rs influencing energy intake and expenditure, and consequently body weight. Re-expressing MC4R expression in Mc4r / mice exclusively in Sim1 neurons, which are primarily found in the PVH and amygdala, restores normal appetite and reduces obesity by 60%, but the altered energy expenditure phenotype is unaffected 31. Conversely, restoration of MC4Rs in Mc4r / mice exclusively in cholinergic neurons is sufficient to normalize energy expenditure and modestly reduce obesity, but it does not affect food intake 32. Thus, MC4R regulation of Sim1 neurons is an important Leptin. It was proposed in the 1950s that circulating signals generated in proportion to body fat stores influence food intake and energy expenditure in a coordinated manner to regulate body weight. The first direct evidence for a hormonal system acting centrally to modulate feeding came from studying two naturally occurring severely obese mouse models, known at the time as obese (ob/ob) and diabetic (db/db) 33. Ob/ob mice were shown to harbor a loss-of-function mutation in the gene encoding a secreted peptide, leptin 34, which is produced by adipose tissue, whereas the db/db phenotype is due to mutations in the receptor for leptin (LEPR) 35, a class 1 cytokine receptor (the mice are now referred to as Lep ob/ob Lepr db/db, respectively). Loss of LEP or LEPR is now the most severe reported monogenic cause of obesity in mouse and man 36. The obesity and neuroendocrine phenotypes are effectively corrected in leptin-deficient humans 37 and mice 38 by treatment with leptin. LEP and LEPR deficiency, however, remain extremely rare causes of profound obesity. Common or diet-induced obesity is of greatest relevance to the modern day obesity crisis, resulting in elevated circulating leptin due to increased adiposity. However, in common obesity, leptin has diminished functional efficacy, a condition typically referred to as leptin resistance. Therefore, although leptin replacement in absolute deficiency is effective in reducing body weight, administering exogenous leptin to individuals with common obesity who have an elevated endogenous supply does not stimulate adequate weight loss 39. These observations suggest that the major energy homeostatic function of leptin is to maintain adequate fat stores for survival during times of energy deficit, rather than to prevent obesity in times of energy excess. This was elegantly demonstrated by Ahima et al., who showed that the broad spectrum of neuroendocrine changes seen in nutritionally depleted mice could be rescued in part by leptin administration 40. Mechanistically, the regulation of acute energy balance by leptin is contingent on its action in the CNS specifically at long-form leptin receptors (LEPRb). Indeed, neuronal specific restoration of LEPRb in Lepr db/db mice entirely rescues the obesity and diabetes phenotype 41. In the brain, one of the best described sites of leptin action in the modulation of energy homeostasis is the melanocortin axis, with leptin positively regulating ARC POMC neurons and negatively regulating AgRP neurons 42 (Fig. 3). The functional importance of this regulation was illustrated through genetic technology by discrete removal of Lepr from POMC or AgRP neuronal populations individually, which was found to increase body weight, and the combined removal of Lepr from both POMC and AgRP populations, which served to further elevate body weight 43,44. The observation that Lepr removal from POMC and AgRP neurons combined does not recapitulate the obesity of Lepr db/db mice indicated that other populations of neurons expressing LEPRbs are also involved in energy balance. Further investigation of critical sites of leptin action implicated the ventral medial nucleus of the hypothalamus (VMN). Removal of Lepr from steroidogenic factor 1 (SF1)-expressing VMN neurons results in a greater obesity than that observed upon Lepr removal from POMC neurons 43,45. Furthermore, a more profound obesity phenotype is observed upon removal of Lepr from SF1 and POMC neurons combined, illustrating the independence of these populations in the regulation of energy balance 45. Another important site of leptin action in energy balance is the lateral hypothalamus, where removal of Lepr from neurotensin-expressing neurons promotes early-onset obesity, modestly increases feeding and decreases locomotor activity 46. Viral approaches have also been useful in identifying critical sites of leptin action. Lepr knockdown by means 1344 VOLUME 15 NUMBER 10 OCTOBER 2012 nature neuroscience

3 r e v i e w Modulation of feeding behaviour Adipocyte signals Leptin Adiponectin Interleukins Hypothalamus Glucose Insulin CCK Gut-derived hormones Ghrelin GLP-1 PYY 3 36 OXM Figure 2 Brain sensing of gut- and adipocyte-derived hormones. Gut- and adipocyte-derived hormones, reflecting short- and long-term nutritional status, respectively, circulate in the periphery and signal to specific receptors in the brain. GLP-1, glucagon-like peptide 1; PYY 3 36, peptide YY residues 3 36; OXM, oxyntomodulin; CCK, cholecystokinin. of adeno-associated virus short-hairpin RNA interference (AAVshRNAi) in the NTS and area postrema results in increased food intake, body weight and adiposity, and knockdown in the ventral tegmental area (VTA) promotes elevated appetite and locomotor activity 47,48. In contrast to suggestions in an earlier report 49, a recent study has revealed that brainstem 5-HT neurons do not express LEPR and are therefore not a fundamental site of leptin action 50. Although the contributions of many populations of LEPRbexpressing neurons have been investigated, including those of the ARC, VMN, LHA, VTA and NTS, none of these sites alone explains the magnitude of leptin s effects on energy balance, indicating that leptin acts through a diverse network of independent systems to regulate appetite and body weight. 5-HT. Central 5-HT synthesis is restricted to nine distinct, but widely ramifying, brainstem raphe nuclei 51. The expansive nature of the central 5-HT system underlies its involvement in many physiologies and behaviors, which are implemented through interaction with 14 different receptor subtypes 51. Pharmacological compounds globally augmenting endogenous 5-HT bioavailability, such as d-fenfluramine, which causes increased release, and sibutramine, which inhibits reuptake, have been at the forefront of modern obesity pharmacotherapy 52. Like the knockout of Agrp discussed above, traditional germline knockout strategies in the developing 5-HT system fail to consistently produce the predicted postnatal metabolic dysregulation, whereas postnatal 5-HT depletion or ablation causes the expected hyperphagia and obesity phenotypes 51,52. Furthermore, pathway manipulation through the genetic inactivation of specific 5-HT receptors shows that perturbation of Htr2c, affecting 5-HT action at the G q -coupled receptor 5-HT 2C R, promotes lifelong hyperphagia, increased locomotor activity and middle age obesity onset in mice 51,52. Similarly, but to a lesser extent, Htr1b / mice, lacking the G i -coupled 5-HT 1B R, exhibit mild hyperphagia and elevated body weight. Consistent with these phenotypes, pharmacological 5-HT 2C R and/or 5-HT 1B R activation reduces food intake and decreases body weight 53,54. Indeed, the 5-HT 2C R agonist lorcaserin has recently been approved for obesity treatment by the US Food and Drug Administration 55. Like leptin, described above, 5-HT also reciprocally regulates the activity of ARC POMC and AgRP neurons. 5-HT increases the activity of anorectic POMC neurons through 5-HT 2C Rs 53 and decreases the activity of orexigenic AgRP through 5-HT 1B Rs 54. Illustrating a divergence in the leptin and 5-HT regulation of the melanocortin system, recent evidence suggests that leptin and 5-HT influence the activity of distinct populations of POMC neurons 56. Furthermore, unlike leptin, genetic studies indicate that modulation of the melanocortin pathway is a primary mechanism by which 5-HT, through 5-HT 2C Rs and 5-HT 1B Rs, influences appetite and body weight. Specifically, expression of 5-HT 2C Rs exclusively on POMC neurons is sufficient to modulate the extent of 5-HT 2C R-regulated energy balance 57. Moreover, genetic inactivation of Mc4r abolishes the anorectic effect of d-fenfluramine and 5-HT 2C R and 5-HT 1B R agonists, and restoration of Mc4r exclusively on Sim1-expressing neurons reinstates the appetite-suppressing actions of 5-HT compounds 54,58. BDNF. BDNF is best known for promoting neuron survival in the peripheral nervous system. It is also widely expressed in the CNS, where it is important in neuronal development 59,60. Genetic studies have also uncovered a BDNF role in the regulation of energy balance. Specifically, mice heterozygous for a targeted deletion of Bdnf 61 or with a specific postnatal deletion of neuronal Bdnf display hyperphagia leading to severe obesity 62. The importance of BDNF s cognate receptor, tropomyosin receptor kinase B (TrkB), in energy balance is demonstrated by mice with hypomorphic expression of Neurotrophic tyrosine kinase receptor, type 2 (Ntrk2), the gene encoding TrkB, which are characterized by increased linear growth and severe obesity 63. The role of BDNF signaling in appetite and body weight in humans has also been revealed in two different rare cases. In one, an 8-yearold girl was reported with a chromosomal inversion encompassing the BDNF gene, presenting with hyperphagia, severe obesity and impairment of cognitive functions 64. In another, a de novo missense mutation in NTRK2 was detected in a patient with severe obesity and developmental abnormalities. The mutation caused an amino acid substitution (Y722C) of one of the evolutionary conserved tyrosine residues in the catalytic domain of TrkB. This change led to an impairment of MAP kinase downstream signaling. Notably, the average food intake of that patient in an ad libitum test meal was comparable to that measured in leptin deficiency 65. Illustrating the therapeutic potential of BDNF in common obesity, injection of BDNF in mice reduces food intake, leading to marked weight loss 66. Several lines of evidence suggest that BDNF-TrkB signaling is a component of the leptin and melanocortin energy balance axes. Specifically, pharmacological blockade of TrkB kinase activity abolishes the anorectic effect of the MC3R and MC4R agonist melanotan-2 (MTII) and leptin 67,68 ; pharmacological MC4R activation stimulates hypothalamic BDNF release in vitro and decreases food intake in vivo 69 ; and brain nature neuroscience VOLUME 15 NUMBER 10 OCTOBER

4 R e v i e w Figure 3 Schematic representation of the hypothalamic nuclei and other relevant higher brain regions. PVN, paraventricular nucleus; DMH, dorsomedial hypothalamus; VMN, ventromedial nucleus; LH, lateral hypothalamus; ARC, arcuate nucleus; VTA, ventral tegmental area; DRN, dorsal raphe nucleus; NTS, nucleus of the solitary tract. Projections between nuclei are indicated with arrows. Sites of action of neuropeptides and neurotransmitters inhibiting ( ) or stimulating (+) food intake are shown. CART, cocaineand amphetamine-regulated transcript; CRF, corticotropin-releasing factor; GLP-1, glucagonlike peptide 1; MCH, melanin-concentrating hormone; OXT, oxytocin; AVP, vasopressin; TRH, thyrotropin-releasing hormone; NA, noradrenaline; A, adrenaline; PrRP, prolactin-releasing peptide; PYY 3 36, peptide YY residues 3 36; OXM, oxyntomodulin; CCK, cholecystokinin. MC3R and MC4R, melanocortin receptors; NPY1R and NPY5R, NPY receptors; LEPRb, long signaling isoform of the leptin receptor; INSR, insulin receptor; GHSR, ghrelin receptor; GLP-1R, GLP-1 receptor. 3V, third ventricle. infusion of BDNF attenuates hyperphagia and body weight gain in mice with impaired melanocortin signaling 63. The discrete population(s) of BDNF neurons specifically involved in these pathways or more generally in energy balance have not Reward Homeostasis Circulating peripheral inputs been fully described. However, efforts to identify specific populations of BDNF neurons influencing appetite and body weight have revealed the VMN and NTS as key sites. Specifically, Bdnf is regulated both nutritionally and by leptin in the VMN and NTS 63,67,70, and stereotaxic injection of BDNF into the VMN or NTS reduces food intake 68,71. Notably, virally mediated knockdown of Bdnf in the VMN promotes obesity owing to hyperphagia 72. Approximately 60% of VMN BDNF neurons are coexpressed with Sf1 (ref. 73). However, genetic evidence suggests that this specific population of BDNF VMN neurons is not required for energy balance, as exclusive deletion of Bdnf in SF1 neurons does not alter energy balance or body weight 45. Collectively, these data suggest that intact BDNF TrkB signaling is of critical importance for maintaining normal body weight and food intake. Evidence indicates that BDNF signaling is physiologically relevant in the VMN and NTS in regulating metabolism, that the BDNF TrkB pathway is a downstream target of both leptin and melanocortin signaling, and that NTS Bdnf expression is regulated by leptin. However, it remains to be determined at which specific sites in the complex regulatory network of body weight BDNF TrkB signaling is integrated; given BDNF s role as a neurotrophin, it is unlikely to adhere entirely to the canonical neuropeptide receptor signaling schema. GWAS and common obesity Although monogenic alterations in each of the pathways described above results in Mendelian obesity and have thereby elucidated crucial pathways controlling appetitive behavior, these cases remain rare. The major burden of disease is carried by those with common obesity, the underlying mechanisms of which have thus far been difficult to elucidate. In contrast to the Mendelian obesity syndromes, common obesity is likely to have a polygenic etiology, with many variations each having a subtle effect. Recent progress into the underpinnings of common obesity has been made with GWAS. Striatum GABA VTA Dopamine GABA Hypothalamus PVN AVP TRH Galanin GLP-1 CRF CART OXT BDNF Pancreas Insulin, amylin DMH CART VMN BDNF NPY ARC POMC CART AgRP Neurotensin + NPY Galanin GABA 3V LH MCH Orexin Neurotensin NPY CART Adipose tissue Leptin, adiponectin, interleukins DRN 5-HT NTS POMC NA/A GLP-1 CART CCK Galanin PrRP NPY Substance P BDNF Stomach, gut Ghrelin, GLP-1, PYY 3 36, OXM, CCK LEPRb GHSR GLP-1R 5-HT 2C R 5-HT 1B R NPY1R NPY5R MC3R MC4R INSR Fat mass and obesity related transcript (FTO). GWAS have revealed that single nucleotide polymorphisms (SNPs) in intron 1 of FTO is robustly associated with BMI and obesity 74. Many subsequent studies, covering more than 22 distinct populations of European, African and Asian ancestries, have confirmed the association of FTO with BMI 75, and confirmed that the major effect of the risk alleles in FTO is increased energy intake and reduced satiety. However, as yet, no conclusive link has been made between the SNPs in intron 1 and expression or function and FTO. Human FTO deficiency results in a complex clinical picture. A Palestinian Arab consanguineous family has been reported with nine members affected by a previously unknown polymalformation syndrome, all of whom were homozygous for a loss-of-function arginine-to-glutamine change at position 316 (R316Q) in FTO 76. The syndromic characteristics observed in the affected individuals include postnatal growth retardation, microcephaly, severe functional brain deficits, facial dysmorphism, cardiac abnormalities, cleft palate and mortality within the first 30 months of life, making it impossible to glean any mechanistic lessons regarding body weight regulation 76. Mice homozygous for a targeted deletion in Fto also display a complex phenotype 77. Consistent with human FTO deficiency, mice are postnatally growth retarded with decreased fat and lean body mass, and more than 50% die by weaning 77. Thus, FTO appears to have a critical, but as yet undetermined role, in the development of several major organ systems, including the CNS and cardiovascular systems. Shortly after FTO was reported, we and others showed by bioinformatics analysis that FTO shares sequence motifs with Fe(II)- and 2-oxoglutarate dependent oxygenases 78. In vitro, recombinant FTO is able to catalyze the Fe(II)- and 2-oxoglutarate dependent demethylation of single-stranded DNA and RNA, suggesting a potential role for FTO in nucleic acid repair or modification. FTO is widely expressed across several tissues, although it is most highly expressed in the brain 1346 VOLUME 15 NUMBER 10 OCTOBER 2012 nature neuroscience

5 r e v i e w and is particularly dense in the hypothalamus 78. Fto is bidirectionally regulated in the ARC as a function of nutritional status, decreasing after fasting and increasing after high-fat diet, and modulating FTO protein levels specifically in the ARC influences food intake 79. Thus, although FTO clearly has a broader biological function, it also acts specifically in the hypothalamus to regulate food intake. Other GWAS genes. A number of meta-analyses following on from the FTO discovery, covering several GWAS and encompassing hundreds of thousands of individuals, have revealed 31 more polygenic obesity loci of varying effect size 80,81. To estimate the cumulative effect of the 32 variants on BMI, Speliotes and colleagues constructed a genetic susceptibility score that summed the number of BMIincreasing alleles 80. For each unit increase in the genetic-susceptibility score, which is approximately equivalent to having one additional risk allele, BMI increased by 0.17 kg per meter of height squared, the equivalent of a g gain in body weight in adults cm in height. Although the authors note that the predictive value for obesity risk and BMI of the 32 variants combined is modest, it is statistically significant, and indicative of a polygenic mode of obesity 80. Several of these identified loci are actually located in or near genes that encode neuronal regulators of appetite or energy balance (see above), a few of which discussed below. After the previously reported variants in FTO, the next strongest association signal (rs ) maps 188 kb downstream of MC4R. The risk allele is associated with increased BMI and odds for severe childhood obesity. Furthermore, overtransmission of the risk allele to obese offspring in 660 families was observed. The SNP location and patterns of phenotypic associations are consistent with effects mediated through altered MC4R function 82. However, as with FTO, and because MC4R is primarily expressed within the CNS, no link has yet been made between the SNP and levels of MC4R expression. SNPs near SH2B1 were found to be associated with adult obesity. SH2B1, a cytoplasmic adaptor protein, binds through its Src homology 2 (SH2) domain to a variety of protein tyrosine kinases, including JAK2 and the insulin receptor. Systemic deletion of the Sh2b1 gene results in metabolic disorders in mice, including hyperlipidemia, leptin resistance, hyperphagia, obesity, hyperglycemia, insulin resistance and glucose intolerance. Neuron-specific restoration of the β isoform of SH2B1 corrected the metabolic disorders seen in the global knockout and also improved JAK2-mediated leptin signaling and leptin regulation of orexigenic neuropeptide expression in the hypothalamus. Moreover, neuron-specific overexpression of SH2B1 dose-dependently protected against high-fat diet induced leptin resistance and obesity. These observations suggest that neuronal SH2B1 regulates energy balance, body weight, peripheral insulin sensitivity and glucose homeostasis at least in part by enhancing hypothalamic leptin sensitivity 83. Like MC4R and SH2B1, neuronal growth regulator 1 (NEGR1) was identified by GWAS to be associated with obesity and has its primary biological function in the CNS. NEGR1 is a cell adhesion molecule known to participate in the establishment and remodeling of neural circuits. Its expression is restricted to the brain, is readily detected at rat embryonic day 16, and gradually increases during development. NEGR1 protein identified through immunohistochemistry was observed in the cerebral cortex and hippocampus, where strongly stained puncta were observed on dendrites and somata of pyramidal neurons 84. Transmembrane protein 18 (TMEM18) is another example of a GWAS gene with no previously known biological function, very much like FTO when the latter was first identified. TMEM18 is a small protein of 140 residues and is predicted to be mostly α-helical, with three transmembrane domains. A recent study showed that TMEM18 localizes to the nuclear membrane and binds to DNA in a sequence-specific manner. The protein binds DNA with its positively charged C terminus, which contains also a nuclear localization signal. As a consequence, the DNA binding by TMEM18 would bring the chromatin near the nuclear membrane. The authors speculate that this localization of TMEM18-bound DNA might result in repressed transcription, although this has not been demonstrated 85. Other genes, such as KCTD15 and GNPDA2 (glucosamine-6- phosphate deaminase 2) are also expressed at high levels in the brain and in particular, in the hypothalamus 81. Of note, the SNPs attributed to the latter gene (potassium channel tetramerization domain-containing 15) actually map to a gene desert between GNPDA2 and GABRA2 (GABA A receptor, α2). It is therefore possible that the effect of the variants might be mediated by GABRA2, which affects addiction behavior 86. There is also evidence for an association of genes in the 5-HT signaling pathway with obesity. Earlier, pre-gwas studies pointed to SNPs in SLC6A14, which encodes an amino acid transporter that regulates tryptophan availability for 5-HT synthesis, as associated with obesity 87. More recently a SNP in HTR2C was also shown to be associated with increased BMI 88. It is probable that GWAS, which are relevant to common obesity, will continue to reveal further regulators or effectors of the critical genes and pathways identified through monogenetic studies and thereby unravel the complex underpinnings of genetic susceptibility to obesity. Conclusions It is clear that the modern obesity epidemic has been driven by lifestyle and environmental changes. An improved understanding of the normal physiology of energy homeostasis and the endocrine control of food intake will have profound implications for the development of effective therapeutic and preventative strategies for human obesity. The study of monogenic causes of obesity has identified crucial pathways and brain regions in the control of feeding behavior. The recent emergence of polygenic clues from GWAS has revealed that, as in rare monogenic forms of obesity, the brain plays a key role in predisposition to common obesity. Acknowledgments G.S.H.Y. is supported by the UK Medical Research Council Centre for Obesity and Related Metabolic Disorders (MRC-CORD) and the European Union (FP7- HEALTH EurOCHIP and FP7-HEALTH Full4Health). L.K.H. is supported by the Wellcome Trust (WT081713; WT090812). COMPETING FINANCIAL INTERESTS The authors declare no competing financial interests. Published online at Reprints and permissions information is available online at reprints/index.html. 1. Stunkard, A.J., Foch, T.T. & Hrubec, Z. A twin study of human obesity. J. Am. Med. Assoc. 256, (1986). 2. Stunkard, A.J., Harris, J.R., Pedersen, N.L. & McClearn, G.E. The body-mass index of twins who have been reared apart. N. Engl. J. Med. 322, (1990). 3. Cone, R.D. Anatomy and regulation of the central melanocortin system. Nat. Neurosci. 8, (2005). 4. Bertagna, X. Proopiomelanocortin-derived peptides. Endocrinol. Metab. Clin. North Am. 23, (1994). 5. Castro, M.G. & Morrison, E. Post-translational processing of proopiomelanocortin in the pituitary and in the brain. Crit. Rev. Neurobiol. 11, (1997). 6. Ollmann, M.M. et al. Antagonism of central melanocortin receptors in vitro and in vivo by agouti-related protein. Science 278, (1997). nature neuroscience VOLUME 15 NUMBER 10 OCTOBER

6 R e v i e w 7. Fan, W., Boston, B.A., Kesterson, R.A., Hruby, V.J. & Cone, R.D. Role of melanocortinergic neurons in feeding and the agouti obesity syndrome. Nature 385, (1997). 8. Cowley, M.A. et al. Integration of NPY, AGRP, and melanocortin signals in the hypothalamic paraventricular nucleus: evidence of a cellular basis for the adipostat. Neuron 24, (1999). 9. Aponte, Y., Atasoy, D. & Sternson, S.M. AGRP neurons are sufficient to orchestrate feeding behavior rapidly and without training. Nat. Neurosci. 14, Krashes, M.J. et al. Rapid, reversible activation of AgRP neurons drives feeding behavior in mice. J. Clin. Invest. 121, Atasoy, D., Betley, J.N., Su, H.H. & Sternson, S.M. Deconstruction of a neural circuit for hunger. Nature 488, (2012). 12. Erickson, J.C., Clegg, K.E. & Palmiter, R.D. Sensitivity to leptin and susceptibility to seizures of mice lacking neuropeptide Y. Nature 381, (1996). 13. Erickson, J.C., Hollopeter, G. & Palmiter, R.D. Attenuation of the obesity syndrome of ob/ob mice by the loss of neuropeptide Y. Science 274, (1996). 14. Qian, S. et al. Neither agouti-related protein nor neuropeptide Y is critically required for the regulation of energy homeostasis in mice. Mol. Cell. Biol. 22, (2002). 15. Tong, Q., Ye, C.P., Jones, J.E., Elmquist, J.K. & Lowell, B.B. Synaptic release of GABA by AgRP neurons is required for normal regulation of energy balance. Nat. Neurosci. 11, (2008). 16. Wortley, K.E. et al. Agouti-related protein-deficient mice display an age-related lean phenotype. Cell Metab. 2, (2005). 17. Bewick, G.A. et al. Post-embryonic ablation of AgRP neurons in mice leads to a lean, hypophagic phenotype. FASEB J. 19, (2005). 18. Gropp, E. et al. Agouti-related peptide-expressing neurons are mandatory for feeding. Nat. Neurosci. 8, (2005). 19. Luquet, S., Perez, F.A., Hnasko, T.S. & Palmiter, R.D. NPY/AgRP neurons are essential for feeding in adult mice but can be ablated in neonates. Science 310, (2005). 20. Xu, A.W. et al. Effects of hypothalamic neurodegeneration on energy balance. PLoS Biol. 3, e415 (2005). 21. Wu, Q., Boyle, M.P. & Palmiter, R.D. Loss of GABAergic signaling by AgRP neurons to the parabrachial nucleus leads to starvation. Cell 137, (2009). 22. Wu, Q., Clark, M.S. & Palmiter, R.D. Deciphering a neuronal circuit that mediates appetite. Nature 483, (2012). 23. Challis, B.G. et al. Mice lacking pro-opiomelanocortin are sensitive to high-fat feeding but respond normally to the acute anorectic effects of peptide-yy Proc. Natl. Acad. Sci. USA 101, (2004). 24. Huszar, D. et al. Targeted disruption of the melanocortin-4 receptor results in obesity in mice. Cell 88, (1997). 25. Yaswen, L., Diehl, N., Brennan, M.B. & Hochgeschwender, U. Obesity in the mouse model of pro-opiomelanocortin deficiency responds to peripheral melanocortin. Nat. Med. 5, (1999). 26. Krude, H. et al. Severe early-onset obesity, adrenal insufficiency and red hair pigmentation caused by POMC mutations in humans. Nat. Genet. 19, (1998). 27. Vaisse, C., Clement, K., Guy-Grand, B. & Froguel, P. A frameshift mutation in human MC4R is associated with a dominant form of obesity. Nat. Genet. 20, (1998). 28. Yeo, G.S. et al. A frameshift mutation in MC4R associated with dominantly inherited human obesity. Nat. Genet. 20, (1998). 29. Farooqi, I.S. et al. Clinical spectrum of obesity and mutations in the melanocortin 4 receptor gene. N. Engl. J. Med. 348, (2003). 30. Kishi, T. et al. Expression of melanocortin 4 receptor mrna in the central nervous system of the rat. J. Comp. Neurol. 457, (2003). 31. Balthasar, N. et al. Divergence of melanocortin pathways in the control of food intake and energy expenditure. Cell 123, (2005). 32. Rossi, J. et al. Melanocortin-4 receptors expressed by cholinergic neurons regulate energy balance and glucose homeostasis. Cell Metab. 13, Friedman, J.M. & Halaas, J.L. Leptin and the regulation of body weight in mammals. Nature 395, (1998). 34. Zhang, Y. et al. Positional cloning of the mouse obese gene and its human homologue. Nature 372, (1994). 35. Chen, H. et al. Evidence that the diabetes gene encodes the leptin receptor: identification of a mutation in the leptin receptor gene in db/db mice. Cell 84, (1996). 36. Montague, C.T. et al. Congenital leptin deficiency is associated with severe earlyonset obesity in humans. Nature 387, (1997). 37. Farooqi, I.S. et al. Effects of recombinant leptin therapy in a child with congenital leptin deficiency. N. Engl. J. Med. 341, (1999). 38. Pelleymounter, M.A. et al. Effects of the obese gene product on body weight regulation in ob/ob mice. Science 269, (1995). 39. Heymsfield, S.B. et al. Recombinant leptin for weight loss in obese and lean adults: a randomized, controlled, dose-escalation trial. J. Am. Med. Assoc. 282, (1999). 40. Ahima, R.S. et al. Role of leptin in the neuroendocrine response to fasting. Nature 382, (1996). 41. de Luca, C. et al. Complete rescue of obesity, diabetes, and infertility in db/db mice by neuron-specific LEPR-B transgenes. J. Clin. Invest. 115, (2005). 42. Cowley, M.A. et al. Leptin activates anorexigenic POMC neurons through a neural network in the arcuate nucleus. Nature 411, (2001). 43. Balthasar, N. et al. Leptin receptor signaling in POMC neurons is required for normal body weight homeostasis. Neuron 42, (2004). 44. van de Wall, E. et al. Collective and individual functions of leptin receptor modulated neurons controlling metabolism and ingestion. Endocrinology 149, (2008). 45. Dhillon, H. et al. Leptin directly activates SF1 neurons in the VMH, and this action by leptin is required for normal body-weight homeostasis. Neuron 49, (2006). 46. Leinninger, G.M. et al. Leptin action via neurotensin neurons controls orexin, the mesolimbic dopamine system and energy balance. Cell Metab. 14, Hayes, M.R. et al. Endogenous leptin signaling in the caudal nucleus tractus solitarius and area postrema is required for energy balance regulation. Cell Metab. 11, (2010). 48. Hommel, J.D. et al. Leptin receptor signaling in midbrain dopamine neurons regulates feeding. Neuron 51, (2006). 49. Yadav, V.K. et al. A serotonin-dependent mechanism explains the leptin regulation of bone mass, appetite, and energy expenditure. Cell 138, (2009). 50. Lam, D.D. et al. Leptin does not directly affect CNS serotonin neurons to influence appetite. Cell Metab. 13, Lam, D.D. & Heisler, L.K. Serotonin and energy balance: molecular mechanisms and implications for type 2 diabetes. Expert Rev. Mol. Med. 9, 1 24 (2007). 52. Garfield, A.S. & Heisler, L.K. Pharmacological targeting of the serotonergic system for the treatment of obesity. J. Physiol. (Lond.) 587, (2009). 53. Heisler, L.K. et al. Activation of central melanocortin pathways by fenfluramine. Science 297, (2002). 54. Heisler, L.K. et al. Serotonin reciprocally regulates melanocortin neurons to modulate food intake. Neuron 51, (2006). 55. Fidler, M.C. et al. A one-year randomized trial of lorcaserin for weight loss in obese and overweight adults: the BLOSSOM trial. J. Clin. Endocrinol. Metab. 96, Sohn, J.W. et al. Serotonin 2C receptor activates a distinct population of arcuate pro-opiomelanocortin neurons via TRPC channels. Neuron 71, Xu, Y. et al. 5 HT2CRs expressed by pro-opiomelanocortin neurons regulate energy homeostasis. Neuron 60, (2008). 58. Xu, Y. et al. A serotonin and melanocortin circuit mediates d-fenfluramine anorexia. J. Neurosci. 30, (2010). 59. Barde, Y.A. The nerve growth factor family. Prog. Growth Factor Res. 2, (1990). 60. Snider, W.D. Functions of the neurotrophins during nervous system development: what the knockouts are teaching us. Cell 77, (1994). 61. Kernie, S.G., Liebl, D.J. & Parada, L.F. BDNF regulates eating behavior and locomotor activity in mice. EMBO J. 19, (2000). 62. Rios, M. et al. Conditional deletion of brain-derived neurotrophic factor in the postnatal brain leads to obesity and hyperactivity. Mol. Endocrinol. 15, (2001). 63. Xu, B. et al. Brain-derived neurotrophic factor regulates energy balance downstream of melanocortin-4 receptor. Nat. Neurosci. 6, (2003). 64. Gray, J. et al. Hyperphagia, severe obesity, impaired cognitive function, and hyperactivity associated with functional loss of one copy of the brain-derived neurotrophic factor (BDNF) gene. Diabetes 55, (2006). 65. Yeo, G.S. et al. A de novo mutation affecting human TrkB associated with severe obesity and developmental delay. Nat. Neurosci. 7, (2004). 66. Pelleymounter, M.A., Cullen, M.J. & Wellman, C.L. Characteristics of BDNF-induced weight loss. Exp. Neurol. 131, (1995). 67. Bariohay, B. et al. Brain-derived neurotrophic factor/tropomyosin-related kinase receptor type B signaling is a downstream effector of the brainstem melanocortin system in food intake control. Endocrinology 150, (2009). 68. Spaeth, A.M., Kanoski, S.E., Hayes, M.R. & Grill, H.J. TrkB receptor signaling in the nucleus tractus solitarius mediates the food intake suppressive effects of hindbrain BDNF and leptin. Am. J. Physiol. Endocrinol. Metab. 302, E1252 E1260 (2012). 69. Nicholson, J.R., Peter, J.C., Lecourt, A.C., Barde, Y.A. & Hofbauer, K.G. Melanocortin- 4 receptor activation stimulates hypothalamic brain-derived neurotrophic factor release to regulate food intake, body temperature and cardiovascular function. J. Neuroendocrinol. 19, (2007). 70. Komori, T., Morikawa, Y., Nanjo, K. & Senba, E. Induction of brain-derived neurotrophic factor by leptin in the ventromedial hypothalamus. Neuroscience 139, (2006). 71. Godar, R. et al. Reduction of high-fat diet-induced obesity after chronic administration of brain-derived neurotrophic factor in the hypothalamic ventromedial nucleus. Neuroscience 194, Unger, T.J., Calderon, G.A., Bradley, L.C., Sena-Esteves, M. & Rios, M. Selective deletion of Bdnf in the ventromedial and dorsomedial hypothalamus of adult mice results in hyperphagic behavior and obesity. J. Neurosci. 27, (2007) VOLUME 15 NUMBER 10 OCTOBER 2012 nature neuroscience

7 r e v i e w 73. Tran, P.V. et al. Diminished hypothalamic bdnf expression and impaired VMH function are associated with reduced SF-1 gene dosage. J. Comp. Neurol. 498, (2006). 74. Frayling, T.M. et al. A common variant in the FTO gene is associated with body mass index and predisposes to childhood and adult obesity. Science 316, (2007). 75. Tung, Y.C. & Yeo, G.S. From GWAS to biology: lessons from FTO. Ann. NY Acad. Sci. 1220, Boissel, S. et al. Loss-of-function mutation in the dioxygenase-encoding FTO gene causes severe growth retardation and multiple malformations. Am. J. Hum. Genet. 85, (2009). 77. Fischer, J. et al. Inactivation of the Fto gene protects from obesity. Nature 458, (2009). 78. Gerken, T. et al. The obesity-associated FTO gene encodes a 2-oxoglutaratedependent nucleic acid demethylase. Science 318, (2007). 79. Tung, Y.C. et al. Hypothalamic-specific manipulation of Fto, the ortholog of the human obesity gene FTO, affects food intake in rats. PLoS ONE 5, e8771 (2010). 80. Speliotes, E.K. et al. Association analyses of 249,796 individuals reveal 18 new loci associated with body mass index. Nat. Genet. 42, (2010). 81. Willer, C.J. et al. Six new loci associated with body mass index highlight a neuronal influence on body weight regulation. Nat. Genet. 41, (2009). 82. Loos, R.J. et al. Common variants near MC4R are associated with fat mass, weight and risk of obesity. Nat. Genet. 40, (2008). 83. Ren, D. et al. Neuronal SH2B1 is essential for controlling energy and glucose homeostasis. J. Clin. Invest. 117, (2007). 84. Funatsu, N. et al. Characterization of a novel rat brain glycosylphosphatidylinositolanchored protein (Kilon), a member of the IgLON cell adhesion molecule family. J. Biol. Chem. 274, (1999). 85. Jurvansuu, J.M. & Goldman, A. Obesity risk gene TMEM18 encodes a sequencespecific DNA-binding protein. PLoS ONE 6, e Morris, H.V. et al. Alpha2-containing GABA A receptors are involved in mediating stimulant effects of cocaine. Pharmacol. Biochem. Behav. 90, 9 18 (2008). 87. Suviolahti, E. et al. The SLC6A14 gene shows evidence of association with obesity. J. Clin. Invest. 112, (2003). 88. Vimaleswaran, K.S. et al. Association between serotonin 5-HT-2C receptor gene (HTR2C) polymorphisms and obesity- and mental health-related phenotypes in a large population-based cohort. Int. J. Obes. (Lond.) 34, (2010). nature neuroscience VOLUME 15 NUMBER 10 OCTOBER