Contents
Contributors
Search

Historically, obesity has been considered a voluntary behavior disorder exacerbated by the ready availability of food and reduced energy expenditure afforded by modern societies. However, the occurrence of spontaneous, monogenic obesity in mice has provided clear proof that single genes can convey fully penetrant obesity with no change in environment. Identification of the responsible genes in rodents has led to the identification of obese humans with mutations in orthologous genes (Table 1). These genes now define the molecular genetic scaffold for the regulation of energy homeostasis in animals and humans, and provide the basis for understanding the complex interactions among genetic constitution, development, and environment in determining body mass and composition.

Spontaneous obesity in rodents is caused by monogenic and polygenic mutations

In 1902, French geneticist L. Cuenot described the obese Yellow ( A^y /a ) mouse, bred and maintained by European mouse fanciers since the 1800s [1]. This was the first report of a spontaneously obese mouse, which prompted investigation of additional spontaneous obese mouse models. In addition, the Yellow mouse led to the subsequent discovery of the well-conserved melanocortin pathway, which may be a relatively common contributor to human obesity.

The autosomal recessive obese (ob) mutation was discovered in 1949-50 in a non-inbred strain (V) maintained by Jackson Laboratories in Bar Harbor, Maine [2]. Sixteen years later a phenotypically similar mouse was discovered [3]. The diabetic state of these animals on the coisogenic KsJ strain background distinguished them from ob and hence the mutation was designated diabetes (db) . Doug Coleman, looking for the molecular predicates of the lipostatic system posited by Kennedy [4] and Hervey [5], performed parabiosis (joined circulation) studies. From these findings, Coleman inferred that the ob mutation occurred in a genetic region that encoded a secreted molecule, and the db mutation occurred in a sequence that encoded its receptor. As described below, these were remarkably prescient insights. In 1990, Coleman and colleagues described additional recessive obese mutations, tubby (tub), and fat (fat) [6, 7].

Positional cloning approaches were employed to identify the genetic mutations in agouti (Yellow) [8, 9], obese [10], and tubby [11, 12]. The diabetes mutation in was isolated using expression cloning with the ob protein as probe [13][14], and a positional candidate approach facilitated the discovery of the mutation in fat [15]. Although human correspondents of monogenic obesities are rare [16], the study of rodent monogenic obesity in mice has proven extremely useful in elucidating new molecules and pathways in the control of energy homeostasis.

Fewer obesity-related spontaneous mutations have been detected in rats, due in part to the absence of explicit screening of large numbers of progeny for such phenotypes, and the greater cost of rat husbandry. The autosomal recessive Zucker obese rat has been studied intensively for mechanistic insights into the genetics of obesity [17]. The molecular mapping of the db mutation [18] permitted a test of the hypothesis that fa and db are mutations of the same gene. Truett et al showed that the fa and db mutations map to syntenic regions of the respective genomes of rat and mouse [19], the first demonstration of likely cross-species effects of the same gene on body composition. This inference was confirmed by the molecular cloning of the db mutation in leptin receptor ( Lepr ) [13, 14] and the identification of mutations in Lepr in the Zucker and Koletsky rats [20-23]. In addition, the OLETF obese rat [24] has a type 2 diabetes phenotype attributed to a mutation in the CCK-A receptor [25].

Spontaneous and genetically manipulated rodent models illustrate fundamental aspects of the molecular physiology of body weight regulation

Body mass and composition reflect the aggregate effects of three processes: energy intake, energy expenditure and the chemical form of energy stored or released because of imbalances of intake and expenditure (partitioning). Energy intake includes all physiological systems that regulate the consumption of calories. Genes that regulate energy intake may be broadly divided into two classes based on the time course over which caloric consumption is affected by their administration. The first class is comprised of long term regulators of energy balance, namely leptin and insulin, that reduce meal size with no compensatory increase in meal frequency, resulting in a decrease in body weight. Leptin and insulin are primary effectors that enter the brain and act on their respective receptors located on secondary effector neurons in the hypothalamus. These neurons express neuropeptides with long term effects to alter food intake and body weight. Neuropeptides that increase food intake, such as neuropeptide Y (NPY), are inhibited by leptin and insulin, while catabolic neuropeptides, such as proopiomelanocortin (POMC) are upregulated, and subsequently act on local receptor populations to exert their physiological effects. The expression of these molecules and their receptors is largely confined to the hypothalamus. The second class consists of short term regulators, such as cholecystokinin (CCK), with primary action lasting a few hours after a localized release from cells lining the gastrointestinal tract. Short term regulators affect meal size with a compensatory alteration in meal frequency, and therefore do not alter body weight.

Basal metabolic rate accounts for about 70% of total energy expenditure (TEE) in sedentary adults, and is determined by body composition, age, wakefulness, and genetic factors [26]. Nonresting energy expenditureis the next largest contributor, averaging approximately 20% of TEE. In sedentary individuals, nonresting energy expenditure has the greatest percentage contribution to TEE during low-level physical activities. Smaller amounts of energy (7%) are accounted for by diet-induced thermogenesis, which represents heat energy released in the process of digestion and disposition of ingested calories.

Adipose tissue is the solution to the need to store energy ingested in excess of immediate need. The energy density of adipose tissue is nearly 10-fold greater than muscle (protein) or liver (glycogen). The ability to store such energy protects against environmental vicissitudes that might result in starvation, loss of fetuses, and inability to provide sufficient breast milk to the young. Therefore, it is likely that evolution has promoted genes/alleles that favor energy storage and conservation. The existence of environments in which excess calories are readily available with minimum or no effort is a very recent occurrence in human evolution. Human genetic "makeup" is presumably designed for the opposite circumstance.

"Partitioning" describes the processes determining molecular form of stored energy and chemical energy released when energy balance is negative. Partitioning of adipose stores is regulated by a variety of genes expressed in adipose tissue, skeletal muscle, and liver that transcribe enzymes, transcription factors, cytokines, receptors, and receptor-related signaling molecules. Deletion and overexpression of these genes often alters accumulation and distribution of adipose tissue and lean mass.

The rodent models discussed here illustrate the roles that individual genes play in energy intake, energy expenditure, and partitioning, and are categorized in Table 2.

Transgenic models of control of body composition

In the early 1980s, transgenic and gene knockout strategies became available that have enabled the analysis of systemic and organ-specific effects on energy homeostasis of one or more genes selected by the investigator [27]. Targeted disruption of genes was first demonstrated in vitro in a fibroblast system [28]. This was accomplished by exploiting the phenomenon of homologous recombination to introduce genetic sequences that could either delete or generate a targeted allelic variant in individual genes while leaving the rest of the genome intact. In subsequent studies, manipulated DNA sequences were introduced into undifferentiated murine embryonic stem (ES) cells to recombine with native DNA, to create a mouse in which a specific gene is inactivated, altered in coding sequence, or expressed in a novel, organ-specific manner. Remarkably, the stem cells retained the developmental program to divide normally when reintroduced into a pre-implantation embryo. The resulting mutant 'knockout' mouse carries a deletion or alteration in the gene of interest.

The advantage of transgenic and knockout approaches is that the role of individual genes may be studied in isolation. However, disruption of a gene during development may affect viability or alter the phenotype in a manner unrelated to its function in the adult animal. In addition, certain genetic manipulations may cause unwanted alterations in gene expression. For example, the inclusion of a drug selectable cassette, often used to identify ES cells that contain a correctly targeted gene, may suppress gene expression on neighboring genes [29].

Another important consideration in mouse models of obesity generated by gene targeting is genetic background effects. Targeting has been most consistently successful in the 129 mouse strain because the ES cells of 129 mice are relatively easy to culture. However this strain exhibits increased levels of anxiety in response to stressors that could potentially distort food intake and metabolic phenotypes [30]. ES cells from C57BL/6 require carefully controlled in vitro conditions, and even then, often fail to transmit the introduced mutation to the germ line. Serial backcrossing of a progenitor 129 or other strains, e.g., C57BL/6, can be used to transfer the mutation to a more suitable "background". For example, mice overexpressing melanin-concentrating hormone (MCH) have an obese phenotype only when backcrossed for 7 generations onto the C57BL/6J background. On an FVB background, they appear to have a normal phenotype with regard to body composition [31]. New strategies such as the insertion of genetic material using transposons (http://www.tosk.com ) and cloning mice from somatic cells [32] may eventually be widely available and could circumvent the need for extensive backcrossing to achieve strain homogeneity.

SPONTANEOUSLY OCCURRING MOUSE MUTATIONS LEADING TO OBESITY

Obese. The obese mouse was identified as an autosomal recessive mutation in a noninbred strain (Stock V) at Jackson Laboratory in 1949. Mice segregating for the obese gene were backcrossed for many generations to generate a congenic line on the commercially available C57BL/6J background strain. Lep^ob mice on C57BL/6J, despite their early-onset obesity, are not diabetic; however, the ob mutation coisogenic on the C57BL/KsJ line results in severe, early-onset type 2 diabetes. Apparently these animals are diabetes-prone because the KsJ strain is a C57BL/6J strain inadvertently contaminated with DNA from the diabetes-prone DBA strain [33].

The hypothalamic infertility of the ob mutation is sensitive to the genetic background on which the mutation is carried. Chehab et al showed that mice with the ob mutation bred onto a BALB/cJ background exhibit enhanced fertility: 41% of C57BL/6J-BALB/cJ- ob/ob males produce offspring [34]. A second loss-of-function mutation in the ob gene, ob^2J , was identified in 1997 on an SM/Ckc-+^DAC background, resulting from a retroviral-like transposon insertion that prevents RNA translation [35]. This recessive mutation causes an obese phenotype virtually identical to that of the original ob mutation.

In parabiosis experiments, Coleman joined the circulatory systems of Lep^ob mice to either wild-type or Lepr^db mice [36]. The Lep^ob mouse became lean when joined to a wild type, but, when joined to a Lepr^db mouse, the Lep^ob mouse died of starvation. These findings led Coleman to hypothesize that a blood-borne factor regulating body weight might be deficient in Lep^ob , but circulating at very high levels in the blood of Lepr^db mice. He suggested that obese was the secreted factor and diabetes its receptor. In 1994, the gene encoding ob was isolated by positional cloning [10], and its product, ob protein, or leptin, was shown to be produced primarily in adipocytes. Lep^ob mice lack circulating leptin by virtue of an R105X mutation that creates a premature stop codon sequence in the leptin gene, resulting in a truncated protein that is rapidly degraded [10].

When leptin is administered to leptin-deficient Lep^ob mice, it reduces food intake and lipid stores, sparing lean tissue [37-39]. Conversely, when leptin-receptor deficient Lepr^db mice are pair-fed to lean mice, Lepr^db mice retain adipose mass, indicating that a lack of leptin signaling allows preferential storage of calories as fat [40, 41]. Lep^ob and Lepr^db mice provided compelling in vivo evidence that fat mass may be profoundly affected by a single gene. Both genes affect all three major components of energy homeostasis (energy intake, expenditure, and partitioning) [41, 42].

The major physiological role for leptin in animals and humans may be to maintain a level of adiposity required for survival and successful parturition. Mice fasted for 48 hours have 50% decreased plasma leptin compared to fed controls, with a concomitant 16% decrease in body weight [43]. Intraperitoneal administration of leptin in doses that restore leptin to pre-fasting levels in fasted mice restores altered neuroendocrine changes in the thyroid and adrenal axes and reduces fasting-induced hyperphagia [43].

In humans, leptin deficiency also elicits a severe obesity phenotype. A rare, recessively inherited LEP mutation was discovered in two related children that are members of a highly consanguineous Pakistani family [44]. As with the Lep^ob mutation in mice, this frameshift mutation resulting from a single guanine deletion at position 133 introduces a premature stop codon that truncates the leptin protein. Daily subcutaneous administration of recombinant leptin selectively reduced body fat from 59% adiposity to 52% [45]. In another report, partial leptin deficiency resulting from heterozygosity for a frameshift in the gene encoding leptin was described in 13 individuals from three different families of Pakistani descent [46]; circulating leptin concentrations were disproportionately reduced relative to fat mass, as reported in Lep^ob/+ mice [47]. In addition, a recent clinical study showed that the metabolic and endocrine changes accompanying a10% reduction in body weight [48] are reversed by low leptin doses that restore thyroid hormones to levels present prior to weight loss [49]. Together these studies demonstrate an important functional role for leptin in the regulation of adiposity and energy expenditure in humans that is analogous to its function in mice.

Diabetes. Diabetes (db) is a spontaneous recessive mutation that was first noted in a C57BL/KsJ ('Kaliss') mouse colony at the Jackson Laboratory, and at least five other instances of mutation at this locus have been reported in mice. The KsJ db mutant is obese and hyperphagic and develops severe type 2 diabetes [3]. Backcrossing db onto a C57BL/6 background attenuates the diabetic phenotype so that C57BL/6J- db is virtually identical to C57BL/6J -ob . As noted earlier, the KsJ strain is segregating modifiers that render obese animals susceptible to diabetes. Congenic lines for db have been made for six inbred strains [3, 50-52]. These strains show wide variability in susceptibility to type 2 diabetes, again pointing to the existence of potent modifiers of susceptibility to diabetes in the context of obesity [53].

The leptin receptor gene was identified by expression cloning [14], and the first genetic mutation in the leptin receptor was identified in the Lepr^db mouse[13, 54, 55], thus substantiating the conceptual model described by Coleman: the obesity of the Lepr^db mouse was due to a mutation that precluded leptin signaling in the receptor that binds the ob gene product. The mutation in the Lepr^db mouse is due to a splicing defect that causes the 3' terminal exon (18a) of leptin receptor isoform a ( Lepr-a ) to be inserted into Lepr-b. A stop codon at the end of this terminal exon prevents transcription of the Lepr-b terminal exon, converting all signaling, Lepr-b isoforms to Lepr-a (Figure 1A)[13, 54, 55]. Because the Lepr^db mouse synthesizes all isoforms of the leptin receptor except Lepr-b (Figure 1B), it is clear that this isoform is critical to the control of energy homeostasis.

Figure 1A. Leptin receptor isoforms. Lepr-a (Ra) encodes the short form of the leptin receptor, and Lepr-b (Rb), encodes the long form. Exon 17 contains a Jak docking site (BOX1) common to Ra, Rb and Rc, while exon 18b contains additional motifs (BOX2 and STAT) required for STAT3 signaling. Soluble LEPR isoform Re lacks a transmembrane domain.

Figure 1b. The db mutation in mice. The Ra isoform in Leprdb differs from the wild-type Ra receptor by a single base pair in the 3'UTR. This creates a new splice donor site, resulting in a 106 bp insertion in the Rb isoform that introduces a premature stop codon (adapted from [415]).

Leptin receptor isoform B, or LEPR-B, is the only one of the six isoforms that signals through a classic Jak-STAT cytokine signaling pathway. Replacement of tyrosine 1138 with serine in exon 18b disrupts STAT3 signaling of the LEPR-B receptor, eliciting a severe obesity phenotype in mice that recapitulates the obesity syndrome of Lepr^db / Lepr^db mice [56]. LEPR-B is coexpressed in the hypothalamus with LEPR-A, which is comprised of the first 17 exons identical to LEPR-B followed by a terminal exon, 18a. LEPR-C is a short form with unique exon 17' after exon 17. Both LEPR-A and LEPR-C are highly expressed in brain microvessels and may have a role in leptin transport [57], although a definitive function for these isoforms has not been elucidated. LEPR-E is a soluble circulating isoform that binds approximately 50% of leptin in lean subjects, and 20% in obese [58], while LEPR-D is found only in mice [55], and LEPR-F is unique to rats [59].

In the periphery, low amounts of Lepr-b mRNA are expressed in white and brown fat depots, liver and kidney, while Lepr-a mRNA is highly expressed in the liver and kidney only, and Lepr-c mRNA is not detected in any of these tissues [60]. LEPR-B is the predominant isoform found in the arcuate nucleus (ARC), a subnucleus of the hypothalamus with a well-documented role in energy balance. In the ARC, Lepr is coexpressed in neurons synthesizing proopiomelanocortin (POMC) and cocaine-and-amphetamine-regulated transcript (CART) as well as those that secrete neuropeptide Y (NPY) and agouti-related peptide (AGRP). These neuropeptides modulate food intake and subsequently decrease body weight and adiposity [61, 62]. Additional Lepr -positive hypothalamic subnuclei that regulate energy homeostasis include the lateral hypothalamic area (LHA), the paraventricular nucleus (PVN), the dorsomedial nucleus (DMN) and the ventromedial hypothalamus (VMH).

Mice heterozygous for the db mutation ( db/+ ) have 29.6% more fat mass and 87.9% elevated leptin levels (adjusted for fat mass) as compared to wild-type littermates [47], allowing them to withstand a prolonged fast much longer than their wild-type littermates [53]. The elevated concentrations of leptin with respect to fat mass apparently reflect blunted leptin sensitivity due to haploinsufficiency in the signaling form of Lepr [47]. In addition, Lepr^db/+ females are prone to gestational diabetes [63], and administration of leptin prevents this syndrome [64]. These findings suggest that haploinsufficiency in leptin signaling may contribute to increased adiposity in humans, although thus far, this association has not been observed.

Five additional murine mutations at the Lepr locus have been identified: Lepr^db-3J [50, 65] , Lepr^db-Pas [66] , Lepr^db-NCSU [67], as well as the now extinct Lepr^db-ad and Lepr^db-2J [68] (' adipose ' or ' adult ')[69]. Lepr^db-3J , Lepr^db-Pas , and Lepr^db-NCSU are unique mutations in the extracellular domain of the leptin receptor resulting in premature truncations before the transmembrane domain, thus ablating all membrane-bound forms of the receptor. Lepr^db-3J is a 17-bp deletion in exon 11 [70], Lepr^db-Pas is a duplication of exons 3, 4, 5 and 6, and the Lepr^db-NCSU mutation is a single base pair deletion in exon 12 [67]. Two Lepr point mutations have been identified in rats: fatty ( Lepr^fa ) [17], which causes a Q269P substitution in the extracellular domain resulting in receptor degradation before the protein is transported to the cell surface [20, 71], and Koletsky ( Lepr^fak ) [23], a Y763X null mutation that produces a premature stop codon before the transmembrane domain [72, 73].

Transgenic overexpression of leptin receptor throughout the central nervous system (CNS) partially corrects the obesity syndrome of Lepr^db-3J mice [74]. Deletion of leptin receptor in the CNS via neuron-specific cre-lox recombination causes 28% reduction in body weight as compared to the Lepr null mouse, a 50% increase over heterozygous littermate controls [75]. These findings indicate that hypothalamic LEPR is clearly required for the regulation of body weight, but that either the models do not completely alter Lepr expression in the hypothalamus, or that peripheral signaling by LEPR may play some role in this regard. Additional studies with deletion and overexpression of LEPR in peripheral tissues will help to address the relative roles of brain versus peripheral LEPR signaling.

Agouti. The Yellow mutation of the agouti locus ( A^y /a ) is also termed 'lethal yellow', since homozygotes for the allele are prenatal lethal. Yellow was bred by mouse fanciers in Europe beginning in the 1800s, and was notable for the dominant inheritance of its striking yellow coat color and obesity proportional to the intensity of the yellow coat [1]. In 1960, another spontaneous agouti mutation was detected in the Jackson Laboratory colony, viable yellow

( A^vy /- ) [76]. [M1]   The agouti gene encodes agouti signaling protein (ASP), a peptide with a high affinity for melanocortin receptors located in both the skin and brain. The yellow coat color of the A^y /a mouse results from agouti overexpression in the skin which blocks alpha-melanocyte-stimulating hormone (a-MSH) signaling at melanocortin-1 receptors in the hair follicle. Sincea-MSH) activates melanocytes to initiate synthesis of eumelanin (black pigment) instead of phaeomelanin (yellow pigment), antagonism ofa-MSH) by agouti protein elicits a yellow coat color. Icv administration of a-MSH) and a-MSH) agonists decreases food intake and body weight via increased melanocortin signaling at GABA interneurons in the PVN [77]. Therefore, overexpression of agouti in the brain of A^y /a mice also antagonizes the anorectic action of a-MSH) signaling, causing hyperphagia leading to adipocyte hypertrophy, hyperinsulinemia, and an obesity that becomes apparent at 8 to 17 months of age. The agouti protein blocks binding of a-MSH) to the melanocortin-4 receptor as evidenced by the 10-fold higher concentration ofa-MSH) required to activate MC4R adenylyl cyclase when agouti is present ( a-MSH) alone: EC ₅₀ = 4.9 x 10^-9 M ; a-MSH) + 0.7 nM agouti protein: EC ₅₀ = 3.3 x 10^-8 M [78]), and its ability to elicit the A^y phenotype.

Obese Yellow mutants are a result of two different mutations, A^y and A^vy , discovered when agouti was cloned at the Oak Ridge National Laboratories [8]. The lethal yellow mutation is a deletion of the Raly gene, which causes a fusion of the constitutively active Raly promoter to the agouti gene, resulting in ectopic overexpression of agouti in all somatic cells. A^vy /a is also the result of ectopic overexpression of agouti , however this mutation results from insertion of a retrovirus-like repetitive intracisternal A particle (IAP) into a noncoding exon of agouti . The resulting splice variant fuses the constitutively active Raly promoter to the agouti gene, allowing constitutive overexpression of agouti in all somatic cells.

The increased body weight of A^y /a and A^vy /a mice results from hyperphagia in the absence of any apparent decrease in energy expenditure [79]. Fat mass and naso-anal length are both increased in these animals [80]. By contrast, Lep^ob and Lepr^db mice have a selective expansion of fat mass due to increased food intake, decreased energy expenditure, and preferential storage of excess calories as fat [42, 81, 82], and Lep^ob and Lepr^db mice are stunted in length resulting in overall reduced lean body mass. These differences in phenotype suggest that MC4R signaling predominantly affects food intake, so that MC4R signaling deficiency results in a general increase in somatic growth.

The increased adiposity of A^y /a mice may also be mediated by peripheral effects of agouti protein to modulate Ca²⁺ concentrations in adipocytes (for a review, see [83]). The location of 10 cysteine residues located on the agouti C terminus is similar to snake and snail venoms that interact with calcium channels, suggesting that ASP may modulate calcium transport in adipocytes. Moreover, in vitro studies demonstrate that ASP increases triglyceride accumulation in adipocytes while reducing lipolysis in a calcium-ion-dependent fashion.

Tubby. The tubby mutation, located 51.45 cM distal to the centromere on chromosome 7, arose spontaneously at Jackson Laboratory in the C57BL/6 strain [7] and has been cloned and characterized [11]. These mice have a mild, late-onset obesity apparent by 8 to 12 weeks of age that is associated with hyperinsulinemia without hyperglycemia. The precise physiological mechanism(s) for the obesity are still unknown [12]. Traits characteristic of the tubby mutation are retinal and cochlear degeneration, resulting in eventual blindness and hearing loss. Alstrom and Bardet-Biedl obesity syndromes resemble some aspects of the tubby mouse phenotype, however no syntenic relationship exists between mouse tubby and these human syndromes.

Tubby results from a G (r) T transversion that interferes with normal intron excision [12]. The result is an aberrant transcript in which a 44-base pair deletion at the 3' end of the gene is replaced with a 24-base pair intronic segment that is usually spliced out. Targeted deletion in the tubby gene allowed for a direct comparison of a loss-of-function model with the spontaneous mutation to verify the identification of the affected gene [84]. Tub-/- mice were phenotypically indistinguishable from tubby , suggesting that the phenodeviant that occurred in the Jackson Laboratory colony had a loss-of-function mutation in the tubby gene. Lack of detectable tubby protein or transcript in tubby mice further supported this finding. By comparison, Lepr^db , Lep^ob , and Mc4r +/- mice all had levels of tubby protein in brain lysates equivalent to those as seen in wild-type animals.

The tubby protein is an apparent transcription factor activated by G-protein-coupled-receptors (GPCRs) [85]. Specifically, tubby protein is bound to the Ga _q subunit of a GPCR complex by phosphatidylinositol 4,5-bisphosphate (PtdIns(4,5)P2) at the cell membrane and is released upon PtdIns(4,5)P2 hydrolysis by phospholipase C-b , allowing tubby to translocate to the nucleus. Lacking functional evidence for a primary physiological role of tubby , some have speculated that GPCRs implicated in the regulation of energy homeostasis, such as 5-HT _2c R, MC4R, MCH1R and dopamine receptors, are candidate pathways that may signal through tubby [85]. Brain expression of both tubby and 5-HT is highest in the PVN and tubby and 5-HT _2c receptor mutant mice have a similar, late-onset obesity syndrome.

Fat. The fat mutation was first identified at the Jackson Laboratory in a colony of inbred HRS/J mice [7]. Cpe^fat mice exhibit apparent hyperinsulinemia as early as four weeks of age followed by obesity at 8 to 12 weeks. Approximately 77% of the measured insulin is proinsulin, thus bioactive insulin levels are normal, and Cpe^fat mice are not insulin resistant. On the C57BLKs/J 'Kaliss' background, Cpe^fat develops transient hyperglycemia of ~250 mg/mL in 80% of males by 12 weeks of age; this hyperglycemia remits spontaneously by 8 months. In contrast, females remain normoglycemic throughout adulthood.

By a positional candidate gene approach, a T-->C point mutation in Cpe^fat , resulting in a serine-to-proline transversion at position 202, was identified in carboxypeptidase E (CPE) [15]. CPE is an enzyme that cleaves C-terminal dibasic residues arginine and lysine in prohormone precursors of insulin, enkephalin, POMC, NPY, MCH, cholecystokinin (CCK) and oxytocin-vasopressin [86]. Since reduced CPE activity was observed in islets and pituitaries of fat mice [15], and aberrant levels of prohormone convertases 1 and 2 were detected in the pituitary and hypothalamus [87], it is possible that any one or all of these hormones are processed abnormally in the Cpe^fat mouse .

Transgenic overexpression of insulin in Cpe^fat mice does not correct the obesity, suggesting that aberrant processing of one of the other peptides processed by CPE contributes to the increased fat mass [88]. Human CPE is located on chromosome 4, and one missense polymorphism in CPE has been identified as a C-->T transversion [89]. This mutation results in a non-conservative R283W amino acid substitution that greatly decreases CPE enzymatic activity and is associated with early onset type 2 diabetes. An analogous G483R mutation has been described in a female proband in whom compound heterozygosity for mutations in prohormone convertase 1 (PC1) was associated with an obese phenotype and evidence of disordered processing of POMC, insulin and gonadotropin-releasing hormone [90]. The patient displayed hypocortisolism and increased plasma POMC, elevated proinsulin with impaired glucose tolerance, and hypogonadism.

Mahogany and mahoganoid. The spontaneous, autosomal recessive coat color mutations, mahogany (mg) and mahoganoid (md), were first reported over forty years ago [91, 92]. When crossed to Yellow A^y /a mice, both mahogany and mahoganoid darken coat color and attenuate obesity. Mahogany , now formally called attractin ( Atrn )is located 15.1 cM centromeric of agouti on chromosome 2 [93]. Positional cloning of mahogany identified a protein with a single transmembrane domain and a putative glycosaminoglycan side chain that may chaperone agouti to melancortin receptors [93]. The extracellular domain of mahogany is orthologous to the human immune-response protein attractin, which accumulates on the surface of activated T cells and subsequently facilitates the interaction between T cells and antigen by "attracting" macrophages [94]. Murine Atrn is normally expressed in widely in the brain, as well as the skin, heart, kidney, liver and lung of wild-type mice [93]. The Atrn^mg mutation in mice is due to a ~5-kb retroviral insertion in intron 11 that disrupts Atrn expression and presumably permits a relative increase in eumelanin expression in the hair follicle, resulting in dark fur, as a consequence of increased melanocortin signaling at the melanocortin-1 receptor (MC1R). Atrn^mg mutants also have 10-15% reduction in body weight and have 20-40% less body fat content than littermate controls, which may result from increased melanocortin signaling at MC4R. Deficiency of Atrn in the A^y /a double mutant may reverse the obesity and yellow coat color by reducing agouti binding to melanocortin receptors [95]. In support of the specificity of Atrn effects, crossing Atrn^mg mice with tubby , fat , diabetes, Mc4r -/- [96] or Lep^ob mice [79] does not attenuate the obese phenotype.

However, the Atrn^mg mutation apparently decreases body weight in A^y /a mice, in part, by increasing locomotor activity and basal metabolic rate (BMR) with no apparent effect on food intake [79]. Since the hallmark of MC4R signaling deficiency in mice and humans is hyperphagia, this unexpected observation suggests that the mechanism of action of Atrn with respect to its interaction with the melanocortin system is not well understood. Atrn^mg mice are resistant to diet-induced obesity [96], due to an elevated BMR and increased locomotor activity [79], and recent evidence indicates that a neurological defect may be the basis for increased energy expenditure observed in Atrn^mg mice [97].

The zitter mutation, which causes hypomyleination in rats, is due to an 8-bp deletion in Atrn at a splice donor site, resulting in a marked decrease in Atrn expression. Histopathological examination of the Atrn^mg brain and spinal cord shows extensive vacuolar formation and aberrant mylein sheaths [98, 99]. It has been suggested that increased motor activity due to the neurologic involvement can account for the increased BMR and decreased body mass of the Atrn^mg mice [97]. However, the specificity of the weight reducing effect of Atrn^mg to the Yellow (and not tub , Lep^ob , Lepr^db or Cpe^fat ) suggests that the effect is specific to the melanocortin pathway rather than reflecting nonspecific motor effects.

The mahoganoid locus, Mgrn1 ( mahogunin, RING finger 1), lies 2 cM from the centromere of chromosome 16 and [100]. Five mutations at this locus have been identified: md , md^3J , md^4J , md^5J , and md^6J . The Mgrn1^md mutation is a 5-kB IAP element intronic insertion between exons 11 and 12. The mutation results in a 10-20 fold attenuation of expression of the protein in kidney, brain, heart, spleen, lung, skin, skeletal muscle and adipose tissue as compared to that of Lep^ob , A^y and C57BL/6J mice. Similar to the mahogany phenotype, the Mgrn1^md mouse is lean, and when crossed with A^y /a mice, the mutation reduces body weight by 80% and darkens the yellow coat color. The mahoganoid protein includes a RING finger domain that suggests that mahoganoid is an ubiquitin E3 ligase that may antagonize melanocortin signaling by impeding ligand binding. Although the spontaneous mutations at the Mgrn1 locus do not cause neurological degeneration, the Mgrn1^md-nc mutation generated by caffeine mutagenesis results in histopathological changes similar to that seen in Atrn^mg and the zitter rat [101].

Polygenic rodent models. Various mouse and rat inbred strains have been used as models genetic control of body mass and composition and response of these to high fat diets. Strain-related differences in adiposity, diet-induced obesity, and other aspects of energy homeostasis can be mapped by phenotyping F2 progeny of interstrain crosses. Thirty such regions, referred to as quantitative trait loci (QTL), have been mapped in rats, and 119 have been mapped in mice [102]. Many of the chromosomal regions in mice correspond to loci associated with human obesity (Figure 2).

Figure 2. Chromosomal location of obesity genes. Ideogram of human karyotype with human monogenic mutations indicated in green, human loci identifed by genome-wide linkage scans indicated in blue, and the potential location of one or more mouse QTLs (based on conserved homology) indicated in red. From [416].

Both Lepr^fa and Obese Long-Evans Tokushima Fatty (OLETF) mutations interact with genes on the background strain upon which they are bred, in a manner that modifies the degree of obesity and the phenotypes associated with obesity [47, 103-105]. Male Lepr^fa rats show increased adiposity, hyperinsulinemia, and hyperglycemia on a Wistar (WKY) background compared to a 13M background [106]. Diabetic males resulting from intercrossing 13M/WKY rats have a QTL contributed by 13M affecting both weight and BMI ( Qfa1 , now renamed Niddm4 ) that corresponds to a murine chromosomal segment near the tubby gene [107]. An entirely different locus on chromosome 12 contributes to increased body weight and BMI in females ( Qfa12 , now Niddm5 ), one of many instances of sex differences in weight gain, adiposity and predisposition to diabetes that are also evident in humans [108-110]. OLETF crosses with nonobese rat strains (Brown Norway and Fischer 344) have uncovered thirteen QTL associated with body weight and adiposity[103-105] (Table 3).

AKR/J and SWR/J mice. Some polygenic models of obesity require a high fat diet to amplify the genetic diathesis towards obesity. Rodent models of diet-induced obesity add to the variety of rodent models that are highly relevant to the study of human obesity and for which the genetic and environmental factors may be controlled.

Large differences in responsiveness to a high-fat diet exist among mouse strains. In a study of obesity induced by a condensed milk diet in nine inbred backgrounds, six strains showed a significant increase in body weight and adiposity in response to the condensed milk diet while three strains did not differ from mice fed control chow. A precise mechanism to account for the strain difference has not been determined [111]. The two strains representing extremes in diet-induced obesity, AKR/J (diet susceptible) and SWR/J (diet resistant) were subsequently crossed to identify three QTL contributing to the obese phenotype: Dob1 [112], Dob2 and Dob3 [113]. None of these QTL were syntenic with db , ob , tub , agouti, or fat loci.

Israeli sand rat (Psammomys obesus). In its native habitat, the Israeli sand rat has adapted to a diet of saltbush, a food source of very low energy density [114]. When presented with normal laboratory chow ad libitum , about 50% of sand rats develop obesity (diabetes-prone, DP) and the other half do not (diabetes-resistant, DR). Two-thirds of the DP rats progress to type 2 diabetes. These animals have been selectively bred to produce DP and DR lines that diverge in their responses to a high-energy diet [115, 116]. The DP line responds to the diet, with hyperglycemia, hyperlipidemia, and hyperinsulinemia and a concomitant depletion of b -cell insulin evident after only 4 to 5 days. The diabetic phenotype is reversed by an overnight fast. Isolated islets from DP and DR rats fed either a low-energy or high-energy diet were perfused with glucose to assessb-cell glucokinase (GK) and hexokinase (HK) activity, insulin secretion and insulin sensitivity.b-cell dysfunction in DP rats results from a 4-fold elevation in GK and HK activity and a subsequent increase in insulin sensitivity in response to elevated glucose levels, which quickly depletes > 90% of insulin from theb-cell and induces b-cell apoptosis [116]. Approximately 40% of b-cell insulin is repleted by an overnight fast [115], presumably due to a curtailment of diet-induced insulin release, although insulin release was not measured after the animals were fasted. To date, no genes specifically associated with obesity or diabetes have been identified in these strains.

Wistar Ottawa Karlsburg W rats. The rat strain Wistar Ottawa Karlsburg with RT1^u haplotype W of the major histocompatibility complex MHC (WOKW) was derived by inbreeding an outbred Wistar rat stock to homozygosity at the RT1^u MHC locus. After fifty generations, the result was a rat that displays phenotypic traits reminiscent of syndrome X, including late-onset obesity, impaired glucose tolerance, hypertension and dyslipidemia [117]. QTL mapping studies of WOKW crossed to Dark Agouti/Karlsburg (DA/K) rats have shown a significant linkage of 30-week body weight ( Wokw1/Q1ms1 ) and BMI ( Wokw2/Q1ms5 ) to regions of chromosomes 1 and 5, respectively [118, 119]. The loci predisposing to increased body weight on chromosome 1 differ between males and females [119].

Goto-Kakizaki rats. The Goto-Kakizaki (GK) rat is a model for type 2 diabetes that is characterized by elevated fasting glucose, insulin resistance as early as 2 weeks of age, and defective glucose-stimulated insulin secretion. This strain was developed by selectively intercrossing Wistar rats with abnormal glucose tolerance for several generations. While the animals are not obese, seven QTL associated with a propensity for increased body weight and adiposity have been identified in F2 progeny of GK x BN or F344 rats [120, 121]. Additional QTL for increased body weight and adiposity uncovered in this study were Niddm3 and Weight1 . Four additional QTL associated with adiposity and body weight, Nidd/gk1 , Nidd/gk5 , Nidd/gk6 and Bw/gk1 , have been identified by successive backcrosses of GK rats with the nondiabetic BN rat [121]. The Niddm1 diabetes susceptibility allele was transferred onto a normoglycemic Fischer 344 strain background [120]. Ten generations of backcrossing were performed to reveal significant association of two GK regions with metabolic traits in the resulting F344.GK congenic. Niddm1b on chromosome 1 retains 28 cM of the GK interval, and is associated with fasting hyperglycemia, fasting hyperinsulinemia, increased body weight and fat mass, and dyslipidemia. Niddm1i , also located on chromosome 1, retains 22 cM of the GK interval and is associated with insulin resistance and impaired insulin secretion. Syntenic regions on human chromosomes 9 and 10 are associated with a predisposition to type 2 diabetes in a subset of the Mexican-American population, although no known candidate genes exist at these locations [122].

New Zealand Obese mice. The New Zealand Obese (NZO) mouse line was created in 1948 because of selective inbreeding for adiposity. NZO mice are not hyperphagic, but become increasingly more inactive after 6 weeks of age. Abdominal obesity is apparent as early as 4 weeks of age [123] accompanied by twice the overall body fat of lean controls and impaired glucose tolerance. Later in adulthood, about half of males develop islet cell failure and overt diabetes. When crossed with the lean, atherosclerosis-resistant S/JL strain, a QTL ( Nob1) associated with BMI at 32 cM on chromosome 5 was identified [124]. The NZO leptin receptor allele ( Lepr^A720T/T10441 ) [125] interacts with the Nob1 to increase body weight, possibly by interfering with LEPR-B STAT3 signaling. This is suggested by an increase in the EC ₅₀ of Lepr^A720T/T10441 as compared to that of the wild-type receptor cDNA in COS7 cells (23.7 ± 4.7 vs 18.0 ± 5.9 nM). Crossing NZO with the Small (SM) mouse strain revealed nine new QTL associated with fat pad size ( Obq7-Obq15 ) [126]. Intercrossing NZO/H1Lt mice with NON/Lt mice (which develop late-onset diabetes) revealed two additional QTL potentially involved in the early onset of obesity and diabetes in NZO mice: Nzoq1 , which accounts for 16% of the variance in body weight, and Nzoq2 , associated with increased fat pad mass, serum leptin concentration and BMI [127](Table 3).

Tsumura Suzuki Obese Diabetes mice. The Tsumura Suzuki Obese Diabetes (TSOD) mouse was developed by selective inbreeding of the outbred, nondiabetic ddY strain [128]. The salient phenotypes are obesity and hyperphagia; males develop diabetes. A selectively inbred nondiabetic, obese line (TSNO) was also generated. TSOD mice were crossed with the nondiabetic nonobese mouse strain BALB/cA [129], and in the resulting F2 male progeny, two QTL were found. One, Nidd5 , is associated with both body weight and circulating insulin concentrations in TSOD mice. However, while Nidd5 contributes to increased body weight, the locus is also strongly associated with a relative decrease in circulating insulin concentrations, although TSOD mice are hyperinsulinemic. A second QTL, Nidd6 , appears to have a strong influence on body weight, and a third, Nidd4 , is associated with postprandial hyperglycemia.