Overview

In mammals, body weight is normally regulated around a set point by coordinated changes in food intake and energy expenditure. These changes are integrated under the influence of specific neural pathways and circulating signals. Almost 50 years ago it was first proposed that circulating signals generated in proportion to body fat stores influence appetite and energy expenditure in a coordinated manner to regulate body weight [1]. Of particular historical value are the classic parabiosis experiments which provided evidence that such circulating signals exist [2].

The positional cloning of the mouse ob gene and identification of its protein product, leptin, suggested that the long sought blood-borne lipostatic factor had been found [3]. Leptin appeared to fulfill the predictions of the lipostatic hypothesis in that it is mainly produced by adipocytes, circulates in proportion to total adipose tissue mass, and interacts specifically with a hypothalamic receptor to reduce food intake and promote weight loss [4, 5]. Paradoxically, almost one decade after the discovery of leptin it becomes clear that leptin alone does not explain all the outcomes of parabiotic studies. The evidence for the existence of additional adipostatic factors has already been reviewed [6, 7, 8]. With obesity as an increasingly important public health focus, a major development in the understanding of energy balance regulation has come with observations made in quite different biological spheres such as whole-body physiology and application of transgenic technology. Given that body weight regulation and food intake represent physiological processes that underpin both cell life and cell death, the presence of back-up mechanisms is not surprising. As with many biological systems, controversies and exceptions are not uncommon, especially to nascent pathways. Even in the age of molecular biotechnology there is a prominent role for physiological experimentation and reasoning in the discovery of important new regulatory and effector molecules.

Great progress has been made in identifying several genes in spontaneous monogenic animal models of obesity as well as in understanding the molecular mechanisms underlying phenotypes of altered body weight, adiposity and fat distribution by creating transgenic animal models (see reviews [9, 10, 11]). Targeted expression and knockout of specific genes has been extremely helpful in establishing the physiological roles of certain genes in the control of energy metabolism in vivo. Transgenic approaches, however, also have limitations [9]. Overexpression or knockout of concrete genes, and the subsequent alterations in the expression of their encoded proteins at different steps of the regulatory pathway of adipogenesis, show the complexity and complementarity of genes involved in energy homeostasis. In this sense, the failure to produce an expected phenotype through transgenesis further reflects the existence of adaptive mechanisms to preserve crucial physiological functions.

The analysis of a number of genetically obese mouse strains has clearly contributed to our understanding of body weight control. In particular, those strains that fail to synthesize either leptin or its functional receptor opened up the field to unravel the distinct metabolic abnormalities related to the development of an obese phenotype [3, 12, 13, 14]. Of interest, the outcome of spontaneous mutations, transgenesis or knockout of specific genes can be subdivided in animal models leading to obesity, mouse strains characterised by leanness and manipulations conferring resistance to obesity (Tables 1, 2, 3). Mutations affecting not only neurotransmitters, neuropeptides and their receptors but also transcription factors, signal transducers, hormones, cytokines, adhesion molecules as well as enzymes and transporters involved in glucose and lipid metabolism play a key role in the development of an obese phenotype.

Table 1. Rodent genetic models leading to obesity
Table 2. Rodent genetic models characterised by a lean phenotype
Table 3. Rodent genetic models conferring resistance to obesity

Inputs to the CNS

Melanocortins

A number of hypothalamic neurotransmitters and neuropeptides—including catecholamines, serotonin, leptin, AGRP, NPY, melanocyte-stimulating hormone (α-MSH), melanin-concentrating hormone, pro-opiomelanocortin (POMC), CART, CRH, and orexins, among others—have been implicated in the regulation of food intake and energy expenditure [5, 120, 121]. Leptin-responsive neurons in the arcuate nucleus include those expressing NPY-AGRP and POMC-CART. Both CART and α-MSH, a cleavage product of POMC that stimulates melanocortin-3 and melanocortin-4 receptors (MC3-R and MC4-R), affect neurons of the paraventricular nucleus to decrease food consumption. Inactivation of MC3-R results in increased fat mass and reduced lean body mass despite hypophagia and normal metabolic rates [39, 40], whereas MC4-R −/− animals are characterised by obesity, mild hyperphagia and reduced energy expenditure [41]. Of interest, the double MC3-R −/− MC4-R −/− mice have a more obese phenotype than MC4-R knockouts [39]. POMC −/− rodents show obesity together with defective pigmentation and adrenal development [49]. Analogously, severe early-onset obesity, red hair pigmentation and adrenal insufficiency are observed in humans with POMC mutations [10].

It has been reported that obesity and yellow coat are two independent phenotypic features. Mice with the e/e genotype, which prevents eumelanin synthesis, are yellow but non-obese, whereas A vy mice with a dominant sombre (e so) gene, which prevents pheomelanin synthesis, are black yet obese [10, 11]. Moreover, inactivation of the MC1-R yields yellow but not obese rodents, whereas knockout of the MC4-R gene results in obese rodents without a yellow coat [51]. These observations suggest that MC4-R, highly expressed in the hypothalamus, is a key element in energy balance regulation and represents the link between the agouti mutation and the development of obesity.

A further piece to the puzzle of the complex relation between obesity and the agouti mutation has been provided by two autosomal recessive mutations, mahogany (mg) and mahoganoid (md). Both mutations are able to revert agouti-induced obesity and restore melanogenesis. Except hyperphagia, homozygous mutations of the mg gene prevent yellow fur coat colour change and obesity in A y mice by restoring normal body weight, adiposity, hyperglycaemia, hyperinsulinaemia, hyperleptinaemia and linear growth [111]. Since mg is not able to reverse the phenotype of the MC1-R or MC4-R mutations, mg action involves molecular events downstream of the A y mutation, but upstream of the melanocortin receptors [10].

Several neurotransmitters play a key role in feeding behaviour [5, 121] as shown for increases in brain serotonin, which decreases food intake. Mice lacking the 5HT2c receptor subtype have a normal body weight at younger ages. However, late-onset obesity with an increase in WAT mass takes place due to hyperphagia [10, 16]. Fenfluramine, a once widely prescribed appetite supressant, stimulates the release of serotonin in the arcuate nucleus of the hypothalamus binding to 5HT2c receptors expressed in POMC neurons. In turn they stimulate the release of α-MSH that subsequently acts on effector neurons expressing MC3-R and MC4-R, which have been identified as critical regulators of food intake and energy homeostasis [122]. Use of gene knockout technology has shown that dopamine is required for hyperphagia in ob/ob mice and that dopamine-deficient mice have a smaller size 2 weeks postnatally with a negative growth rate leading to death at around 1 month of age in relation to decreased eating and drinking behaviour [104].

Neuropeptide Y

NPY is a widely expressed neuropeptide with biological functions including food intake stimulation and control of autonomic and endocrine actions aimed at sparing energy and regulating the cardiovascular, cortical and sympathetic nervous systems [10]. Lack of NPY results in normal weight animals with normal appetite under physiological circumstances, but with a hyperphagic response to fasting. Mice remained leptin-responsive even with hypersensitivity to leptin-mediated inhibition of feeding. A role for NPY in leptin action was shown by the generation of a double knockout lacking both NPY and leptin [112]. Compared to ob/ob mice the simultaneous deficiency of NPY resulted in an intermediary obese phenotype showing reduced adiposity, and food intake together with improved fertility, increased oxygen consumption, locomotor activity and body temperature [112]. Thus, both anabolic NPY-containing and catabolic POMC-containing neurons are direct targets of leptin action. The orexigenic effects of NPY have been reportedly assigned to binding to Y1 and Y5 receptors [9]. However, Y1, Y2, Y4 and Y5 receptor knockouts show obesity [9, 58, 59, 60, 61, 82]. Y1R −/− mice show a markedly reduced feeding response to fasting and slightly decreased daily food intake and NPY-stimulated feeding while Y5R −/− rodents have a normal feeding response to fasting and leptin but markedly reduced to NPY stimulation. They develop late-onset obesity associated with hyperphagia in the case of Y5R-deficient animals, or lowered metabolic rate associated to hypoactivity or changes in UCP gene expression in the case of Y1R mutants [9, 61]. Mice lacking Y2R also develop mild obesity due to hyperphagia, therefore normal feeding behaviour, body weight and leptin response require Y2R [59].

Others

A number of factors such as nescient helix-loop-helix 2 (Nhlh2), single-minded 1 (SIM1), steroidogenic factor 1 (SF-1) and nerve growth factor-induced protein (VGF) involved in CNS development have also been implicated in body weight regulation [10]. Homozygotes for Nhlh2 suffer from hypogonadism associated to a reduced fertility [44]. Approximately 10 to 15 weeks after birth extreme obesity becomes evident. This characteristic phenotype shows that the Nhlh gene is necessary for the adequate development of the hypothalamic-pituitary axis and that part of the obesity seems to be independent of the hypogonadism being directly linked to a defect in hypothalamic development [10]. Mice in which SIM1 is deleted lack CRH, oxytocin, vasopressin, somatostatin and thyrotropin-releasing hormone and die shortly after birth. In humans haploinsufficiency of the SIM1 gene has resulted in early onset obesity with increased adiposity and hyperphagia despite apparently normal metabolic and hormonal variables [52]. SF-1 knockout mice suffer from infertility and inadequate development of the hypothalamic ventromedial nucleus and the adrenal glands [51]. Mice lacking the expression of Vgf are of a small size a few days after birth due to a decrease in body fat, which is not attributable to decreased food intake [101]. Characteristic features of Vgf-null mice are hyperactivity, augmented energy expenditure and infertility with low plasma concentrations of glucose, insulin and leptin.

Transcriptional regulation

Numerous transcription factors have been reported to regulate adipocyte proliferation and differentiation [10, 11] (Fig. 1).

Fig. 1.
figure 1

Transcriptional regulation of proliferation and differentiation of adipocytes

CCAAT/enhancer-binding protein

Knockouts of the CCAAT/enhancer-binding protein (C/EBP) α, β, and δ result in partial or complete lack of adipocytic lipid accumulation [72, 73]. C/EBPα is required for the differentiation of preadipocytes to mature adipocytes in most WAT depots. However, it is not indispensable for BAT and mammary gland WAT development. Mice lacking C/EBPβ and/or δ show a severe phenotype with decreased fat pad weights due to a lower number of adipocytes tributary to the incompetence of embryonic fibroblasts to differentiate into adipocytes. Furthermore, fibroblasts derived from C/EBPβ and/or δ null mice do not express C/EBPα and PPARγ [11]. These animal models show the ocurrence of cross-talk between C/EBPα and PPARγ and suggest the existence of alternative pathways in vivo to compensate for the lack of C/EBPβ and/or δ in null mice. The role of C/EBP in adipocyte development is further supported by A-ZIP/F transgenic mice. Overexpression of the A-ZIP/F gene whose product inactivates, among others, the C/EBP family, renders a phenotype that is virtually fat-depleted, thus showing very low energy reserves with no WAT and a reduced amount of BAT [69].

Peroxisome proliferator-activated receptor

Many lipid factors such as ceramides, gangliosides, sphingolipids, prostaglandins, eicosanoids, and sterols are known activators or ligands of the PPARs. The study of their molecular and physiological characteristics has shown their involvement in many physiological functions including energy balance and adipocyte differentiation. Thus the lack of PPARα disrupts fatty acid oxidation in the liver leading to an increased WAT deposition [50], whereas the ubiquitously expressed PPARβ, also termed δ, seems to play a role in adipocyte differentiation with null mice showing a normal development except for a smaller size [11, 123]. The reduction in fat-pad mass observed in PPARβ −/− mice is due to a loss of its expression in cell types different to adipocytes as the selective depletion of PPARβ in WAT did not yield different phenotypes between transgenic and wild-type rodents [11]. The homozygous deletion of PPARγ is not compatible with life as the protein is essential for placental and cardiac development, in addition to adipocyte differentiation [10, 123]. PPARγ +/−, however, are resistant to adipocyte hypertrophy and insulin resistance associated to a high-fat diet supporting a direct modulatory effect of PPARγ in adipocyte development in response to dietary stimulation [115].

PPARs bind as heterodimers with retinoid X receptor (RXR). Upon binding an agonist, the PPAR conformation is changed and stabilised creating a binding cleft that enables recruitment of transcriptional factors leading to an increase in gene transcription. RXRα is expressed at high levels in WAT and its deletion is accompanied by in utero mortality [9]. Specific adipose tissue ablation of RXRα yielded mice that do not develop obesity under a high-fat diet and have an impaired increase in plasma NEFA during fasting. These findings provide evidence that RXRα plays a role in both fat accumulation and mobilization [10].

Markedly reduced subcutaneous and visceral WAT with simultaneous enlargement of the interescapular fat depot is observed in overexpression of SREBP-1c. These mice show a normal size at 6 weeks of age while shortly after birth they are smaller than wild type littermates [9]. Accompanying features of these rodents are insulin resistance, hypertriglyceridaemia, lipid-laden internal organs, difficulty in thriving and early death [67]. SREBP-1c has been shown to stimulate both lipogenesis and adipogenesis in cellular models. Absence of SREBP-1 ameliorates fatty livers, but not obesity or insulin resistance in ob/ob mice [110].

Tubby-like proteins

The protein encoded by the tubby gene is a member of the tubby-like proteins gene family [10]. Structure-based functional analysis has showed the participation of these proteins as transcription factors and in G protein signalling. The absence of the protein encoded by the tubby gene results in an increased body weight accompanied by normoglycaemia, hyperinsulinaemia and hyperleptinaemia together with retinal and cochlear degeneration [10, 25]. Homozygous mice for the mutation are fertile with the possibility of producing litter before the onset of obesity, which happens around 3 to 4 months in male mice and 4 to 6 months of age in female mice [10].

Other factors promoting adipocyte differentiation

Consistent with a role in development, high mobility group protein I-c (HmgI-c) is only expressed during embryonic and fetal stages. Ubiquitous expression of a truncated Hmgic leads to an increased development of WAT early in life while disruption of one or both alleles of the gene confers resistance to obesity [81]. In addition, lack of Hmgic expression in leptin-deficient mice results in decreased fat pads due to hypoplasia with no evident effect on adipocyte gene expression. Thus, Hmgic seems to play a role in adipocyte precursor proliferation and commitment [11, 81].

Forkhead box C2 (Foxc2) is a winged helix gene expressed in differentiated adipocytes that counteracts obesity, hypertriglyceridaemia, and diet-induced insulin resistance [77]. The relevance of this transcription factor becomes evident as mice lacking Foxc2 die during the embryonic or perinatal period. Transgenic mice with selective overexpression in adipose tissue have a lean phenotype with a decreased intraabdominal WAT depot [11]. Adipocytes showed features of brown fat cells with a diminished size, multilocular lipid droplets, and increased number of mitochondria resulting in an augmented thermogenic capacity. Changes in expression of genes with a role in energy dissipation such as PGC-1 and UCP1 has been observed. Moreover, C/EBPα, PPARγ and SREBP-1c, genes closely related to adipogenesis, were also increased [77]. Thus, Foxc2 seems to be involved in the maintenance of the adipocyte phenotype in addition to protecting against overweight and diet-induced insulin resistance.

Information concerning the role of the cell cycle regulator E2F family members has shown that E2F1–3 play a role in the induction of cellular proliferation while E2F4 and 5 appear to be particularly relevant for the transition from cell proliferation to differentiation [106]. The E2F family has been reported to induce PPARγ transcription during clonal expansion (through E2F1) while repressing PPARγ expression during terminal adipocyte differentiation (through E2F4) [106]. Thus, E2Fs play a key role in coordinating the transition between cell proliferation and terminal differentiation because the different proteins of the family have opposing effects on the regulation of PPARγ expression.

Endocrine control

Insulin signalling

Transgenic and knockout mice have helped to untangle the complex participation of insulin signalling in body weight control. Targeted disruption of the insulin receptor (IR) gene leads to diabetic ketoacidosis and profound postnatal growth retardation followed by neonatal lethality probably in relation to the pleiotropic effects of IR activation [124]. IR null mice have a markedly decreased amount of fat characterised by a reduction of fat cell volume rather than due to a diminished fat cell number [11]. Unlike in mice, patients with leprechaunism, in which the IR is mutant or missing, suffer only from relatively mild hyperglycaemia [124]. IRS-1 −/− mice do not show diabetes although they develop beta-cell hyperplasia together with mild insulin resistance attributable mainly to skeletal muscle. In contrast, IRS-2 −/− mice show an increased food intake leading to obesity and develop overt early-onset diabetes due to frank hepatic insulin resistance and absent pancreatic compensation [36, 37, 124]. Simultaneous disruption of IRS-1 and IRS-2 by intercrossing renders mice with cells that are completely unable to differentiate into adipocytes [11].

The lethal phenotype of IR inactivation led researchers to focus on tissue-specific elimination of IR in peripheral organs. Muscle-specific insulin receptor knockout (MIRKO) mice do not show apparent disturbances in glucose homeostasis as evidenced by normal glucose and insulin concentrations [42]. However, the mutant mice have increased serum triglycerides and NEFA together with increased visceral fat mass. Although MIRKO mice have a decreased insulin-stimulated muscle glucose uptake in vitro and during clamp studies, they achieve a near normal skeletal muscle glucose uptake during a glucose tolerance test [42, 124]. However, an impaired insulin activation of muscular glycogen synthase that leads to a diminished glycogen content in the muscles of MIRKO mice has been observed. These observations suggest that although insulin is necessary for muscle glucose storage in the form of glycogen, muscle IR signalling is not needed to maintain post-prandial glucose disposal [124]. Selective pancreatic beta-cell insulin receptor knockout (βIRKO) mice results in diabetic and obese mice with decreased pancreatic islet size and insulin content as well as loss of glucose-induced first-phase insulin release and age-dependent glucose intolerance [125]. Surprisingly, βIRKO-MIRKO mice have a phenotype in which only a mild glucose intolerance is apparent [126]. Mice with tissue-specific disruption of the IR gene in neurons (NIRKO) have an increased food intake leading to diet-induced obesity and insulin resistance [45]. This animal model further supports a role for insulin in the CNS control of body weight by providing a negative feedback loop for post-prandial inhibition of food consumption. NIRKO mice also show an impaired reproduction attributable to dysregulation of pituitary luteinizing hormone secretion, leading to altered ovarian follicle maturation in female mice and perturbed spermatogenesis in male mice [45, 124]. These findings provide additional evidence of the connection between central insulin action, energy allocation and reproductive performance. Liver-specific insulin receptor knockout (LIRKO) mice show fasting hyperglycaemia and severe insulin resistance probably secondary to increased hepatic glucose production rather than to a decreased muscle glucose uptake [127]. The physiological role of insulin signalling in WAT and BAT has been investigated by creating a fat-specific insulin receptor knockout (FIRKO) animal model [76]. Adipocyte-specific inactivation of the IR gene resulted in almost complete protection against age-induced and hyperphagia-induced obesity and the impaired glucose tolerance associated to these conditions [76]. FIRKO mice have a lower body weight due to a reduced fat mass with a disruption of the plasma leptin to adipose mass relation together with a normal insulin sensitivity and glucose homeostasis. Abrogation of the IR in BAT only (BATIRKO) results in age-dependent impaired glucose tolerance without insulin resistance [128]. These observations suggest that the knockout of the IR in WAT overcomes the detrimental effects on glucose metabolism following lack of insulin signalling in BAT [127, 128]. Tissue-specific disruption of the IR gene has provided a powerful approach to dissect the direct and indirect effects of the complex insulin signalling pathways showing that insulin is an essential regulator of intermediary metabolism producing a broad spectrum of actions in almost all organs.

Fetuin inhibits insulin-induced IR autophosphorylation and tyrosine kinase activity. Consistent with the participation of the fetuin gene (AHSG) in the susceptibility to develop Type 2 diabetes and the metabolic syndrome, fetuin knockout mice show an improved insulin sensitivity and resistance to weight gain with a decreased body fat when challenged with a high-fat diet [129]. These findings suggest that fetuin plays a role in postprandial glucose disposal, insulin sensitivity, weight gain and fat accumulation.

Insulin-like growth factor

The IGF system, encompassing ligands (IGF-I, IGF-II), receptors (IGF-1R, IGF-2R) and binding proteins (IGFBP) plays a key role during both intrauterine and postnatal development [11]. Mice lacking the IGF genes weigh approximately 60% less than wild type littermates due to intrauterine growth retardation. Although IGF1-R null pups suffer from reduced weight associated to postnatal birth, the IGF2-R −/− phenotype shows a fetal overgrowth syndrome leading to death.

Carboxypeptidase E

The fat mutation was originally identified in obese mice and mapped to chromosome 8 [25]. Based on chromosomal localization and on the observation that mice with this recessive mutation have an increase in pancreatic proinsulin lead to the identification of a point mutation in carboxypeptidase E (CPE) being responsible for the phenotype of fat mice [26]. Mutant animals are fertile before the onset of obesity, which becomes evident at about 8 weeks of age [10].

Growth hormone

A rodent genetic model of GH deficiency, the little mouse, shows the participation of GH in somatic growth and body composition [38]. These mice show at the same time a dramatic decrease in body weight and skeletal muscle development as well as a marked increase in adiposity. The changes in body composition take place in the context of an unchanged food intake relative to body weight [10]. The hypothalamic GHS-R regulates GH secretion, feeding, and adiposity [78]. The partial knockout of GHS-R renders rodents with lower body weight and less adipose tissue together with a reduced daily food intake. On the contrary, GH overexpression in rats results in animals with increased amounts of abdominal fat [11]. The unexpected outcome of increased percentage of body fat in the presence of increased GH concentrations show the potential opposite effects of short-term and long-term stimulation of the hormone on adipocytes. Furthermore, the continuous secretion of GH by transgenic mice has inhibited the physiological pulsatility and secretory pattern of the hormone resulting in decreased overall mean plasma GH concentrations [130]. The transgenic rodents developed severe obesity accompanied by increased blood concentrations of glucose, insulin, triglycerides and NEFA. Despite an evident hyperleptinaemia a defective leptin transport from peripheral blood to the cerebrospinal fluid explains that the mutants show an increased food intake [11, 130].

Glucocorticoids

Energy balance and appetite are under the control of the hypothalamic-pituitary-axis, which regulates, among others, glucocorticoid production. To dissect the central and peripheral effects of glucocorticoid receptors (GR) the selective inactivation of the GR gene in the nervous system has been studied [80]. A marked growth retardation already from the suckling period was observed with specific reduction in lean body mass accompanied by an increased body-fat content until weaning, followed by a reduced fat accumulation after weaning. These mutant mice are smaller and leaner than their littermates due to decreased food intake and diminished metabolic efficiency. The increased plasma glucocorticoids and hypothalamic CRH showed by the selective CNS knockout of the GR gene could function as catabolic signals probably leading to a reduced energy accumulation [80]. The relevance of glucocorticoids in the periphery has been addressed by means of other transgenic animal models. It has been shown that 11β HSD-1 plays a key role in glucocorticoid reactivation in visceral adipose tissue in obesity [131, 132]. Evidence for this action is further provided by transgenic mice overexpressing 11β HSD-1 in adipose tissue, which have an increased adipose level of corticosterone, hyperlipidaemia, insulin resistance, hyperphagia despite hyperleptinaemia and visceral obesity [15]. These findings strongly support a key role for local production of active glucocorticoids in the development of visceral obesity.

Estrogens

Two complementary gene knockout models have addressed the contribution of estrogens in body weight regulation [20, 62]. Development of ER-α knockout (αERKO) mice showed that ER-α absence results in marked increases in WAT, but not BAT, insulin resistance and impaired glucose tolerance in both sexes with an added effect of reduced energy expenditure in male mice [62]. Estrogen deficiency can be further attained by targeted disruption of the aromatase gene since this enzyme catalyses the final step in the biosynthesis of C18 estrogens. Aromatase-knockout mice progressively accumulate increased intra-abdominal adipose tissue [20]. The increased adiposity was not due to hyperphagia or reduced resting energy expenditure, but was associated with diminished spontaneous physical activity, decreased glucose oxidation and reduced lean body mass.

Prolactin

Indirect roles for lactogenic hormones in adipose tissue growth and metabolism has received less attention considering the extremely low expression of prolactin receptors in WAT under physiological circumstances other than pregnancy and lactation. Transgenic mice overexpressing the prolactin receptor show a reduced retroperitoneal fat pad, while no changes in parametrial depots were observed [133]. Prolactin-receptor null mice show a mild reduction in body weight mainly in females and a marked decrease in abdominal fat mass [134].

Gastrointestinal factors

Endocrine cells scattered throughout the epithelium of the stomach and small intestine secrete numerous peptides that can be released to act locally in the periphery in a paracrine manner, secreted into the bloodstream to act in an endocrine fashion, or released in the brain to act centrally. Cholecystokinin (CCK), glucagon-like peptide-1 (GLP-1), and glucose-dependent insulinotropic peptide (GIP), among others, all known to exert an effect on energy homeostasis at a central and peripheral level have been implicated as putative satiety signals [135]. More recently the involvement of both leptin and ghrelin as gut-derived peptides is being studied [136, 137, 138].

Cholecystokinin

The application of transgenesis to the study of CCK action has yielded the CCK gene knockout as well as the models comprising the deletion of either the CCK-A or CCK-B receptor [10]. Despite the well-established effect on satiety of CCK all three knockout models show normal mature weights further supporting the existence of a highly redundant system warranting energy homeostasis pathways. Transgenic overexpression of pancreatic polypeptide is accompanied by a 50% postnatal lethality due to decreased milk intake with the surviving mice showing a reduced weight gain and fat mass attributable to a decreased food intake together with a slower gastric emptying rate of solid meals [93].

Bombesin

Mice lacking bombesin receptor subtype-3 (BRS-3 −/−) develop an obesity syndrome, with a 50% increase in adipose mass as well as concomitant fasting hyperglycaemia and hyperinsulinaemia [9, 23]. BRS-3 −/− further showed a diminished energy expenditure together with a decreased ability to maintain core body temperature upon cold exposure.

Glucose-dependent insulinotropic peptide

Genes that promote the efficient storage of ingested energy in the form of body fat for use during periods of food scarcity are a common feature in evolution-based selection. In this sense, GIP has been recently shown to directly link overnutrition to obesity [107, 139]. Wild-type mice fed a high-fat diet show hypersecretion of GIP, which increases nutrient uptake and triglyceride accumulation in adipocytes leading to extreme visceral and subcutaneous adiposity accompanied by insulin resistance, whereas mice lacking the receptor (Gipr −/−) showed a protection against all the mentioned detrimental effects. This protection is attained with a normal oxygen consumption and a reduced respiratory quotient indicating that in Gipr −/− mice fat is used as the preferred energy substrate and, therefore, cannot be effectively accumulated in adipocytes [107]. Crossbreeding of Gipr −/− with ob/ob mice generated a double homozygous model (Gipr −/−, Lep ob /Lep ob) with a reduced body weight and fat content compared to ob/ob mice, thus confirming that GIP signalling through its receptor takes place in the absence of leptin [107].

Glucagon-like peptide-1

The hormonal factors participating in transmission of signals from the gut to pancreatic beta cells were originally referred to as incretins. GIP and GLP-1 were initially identified as candidate incretins and were thought to be crucial in glucose homeostasis by immediately promoting insulin secretion in response to meal ingestion. Analogously to Gipr −/− mice rodents lacking the GLP-1 receptor (GLP-1R −/−) show glucose intolerance after glucose loading [108]. While high-fat feeding does not alter glucose tolerance in GLP-1R −/− mice, glucose homeostasis in GIP receptor-deficient rodents is altered [140]. Furthermore, GLP-1R −/− mice show a resistance to diet-induced obesity in females, but not in males [108]. These findings suggest that although GIP and GLP-1 act as incretins and belong to the same gastrointestinal family, their physiological roles are not superimposable.

Ghrelin

Ghrelin was first characterised as the natural ligand of the growth hormone (GH) secretagogue receptor, which is expressed by the somatotrophs of the anterior pituitary gland and by neurons in the arcuate nucleus in the basal hypothalamus [137]. The greatest amount of ghrelin-immunoreactivity has been found in neuroendocrine cells of the gastric fundus. In accordance with this localisation ghrelin has been shown to participate in the complex entero-hypothalamic control of food intake signalling. Central or peripheral administration of ghrelin to rodents has been reported to increase food intake and body weight in addition to stimulating gastric motility and acid secretion [138]. The orexigenic effect of ghrelin seems to be mediated partly through activation of NPY/AGRP neurons in the arcuate nucleus followed by the increased expression of NPY and AGRP expression in the hypothalamus [138].

Signal transduction

Obvious targets for unravelling the potential factors participating in body weight regulation include the study of genes involved in signal transduction control. Adenylate cyclase represents one of the principal elements in transmembrane signalling with regard to energy balance since adenylate cyclase (AC) activity determines the levels of intracellular cAMP, which in turn are crucial for lipolysis stimulation. Known modulators of the activity of AC are adrenergic receptors (AR), GTP-binding proteins, protein kinase A and phosphatidylinositol 3-kinase, among others.

Adrenergic signalling

Of the nine different AR identified to date four, α2 and β1–3, are known to be expressed in adipocytes regulating WAT and BAT development and metabolism through the use of opposite transduction pathways [11, 141]. In response to a high-fat diet transgenic mice overexpressing the human β1-AR in WAT have a reduced weight gain and adipose tissue accumulation than their control littermates with brown adipocytes scattered throughout subcutaneous WAT [119]. These findings support the participation of β1-AR in lipolysis stimulation as well as in increased energy expenditure through heat production. The β3-AR has been postulated to be the major regulator of adrenergic responses in spite of its lower affinity for endogenous catecholamines [141]. Administration of β3-AR agonists is followed by an increase in metabolic rate, a decrease in food intake with reduction of fat stores in obese rodents [10]. Surprisingly, mice lacking β3-AR have normal body weight and food intake compared to wild type animals though an increase in total body fat can be observed [63]. Moreover, when challenged with a high-fat diet β3-AR knockout mice show only a modest tendency towards obesity. These findings suggest that compensatory mechanisms take place in β 3 -AR −/− mice to maintain energy homeostasis as evidenced by an upregulation of β1-AR expression in the WAT and BAT of these mice [10]. The β 3 -AR −/− mice line has been further used to produce transgenic rodents that express the β3-AR exclusively in WAT or BAT or both [10, 11]. The majority of effects attributable to β3-AR activation, such as increased insulin sensitivity and decreased food intake, requires expression of the receptors in WAT, whereas the effect on augmented oxygen consumption depends on β3-AR expression in BAT. Altogether, β3-AR in adipose tissue participates in short-term regulation of energy expenditure while its role in the long-term control can be overtaken by other compensatory mechanisms.

It has been recently reported that β-less mice, i.e. animals lacking all three β-AR, fed on a standard chow diet have a reduced metabolic rate and become slightly obese but when challenged with a high-fat diet they develop massive obesity attributable to a failure in activating diet-induced thermogenesis [142]. These findings provide evidence that β-AR are necessary for diet-induced thermogenesis and that the absence of a single β-AR subtype can be compensated by the remaining AR. In addition, β-less mice are intolerant to cold exposure suggesting that the sympathetic nervous system and the β-AR signalling pathways overlap in heat production regulation in response to cold and diet.

A large body of evidence shows that rodents and humans are not comparable in their adrenergic responses based on different receptor tissue distribution, relative AR subtypes ratios and expression levels [11, 141]. While rodents present high levels of β3-AR and very low expression of α2-AR in WAT, in humans β3-AR mRNA is expressed only in the scarce brown adipocytes with little or no expression in WAT. Thus, although β3-AR agonists have been described as effective anti-obesity drugs in rodents, their potential application for treating human obesity remains questionable. In addition, in humans the β-adrenergic response in WAT can be completely counteracted by the α2-adrenergic pathway [11]. In fact, the α2/β-AR ratio in fat depots has been shown to determine the lipolytic rate and to be closely associated with adipose tissue enlargement in obese people [141]. To establish the relative importance of the different AR distribution in humans and rodents a transgenic model mimicking the human characteristics was obtained by targeted expression of human α2-AR in the adipose tissue of β3-AR knockouts [143]. These "humanised" mice developed high-fat diet-induced obesity associated to adipocyte hyperplasia without signs of insulin resistance. Since transgenic mice expressing α2- and β3-AR did not become obese in response to a high-fat diet it can be concluded that obesity requires the presence of α2-AR and the lack of β3-AR in the context of a dietary challenge such as a high-fat diet [11].

G proteins

Heterotrimeric GTP-binding proteins are key elements in transmembrane signalling due to their ability to be coupled with several different receptors. Expression of regulatory G proteins is essential for both catecholamine and insulin signalling in adipocytes. The stimulatory and inhibitory G protein alpha-subunits (Gsα and Giα) are able to stimulate and inhibit AC activity, respectively. Albright hereditary osteodystrophy, an autosomal dominant syndrome that includes short stature and obesity, is associated with mutations in Gsα [10]. Mice with a homozygous disruption of the ubiquitously expressed Gsα gene are not viable due to embryonic lethality, whereas heterozygotes have different phenotypes depending on the parental inheritance of the allele [144, 145]. Mice inheriting the Gsα-null allele from the mother (m−/+) become obese in the early adulthood whereas those with a Gsα-null allele derived from the father (+/p−) are leaner than the control mice. Differences in body weight are not attributable to changes in food intake. Tissue-and sex-specific imprinting of the Gsα gene has been observed with low expression of Gsα in WAT and BAT of m−/+ mice leading to an increase in both fat depots while +/p− animals show a substantial decrease in epididymal WAT and interscapular BAT as a consequence of reduced lipid accumulation in both tissues. Analogously, the effects on energy homeostasis are also characterised by a parental inheritance; whereas m−/+ mice are hypometabolic and have a decreased locomotor activity, +/p− animals are hypermetabolic and more active. The underlying mechanisms for these phenotypes could depend on a decreased Gsα-induced lipolysis and thermogenesis stimulation in m/+ genitors and an increased total sympathetic activity in +/p− mice [10, 11].

The lack of Giα signalling does not affect body weight but has been reported to participate in insulin action [146]. Because the induction of Giα2-specific antisense RNA in vivo inhibits neonatal growth, a strategy was designed in which the transgene becomes active at birth [11]. These transgenic mice develop Type 2 diabetes with hyperinsulinaemia, impaired glucose tolerance and insulin resistance. Targeted expression of Giα2 in fat and muscle further supported the notion that Gi mimics insulin action unravelling that Giα2 enhances insulin signalling via suppression of protein-tyrosine phosphatase 1B (PTP-1B) [147]. In turn, mice lacking the PTP-1B gene show resistance to obesity and an increased insulin sensitivity even on a high-fat diet [116]. Intercrossing of PTP-1B −/− rodents with leptin-deficient mice also conferred resistance to obesity with an attenuated weight gain and decreased adipose tissue mass in relation with an increased resting metabolic rate [117]. Induction of Gαq-specific antisense RNA resulted in increased body mass and adiposity due to a reduced lipolytic response [148].

Protein kinase A

In mice protein kinase A, a ubiquitously expressed cAMP-dependent kinase, is formed by four different regulatory and two catalytic subunit genes that are expressed in a tissue-specific pattern [10]. The RIIβ subunit is abundant in WAT, BAT and certain brain regions and its targeted disruption yields normal-weight mice with approximately 50% less WAT mass [94]. RIIβ-null mice show a slight hyperphagia with normal plasma concentrations of glucose, insulin and lipids and are resistant to a high-fat diet challenge. The lean phenotype is achieved at the expense of an increased lipolytic activity, metabolic rate and body temperature resulting from a compensatory rise in the RIα subunit in WAT and BAT, which is associated with an increased expression of UCP1 [11]. A direct central effect of protein kinase A RIIβ on energy homeostasis should not be completely discarded. RIα-null mice suffer early in utero lethality due to severe developmental alterations thus demonstrating the participation of the RIα subunit in cAMP-activated cellular responses essential for life [11].

JAK-STAT signalling

Class I cytokine receptors are known to act through Janus kinases (JAK) and signal transducers and activators of transcription (STAT). JAK proteins are associated with membrane-proximal sequences of the receptor intracellular domain, which is phosphorylated upon ligand binding. The phosphorylated intracellular domain then provides a binding site for a STAT protein, which is activated, translocates to the nucleus and stimulates transcription. Many factors, including erythropoietin, GH, leptin, prolactin and interleukins 2 and 3 (IL-2, IL-3) mediate ligand-induced activation of STAT proteins [4]. Targeted disruption of STAT5a and b has been accomplished to elucidate their contribution in energy balance regulation. All Stat5 mutants showed reduced WAT depots at young ages, with Stat5b −/− mice developing adult obesity associated with increased epidydimal and abdominal fat amounts [54]. Stat5b −/− male mice show a decreased body growth rate thus providing evidence for the requirement of STAT5b for sexual dimorphism of body growth rates. Consistent with the role of STAT5 in the effects of GH, erythropoietin, prolactin, and a number of hormones and cytokines, all STAT5 mutants presented further alterations such as growth retardation, disrupted haematopoiesis, as well as mammary gland and ovarian development defects [10].

Adipocyte-derived influences

The adipocyte is no longer considered a passive bystander as fat cells actively secrete a large number of cytokines, adhesion molecules, vasoactive factors, as well as regulators of glucose and lipid metabolism, among other signals, which influence peripheral fuel storage, mobilisation and combustion, as well as energy homeostasis [8, 149].

Leptin

Defective leptin signalling due to either leptin deficiency or dysfunctional leptin receptors leads to early onset severe obesity due to increased adiposity characterised by hyperphagia, decreased energy expenditure, lower body temperature, defective thermogenesis, infertility as a consequence of hypogonadotropic hypogonadism, hyperglycaemia, hyperinsulinaemia, dyslipidaemia, hypothyroidism, and hypercortisolism [3, 4, 5, 6, 7]. Although some of the broad array of metabolic and neuroendocrine alterations such as infertility are normalised in transgenic ob/ob mice expressing about 50% of the physiological leptin concentrations under the control of an adipocyte promoter, the animals continue to show a moderately obese phenotype [150]. Meanwhile, chronic hyperleptinaemia achieved in normal mice by the same transgenic approach yields rodents that accumulate adipose mass at an older, but not younger age [151]. In contrast, transgenic skinny mice overexpressing leptin by an hepatic promoter results in accelerated puberty, high blood pressure, increased glucose metabolism and insulin sensitivity accompanied by complete disappearance of WAT and BAT [97]. The selective deletion of leptin receptors in neurones leads to obesity in a lesser extent than that observed in leptin receptor-null mice and only to slight glucose intolerance in some cases suggesting that the peripheral effects of leptin contribute to the full manifestation of the syndrome [47]. To unravel the concrete contribution of leptin in the periphery a similar approach to that carried out with the IR with targeted disruption of the leptin receptor in adipocytes, liver, skeletal muscle, etc. has to be undertaken.

Interleukin-6

IL-6 is considered to be an inflammatory mediator as well as a stress-induced cytokine with pleiotropic effects on a variety of tissues, including stimulation of acute phase protein synthesis, increase in thermogenesis as well as increased activity of the hypothalamic-pituitary axis and down-regulation of adipocyte lipoprotein lipase (LPL) [8]. Although increased concentrations of IL-6 have been detected in obese subjects, IL-6-deficient mice develop mature-onset obesity accompanied by an increased subcutaneous fat mass with the obese phenotype being only partly reversed by IL-6 replacement [35]. Since the lack of this immune-modulating cytokine confers resistance to the decrease in food intake caused by tumour burden or infection it has been suggested that IL-6 plays a role in the regulation of food intake in pathological states [10].

Interleukin-1

IL-1 has also been proposed as a potential mediator of cancer cachexia based on the transgenic mice obtained either lacking or overexpressing the IL-1 receptor antagonist (IL-1ra), the endogenous inhibitor of IL-1 [83]. Consistent with the participation of IL-1 in body weight regulation, IL-1ra −/− mutants weigh less than wild type control mice. In addition, mice lacking IL-1β converting enzyme, which produces biologically active IL-1β, are resistant to the anorectic effects of lipopolysaccharide [10]. Paradoxically, overexpression of IL-1ra yields mice with similar body weights to those of wild type animals.

Tumour necrosis factor-α

TNF-α inhibits the expression of two master regulators of adipose differentiation, CEBPα and PPARγ2 [8]. This suppression could result in the subsequent down-regulation of many adipocyte specific proteins such as LPL, aP2, FAS, ACC, and GLUT4 among others. Furthermore, mature adipocytes are stimulated to mobilise lipids upon TNF-α exposure, possibly via hormone sensitive lipase (HSL) activation. Among the multiple properties of TNF-α the antiadipogenic effect is especially relevant as adipose conversion of fat cell precursors is potently inhibited by TNF-α [8]. Several genetic models lacking different elements of the TNF-α system, such as TNF-α-deficient mice and animals lacking the expression of either one or both TNF-α receptor subtypes, have been generated [10]. While one study [96] found that younger mice lacking TNF-α expression tended to weigh less than wild type animals with significant changes becoming evident at older ages, another study [152] reported no difference in body weight of TNF-α −/− on either a standard or a high-fat diet. TNF-α-deficient mice showed decreased glucose, insulin and leptin concentrations showing an impaired glucose clearance when challenged with a high-fat diet as evidenced by increased circulating glucose and insulin, which did not reach the concentrations of wild-type control mice [10, 96, 152]. Disruption of TNF-α receptors expression had no effect on body weight or glucose homeostasis when mice were fed on a standard diet [113]. However, on a high-fat diet only mutants lacking the p75 receptor showed a lower body weight together with decreased leptin concentrations compared to wild types. Absence of both receptor subtypes (p55 and p75) resulted in severe hyperinsulinaemia on a high-fat diet, especially after a short fasting period [113]. The lack of TNF-α receptors has been further studied in leptin-deficient ob/ob mice evidencing that while the absence of p75 does not affect insulin resistance, the deficiency of p55 improves insulin sensitivity [153]. Thus, TNF-α could play a role in obesity-related insulin resistance though with the participation of other factors in the development of this syndrome. An unexpected role for TNF-α in thermogenesis control in genetic and dietary models of obesity can be inferred from the findings that in obese mice lacking either TNF-α or its receptors an increase in BAT UCP1 and β3-AR expression are observed in association to a rise in multilocular functional brown adipocytes [154].

Resistin

By screening for genes that were induced during the differentiation of adipocytes but were down-regulated in mature adipocytes exposed to an insulin-sensitizing drug, a new peptide hormone that belongs to a family of tissue-specific resistin-like molecules originally named for its resistance to insulin was identified [155]. Although the seminal proposal suggested resistin to be a hormone that links obesity to diabetes, several subsequent studies support the concept that insulin resistance and obesity are actually associated with a decreased resistin expression [155]. The way resistin was measured and the differences between serum concentrations and mRNA and protein levels probably contribute to the inconsistency. A variety of agents and hormones, including thiazolidinediones, insulin, TNF-α and GH have been shown to participate in the regulation of resistin expression. However, in this case too contradictory results have been reported [155]. The apparent inconsistency of current experimental evidence does not rule out the possibility that resistin plays a role in metabolic disorders. In this sense, the generation of the resistin knockout model could prove helpful in establishing the contribution of resistin in the development of obesity and insulin resistance.

Adiponectin

The cDNA encoding adipocyte complement-related protein of 30 kDa (Acrp30), also termed as adiponectin, was isolated by a substractive hybridization screen comparing 3T3-L1 adipocytes with undifferentiated preadipocytes [156]. The Acrp30 mRNA was induced over 100-fold during differentiation and was found to be expressed exclusively in adipocytes. Contrary to what is observed with most adipocyte-derived factors, which in general are proportional to overall adipose mass, circulating adiponectin concentrations are lower in obese than in lean subjects [156]. Adiponectin has been shown to decrease the postprandial rise of plasma NEFA, to improve postabsorptive insulin-mediated suppression of hepatic glucose output with a strong correlation between plasma adiponectin and systemic insulin sensitivity being evident. Furthermore, adiponectin has been shown to have effects on monocyte adhesion to endothelium, myeloid differentiation and macrophage cytokine production and phagocytosis, processes related to atherosclerotic plaque formation [156]. At present, it is not known if the main role of adiponectin is a direct anti-atherosclerotic effect or if it exerts its activity through the modulation of lipid metabolism and/or regulation of insulin sensitivity.

Nitric oxide

Inflammatory cytokines produce inducible nitric oxide synthase (iNOS) stimulation in skeletal muscle and WAT, which in the long-term has been proposed to cause insulin resistance in muscle. Increased iNOS expression in myocytes and adipocytes of genetic and dietary models of obesity have been described [157]. Moreover, targeted disruption of iNOS has been shown to protect against dietary-induced insulin resistance in muscle. While both wild type and iNOS −/− mice develop obesity associated to hyperphagia when put on a high-fat diet, the latter show an improved glucose tolerance, normal muscle insulin sensitivity and insulin-stimulated muscular glucose uptake despite the increased body weight and adiposity [157]. Obese iNOS −/− mice remained glucose intolerant as evidenced by increased fasting glycaemia in spite of the improvement in muscle insulin sensitivity, which can be explained by the lack of effect of iNOS disruption in preventing defective hepatic and WAT insulin signalling. Endothelial NOS (eNOS), which is expressed in endothelium and skeletal muscle, is also involved in mediating insulin sensitivity with eNOS −/− mice showing fasting hyperinsulinaemia, hyperlipidaemia and defective insulin-stimulated glucose uptake besides hypertension [158].

Other vasoactive factors

Compelling evidence about the role played by adipocytes in cardiovascular physiology has been gathered over the last years. An increasing number of products secreted by adipocytes such as angiotensinogen (AGT), plasminogen activator inhibitor-1 (PAI-1), adhesion molecules, tissue factor and transforming growth factor-β (TGF-β), promote the consideration of adipose tissue as a source of vasoactive factors. Blood vessels express receptors for most of the adipocyte-derived factors, thus drawing attention to the existence of a network of local and systemic signals. AGT expression is increased in obesity and adipocytes have been shown to be able to secrete AGT itself as well as angiotensin converting enzyme [8]. During fasting a decrease in AGT mRNA above control has been observed, which increases upon refeeding. Furthermore, these changes in gene expression are paralleled by fluctuations in AGT secretion from isolated adipocytes [8]. AGT-deficient mice show impairment of diet-induced weight gain with alteration in adipose tissue development and increased locomotor activity [103]. In contrast, overexpression of AGT in wild type mice produces an increase in body weight due to preadipocyte proliferation and differentiation through FAS gene induction [19].

Among the multiple mechanisms that may explain the relationship between obesity and cardiovascular disease, disorders of the fibrinolytic system are very plausible candidates [8]. In fact, increased plasma concentrations of PAI-1 have been found in obese subjects and a close correlation with an abdominal pattern of adipose tissue distribution in both men and women as well as a positive association with other components of the insulin resistance syndrome have been reported [8]. Insulin, TNF-α, IL-1β as well as TGF-β have a stimulatory effect on PAI-1 protein secretion and could contribute to the augmented PAI-1 concentrations observed in obesity and insulin resistance. The transgenic approach has contributed with the observation that disruption of the PAI-1 gene reduces the adiposity and improves the metabolic profile of genetically obese and diabetic ob/ob mice [114].

TGF-β is a multifunctional cytokine produced by a variety of cells, which is capable of regulating the growth and differentiation of numerous cell types [8]. It has been implicated in a number of biological processes including cell adhesion and migration, extracellular matrix production, tissue remodeling, and wound repair. TGF-β mRNA expression has been shown to be higher in adipose tissue of ob/ob and db/db mice compared to their lean littermates [8]. The increased gene expression of TGF-β in adipose tissue could have broad implications in the pathophysiology of obesity and its associated complications since TGF-β has been shown to increase preadipocyte cell proliferation, thereby contributing to the increased cellularity of fat depots related with the obese phenotype. Both obesity and Type 2 diabetes are also associated with characteristic long-term complications, including microvascular kidney disease. Intercrossing of ob/ob mice with TGF-β1-overexpressing animals reverses the obese phenotype, but results in a lipodystrophy-like syndrome accompanied by liver fibrosis, hepato-splenomegaly and glomerulosclerosis [90].

A further class of obesity-related genes encodes leukocyte adhesion molecules as well as their receptors. Mice lacking either leukocyte integrin αMβ2 or its receptor, intercellular adhesion molecule-1 (ICAM-1), develop an obese phenotype at an old age when fed a standard chow diet or at a young age when challenged with a high-fat diet [34]. The weight gain takes place due to an increased adiposity predominantly of the subcutaneous depot without an augmented food intake suggesting a defective energy utilization mechanism. ADAM, a disintegrin and metalloproteinase protein family member involved in the cleavage of membrane anchored precursor of TNF-α, has been also shown to participate in adipose mass control. Overexpression of ADAM 12 protease reportedly induces adipogenesis leading to increased body weight and adiposity in female mice and slight overweight in male mice [17], whereas disruption of the gene encoding ADAMTS-1, a metalloproteinase-disintegrin essential for normal growth, fertility, and organ morphology and function, result in growth retardation and adipose tissue malformation [159].

Metallothioneins (MT) comprise a family of highly conserved metal binding proteins that have been proposed to participate in metal homeostasis through zinc and copper detoxification by scavenging free radicals with a potential involvement in oxidative stress protection [10, 11]. In addition to suffering an increased sensitivity to the toxic effects of metals, mice lacking the ubiquitously expressed MT-I and MT-II became heavier than the controls at about 5 weeks of age [43]. Hyperphagia and obesity were accompanied by a substantially increased epididymal fat depot as well as by hyperinsulinaemia and hyperleptinaemia in MT-I and -II null mice [43]. The underlying mechanisms linking obesity to the absence of MT expression have not been fully elucidated though increased fat accretion associated with increased LPL and C/EBPα mRNA expression have been reported.

Transport molecules and enzymes governing glucose and lipid metabolism

The role in energy balance control of numerous elements participating in intermediary metabolism has been shown through a large number of transgenic and knockout mouse models. Transport molecules and enzymes are step-limiting regulators of certain reactions controlling glucose and lipid metabolism as well as thermogenesis, which have been proven to exert a direct influence on body weight control.

Uncoupling proteins

UCP1 plays a decisive role in thermoregulation and in protecting against diet-induced obesity in rodents by uncoupling oxidation of fuels from ATP production in BAT mitochondria, resulting in heat generation instead. Genetic ablation of BAT leads to obesity and an increase in total body lipid owing to hyperphagia, lower core body temperature and decreased metabolic rate [57]. In addition, transgenic mice are hyperglycaemic, hyperinsulinaemic and hyperleptinaemic with a normoresponsive HPA axis as evidenced by normal somatic growth and fertility. Unexpectedly, Ucp1 knockout mice do not develop hyperphagia and obesity but are sensitive to cold [160], thus providing evidence for the existence of additional BAT-derived factors responsible for diet-induced thermogenesis. Ucp1-deficient mice have a normal resting metabolic rate but have a blunted β-adrenergic stimulated oxygen consumption. Sensitivity to cold depended on the genetic background of the animals. While in the initial study 85% of the mice with a mixed 129/SvPas and C57BL/6J background were sensitive to cold, animals on a 129/SvPas x C57BL/6J F1 hybrid background were resistant to cold despite the absence of Ucp1. The difference in cold-sensitivity between the diverse mouse lines is not attributable to compensation by the other UCPs, but represents an example of hybrid vigour or heterosis. The phenotypic discrepancies regarding obesity between mice lacking BAT and Ucp1 knockouts suggests the existence of a BAT-brain signalling system involving the potential secretion of a centrally acting factor. Constitutive expression of Ucp1 in WAT and BAT confers resistance to both genetic-related and diet-induced obesity with a special reduction of the subcutaneous fat depots [68, 99, 161]. Transgenic expression of Ucp1 in skeletal muscle also prevents diet-induced obesity due to an increased resting metabolic rate, and insulin resistance as a consequence of augmented respiratory uncoupling [100]. On the contrary, ectopic overexpression of Ucp3 provides resistance to obesity despite hyperphagia [98]. Targeted disruptions of Ucp2 or Ucp3 do not affect adipose tissue development or cold responsiveness [11]. The lack of evident alterations in body weight in all three knockout models is probably related to compensatory mechanisms.

Eukaryotic translation initiation factor

An increased thermogenesis has also been observed in relation to the inactivation of the eukaryotic translation initiation factor 4E-binding protein 1 (eIF4E-BP1) leading to smaller WAT pads and an increased metabolic rate in Eif4ebp1 −/− mice [75]. Of interest, the decreased adiposity is accompanied by a shift in the characteristic features of white adipocytes to those of brown fat cells as evidenced by a multilocular aspect and a marked increase in UCP1 mRNA expression owing to an increased PGC-1 translation in WAT, thus unravelling a new molecular mechanism of energy expenditure control at the translational level [75].

Glucose metabolism

The glucose transporter 4 (GLUT4) is primarily expressed in tissues responding to insulin-stimulated glucose uptake such as WAT, BAT, skeletal and cardiac muscle [10]. Two transgenic models overexpressing GLUT4 specifically in adipose tissue [32] or in fast-twitch skeletal muscle [162] have been generated. GLUT4-null mice have also been produced [79, 163] and intercrossed with the animals overexpressing GLUT4 in skeletal muscle fibres to yield a line that expresses GLUT4 only in fast-twitch muscles [162]. Increased body weight and total lipid content due to adipose cell hyperplasia together with an enhanced glucose disposal were observed in transgenic mice overexpressing GLUT4 selectively in adipose tissue [32]. Overexpression of GLUT4 specifically in fast-twitch muscles did not affect body weight but lead to decreased fasting glycaemia and an increase in insulin-stimulated glucose clearance [162]. GLUT4 −/− mice are characterised by a decreased body weight due to a marked WAT reduction accompanied by normoglycaemia and a normal response to a glucose load [79]. However, mice deficient in GLUT4 show increased postprandial glycaemia together with a diminished insulin sensitivity in an insulin tolerance test [79]. Further evidence for the compensatory mechanisms operating in part in GLUT4 −/− mice is provided by GLUT4 heterozygous mutants, which develop diabetes with hyperglycaemia and hyperinsulinaemia 2 months after birth, when GLUT4 expression starts to decline [164]. Muscle-specific transgenic complementation of GLUT4-deficient mice yielded animals with decreased body weight and adiposity showing normoglycaemia and insulinaemia as well as a normal glucose clearance in response to insulin [162]. The degree of glucose intolerance and insulin resistance in GLUT4 −/− mice is similar to that seen in mice with muscle-selective ablation of GLUT4 [163], thereby suggesting distinct and complementary roles of adipose tissue and skeletal muscle in mediating glucose disposal in vivo. Selective inactivation of the GLUT4 gene in adipose tissue had no effect on growth, body weight, fat mass and cardiac development, but impaired insulin action in muscle and liver leading to glucose intolerance and insulin resitance [165]. Thus, glucose transport in adipose tissue plays a critical role in glucose homeostasis. The adipose-selective downregulation of GLUT4 seen in human obesity and Type 2 diabetes mellitus could contribute to insulin resistance and to the risk of developing Type 2 diabetes.

Lipid metabolism

Transgenesis has helped to provide insight into the recognised but partly misunderstood participation of cellular lipid metabolism in energy homeostasis (Fig. 2). Numerous mouse models with disruptions in pathways relevant to fatty acid and triglyceride metabolism have been generated. Fatty acids circulate in the bloodstream as nonesterified molecules, bound to albumin, or in the core of lipoproteins. VLDL receptor −/− mice are resistant to both genetic and diet-induced obesity caused by a decreased peripheral and whole-body uptake of NEFA with no alteration in either food intake or fat absorption [118]. The reduction in adipocyte triglyceride storage as shown by a decreased average fat cell size in VLDL receptor-deficient rodents implies an impaired fatty acid delivery to adipose tissue in the absence of this lipoprotein receptor [118]. Analogously, hepatic overexpression of human apolipoprotein C-I leads to hyperlipidaemia accompanied by decreased visceral fat depots and lack of subcutaneous WAT in mice [166]. On the contrary, overexpression of apoA-II, the second most abundant HDL component, results in increased adiposity and insulin resistance in relation to a decreased skeletal muscle glucose uptake [167]. In the capillaries of skeletal muscle and adipose tissue, LPL catalyses the rate-limiting step in the hydrolysis of triglycerides from circulating VLDL and chylomicrons. Thus, LPL plays an important role in directing fat partitioning. In fact, LPL deficiency results in minimal amounts of tissue lipids leading to neonatal death due to marked hypoglycaemia and hypertriglyceridaemia [168]. In heterozygotes only mild hypertriglyceridaemia with impaired LDL clearance and mild hyperinsulinaemia accompanied by an approximately 20% decrease in fasting glucose concentrations was observed. Exclusive LPL deficiency in adipose tissue on a standard genetic background renders a normal growth and body composition. This implies that although LPL controls fatty acid entry into adipose tissue, fat mass is preserved by endogenous synthesis [109]. When the same lack of LPL in adipose tissue is generated on an ob/ob background a diminished weight gain is attained as a consequence of an impaired lipid accumulation in adipocytes. On the contrary, targeted overexpression of LPL in skeletal muscle or liver has no effect on body weight, but produces an increase in fatty acid uptake into the respective tissue adversely affecting glucose metabolism [169].

Fig. 2.
figure 2

Schematic representation of the main triglyceride metabolism pathways in adipocytes

Acylation-stimulating protein (ASP), an adipocyte-derived cleavage product of complement C3, stimulates fatty acid reesterification and glucose transport at the same time as inhibiting HSL and, hence, lipolysis. Since mice lacking complement C3 are unable to produce ASP, C3 gene knockout rodents are consequently ASP-null animals [10, 11]. ASP-deficient mice show a reduced WAT weight, distributed evenly through all depots in females while primarily affecting gonadal and perirenal localizations in males [70]. The reduced adipose mass was accompanied by hypoleptinaemia with a modest increase in food intake. When fed on either a low-fat or high-fat diet male ASP −/− mice show no differences in body weight compared to wild type animals. Female ASP −/− mice, however, had a decreased body weight compared to their wild type counterparts. The relatively mild phenotype of these mice suggests a minor role of ASP in fat storage and lipolysis control or the existence of compensatory mechanisms rescuing part of the physiological functions of ASP.

The transport of NEFA into cells is facilitated by another fatty acid transporter (FAT), the CD36/FAT molecule, which is expressed in tissues with a high metabolic capacity for fatty acids such as WAT, skeletal muscle and heart [11, 169]. CD36 is a class B scavenger receptor that binds multiple ligands, including fatty acids, and has been proposed to function as a transporter of long-chain NEFA acting as a gatekeeper. Fat cells of CD36 null mice show a decreased capacity to incorporate long-chain NEFA into triglycerides [170]. Although neither the effects of CD36 deficiency on adipose mass, adipocyte size or insulinaemia have been reported, an approximately 30% reduction in plasma glucose concentrations has been observed. Transgenic mice overexpressing CD36 in skeletal muscle showed increased glucose and insulin concentrations together with decreased plasma concentrations of NEFA and triglycerides at the same time as having a markedly increased fatty acid oxidation in muscle [87].

Gene targeting has revealed that mice lacking Dgat1 are capable of synthesizing triglycerides and reach a normal body weight on a standard chow diet although with an approximately 50% reduction in fat mass, adipocyte size and leptinaemia [105]. In addition, after a glucose load Dgat1-deficient rodents have a decreased plasma glucose response, consistent with an increased insulin sensitivity that correlates with a diminished muscular and hepatic triglyceride content [169]. Surprisingly, DGAT1 deficiency has no effect on plasma triglycerides and NEFA, probably due to a compensation by DGAT2 or alternative pathways of conversion of diglycerides to triglycerides [171]. Dgat1 −/− mice, however, are resistant to diet-induced obesity, which is related to an increased energy expenditure independent of increased lean body mass or changes in cold-induced thermogenesis.

Malonyl-CoA is a key metabolite generated by ACC1 and ACC2 that plays a pivotal role linking fatty acid and carbohydrate metabolism through fatty acid oxidation and synthesis in response to hormonal and dietary influences. ACC1 is highly expressed in the cytosol of adipocytes and hepatocytes, while ACC2 is localised in the mitochondria being predominantly expressed in skeletal muscle and heart. Deciphering the roles of the carboxylases in energy homeostasis in lipogenic and non-lipogenic tissues has been approached by the generation of Acc2 −/−mice [65]. In comparison to wild type controls, mutant mice had a 30% higher fatty oxidation rate and accumulated 50% less fat in their adipose depots. Mutants, however, had a normal growth rate despite consuming 20 to 30% more food than the control mice.

Although the fibroblast growth factor (FGF) proteins have been primarily related to fibroblast and epithelial cell proliferation, more recently relevant roles in physiological homeostatic control have been attributed to some of the newly described FGFs. In this context, it is interesting to point out that transgenic mice overexpressing FGF19 show a specific reduction in fat mass resulting from an increased energy expenditure due to an increased BAT mass and a decreased hepatic ACC2 expression [172]. Furthermore, FGF19 transgenic mice do not develop obesity or diabetes on a high-fat diet.

Triacylglycerol stores play a critical role in the ability to withstand fuel deprivation through the release of glycerol and NEFA by lipolysis. The lipid droplets contained in adipocytes are coated by structural proteins, such as perilipin, that stabilise the single fat drops and prevent triglyceride hydrolysis in the basal state. The phosphorylation of perilipin after adrenergic stimulation or other hormonal inputs induces a structural change of the lipid droplet that allows the hydrolysis of triglycerides. The participation of perilipin in body weight regulation has been addressed by knockout approaches from different experimental groups [91, 92]. Perilipin-deficient mice show an increased lean body mass accompanied by an increased metabolic rate together with an increased basal lipolytic rate, which confer resistance to diet-induced obesity. Although absence of perilipin results in leanness and reversal of obesity with an approximately 50% decrease in adipose mass and adipocyte size, it does not enhance glucose intolerance [91], even with perilipin ablation being able to worsen glucose disposal [92].

HSL has been typically considered the key enzyme catalysing the rate-limiting step of lipolysis. Mice lacking HSL show normal growth rates and body weights [173]. While the epididymal, retroperitoneal and femoral WAT depots of HSL-deficient rodents remain unchanged, these mice show a 65% increase in BAT mass compared to wild type control mice. Although catecholamine-induced lipolysis was markedly blunted, the basal lipolytic activity was unaltered. The findings concerning the lack of obesity and mild adipocyte hypertrophy observed in HSL-null mice suggests that other lipases could also play a relevant role in fat mobilisation. The changes observed in triglyceride-rich lipoprotein metabolism have been attributed to a downregulation of hepatic VLDL synthesis and an upregulation of LPL activity in WAT and skeletal muscle [174].

Lipolysis activation could also depend on proteins not directly participating in the catalytic process. Two proteins have been shown to interact with HSL, namely adipocyte lipid binding protein (ALBP or aP2) and lipotransin [175]. HSL is not associated with the lipid droplet and is probably tethered to lipotransin, while perilipin coats the lipid droplet and hinders the access of HSL to it. After hormonal stimulation, HSL and perilipin are phosphorylated and HSL translocates to the lipid droplet. ALBP then binds to the N-terminal region of HSL, preventing fatty acid inhibition of the enzyme's hydrolytic activity. Consistently, aP2 −/− mice show an approximately 40% decreased basal and isoproterenol-stimulated lipolytic rate [176, 177]. These ALBP-deficient rodents show a normal growth rate, body weight and body composition compared to wild type littermates due to a functional compensation by the fatty acid binding protein of keratinocytes [46, 178]. However, when exposed to a high-fat diet aP2 −/− mice develop diet-induced obesity reaching a greater total weight gain than the control mice as a consequence of an increased fat pad weight. Both lean and obese aP2 −/− mutants show normoglycaemia and normoinsulinaemia providing evidence for an uncoupling of obesity from insulin resistance through ALBP deficiency [46, 176].

Glycerol 3-phosphate dehydrogenase (GPDH) is a ubiquitously expressed enzyme participating in triglyceride synthesis as well as in shuttling NADH into mitochondria for oxidative metabolism. Transgenic mice overexpressing GPDH show a normal body weight but with an increased interscapular brown fat depot and virtually no WAT [179]. On the contrary, animals lacking mitochondrial GPDH show a decreased body weight compared to wild type mice with no reported effect on adipose tissue depots but a marked insulin release defect when the malate-aspartate shuttle is blocked. This indicates an important role of the NADH shuttle or glycerol phosphate system in glucose-induced activation of mitochondrial metabolism and insulin secretion [88]. Therefore, the two mouse models with altered expression of GPDH provide evidence for an involvement of this enzyme in lipogenesis and WAT development as well as in the regulation of the glycolytic pathway.

The synthesis of triglycerides and glycerophospholipids starts with the acylation of glycerol-3-phosphate by glycerol-3-phosphate acyltransferase (GPAT) to form lysophosphatidic acid. Mammals have two isoforms of the enzyme, located in the outer mitochondrial membrane (mtGPAT) and the endoplasmic reticulum (microsomal GPAT). It has been recently reported that mtGpat −/− mice have reduced weight and fat pad mass accompanied by lower liver and plasma triacylglycerol, together with a lower VLDL secretion rate and an altered glycerolipid fatty acid composition [180].

During the last decade the generation of transgenic mouse models has begun to provide valuable though fragmented insight into the mechanisms by which cellular lipid metabolism modulate body weight regulation. Accumulating evidence for important functions of enzymes involved in the anabolic and catabolic processes of lipid metabolism has been obtained through targeted disruption in rodent models of genes controlling triglyceride metabolism. However, direct links between abnormal expression or genetic variations and human disorders such as obesity, hyperlipidaemia, insulin resistance, and Type 2 diabetes await further clarification.

Concluding remarks

The slow progress in understanding body weight regulation shows the multifactorial and highly intricate adaptive mechanisms developed over time against weight loss. Current evidence for the involvement of several different factors in energy balance regulation indicates that body weight homeostasis and food intake control are far more complex than initially thought from the predictions of a single circulating satiety factor derived from the classic parabiosis studies. The identification of genes that cause obesity, leanness or provide resistance against obesity development have added new pieces to the puzzle of body weight control. In addition, transgenesis has stimulated the impetus for the search of potentially unsuspected physiological consequences of already known gene products. However, it has to be stressed that studying single genes independently through the analysis of genetically modified animal models provides only a fragmented view of what has to be considered -in most cases—a multigenic pathology such as obesity. Thus, it is vital to avoid oversimplified views. In this sense, information derived from traditional physiological studies continues to deliver relevant clues for a better understanding. Trying to integrate findings obtained from both whole-body physiology as well as transgenic experiments represents a useful approach to explore the equilibrium achieved to compensate for fluctuations in body weight. Control of gene expression from a quantitative, qualitative and spatiotemporal perspective together with functional genomics embodied by DNA microarrays will provide new clues to characterise modified and modifier genes that influence transgene-dependent phenotypes. Such complementary approaches could prompt the design of new drugs for the treatment of obesity as well as open up the field of pharmacogenomics.

Sources

This review is based on the relevant literature published in the English language during the period 1990–2002, and seminal prior contributions. The sources available to the authors were integrated with sources identified through PubMed searches for "body weight, obesity, leanness or adipose tissue" combined with searches for "knockout, transgenic or overexpression".