TO OBTAIN A DOWNLOAD OF THIS CHAPTER IN WORD OR PDF FORMAT, CLICK HERE

Obesity Due Primarily to Increased Food Intake

Neuropeptide Y. Neuropeptide Y (NPY) is one of the most intensively studied neuropeptides affecting body weight. The 36-amino-acid protein is cleaved from the N terminus of a 94-amino acid precursor, proNPY. NPY is a member of a family of peptides characterized by having a single carboxy-terminal tyrosine (Y), hence the designation neuropeptide Y. Fasted animals show a highly reproducible and significant increase in hypothalamic NPY mRNA and protein expression after a 24-h fast, and mice are acutely hyperphagic in response to an intracerebroventricular injection of 5μg of NPY. A similar effect is observed in a variety of species.

Only an estimated 5% of brain NPY is localized to hypothalamic subnuclei that regulate body weight, with the remaining 95% of brain NPY apparently serving other physiological processes including cardiovascular physiology, excitatory transmission of neural impulses, and anxiety. Considering NPY's important physiological roles, it was surprising that congenitally NPY-deficient animals survived to adulthood and were grossly normal, with no metabolic phenotype attributable to a lack of NPY apart from increased sensitivity to leptin-mediated reduction in food intake [130]. The lack of the anticipated lean phenotype in NPY-deficient animals suggests the presence of compensatory "backup" systems for the critical physiological processes in which NPY participates. AGRP is a candidate compensatory hypothalamic neuropeptide that increases food intake and body weight, and is colocalized to NPY neurons; however, AGRP mRNA expression in the ARC is not elevated in NPY-deficient mice as compared to controls [131]. And, Npy-/-;Agrp-/- double mutants have no grossly apparent metabolic or behavioral abnormalities [132].

Compelling evidence for a role for NPY in the leptin signaling pathway was obtained by crossing Npy-/- animals to Lep ^ob /Lep ^ob mice [133]. NPY deficiency attenuated the elevated hypothalamic NPY levels and obesity syndrome of Lep ^ob /Lep ^ob mice; the double mutants weighed 40% less than Lep ^ob /Lep ^ob mice and had a 51% reduction in the combined mass of the inguinal, reproductive and retroperitoneal fat pads as compared to Npy-/- mice. Follow up studies found that NPY-deficient mice show a modest reduction in refeeding hyperphagia after a 24-hour fast [134], suggesting that NPY may act as a stimulator of food intake after prolonged food deprivation. It is possible that a temporally conditional NPY knockout might elicit the predicted, lean phenotype if physiological compensation for NPY deficiency is prevented during development; however, this model has not been reported.

NPY receptors. NPY receptors are G-protein coupled receptors coupled to nuclear gene expression by cAMP-mediated effects on cAMP response elements (CRE) in regulated genes, as well as other effects of PKA activation in neurons with NPY receptors [135]. Six NPY receptors have been identified: Y1, Y2, Y3, Y4, Y5, and y6. The greatest conservation of amino acid sequence occurs in the seven transmembrane domains characteristic of G-protein coupled receptors.

Although the Y2 receptor has only 30% homology to the other NPY receptors, ligand binding is conserved; Y1, Y2, and Y5 are equally activated by NPY. The Y3 receptor has not been cloned, while Y4 may be identical to the pancreatic polypeptide receptor 2 (PP2), and the y6 receptor (which was referred to as Y5 in the first report) has been cloned and detected in the hypothalamus of mice [136] and other species [137], but not primates or rats. NPY receptors are widely distributed in the brain and intestines; Y1 and Y5 are localized to the hypothalamus and have putative roles in the regulation of food intake and body weight. The murine genes encoding the Y1 and Y5 receptors are located 10 kB apart on chromosome 8, and Y4 and Y6 are located on chromosomes 14 and 18. A polymorphism in the human Y5 receptor associated with obesity in Pima Indians [138].

Intracerebroventricular administration of Y1- and Y5 receptor antagonists reduce food intake; these receptors are present in the paraventricular nucleus where they can potentially bind NPY ligand released from the synapses of NPYergic neural projections that emanate from the arcuate nucleus. Although mice with targeted Y1 disruption show a 45% decrease in 3-h food intake and a blunted refeeding response after a 24-hour fast, female Y1-deficient mice are 30% heavier than their wild-type littermates, and have a two-fold increase in percent body fat [139]. These differences are attributable to a 20% decrease in metabolic rate and a significant reduction in locomotion during the active, nocturnal phase of their diurnal cycle. It has been suggested that these paradoxical effects may be a result of overcompensation by Y5 mechanisms [139]. However, Y5-deficient mice also develop mild obesity, without any measurable compensatory increase in Y1 mRNA in whole brain [140].

Previous studies have suggested that the anorectic action of leptin is partially mediated via downregulation of hypothalamic NPY and NPY signaling; this inference is supported by the observation that Npy-/- mice are more sensitive to the anorexigenic effects of intracerebroventricular administration of leptin than +/+ littermates. [M2]   To determine if leptin's effects are mediated by inhibition of NPY signaling, ob/ob;Y5 -/- mice were generated; no effect on the obese phenotype of Lep ^ob /Lep ^ob mice was observed, suggesting that leptin action is not modulated by the Y5 signaling pathway, or that compensation by other NPY receptor subtypes masked any contribution made by Y5. In addition, both Y1- and Y5-deficient mice respond with hyperphagia to centrally administered NPY, suggesting the existence of an additional NPY receptor or receptors that regulate food intake.

Hypothalamus-specific deletion of the Y2 receptor has been achieved by adenoviral delivery of CRE recombinase to mice with loxP sites flanking the Y2 gene [141], and a significant decrease in both food intake and body weight was observed. The endogenous peptide YY _1-36 , a Y2 ligand, stimulates food intake; however, its cleavage product peptide YY _3-36 (PYY _3-36 ), a Y2 agonist primarily secreted from endocrine cells lining the gastrointestinal tract, decreases food intake [142]. Thus, the Y2 receptor may have dual functionality that is determined by the PYY moiety that binds to it. An analogous ligand for Y1 and/or Y5 could account for either of the paradoxical body weight phenotypes of Y1-/- and Y5-/- mice. It may be inferred from the NPY receptor knockout mouse models that receptor type, anatomic location, and presence during development may all play a role in determining receptor function with respect to energy balance in this redundant and functionally diverse pathway.

Melanin-concentrating hormone. Melanin-concentrating hormone (MCH) is an orexigenic neuropeptide first isolated from salmon pituitary [143]. MCH darkens skin pigmentation in fish and amphibians [144]. A role for MCH in the regulation of energy balance was discovered by examining differential expression of hypothalamic transcripts in Lep ^ob mice as compared to their lean littermates [145]. MCH-containing neurons are located primarily in the LHA. Ablative LHA lesions result in hypophagia, suggesting that the intact nucleus plays an important role to maintain adequate caloric intake [146]. Like NPY, MCH levels are elevated after fasting, intracerebroventricular injections of MCH increase food intake, and MCH mRNA and protein are upregulated in ad libitum -fed Lep ^ob /Lep ^ob mice [145, 147]. However, unlike NPY deficiency, targeted deletion of the MCH gene produces a lean, hypophagic phenotype [148]. Male and female MCH-/- mice are 28% and 24% leaner than wild-type littermates, and ingest 12% fewer calories during their active, nocturnal cycle, when rodents usually consume 90% of their food. This reduction in energy intake is also reflected in fat mass, which is reduced by nearly 50%, while the lean compartment of body mass is reduced by less than 2%. Oxygen consumption in

MCH-/- mice is not significantly different from their +/+ littermates. However, MCH-/- mice have 20% increased oxygen consumption after normalization to body mass, suggesting that they resist decreased energy expenditure despite reduced food intake and suppression of circulating leptin.

To investigate the CNS pathways responsible for the lean phenotype, ARC NPY, AGRP and POMC and LHA orexin mRNA have been analyzed in MCH-/- mice by in situ hybridization. ARC POMC mRNA was reduced, probably secondary to reduced food intake and attenuated plasma leptin levels, rather than a primary response to MCH deficiency. Transgenic overexpression of MCH in the lateral hypothalamus did not cause obesity in FVB mice, but obesity occurred when MCH transgenic mice where backcrossed onto a C57BL/6J background for seven generations [31]. On standard chow, male MCH-overexpressing mice on a C57BL/6J background had a 6.8% increase in food intake, which produced a 10% increase in body weight at 20 weeks of age. On a high-fat diet, however, the same mice had a 10% increase in food intake, which elicited a 12.6% increase in body weight and a 31% increase in total fat mass. It is likely that the FVB background obscures differences in responsiveness to a high-fat diet between MCH overexpressors as compared to wild-type mice, since the C57BL/6J mice are prone to diet-induced obesity [149]. It would be of interest to know which genes are responsible for the distinct phenotypes of the FVB and B6 mice segregating for the overexpression of MCH.

The ligand for the orphan GPCR, SLC-1, was identified by transfecting the receptor into a CHO cell line and introducing high performance liquid chromatography fractions of rat brain extracts into the medium. Treated CHO cells were subsequently assayed for changes in calcium flux that might indicate receptor activation, and MCH was identified as the ligand [150]. Genetic ablation of the Slc1 , or Mch1 receptor, recapitulates the lean phenotype seen in MCH-deficient mice [151], however Mch1r-/- mice are markedly hyperphagic, a phenotype that is counteracted by increased motor activity and elevated metabolic rate. Mch1r-/- mice are lean, with 50% less fat mass and 7% more lean mass than +/+ littermates. When presented with a high fat diet, Mch1r-/- mice consume the same amount of food as wild-type, and thus appear resistant to diet-induced weight gain. Hypothalamic NPY and AGRP signaling pathways are intact in Mch1r-/- mice, as shown by an increase in food intake following lateral ventricle administration of either NPY or AGRP; however chronic administration of MCH has no effect, demonstrating that the orexigenic effects of MCH are mediated solely through the MCH1R.

Serotonin receptor 2c. Serotonin (5-HT _2c ) receptors are expressed in the PVN, LHA, and anterior hypothalamic nucleus (AH) of the hypothalamus. 5-HT is anorexigenic when injected into the PVN of rodents [152], and the weight-reducing effects of sibutramine are mediated by inhibition 5-HT uptake [153]. Deletion of the 5-HT _2c receptor produces a grossly normal mouse on a mixed background of 129/Sv, DBA and C57BL/6J [154]. However, when backcrossed for eight generations to the C57BL/6J strain, male 5-HT _2c null mice developed obesity at 5 months of age, had a 30% increase in body weight, and a 40-64% increase in fat mass by 10 months of age as compared to wild-type littermates [155]. These effects were due primarily to hyperphagia. Administration of a 17% fat diet (vs 4% kcal fat of control chow) for 9 weeks increased the rate of weight gain. Body composition changes in 5-HT _2c -deficient mice were accompanied by hyperinsulinemia, glucose intolerance, and hyperleptinemia without hypercortisolemia. Therefore, 5-HT _2c deficiency predisposes C57BL/6J mice to a mild obesity syndrome that is accelerated by exposure to a high-energy diet. Pairfeeding studies suggest that the increased body weight and adiposity of 5-HT _2c mice is purely a function of increased food intake, since pairfed 5-HT _2c mice have body weights and fat content indistinguishable from their wild-type littermates. Therefore, the 5-HT _2c receptor knockout mouse represents a model of obesity due largely, or solely, to hyperphagia.

Melanocortin receptors. The melanocortin-4 receptor (MC4R) is a G-protein coupled receptor that is potently activated by a-MSH with an EC ₅₀ value of 18 nM for rat MC4R [156] and an EC ₅₀ value between 1-7.9 nM for the human receptor [157-160]. MC4R is expressed in the PVN, DMH, VMH and LHA [161], and is constitutively active in vitro , as demonstrated by increased basal cAMP production that is proportional to receptor number [162]. AGRP induces hyperphagia by suppressing MC4R constitutive signaling via inverse agonism [162, 163] and by blocking a-MSH [164].

Female mice homozygous for a targeted deletion of Mc4r consume 46% more calories than their +/+ littermates and have a 3- to 5-fold increase in adiposity and a 50 and 100% increased body weight (males and females, respectively) [165, 166] while maintaining the same absolute lean body mass as +/+ littermates [167]. Heterozygosity for the Mc4r mutation elicits an intermediate phenotype [165]. By comparison, male A ^vy /a mice, which have impaired melanocortin receptor signaling due to antagonism by agouti (ASP) protein, have only a 19% increase in calorie intake that produces a nearly 5-fold increase in percentage body fat as compared to a/a littermates, and a 55% increase in body weight [80]. The fact that the Mc4r knockout has a more marked hyperphagia than Ay/a mice further supports the hypothesis that MC4R is constitutively active, and suggests that A ^vy /a and AGRP-overexpressing mice are obese due to the presence of inverse agonists that may incompletely block melanocortin signaling.

Mc4r-/- mice also have increased linear growth as is characteristic of other mice with reduced melanocortin signaling, e.g. A ^y /a and AGRP overexpressors, which is also evident in the human MC4R deficiency syndrome [168]. Mc4r -deficient mice are able to maintain core body temperature when exposed to a cold challenge [166], indicating that sympathetic tone is not reduced as it is in Lep ^ob and Lepr ^db mice. This suggests that homeostatic regulation of core body temperature is apparently mediated downstream of leptin signaling via a pathway other than the melanocortin signaling pathway. Pairfeeding Mc4r+/- mice to their +/+ littermates normalized the body weights of male Mc4r+/- mice but not females, suggesting a purely hyperphagia-driven obesity in males and an additional metabolic defect in females. Oxygen consumption of Mc4r-/- mice is reduced by 20% as compared to +/+ controls of the same weight, indicating that the melanocortin pathway mediates its effects on body weight regulation via both increased food intake and a reduction in energy expenditure.

A recent study indicates that approximately 4% of morbid human obesity (as defined by a maximal BMI > 40 kg/m ² ) may be due to mutations in the melanocortin receptor [169], and this finding has been further supported by additional reports [168, 170]. Most obesity associated with MC4R mutations has been attributed to heterozygosity for such mutations [170]. Currently 29 MC4R mutations have been reported in a total of 71 individuals [102]; these are, therefore, the most frequently published human single-gene mutations related to obesity. Severe childhood obesity results from a null MC4R receptor, generated by missense, frame shift, deletion, and nonsense mutations in 5.8% of obese children (29/500) [168]. MC4R mutations are codominantly inherited [168], and as a result, heterozygous family members are overweight [171].

Centrally administered AGRP causes hyperphagia in Mc4r-/- mice [131], supporting a role for of an additional brain melanocortin receptor in the regulation of body weight. The MC3R is found in the ARC, DMH and VMH [161]. In comparison to the MC4R, the MC3R has reduced affinity for AGRP [164], and increased affinity for a-MSH [172]. Chen et al. showed that Mc3r-/- mice develop late onset obesity beginning at 26 weeks of age, when their body weight diverges significantly from their wild-type littermates, accompanied by a 100% increase in fat mass [173]. However, in a second study, female Mc3r-/- mice weaned onto a high-fat chow were slightly heavier than heterozygous littermates at 50 to 60 days of age (p < 0.05) [174]. Mc3r-/- males were not significantly heavier than littermates after weaning onto a high-fat diet, but showed a 40% increase in adiposity by 15 to 20 weeks of age [174]. These effects are considerably smaller than the 3- to 5-fold increased adiposity observed the Mc4r null mouse.

The Mc3r-/- mouse displays normal food intake, reduced locomotor activity and significantly raised respiratory quotient on high-fat chow as compared to wild-type littermates, suggesting a defect in energy partitioning when challenged with a high-fat diet [173, 174]. The Mc3r-/- mouse of Chen et al. exhibits increased feed efficiency despite hypophagia in female Mc3r-/- mice, and normal food intake in males [173]. These Mc3r-/- mice have a normal metabolic rate by indirect calorimetry, in contrast to the Mc3r-/- mice of Butler et al., but have reduced ambulatory activity in females only. These findings suggest a defect in energy partitioning that reduces lean mass and increases fat mass in both sexes.

A severe form of human obesity associated with a novel heterozygous MC3R mutation was recently reported in an obese 13-year-old girl and her father [175]. The mutation is a single amino acid transversion [I183N]. However, a subsequent study showed no significant increase in the frequency of MC3R sequence variants in a group of morbidly obese subjects [176], suggesting MC3R mutations are not frequently associated with genetic obesity.

Proopiomelanocortin. Proopiomelanocortin (POMC) is expressed in both the anterior pituitary and the hypothalamus. In the anterior pituitary, POMC is processed enzymatically to ACTH and b-lipotropin. In the intermediate lobe of the pituitary, and in the hypothalamus, ACTH is processed further to a-MSH and CLIP. A defect in the processing of POMC, and the production of a-MSH, an MC4R agonist, has been identified in humans. Human subjects have been described that are 1) compound heterozygous for mutations in exon 2 of POMC that result in premature termination of transcription as well as a frameshift mutation that disrupts the common binding site of a-MSH and ACTH, or 2) homozygous for a nucleotide transversion mutation in exon 3 that truncates POMC protein at codon 79, resulting in trace or undetectable amounts of circulating a-MSH and ACTH [177]. These individuals exhibit early onset obesity and red hair because of the a-MSH deficiency, and are adrenal insufficient due to a lack of circulating ACTH.

Pomc-/- mice generated by gene targeting weigh 100% more than wild-type littermates at 12 weeks of age, and are hyperphagic when presented with either standard or high-fat chow [178]. Daily intraperitoneal injection of a-MSH to Pomc-/- mice caused a 46% weight loss over a 2-week period with a concomitant darkening of coat color, while wild-type littermates were unaffected. Unlike the Cpe ^fat mouse, this model represents POMC deficiency in isolation from other hormone deficiencies. To investigate specifically the role of a-MSH in body weight regulation, Pomc null mice have been rescued by transgenic overexpression of POMC in the pituitary but not the brain [179]. Homozygous rescue mice are 33% heavier than Pomc null mice, indicating that a-MSH deficiency in the brain causes the obesity phenotype and that the increased concentrations of circulating glucocorticoids have an additive effect to increase body weight.

Prohormone convertase-1 (PC1, or PCSK1) cleaves ACTH from POMC in the anterior pituitary, and is required for cleavage of a-MSH from POMC in the hypothalamus. The cleavage site of prohormone convertases is (K/R)--(X)n--(K/R), where n = 0, 2, 4, or 6 and X is any amino acid (for a review, see [180]). Human PC1 deficiency caused by missense and splice site mutations in the PC1 gene results in a disorder characterized by obesity and hypocortisolemia as well as hypogonadism due to aberrant processing of progonadotropin release hormone [90, 181]. Additional prohormones that are incompletely processed due to PC1 deficiency include proinsulin, prothyrotropin release hormone, progastrin, proneurotensin and prodynorphin [88], which may all potentially contribute to the obesity syndrome. Similarly, CPE enzymatically "trims" arginine and lysine residues from the C-terminus of the cleaved hormone or neuropeptides, eliciting the obese phenotype of Cpe ^fat mice [15] and humans [89].

Agouti-related protein. Agouti signaling protein (ASP), which is ectopically overexpressed in obese A ^y /a and A ^vy /a Yellow mice, contributes to the phenotype by blocking melanocortin signaling [182]. An homology search was undertaken to identify a protein with a comparable normal physiological function in the brain, and a candidate with 25% amino acid homology to ASP was identified. Direct evidence supporting a similar mechanism for AGRP was later demonstrated. In vitro binding studies showed that AGRP binds the MC1, MC3 and MC4 receptors, and mice globally overexpressing AGRP or agouti protein are hyperphagic and are twice as obese as nontransgenic littermates [164]. AGRP is unlike agouti protein in that it does not block eumelanin synthesis to elicit a yellow coat color when transgenically overexpressed in mice [164].

In support of a physiological role for the orexigenic action of AGRP, fasted animals show increased expression of AGRP mRNA in the ARC. AGRP is colocalized with NPY in ARC neurons, and exogenous intracerebroventricular administration of AGRP to rats elicits a long-lasting hyperphagic response [183]. In addition, Lep ^ob and Lepr ^db mice overexpress AGRP in the arcuate nucleus [164]. However, deletion of AGRP by homologous recombination results in a normal phenotype [132], although MCH expression in the LHA is significantly elevated (p < 0.05), suggesting a compensatory MCH response. Npy-/-;Agrp-/- mice are also metabolically identical to their +/+ littermates [132]. As with the Npy-/- mouse, these findings suggest that compensatory mechanisms defend energy homeostasis.

Syndecans. Cell surface heparan sulfate proteoglycans (HSPGs) modulate ligand-receptor interactions at neural synapses. In vitro studies suggest that HSPG syndecan-1 may bind to AGRP, facilitating AGRP binding to MC4R. In accord with this model, transgenic mice overexpressing syndecan-1 exhibit late-onset obesity [184]. The endogenous hypothalamic analogue of syndecan-1 is syndecan-3, and fasted mice show a four-fold induction of syndecan-3 mRNA in hypothalamic areas involved in energy balance [184]. When challenged by a 16-hour fast, syndecan-3-/- mice had a blunted refeeding response, presumably due to the decreased binding of AGRP to MC4R. These studies reveal a novel role for HSPGs in the regulation of peptidergic synapses that control body weight. The effects of null mutations of syndecan-3 on the phenotypes of mice with mutations of the leptin or melanocortin axes have not been reported. [M3]  

Insulin receptor. Like leptin, insulin circulates in proportion to fat mass, and alters neuropeptide expression in the hypothalamus via receptors located in the ARC, PVN, and DMH [185]. Insulin signaling in the hypothalamus shares key transduction pathways with leptin, notably, phosphatidylinositol-3-OH kinase (PI(3)K) and MAP kinase, and alters neuropeptide expression in favor of reduced energy intake [186, 187]. Mice with insulin receptor deficiency in the brain are more susceptible to diet-induced obesity than wild-type mice [188], although on a standard chow diet, only females show a modest 10% elevation in body weight and adiposity as compared to control females. Insulin receptor substrate-2 (IRS-2), a member of the insulin signaling cascade that couples the insulin receptor to downstream signaling molecules, is expressed in the hypothalamus. Global deletion of IRS-2 results in a 20% increase in body weight, and a 30% increase in food intake, causing a doubling of body fat in mutant mice [189].

To investigate the role of insulin signaling in the periphery, deletion of insulin receptor has been performed in a host of other tissues, including pancreatic b-cells, liver, and adipose tissue. Pancreatic-specific deletion of insulin receptor has significant effects on body weight and adiposity [190], and insulin receptor deficiency in skeletal muscle causes a hyperinsulinemic, normoglycemic phenotype in male mice, with an increase in percentage body fat from 27.5 to 38% [191, 192]. This increased adiposity is associated with a decrease in insulin-stimulated glycogen storage in muscle and a concomitant increase in glucose uptake in adipose tissue.

Protein tyrosine phosphatase-1B (PTP-1B) is a negative regulator of insulin signaling. Global disruption of this gene causes sustained activation of the insulin signaling pathway [193]. PTP-1B-/- mice have normal body weight [194] and are resistant to weight gain on a high-fat diet. Klaman et al. subsequently reported that their independent line of PTP-1B-/- mice were 16% leaner on a chow diet, and that male mice had 50% reduced adipose mass on a high fat diet [195]. Both groups found that PTP-1B-/- mice were protected from high-fat diet-induced insulin resistance. This protection from insulin resistance appears to be a product of increased energy expenditure as determined by indirect calorimetry [195] and an increased sensitivity to both insulin and leptin signaling [194, 195][196]. Since intracerebroventricular administration of insulin upregulates POMC mRNA expression [187], deletion of PTP-1B in hypothalamic neurons expressing POMC could potentially provide a model to elucidate the contribution of brain insulin signaling to diabetes.

Corticotropin-releasing hormone. During classical stress tests such as the 10-minute restraint and exposure to open field, the release of CRH from the PVN is increased in mice and rats [197]. CRH, in turn, increases glucocorticoid secretion via the hypothalamic-pituitary-adrenal axis. Centrally administered CRH is associated with decreased food intake and weight loss [198], and metallothionein-driven transgenic overexpression produces a Cushing's syndrome phenotype with increased central adiposity due to corticotrope stimulation [199]. Transgenic inactivation of CRH impairs the response to restraint stress, but has no impact on energy homeostasis [200].

CRH1 and CRH2 receptors are G-protein-coupled receptors that share 71% amino acid sequence homology [201]. CRH1 is found primarily in the brain in the cerebral cortex, cerebellum, medial septum and anterior pituitary, while CRH2 is expressed mainly in the periphery, in the heart, skeletal muscle, vasculature and gastrointestinal tract [202], but can be found in the ventromedial hypothalamus [203]. Antagonism of either CRH1 or CRH2 leads to increased body weight via increased food intake (CRH1) [204] or decreased energy expenditure (CRH2) [205]. However, all models of CRH receptor deficiency ( Crh1-/- [206], Crh2-/- [207] or Crh1-/- plus Crh2-/- [202]) have normal regulation of body weight, despite altered responses to stress.

At least two explanations are possible for the discrepancies between pharmacologic and gene knockout studies. Mice have elevated CRH in response to a fast, while rats show a decrease, and studies in which CRH was exogenously administered have been largely carried out in rats, due to the technical constraints of performing central injections in mice. In mice, therefore, it cannot be ruled out that the role of CRH in the stress response overshadows its role as a modulator of food intake, since even a short fast jeopardizes the survival of mice.

In addition, urocortin and urocortin III have 45% and 26% amino acid homology, respectively, to CRH, and are potent ligands of the CRH2 receptor [208]. Urocortins are endogenously expressed in the Edinger-Westphal nucleus located in the midbrain, and intracerebroventricular administration of urocortin suppresses food intake [209]. No mouse models of urocortin deficiency or overexpression exist to date; however, mice with overexpressed [210] or deleted CRH binding protein (CRH-BP) [211] in the brain show increased and decreased food intake, respectively. Since CRH-BP binds urocortin in addition to CRH, it is possible that alteration in both urocortin as well as CRH is necessary to elicit changes in body weight, perhaps via an as yet unidentified receptor.

Glucocorticoid receptor. CRH alsoregulates somatic energy partitioning via the HPA axis. CRH-stimulated ACTH release causes glucocorticoid secretion, and pathological upregulation of this axis promotes concomitant fat storage and skeletal muscle catabolism, and an increase in visceral adiposity, as seen in Cushing's syndrome [212]. However, under normal physiological conditions, glucocorticoids feed back on CRH neurons to downregulate CRH release. Neuron-specific deletion of the glucocorticoid receptor (GR) has been achieved by crossing animals with a nestin (neuron)-expressed CRE recombinase to mice targeted for loxP sites flanking exon 3 of GR [213], which is common to all GR isoforms. These animals overproduce glucocorticoids, but do not exhibit visceral obesity due to a lack of negative feedback inhibition via glucocorticoid receptors on CRH neurons of the PVN [214]. The resulting overexpression of CRH may be the mechanism by which an increase in visceral adiposity is prevented, since intracerebroventricular administration of CRH decreases food intake and body weight in rodents [215]. Adult animals are therefore 50% leaner, with no change in energy expenditure when corrected for their 16% decreased linear growth as compared to wild-type.

By contrast, human visceral obesity is associated with elevated cortisol secretion due to increased local activity of 11-b--hydroxysteroid dehydrogenase type 1 (11bHSD-1) in adipose tissue [216], but normal levels of circulating glucocorticoids [217]. 11 bHSD-1 is the enzyme that catalyzes the last step in glucocorticoid and mineralocorticoid synthesis. This tissue-specific overexpression is similar to a form of hereditary hypertension, in which overproduction of renal 11 bHSD-2 causes mineralocorticoid excess in the kidney. It was therefore hypothesized that adipose-specific overexpression of 11 bHSD-1 might contribute to obesity by catalyzing the conversion of cortisone to cortisol, which in turn upregulates adipocyte differentiation of stromal cells to mature adipocytes, as demonstrated in vitro [218]. A mouse model of visceral obesity was generated by transgenic overexpression of 11 bHSD-1 driven by the aP2 promoter to confer adipose tissue specificity in the milieu of normal levels of circulating corticosterone [219]. 11 bHSD-1 overexpressing mice of one line consumed 17.1% more calories than lean controls, gained 16% more weight by 9 weeks of age, and weighed 21% more than controls when administered a high-fat diet. A 3.7-fold increase in the mesenteric (visceral) fat pad weight was accompanied by significant elevations in serum free fatty acids, triglycerides and leptin. Adipocyte size rather than adipocyte number was increased in both the subcutaneous and mesenteric fat pads, with a preferential 3-fold increase in fat mass the latter, possibly as a direct consequence of local glucocorticoid action. Transgenic mice were also hyperglycemic, hyperinsulinemic and glucose intolerant, recapitulating the human syndrome of visceral obesity. These findings suggest that increased production of glucocorticoids in adipose tissue drive the visceral obesity syndrome via increased adipocyte differentiation.

Interleukin-1 and Interleukin-6. Elevations in plasma and brain cytokines are associated with negative energy balance. Intracerebroventricular concentrations of interleukin-1b (IL-1b ) increase in anorexic tumor-bearing rats, and central injections of recombinant IL-1b elicit a reduction in food intake [220]. Targeted deletion of IL-1b converting enzyme (ICE), which is essential for IL-1b activity, prevents lipopolysaccharide (LPS)-induced anorexia in mice [221], although baseline food intakes do not differ between ICE-/- and ICE+/+ mice. Mice with constitutively increased IL-1 receptor signaling induced by targeted deletion of the endogenous IL-1 receptor antagonist IL-1ra have 18% and 29% decreased body mass in males and females, respectively, as compared to wild-type littermates [222]. In agreement with these observations, reduced cytokine signaling in IL-6-/- mice is associated with a 20% increase in body weight and a 50-60% increase in subcutaneous fat pad mass by 36 weeks of age [223].

Plasma cytokine concentrations are normal or reduced in obese individuals as compared to lean controls [224], although they are significantly elevated in adipose tissue [225], and are thought to stimulate adipocyte growth. Site-specific deletion and transgenic overexpression of cytokines may further elucidate their putative paracrine function in adipose tissue.

Single-minded-1. In mice, Sim1 encodes a transcription factor required for the development of the PVN, an integrative hypothalamic nucleus in which a-MSH, NPY and 5-HT neuromodulators are released. Its constituent neurons contain MC-4R, NPY-Y1 and 5-HT _2c receptors, and neurons secreting the anorexigenic peptide CRH. An underdeveloped PVN has the potential to alter energy balance, since PVN ablative lesions in rodents produce a profound hyperphagia [226].

Human Single-minded-1 (SIM1) deficiency was discovered by karyotyping in three case studies of young obese patients with small deletions or translocations at the human SIM1 locus on chromosome 6 [227-229]. In an overweight female patient with a balanced translocation between the short arm of chromosome 1 and the long arm of chromosome 6, hyperphagia was documented as early as 4 years of age, and was accompanied by increased height and a normal TEE:BMR ratio as compared to girls of her age [227]. By age 6 years the patient weighed nearly 50 kg and had 52% body fat. All laboratory values were normal. Cytogenic studies and fine mapping revealed that the translocation was complete with the exception of a C (r) T transversion in the SIM1 gene. The SIM1 mutation predicted a fusion transcript that encodes a dominant-negative protein; however, this gene product was not detected in proband RNA isolated from lymphoblastoid cells. The transversion is a common polymorphism in obese and lean children that may be more fully expressed when accompanied by certain genetic and/or environmental variables. Therefore the contribution of SIM1 to the obese phenotype is not clear.

The PVN regulates a variety of homeostatic processes including blood pressure regulation and respiration, and thus homozygous deletion of Sim1 is embryonic lethal in mice. Sim1 +/- mice are normal until 4 weeks of age, when they develop hyperphagia-induced obesity and surpass the energy intake of their littermate controls by more than 30% [230]. At 24 weeks, males and females weigh 43% and 68% more than wild-type males, respectively, with an approximate 50% increase in retroperitoneal fat pad mass in both males and females. Immunohistological examination revealed that vasopressin, oxytocin, galanin, CRH, TRH, 5-HT _2c, and tubby protein were still expressed in the PVN. However, the number of neuronal nuclei were reduced by 24%, with a proportional decrease in the PVN area. It is possible that hypocellularity of the PVN could lead to reduced neuropeptide production eliciting the obese phenotype. Surveys of obese human populations have indicated that the occurrence of the SIM1 mutation in obesity is exceedingly rare, however, it is possible that the more common, milder human obesities may involve factors affecting PVN development.

Steroidogenic factor-1. Like Sim1 , steroidogenic factor 1 ( Sf1 ), is a transcriptional modulator expressed in the VMH, a discrete hypothalamic nucleus implicated in the regulation of body weight [231]. The VMH contains neurons that express receptors that bind anorexigenic molecules (LEPR, CRH2R, MC3R) and are positive for tubby protein expression. Although the Sf1 -deficient mouse was first described in 1994 [232, 233], early death due to adrenal insufficiency prevented characterization of this mouse in adulthood. By performing adrenal transplantation, it was possible to observe a late-onset obesity in Sf1 -deficient mice that resembles the mild obesity phenotypes of MC3R deficiency and the spontaneous tubby mutation [234]. Sf1 -deficient, adrenal-transplanted mice appear normal until 8 weeks of age, when their body weights diverge from their wild-type littermates. By 6 months, Sf1-/- mice are 72% heavier than controls due primarily to increased body fat, with no differences observed in linear growth. A systematic study of energy expenditure has not been performed on SF-1-deficient mice, although they exhibit a decrease in locomotor activity as compared to wild-type mice.

Necdin. Prader-Willi syndrome (PWS) is a neurogenetic disorder characterized by postnatal hypotonia and failure to thrive until approximately two years of age, when a pronounced hyperphagia develops ultimately resulting in severe obesity. PWS arises from a lack of expression of paternally derived genes at locus 15q11-q13 [235]. Five candidate genes for PWS have been identified in this region: Snurf-Snrpn , Mkrn3 , Ipw , Ndn and Magel2 . Orthologs of these genes are located in an imprinted region of mouse chromosome 7C. Mice with deletions in each of these genes have been generated, but no obesity phenotype is apparent in any if them. However, four of the five models are lethal in the first week of life due to respiratory distress or hypotonia, consistent with human PWS. Necdin (Ndn) , a neuron-specific cell growth suppressor (for a review, see [236]), is a likely candidate gene in this chromosomal region since it is inactivated in PWS patients and is expressed throughout the brain, including subpopulations of hypothalamic neurons. Only 20% of Ndn-/- mice survive the neonatal period, and these animals are not obese. The major behavioral phenotype of Ndn null mice is improved spatial learning and increased skin scraping activity in the open field as compared to littermate controls, suggestive of the high aptitude for visual-spatial integration and stress-induced skin-picking behaviors exhibited in PWS [237]. These mice also exhibit suppression of hypothalamic oxytocin and luteinizing hormone-releasing hormone (LHRH) expression, characteristic of PWS [238, 239].

Interestingly, penetrance of the Ndn-/- phenotype is accentuated on a C57BL6/J background [237]. Since PWS is thought to be a contiguous gene disorder, the absence of obesity in Ndn-/- mice suggests that neighboring genes may also be candidate genes that interact with one another, or may act in concert with Ndn . It is possible therefore that an obese phenotype may become evident in double or triple mutants of the contiguous genes.

Brain-derived neurotrophic factor. Brain-derived neurotrophic factor (BDNF) is a member of the neurotropin family, a group of four proteins that promote neuronal survival. While other neurotrophins have specific roles in the peripheral nervous system, BDNF has a wide variety of functions in the central nervous system. In the hippocampus, BDNF alters plasticity of synaptic connections and synaptic input in the long-term potentiation component of learning and memory, but in the hypothalamus its role is less clear. BDNF is a secreted protein expressed in neurons of the LHA and the PVN, and is upregulated by osmotic or immobilization stress. Central infusion of BDNF elicits a 10-15% weight reduction in rats [240] and mice [241]. Since BDNF-/- mice die shortly after birth, body weight regulation has been studied in heterozygotes [241, 242]. BDNF+/- mice are obese and hyperphagic, showing a 47% increase in food intake and a 4-fold increase in adiposity as compared to wild-type mice. While BNDF mRNA was significantly reduced in BDNF+/- mice, as determined by in situ hybridization, no differences in hypothalamic NPY, CART or LEPR mRNA were apparent, suggesting that BDNF may regulate body weight via other pathways.

Mice with conditional deletion in BDNF have been generated to circumvent embryonic lethality associated with complete BDNF deficiency [243]. Conditional deletion of BDNF in the brain using a Cre recombinase driven by the CamKII a promoter allowed mice to survive to adulthood. Whereas BDNF+/- male and female mice were 50% and 27% heavier than lean littermates, BDNF-/- mice showed 80% and 150% increases in body weight that were apparent after 8 weeks of age, and were attributable to hyperphagia. In addition, BDNF-/- mice showed a significant elevation of basal POMC in the ARC probably secondary to elevated leptin levels that accompany increased adiposity. BDNF+/- mice have reduced hypothalamic cFos activation in response to dexafenfluramine, a 5-HT-release reuptake inhibitor [244], suggesting that the observed obesity is due to BDNF-deficiency-induced suppression of serotonin response. However, administration of fluoxetine, a 5-HT reuptake inhibitor, did not reverse the obesity syndrome of BDNF-deficient mice [243], indicating that decreased endogenous serotonin concentration is probably not a major contributor to this phenotype.

Dopamine. The role of dopamine in the regulation of food intake is controversial. Lesions of the lateral hypothalamus in rodents cause hypophagia [146], presumably due to the ablation of neurons containing the orexigenic neurotransmitters MCH and orexin as well as dopamine. In support of this observation, centrally administered dopamine agonists increase food intake [245], and peripherally administered antagonists or central dopamine depletion causes hypophagia [246]. Yet intraperitoneal administration of bromocriptine, a dopamine agonist, attenuates obesity in Lep ^ob mice [247].

Mice that lack tyrosine hydroxylase are deficient in dopamine, noradrenaline and adrenaline, and die between embryonic day 11.5 and 15.5 [248]. Replacing tyrosine hydroxylase in noradrenergic neurons by transgenesis generates viable dopamine-deficient mice that synthesize noradrenaline and adrenaline [249]. Dopamine-deficient pups nurse normally until 2 weeks of age, but thereafter fail to thrive due an inability to wean themselves onto solid food unless supplemented with L-DOPA, suggesting that dopamine is required for normal ingestive behavior. However, ingestive behavior data that describe dopamine as a stimulator of food intake may be confounded by the roles of dopamine in the initiation of motor activity and reward mechanisms. For instance, the potency of dopamine antagonists may be dependent upon mode of feeding, since direct delivery of nutrients via intraoral catheter, which does not require motor activity or motivation, attenuates the hypophagic response to the dopamine antagonist SCH 23390 [250].

Ghrelin. Ghrelin is a recently discovered hormone released from cells in the mucous membrane of the stomach, duodenum, ileum, cecum and colon [251], and is postulated to be an initiator of food intake [252]. Ghrelin is a growth hormone secretogogue [253] that increases growth hormone (GH) secretion via a GHRH-independent pathway. Diurnal release of ghrelin into the circulation coincides with the initiation of meals, and decreases over the course of each meal [254].

Intracerebroventricular administration of ghrelin over seven days increases body weight and adiposity, while fasting increases ghrelin levels, suggesting a central role in the regulation of energy balance [252]. These effects occur independently of GH secretion, since GH-deficient mice also exhibit an increase in adipose mass after ghrelin administration [252]. Ghrelin activates vagal afferents that link the gut to the brain, thereby increasing cFos expression of NPY-positive neurons in the ARC [255]. Ghrelin-induced feeding and activation of NPY neurons is abolished by bilateral subdiaphragmatic/gastric branch vagotomy or perivagal capsacin treatment. Moreover, receptors that bind ghrelin have been colocalized with ARC NPYergic and AGRPergic neurons [256].

An association of obesity with oversecretion of ghrelin has only been observed in the instance of extreme hyperphagia present in Prader-Willi patients [257]. In all other human populations and mouse models studied to date, obesity is associated with reduced ghrelin levels as compared to lean controls, and increases when body weight is reduced [258]. Roux-en-y gastric bypass surgery is an effective treatment for morbid obesity that is associated with a profound suppression in serum ghrelin concentrations. This decrease may account for some of the reduced hunger reported by these patients [258].

The recently generated ghrelin-deficient mouse exhibits no gross defects in body weight indicating that compensatory mechanisms may exist similar to those present in NPY- and AGRP-deficient mice [259]. Additional studies to determine if ghrelin deficiency alters the phenotype of obese mouse models will be necessary to delineate further its role in energy metabolism.

Cholecystokinin. Cholecystokinin (CCK) is a gastrointestinal hormone secreted by endothelial cells lining the jejunum in response to ingestion of a meal. CCK induces a transitory sensation of satiety, secretion of pancreatic enzymes and gallbladder contraction. CCK-A receptors are located on vagal afferents of the stomach and the liver and transduce signals via the vagal nerve to satiety centers in the brainstem, eliciting a brief reduction in food intake (for a review, see [260]). CCK-B receptors are located diffusely throughout the brain, but their role in the satiety effect of CCK has not been demonstrated. The Otsuka Long-Evans Tokushima Fatty (OLETF) rat is an outbred strain of Long-Evans rats used experimentally as a model of type 2 diabetes. This animal has a significant 34% increase in food intake resulting from larger meal size, accompanied by a 23% increase in body weight at 15 weeks as compared to lean Long-Evans rats [261]. In 1994, a mutation in the CCK-A receptor of OLETF rats was identified and characterized [262], and it has since been established that this 6847-base-pair deletion disrupts the CCK-A receptor promoter [25]. This mutation reduces receptor expression on enzyme-secreting pancreatic acini, and elicits a subsequent attenuation of CCK-A receptor signaling as determined by reduced amylase secretion [263].

While CCK decreases meal size and duration, compensatory increases in meal frequency cause CCK administration to have no long term effects on total food intake or body weight. By contrast, acute central administration of leptin decreases body weight and food intake by suppressing meal size, with no significant change in meal duration or frequency [264]. Interestingly, a polymorphism in the CCK-AR gene is also associated with obesity in humans [265], however deletion of the CCK-A receptor in mice does not elicit an obese phenotype [266]. This suggests that the OLETF phenotype results from a number of identified polymorphisms associated with increased adiposity in addition to the CCK-A receptor deficiency [103-105]. When the OLETF rat is backcrossed to the inbred, nondiabetic Fischer 344 rat strain, several QTL associated with increased adiposity are revealed. Therefore, it seems likely that these QTL make significant contribution to the obese phenotype independent of the CCK-A receptor mutation, although the fact that littermates of the OLETF rat nonmutant for CCK-AR are lean indicates that the background itself is not sufficient to cause an obese phenotype.

Glucagon-like-peptide-1. Glucagon-like-peptide-1 (GLP-1) regulates blood glucose and satiety, and functions as an incretin, or stimulator of insulin secretion, upon its release from L-cells of the duodenum after nutrients enter the intestine. Recently, GLP-1 has also been implicated in the central regulation of body weight, since intracerebroventricular injections of GLP-1 potently suppress food intake in rats, and the GLP-1 receptor antagonist, exendin (9-37), increases food intake 2-fold over saline-injected controls [267].

Glp1r-/- mice were generated to investigate its effects on glucose metabolism, as well as its actions as an incretin, as a suppressor of glucagon secretion, and as a putative central regulator of body weight [268]. In an oral glucose tolerance test, female mutant mice had significantly increased blood glucose at 30 and 60 minutes after glucose administration. Both male and female mutants had suppressed insulin levels with a normal range of plasma glucagons, suggesting that GLP-1R deficiency causes a reduction of glucose-stimulated insulin secretion, but does not alter glucagon levels. Interestingly, intraperitoneal administration of glucose to Glp1r-/- mice also elicited hyperglycemia as compared to littermate controls, demonstrating that GLP-1 modulates blood glucose when no nutrients are in contact with the lumen of the intestine. However, body weight and food intake were unaffected by ablation of GLP-1R, suggesting that compensatory systems offset GLP-1R deficiency. Binding studies on brain sections were performed to rule out the possibility that a second GLP-1 receptor, GLP2-R, mediates GLP-1 action in the absence of GLP-1R, however no binding was observed in GLP-1R null mice. Moreover, crossing Glp1r-/- with Lep ^ob /Lep ^ob mice did not increase the obesity or sensitivity to leptin characteristic of leptin deficiency.

Clinical trials are currently underway to determine the utility of GLP-1 in the treatment of type 2 diabetes and obesity [269-271]. Recent clinical trials show that six weeks of continuous subcutaneous GLP-1 treatment to type 2 diabetic patients reduced body weight by an average of 2 kg [272], however it remains to be seen if this weight loss can be sustained over long periods of time.

Obesity Primarily Due to Decreases in Energy Expenditure

Uncoupling proteins. UCP1 is present in BAT mitochondria and allows dissipation of the electrochemical gradient across the inner mitochondrial membrane, with release of potential chemical energy as heat. BAT ablation, by preventing such thermogenesis, should increase energy stored in adipocytes as acylglycerides. This is the result when a transgene expressing diphtheria toxin A driven by the UCP1 promoter is introduced into mice [273], and the obesity develops in the absence of primary hyperphagia. The obese phenotype is completely prevented when mice are housed in a thermoneutral environment at 35˚C from 3 weeks of age to adulthood, a temperature at which BAT thermogenesis is minimal in normal mice. This reduces the energy requirements of UCP-dta mice to that of wild-type littermates [274].

Mice with a targeted disruption of Ucp1 are not obese, although they exhibit impaired thermogenic response to a cold challenge or administration of a beta-3-adrenergic-receptor agonist [275]. While this inability to adapt to perturbations in metabolic equilibrium apparently reflects reduced response to sympathetic activation, the basal metabolic state of these animals is not compromised by loss of UCP1. The lack of an overt phenotype suggests a novel thermogenic pathway. Ucp2 sequence was found in a homology search for isoforms of Ucp1 , and subsequently cloned from lung and skeletal muscle cDNA. Ucp2 is 59% homologous to Ucp1 , and is expressed in brown adipose tissue, kidney and heart [276, 277] as well as pancreatic b-cells [278]. UCP2 has also been localized to hypothalamic regions that regulate body weight [279], suggesting a possible role for UCP2 in the regulation of local brain temperature in response to sudden environmental alterations in ambient temperature [280].

Although the two mouse models of UCP2 deficiency have shown no differences in body weight as compared to wild-type controls [281, 282], lack of UCP2 prevents uncoupling in islets and increases islet ATP levels [282], resulting in increased b-cell mass and islet insulin content [283]. When Ucp2-/- mice are exposed to high-fat feeding conditions, they demonstrate increased insulin sensitivity despite a significant increase in body weight gain as compared to Ucp2-/- mice fed control diet [283]. These findings indicate that inhibition of UCP2 activity is a potential therapeutic target for the prevention and treatment of type 2 diabetes.

The short and long isoforms of UCP3 were isolated by screening the human skeletal muscle cDNA library [284]. The amino acid sequence of UCP3 is 57% homologous to UCP1, which instigated numerous studies exploring the uncoupling properties of UCP3. Ucp3 deletion by homologous recombination has no effect on body weight regulation [285, 286]. However, mice with transgenic overexpression of UCP3 in skeletal muscle are leaner than wild-type littermates despite hyperphagia, due to a 25% elevation in metabolic rate [287]. To explore these discrepancies, the uncoupling capacity of UCP3 was assessed in humans using phosphocreatine (PCr) resynthesis as an indicator of mitochondrial function [288]. Upregulation of UCP3 protein expression occurred independently of PCr resynthesis induced by voluntary anoxic skeletal muscle contraction in human subjects fed a high-fat diet. This strongly suggests that UCP3 does not participate in mitochondrial uncoupling, but instead as a mitochondrial fatty acid anion exporter when influx of fatty acids exceeds mitochondrial capacity.

No evidence to date suggests that changes in UCP expression in any tissue are associated with body mass. Instead, it may be proposed that any physiological condition that promotes increased circulating free fattty acids (FFAs) induce UCP expression. For instance, UCP2 and UCP3 mRNA expression in skeletal muscle is upregulated in fasting lean individuals [289], and likewise skeletal muscle UCP3 expression is increased in human subjects by 44% after one week on a high-fat diet [288].

Several human coding variants for UCP1 , UCP2 and UCP3 associated with increased BMI and body fat have been described. UCP1 allele polymorphisms occur with higher frequency in individuals with a propensity for weight gain [290, 291], and one study found an association between a common UCP2 polymorphism and obesity [292]. Although several reports of mutations in UCP3 do not show any correlation with human obesity [293][294], morbid obesity in one subject has been associated with a splice donor junction polymorphism in exon 6 of UCP3 [295], and a severely obese and diabetic subject was heterozygous for an R70W transversion in a highly conserved region of UCP3 [296].

Beta-adrenergic receptors. While b₃ -ARs are the most abundant b-AR transcripts present in WAT, BAT and skeletal muscle tissue in rodents, determination of receptor abundance by Northern blot [297, 298], RT-PCR [298] and Western blot [299] in human fat and muscle tissue has yielded inconclusive results. Heterozygosity for a W64R polymorphism in human b₃ -AR is associated with obesity and type 2 diabetes [300-302], however this association was not reproduced by all investigators [303-305].

Yet many rodent studies have implicated b₃ -ARs in BAT-mediated thermogenic responses. In both rodents and humans, b₃ -AR agonists induce lipolysis and fat oxidation in WAT, and increase insulin sensitivity. The atypical class of b-AR agonists, such as CL 316,243, elicits these effects in humans via b₃ -ARs without the anxiogenic profile characteristic of b₁ -ARs or b₂ -ARs. However, clinical trials have not revealed a practical role for these compounds in weight loss, since chronic administration of b₃ -AR agonists has no effect on energy balance [306, 307].

Leptin deficient and insensitive rodent models of obesity, such as Lep ^ob and Lepr ^db mice,and Lepr ^fa rats, exhibit a profound downregulation of b₃ -AR mRNA in both WAT and BAT, and are unable to maintain core body temperature when subjected to a cold challenge. Reduced b₃ -AR expression may account for some of the reduced sympathetic nervous response of these animals. Ablation of b₃ -ARs using gene targeting was one of the first genetically engineered models of rodent obesity [308]; however, the resulting phenotype was not as severe as predicted by the observed downregulation of b₃ -AR in other obese mouse models. Female mice show a significant 131% increase in total body fat stores as compared to wild-type mice, while male mice are less affected, with only a 34% increase in fat mass. Food intake and serum levels of blood glucose, insulin and FFA are largely unaffected by b₃-AR ablation. b₃-AR-deficient mice exhibit normal degrees BAT hypertrophy after a 3-week cold challenge and respond normally to the nonselective b-AR agonist isoproterenol. It is possible that the FVB/N background strain used in this study may have attenuated the effects of the mutation; however, mice with simultaneously targeted disruption of b₁ -AR, b₂ -AR, or b₁ -AR and b₂ -AR have normal phenotypes [309, 310].

b-less mice have a global targeted deletion of all three receptor isoforms, and have an obese phenotype that is apparent after 8 weeks on a 58% kcal fat diet [311]; b-less mice are also 50% heavier than wild-type littermates, as a result of increased body fat (22.2 % ± 0.9 as compared to 16.2% ± 1.9 for wild type controls). In contrast, mice with b_1,3 -AR ablation are grossly normal, and have BAT and WAT adipocytes that are visually indistinguishable from wild-type littermates [311], suggesting that diet-induced thermogenesis in mice is maintained by the interaction between or compensation by b-ARs and is prevented only in the absence of all adrenergic signaling.

Protein kinase A regulatory subunit IIb. Increased sympathoadrenal activity in adipose tissue activates a signaling cascade that induces phosphorylation of regulatory subunits of protein kinase A (PKA), which in turn inhibits lipogenesis and increases lipolysis. Deletion of the regulatory subunit II b of PKA (RIIb) , found mainly in WAT, BAT and brain, causes a compensatory increase in RIa , a subunit isoform that constitutively upregulates PKA activity [312]. RIIb -/- mice therefore have constitutively increased cAMP in response to sympathetic activation in adipose depots, with secondary elevation of metabolic rate and body temperature, and a 50% reduction in WAT pad weight despite a mild hyperphagia. Administration of a 58% fat diet for 4 months does not increase adiposity or body weight as compared to wild-type mice administered the same diet [312], and RIIb deficiency improves insulin sensitivity on a high-carbohydrate, diabetogenic diet [313].

Noradrenaline. Sympathetic activity transduced by noradrenaline stimulates biochemical pathways contributing to resting energy expenditure and activates nonshivering thermogenesis in brown adipose tissue in response to a decrease in ambient temperature or an excessive intake of calories [314]. Adrenaline, secreted by the adrenal medulla, promotes the breakdown of storage molecules such as triglycerides and glycogen when energy is required to maintain core body temperature and provide metabolic fuels to critical organs such as the brain. A deficiency in the sympathetic neurotransmitters noradrenaline and adrenaline might contribute to excess energy stored as fat due to reduced energy expenditure and decreased catabolism of triglycerides.

To investigate the role of noradrenaline and adrenaline in the regulation of energy balance, mice were generated with a deletion in b-hydroxylase (DBH), the enzyme that catalyzes the conversion of dopamine to noradrenaline and adrenaline [248]. To prevent embryonic lethality, maternal drinking water was supplemented with the noradrenaline precursor L-threo-3, 4-dihydroxyphenylserine (DOPS) from embryonic day 9.5, and withdrawn at parturition. At 4 to 5 months of age, these noradrenaline- and adrenaline-deficient animals had normal WAT stores but BAT mass was doubled. In a cold challenge, dbh-/- mice showed no upregulation in UCP1 mRNA, and their core body temperature rapidly declined, indicating that UCP1 in the hypertrophied BAT of dbh-/- was hypofunctional as a result of decreased sympathetic activation. In addition, two mechanisms that conserve body heat, piloerection and vasoconstriction, were impaired in dbh-/- deficient mice. Although food intake was significantly increased, body weight was slightly reduced in mutant mice, due to a 24% increase in energy expenditure. Shivering thermogenesis, thyroid hormones, and UCP2 mRNA levels in BAT, WAT, heart, skeletal muscle and liver were all unchanged by DBH deficiency; the mechanism for paradoxical increased metabolic rate is unknown.

Noradrenaline or the a-adrenergic receptor agonist clonidine decrease metabolic rate when injected directly into the PVN [315]. The reduction in metabolic rate suggests that molecules other than noradrenaline may play a role in the regulation of cold-induced thermogenesis and basal metabolic rate. Alternatively, reduction in noradrenergic tone characteristic of obesity may be mediated by local attenuation of noradrenergic signals at the fat pad only. Therefore, the central and peripheral catecholamine deficiency of the dbh-/- mouse may add a CNS component to upregulate energy expenditure that overrides any decrease in energy expenditure that the peripheral deficiency may exert.

Obesity Primarily Due to Altered Partitioning of Food Substrates

ADIPOSE TISSUE 

Formation of adipocytes

CCAAT-enhancer binding proteins. Peripheral regulation of body fat content may be mediated by effects on adipocyte differentiation and/or proliferation. Transcription factors affecting adipocyte differentiation include at least three CCAAT-enhancer binding proteins, a , b, and d (C/EBPa, C/EBPb, and C/EBPd, respectively), that have been deleted by homologous recombination in mice. These C/EPB isoforms share 90% homology and contain a leucine zipper motif that binds to DNA to initiate transcription. Homo- or heterodimerization of C/EBP isoforms is prerequisite to DNA binding, and all combinations of C/EBPa, C/EBPb, and C/EBPd occur.

The first transcription factor identified that binds to the leptin promoter was C/EBPa [316]. Co-transfection of 3T3-L1 cells with C/EBPa activates transcription of LEP mRNA driven by the murine Lep [316][317] or human LEP [318] promoter. Neonatal C/ebpa-/- mice are unable to accumulate lipid in WAT and BAT [319, 320], and show downregulation of Ucp1 expression in BAT mitochondria. Homozygous C/EBPa deficiency is lethal within 8 hours of birth. However, these mice can be rescued for up to 32 hours by repeated injection of glucose. At 32 hours postpartum, C/ebpa-/- mice had no hepatic glycogen, and no accumulation of lipid in either white (inguinal) or brown (interscapular) adipose tissue.

Overexpression of C/EBPb and C/EBPd in 3T3-L1 cells initiates adipoblast differentiation. Homozygous deletion of these transcription factors in mice was achieved by gene targeting [321]. Two-thirds of C/EBPb-deficient mice survive to adulthood, and show reduced ability to accumulate acylglycerides in BAT, while all C/EBPd mice survive to adulthood and exhibit little if any alteration in BAT as compared to wild-type. Double homozygous knockout of C/EBPb and C/EBPd , however, is 85% lethal and lipid accumulation in WAT and BAT is absent due to inhibition of differentiation at the pre-adipocyte stage. C/EBPa expression is the same as wild-type and thus does not compensate for the loss of the other isoforms. This double knockout recapitulates the phenotype of C/EBPa-deficient mice, indicating that full expression of all three isoforms is necessary for normal adipocyte differentiation.

When both WAT and BAT depots are ablated by using the adipocyte-specific fatty acid binding protein promoter (aP2) to drive the expression of diphtheria toxin A, the phenotype of the resulting mouse is strikingly similar to the C/ebpa-/- mouse [322], and the animals die shortly after birth. Animals expressing lower levels of this transgene survive to adulthood, and are resistant to obesity, and infertility induced by monosodium glutamate ablation of cell bodies in the arcuate nucleus and surrounding hypothalamic subnuclei is attenuated.

Adipocyte-specific expression of A-ZIP/F, a protein with dominant-negative effects on transcription of genes required for adipogenesis, prevents the binding of C/EBP isoforms as well as sterol-regulator element binding proteins (SREBPs) at leucine zipper regions on DNA. Differentiation of adipoblasts into WAT is thus prevented, resulting in a WAT-less mouse with a 30% survival rate after 4 weeks of age [323]. The mice that survive to adulthood develop lipoatrophic diabetes and a profound dyslipidemia that resembles Seip-Berardinelli syndrome, although the human genetic mutations associated with this disease are not syntenic to mouse genes associated with obesity [324, 325]. A-ZIP overexpressing mice are hypoleptinemic due to the absence of WAT, and circulating leptin concentrations are restored in a dose-dependent manner by surgical implantation of wild-type adipose tissue [326]. Fat pad transplantation is accompanied by a partial restoration of circulating leptin; conversely, implantation of adipose tissue from Lep ^ob mice has no effect on the A-ZIP overexpressors phenotype [327]. Transgenic overexpression of leptin in A-ZIP mice more effectively improves insulin resistance than fat transplantation, and additionally reduces fatty deposition in the liver [328]. Similar effects have also been recently observed after leptin injection in lipodystrophic humans [329].

Peroxisome-proliferator activated receptor alpha, gamma. The PPAR family ofnuclear transcription factors modulates genes encoding proteins involved in lipid homeostasis. PPARa agonists (fibrates) are used clinically to lower circulating triglycerides and free fatty acids via induction of hepatic genes that regulate mitochondrial uptake and beta-oxidation of free fatty acids. Female mice with a targeted deletion in PPARa exhibit obesity at 10 weeks of age, and by 31 weeks, are 40% heavier than wild-type females when maintained on standard mouse chow [330]. Fat pad weight in Ppara-/- females increases concomitantly with triglyceride and VLDL levels. Food intake increases proportionately with body weight and therefore no increase in feed efficiency is observed, suggesting that the abundance of circulating lipids promotes fat depot uptake due to constitutively impaired lipid oxidation.

Hepatic steatosis is present in male Ppara-/- mice on a control diet by 11 months of age [330], and may be elicited in young mice by a 24-hour fast or several weeks on a high-energy, 39% fat diet [331]. Ppara-/- mice have 50% increased rate of hepatic glycogen depletion as compared to wild-type controls, which significantly impairs their ability to perform endurance exercise on a treadmill [332]. This lack of readily available fuel also reduces metabolic rate by 30% (p < 0.05) in 24-hour fasted Ppara-/- mice as compared to fasted wild-type controls [331]. Taken together these observations suggest partitioning in favor of fat deposition resulting from a reduction in PPARa-dependent hepatic fatty acid oxidation.

PPARg is expressed primarily in adipose tissue, and promotes adipocyte differentiation via C/EBP and SREBP1 and storage of FFAs as triacylglycerols in adipose depots resulting in lower circulating concentrations of FFAs, ultimately improving insulin sensitivity. PPARg agonists (thiazolidinediones, TZDs) are widely used in the treatment of insulin resistance associated with type 2 diabetes. Since targeted knockout of PPARg is embryonic lethal, Pparg +/- mice were generated to study the effects of partial PPARg deficiency [333]. After 15 weeks on a high-fat diet Pparg +/- mice have lower insulin levels, weigh 14% less than wild-type C57BL/6 mice, a have a 70% inhibition of WAT mass accumulation. This is a result of an elevated metabolic rate despite a decrease in food intake. In addition, Pparg +/- mice have increased GLUT4 mRNA in epididymal WAT [333], and improved insulin sensitivity (as measured by OGTT and glucose clamp [334]). Thus, PPARg deficiency may improve insulin response secondary to weight reduction; by contrast, TZDs improve insulin sensitivity while concurrently increasing adiposity.

Administration of the TZD pioglitizone increases adiposity in Pparg +/- vs wild-type mice on both standard and high-fat diets, but decreases insulin sensitivity in the Pparg +/- mice. This unexpected attenuation of insulin action may be explained possibly by the unique inability of PPARg to initiate adipoblast differentiation in Pparg +/- mice, a process that improves insulin sensitivity [333]. Pparg +/- mice on a high-fat diet have elevated serum leptin levels and increased adipocyte expression of leptin mRNA relative to fat mass as compared to wild-type mice.

Interestingly, a PPARg Pro12Ala polymorphism in a Japanese cohort of 896 subjects [335], and in a Scandinavian population of over 3,000 individuals [336] was found to be more prevalent in nondiabetics as compared to diabetics. Among nondiabetics, those individuals with the Ala12 allele were less insulin resistant than those without. In vitro studies indicate that the Pro12Ala polymorphism reduces PPARg transcriptional activity as evidenced by a suppression of lipoprotein lipase and acyl-CoA oxidase peroxisome-proliferator response element transactivation [337], a paradoxical finding given that upregulation of PPARg by TZDs increases insulin sensitivity. This result suggests a novel pathway is responsible for the insulin-sensitizing effects observed in Pro12Ala probands and Pparg +/- mice.

cAMP response element binding protein. cAMP response element binding protein (CBP) is a co-activator for SREBPs, C/EBPs, and PPARs, and therefore has the potential to regulate adipocyte differentiation at several levels. Since CBP-deficient mice ( Crebbp-/- ) die during embryogenesis, Crebbp+/- mice were generated to study the effects of reduced expression of CBP [338]. Crebbp+/- mice are similar to A-ZIP/F, SREBP-1c transgenic and C/EBP knockout mice by virtue of an atrophic WAT compartment. In contrast, however, Crebbp+/- mice have normal BAT, are more insulin sensitive and glucose tolerant than wild-type mice, are resistant to weight gain on a high-fat diet, and show improved glucose tolerance on a high-carbohydrate diet. Similar to Pparg +/- mice, serum leptin levels are four times higher in Crebbp+/- mice versus wild-type controls, although PPARg levels in WAT, skeletal muscle, liver and BAT mRNA remain unchanged. These findings suggest that CBP is a cofactor that promotes storage of energy in WAT, perhaps via the PPAR pathway, but is not essential for adipocyte differentiation or function.

Tumor necrosis factor alpha. Obesity is associated with overexpression of tumor necrosis factor alpha (TNFa) in adipose tissue, with a concomitant modest elevation of this cytokine in the circulation [339]. Obese humans show increased levels of TNFa and IL-6 mRNA and proteinin adipose tissue; there molecules may contribute to decreased insulin sensitivity associated with obesity [340]. One possible mechanism for these effects is the inhibition of insulin receptor and IRS-1 phosphorylation (reviewed in [341]).

TNFaa-immunoglobulin G fusion protein neutralizes TNFa signaling, and improves insulin sensitivity when administered to obese Zucker ( fa/fa ) rats [342]. Mouse models of TNFa deficiency are partly protected from insulin resistance associated with genetic, gold-thioglucose (GTG) and diet-induced obesity [343]; improvements in fasting glucose (13% lower) and insulin (67% lower) are observed in TNFa-/- mice with GTG-induced obesity as compared to TNFa +/+ GTG controls, although glucose and insulin concentrations remain far above normal range. Similar findings are observed in Lep ^ob /Lep ^ob mice null for two isoforms of TNFa receptors (p55 and p75) [344]. Diet-induced obese TNFa-/- and TNFa +/+ mice show no differences in body weight and only slight differences in adiposity [344]. Insulin sensitivity was restored in ob/ob ;p55-/- p75-/- mice [344], however another study reports the opposite effect on insulin sensitivity in diet-induced obese mice lacking both p55 and p75 [345]. In addition, a worsening of glucose homeostasis has been observed in db/db ;p55-/- p75-/- mice [345].

The observation that significant improvements in obesity-induced insulin resistance are seen in a model of TNFa deficiency, but not TNFa receptor deficiency, suggests the possibility that an unidentified TNFa receptor exists. Alternatively, it is possible that a difference in background strains between the studies contributes to the observed phenotypes, since TNFa-/- mice had a mixed 129/C57BL/6 background while the receptor knockouts were backcrossed to C57BL/6 for several generations. Notably, none of these studies includes an adipocyte-specific deletion of TNFa , and therefore the possibility that TNFa may have local effects on fat stores has not been addressed.

High mobility group IC. HMGIC is a member of the high mobility group I (HMGI) family of architectural transcription factors that includes HMGI and HMGI(Y). HMGIC is expressed in undifferentiated mesenchymal cells [346] and fat-cell tumors [347]. HMGIC is highly expressed in fat pads of mice fed a high-fat diet for 1 week, and those of Lep ^ob mice [348]. Hmgic-/- mice have a 47% decrease in body weight and reduced fat mass as compared to wild-type mice [349], and do not gain weight on high-fat diet despite consuming the same amount of food as control mice [348]. This finding suggests that HMGIC regulates adiposity at the level of the fat pad rather than by centrally mediated mechanisms.

Crossing Lep ^ob / Lep ^ob mice with Hmgic-/- mice attenuates the body weight of the obese mice by 73%; however, hyperinsulinemia and hyperglycemia persist despite a 10-30% reduction in fat mass. Taken together, these findings suggest that HMGIC exhibits properties similar to PPARg in that it promotes adipocyte proliferation, however unlike PPARg , it lacks the ability to initiate adipocyte differentiation and does not lower circulating glucose and insulin concentrations.

Sterol-regulatory element-binding protein-1c. Sterol-regulatory element-binding protein-1c ( SREBP-1c) is one of three transcription factors in the SREBP family that increases triglyceride and cholesterol content of adipocytes by upregulating expression of enzymes that synthesize unsaturated fatty acids and cholesterol. In addition, SREBP-1c also promotes adipoblast differentiation by activating genes encoding ligands for PPARg . A lipodystrophic phenotype is exhibited in mice with overexpression of the truncated, biologically active form of SREBP-1c in adipocytes [350]. A 30-40% reduction in epididymal WAT depots is observed in these mice, while the interscapular BAT compartment is hypertrophic and comprised primarily of immature white adipocytes. As in the dominant-negative A-ZIP/F overexpressing mice described above, frank insulin resistance and diabetes with hyperglycemia are present. SREBP-1c overexpressing mice also exhibit a 90% suppression of leptin mRNA expression in WAT, possibly due to a reduction in adipocyte size [351]. Leptin administration reduces food intake and normalizes plasma glucose and insulin, however the size of adipose depots are unaffected by this treatment [352].

Lipodystropic patients treated twice daily with leptin for three months showed normalization of blood glucose, significant lowering of triglycerides and insulin, and reversal of hepatic steatosis characteristic of the disorder [329]. These changes may be elicited by improved hepatic and muscle insulin sensitivity, as indicated by suppression of glucose production and an increase in peripheral glucose uptake.

Acyl CoA:diacylglycerol transferase. Because Acyl CoA:diacylglycerol transferase (DGAT) catalyzes the committed step in the formation of triglycerides, mice deficient in DGAT1 might be expected to have a reduced capacity to store FFAs as triglycerides. Consequently, although body weight is normal in Dgat1-/- mice, fat pad weight is reduced by 50% secondary to decreased triglyceride content per adipocyte [353]. On standard chow, Dgat1-/- mice also show increased energy expenditure that is not attributable to physical activity, but may instead be a by-product of upregulated uncoupling protein expression (UCP1 and UCP3) activated by an apparent increase in insulin and leptin sensitivity [354] that is secondary to a reduction in adipocyte triglyceride content. In addition, PPARa is upregulated and PPARg reduced in WAT of Dgat1-/- mice, consistent with a profile of downregulated adipogenesis [354].

DGAT1-deficient mice are resistant to weight gain on a high fat diet, partly due to increased locomotor activity that is not apparent on standard chow [353]. Energy expenditure of Dgat1-/- mice on a high-fat diet appears otherwise normal, since serum leptin, serum thyroxine and thermogenic response to cold exposure are not altered in comparison to mice on standard chow [353].

DGAT1 deficiency corrects the obesity of A ^y /a mice, but has no effect in leptin-deficient Lep ^ob /Lep ^ob mice, possibly due to compensation via an observed upregulation in DGAT2 expression in the Lep ^ob /Lep ^ob mice that is not evident in A ^y /a mice. DGAT2 is an isoform of DGAT1 with similar biological activity and tissue distribution [355]. The differential response of Lep ^ob /Lep ^ob and A ^y /a mice suggests that leptin is required for suppression of DGAT2. Since DGAT1 deficiency does not promote hyperphagia and increases energy expenditure while selectively decreasing fat stores, inhibitors of DGAT have potential clinical application.

Acetyl-CoA carboxylase 2. Acetyl-CoA carboxylase (ACC) mediates fatty acid synthesis by converting an end product of glycolysis (acetyl-CoAs) to a molecule that is a precursor for fatty acids (malonyl CoAs). ACC1 is present in the cytosolic fraction of hepatocytes and WAT, while ACC2 is active in mitochondria of nonlipogenic organs including heart and skeletal muscle. It was hypothesized that deletion of ACC2 in mice would block fatty acid synthesis in nonlipogenic organs and favor fatty acid oxidation, which is triggered by an increase in acetyl CoA. Acc2-/- mice have reduced triglyceride and glycogen storage in the liver, decreased plasma glucose, and weigh 10% less than wild-type controls, despite a 20% increase in food intake [356]. Acc2-/- mice are also leaner, as evidenced by a 50% reduction of epididymal fat pad mass, and have 30% lower leptin levels than their wild-type littermates. However, while plasma FFAs are decreased, triglycerides are increased by 30%, presumably due to increased mobilization of fuels from fat and liver, respectively, to FFA-dependent heart and muscle tissue. These findings suggest that prolonged upregulation of fatty acid oxidation induced by ACC2 deficiency in nonlipogenic tissues may promote hyperphagia by suppressing circulating leptin concentrations [356], or by increasing ARC NPY expression associated with decreased malonyl-CoA accumulation [357]. Upregulated fatty acid oxidation may furthermore increase plasma triglycerides transport from lipogenic tissues to the heart and muscle.

Hormone-sensitive lipase. HSL hydrolyzes triacylglycerols to FFA and glycerol in WAT for transport into the circulation, and also hydrolyzes neutral cholesterol esters. Insulin released after a meal downregulates hormone-sensitive lipase (HSL) and reduces lipolysis. Remarkably, no significant change in WAT mass is observed in the Hsl -/- mouse, although heterogenous adipocyte size suggests that a subset of adipocytes may be sensitive to lipid accumulation. In addition, absence of HSL activity has no apparent impact on cold-induced thermogenesis. However, Hsl -/- mice show a 1.65-fold increase in BAT mass resulting from an increase in BAT triglycerides, even though triacylglycerol lipase activity in BAT is not significantly reduced by HSL deficiency. Moreover, plasma FFA and glycerol are reduced only 44% and 62% respectively in Hsl-/- mice as compared to wild-type controls, suggesting that the residual 40% non-HSL triacylglycerol lipase activity present in the WAT of mutant mice probably contributes to the relatively normal phenotype observed. Cultured embryonic fibroblasts from HSL-deficient mice differentiate normally into adipocytes and hydrolyze triglycerides in response to isoproterenol and forskolin, but exhibit no cholesterol ester hydrolase activity as compared to wild-type controls [358]. These in vitro observations further support the existence of an unidentified non-HSL triacylglycerol lipase acting via a pathway similar to HSL [358].

Perilipin. Perilipin has 65% homology with adipocyte-differentiation-related protein (ADRP) at the amino acid level, and is found at high levels in adipose tissue. Cyclic AMP in adipose tissue upregulates lipolytic activity via protein kinase A (PKA)-mediated phosphorylation of lipolytic enzymes. Perilipin isoforms A and B each contain consensus A-kinase sites, and can be phosphorylated at six unique serine residues by PKA. Since hormone-driven phosphorylation of lipolytic enzymes in adipose tissue regulates lipid trafficking and packaging, phosphorylated perilipin may have a parallel role in these processes. In support of this hypothesis, in vitro studies reveal that lipolytic activation of adipocytes phosphorylates perilipin protein which in turn adheres to the surface of lipid storage droplets where it facilitates the hydrolysis of triacylglycerols by hormone-sensitive lipase (HSL) [359]. In its dephosphorylated basal state, perilipin prevents HSL activity.

Three isoforms of perilipin, perilipin A, B and C have similar function in different tissues: perilipin A is found in both adipose tissue and steroidogenic cells, while perilipin B is found only in adipocytes and perilipin C, only in steroidogenic cells. Perilipin-deficient ( Plin-/- ) mice were generated by targeted disruption of the gene encoding these three splice products [360]. Plin-/- mice on a C57BL/6 background resemble their wild-type littermates in overall size and body weight but are hyperphagic and have 21% more total lean tissue than control mice, and 11% increased total oxygen consumption as measured by indirect calorimetry. Plin-/- mice are resistant to obesity induced by a high-fat diet (55% kcal from fat) and have a 75% reduction in epididymal fat pads [361]. Lipolysis is increased, as measured by increases in basal HSL activity and beta-3-adrenergic-stimulated glycerol release to 287% and 375%, respectively, over that of control mice. This results in a decrease in the size of individual adipocytes that has a significant effect to reduce fat stores [360]. In addition, perilipin deficiency, like SREBP-1c deficiency, reduces adipose tissue mass of Lepr ^db /Lepr ^db mice to that of their wild-type littermates by increasing metabolic rate, although other parameters such as food intake were not evaluated.

Obesity Due Primarily to a Defect in Partitioning: Musle Tissue

Myostatin. Myostatin is a member of the transforming growth factor b superfamily and is a negative regulator of skeletal muscle mass. Myostatin-deficient cattle have increased muscle mass, and two such breeds of cattle, Belgian Blue and Piedmontese, have known inactivation of mutations in the myostatin gene [362]. The Belgian Blue breed has an 11-bp deletion in the myostatin sequence, and the Piedmontese myostatin gene has a missense mutation in exon 3. A knockout mouse model of myostatin deficiency, Mstn-/- , exhibits a similar phenotype, with an increase in skeletal muscle mass [363], and a 70% reduction in total adipose mass relative to Mstn+/+ controls that is maintained throughout the lifespan [364]. Mstn+/- mice exhibit effects on skeletal muscle that are intermediate to wild-type and homozygous mice. Mstn-/- mice consume the same amount of energy per gram of body weight as do Mstn+/+ mice, but have slightly lower energy expenditure when total and resting oxygen consumption are expressed as a function of body weight (mL VO ₂ /kg/hour). BAT UCP1, UCP2 (liver, kidney, spleen, WAT, BAT, triceps and gastrocnemius) and UCP3 (BAT, triceps, gastrocnemius) mRNA expression and cold tolerance are similar to wild-type mice. Myostatin may act directly on adipocytes to reduce fat stores, since myostatin is expressed in WAT, and in vitro studies suggest that myostatin may inhibit adipocyte differentiation [365]. Another possibility is that myostatin deficiency in muscle may prevent fat accumulation indirectly through a hypothetical second messenger released from muscle tissue, or via a CNS pathway that regulates the size of adipose stores. Crossing Mstn-/- mice to A ^y/a or Lep ^ob /Lep ^ob mice attenuates obesity and glucose intolerance by selectively decreasing individual fat pad weights and increasing muscle mass as compared to A ^y /a;Mstn+/+ and ob/ob;Mstn+/+ mice.

CONCLUSIONS

The mice and rats described in this essay provide proof that body weight and composition are regulated by specific genes that participate in complex neural and metabolic pathways that determine the performance of only three major categories of control: energy intake, expenditure and partitioning of stored calories. The identification of molecules responsible for the single gene obesities in these animals has expedited the discovery of many other molecules and pathways that constitute the still only incompletely understood mechanisms for energy homeostasis that interact with developmental and environmental processes to determine body mass and composition. Their existence provides definitive refutation of vitalist/psychological notions that have permeated the field of energy intake and metabolism, and provides the heuristic, reductionist framework in which ongoing research on these questions should be conducted. It is likely that major genes and their modifiers, as well as allelic variants of a larger number of genes with lesser individual impact, will eventually account for the critical phenotypes in rodents and humans. As this chapter demonstrates, mice and rats provide a powerful resource for the discovery and study of the constituent molecules, and for hypothesis generation regarding the same processes in humans. The ability to refine the characterization of the behavioral and metabolic phenotypes that are controlled by these genes --in rodents and humans-- will add greatly to the power of genetics to reduce the complex continuous phenotypes that are the physiologic "stuff" of energy homeostasis to their constituent molecular events.