Report

Independent and Additive Effects of Central POMC and Leptin Pathways on Murine Obesity

See allHide authors and affiliations

Science  28 Nov 1997:
Vol. 278, Issue 5343, pp. 1641-1644
DOI: 10.1126/science.278.5343.1641

Abstract

The lethal yellow(AY/a) mouse has a defect in proopiomelanocortin (POMC) signaling in the brain that leads to obesity, and is resistant to the anorexigenic effects of the hormone leptin. It has been proposed that the weight-reducing effects of leptin are thus transmitted primarily by way of POMC neurons. However, the central effects of defective POMC signaling, and the absence of leptin, on weight gain in double-mutant lethal yellow(AY/a) leptin-deficient (lepob/lep ob) mice were shown to be independent and additive. Furthermore, deletion of the leptin gene restored leptin sensitivity toAY/a mice. This result implies that in the AY/a mouse, obesity is independent of leptin action, and resistance to leptin results from desensitization of leptin signaling.

Serum concentrations of the hormone leptin (1) correlate well with body mass index in both humans and rodents (2). The long form of the leptin receptor has been detected in multiple hypothalamic regions including the arcuate nucleus (3), where it completes a feedback loop that sends information on peripheral energy stores to the hypothalamus, altering both food intake and metabolic rate. Absence of leptin results in extreme obesity in the obese(lepob /lepob ) mouse.

Obesity in another rodent model, the lethal yellow(AY/a) mouse (4), is caused by a dominantly inherited promotor rearrangement at theagouti locus that results in constitutive ectopic expression of the Agouti peptide (5). Agouti is a potent antagonist of the hypothalamic melanocortin-4 receptor (MC4-R) (6), and interruption of signaling at MC4-R increases feeding behavior in mice (7-9). Thus, desacetyl-α–melanocyte-stimulating hormone derived from arcuate nucleus POMC neurons, the primary source of ligand for MC4-R, appears to play a tonic inhibitory role in feeding and energy storage. Leptin levels in theAY/a mouse, as well as other rodent obesity models, are elevated (2), reflecting the increase in adipose tissue. The AY/a mouse is also resistant to leptin administered peripherally or intracerebroventricularly (10). Peripheral resistance to leptin occurs in other obese-rodent models (2,11), and it has been argued that both obesity in theAY/a mouse as well as common forms of human obesity may result from genetically determined resistance to leptin feedback (10, 12)

To study the relation between leptin and POMC signaling pathways on weight homeostasis and the potential dependence of leptin action on POMC signaling, we generated a lethal yellow obese(AY/a lepob/lep ob) double-mutant mouse. C57BL/6J mice with either theAY/a orlepob/+ genotypes were crossed to create AY/a lepob/+ breeders. These animals were bred to create the double-mutant AY/a lepob/lep ob mice, which were identified by their yellow coat color and obese phenotype, and their genotype was confirmed by allele-specific oligonucleotide hybridization (13).

If defective POMC signaling in theAY/a animal causes obesity solely by blocking the anorexigenic leptin signal, then introduction of theA Y allele into the leptin-deficientlepob/lep obbackground should have no added effect on weight gain or metabolism in this model. However, because of the extreme rate of weight gain that results from leptin deficiency, the modest effects of theA Y allele on weight might be difficult to detect in the intact lepob/lep obanimal. Leptin deficiency in thelepob/lep obmouse markedly increases glucocorticoid levels, which indirectly are reponsible for a large percentage of the obesity phenotype in these animals. To examine the direct effects of leptin and POMC pathways in the brain, we first measured weight gain in animals adrenalectomized and then maintained on normal levels of corticosterone.

The weight gain of female mice from each genotype, fed normal Chow ad libitum, was measured over a period from 6 to 31 weeks after birth. Before the experiment, animals were adrenalectomized and placed on maintenance glucocorticoids supplied in drinking water (14). The presence of the A Y allele increased weight gain to a similar extent in both the wild-type and leptin-deficientlepob/lep obbackgrounds, indicating that the obesity-inducing actions of defective POMC signaling are leptin-independent (Fig.1A). Increased linear growth, another phenotype induced by the A Y allele (15), did not appear to explain the difference in weight, because the adrenalectomized AY/a ,AY/a lepob/lep ob , andlepob/lep obmice all demonstrated the same 10% increase in linear growth relative to control a/a mice (16).

Figure 1

The effects of leptin deficiency and Agouti antagonism of POMC signaling on weight gain and serum insulin concentrations. (A) Weight curves. (B) Fasting serum insulin and (C) weight versus serum insulin scatter plot. All data are from adrenalectomized C57BL/6J female wild-type (a/a +/+) (n = 7), lethal yellow(AY/a +/+) (n = 7),obese (a/a lepob/lep ob) (n = 7), and double-mutant lethal yellow obese(AY/a lepob/lep ob) (n = 5) mice (17). Data are reported as the mean ± SE. Weight curves were compared by two-way analysis of variance (ANOVA). All curves were significantly different (P < 0.0001). Serum insulin concentrations in lethal yellow obese mice were compared with those in obese mice by the two-tailedt test (P < 0.05 at 4 months and P< 0.01 at 5, 6, and 7 months). Linear regression of insulin versus weight in the obese mice was significant (P = 0.02) with r = 0.46. Linear regression of insulin versus weight in lethal yellow obese mice was not significant.

Fasting serum insulin and glucose concentrations were determined monthly during the course of the experiment in the adrenalectomized female mice (17). A Y alone produced a mild late-onset hyperinsulinemia (18, 19), whereas leptin deficiency produced an early, more significant rise in serum insulin. The effects of defective POMC signaling and leptin deficiency on serum insulin also appeared to be additive (Fig. 1B). Fasting serum glucose concentrations in theAY/a lepob/lep oband a/a lepob/lep ob mice, however, were not significantly different (16). When mice of similar weights were compared (Fig. 1C), the serum insulin concentrations in the AY/a lepob/lep ob were increased compared with those in a/a lepob/lep ob animals. Additionally, a/a lepob/lep ob animals demonstrated a linear relation between weight and serum insulin, suggesting that the increase of insulin levels in the absence of leptin is partially a function of the increased adipose mass. In contrast, introduction of the A Y allele into thelepob/lep obbackground removed any significant correlation between weight and insulin levels, suggesting that Agouti inhibition of the POMC signal causes a dysregulation of insulin by a second mechanism independent of adipose mass.

If the A Y allele induces obesity in a leptin-independent manner, implying that MC4-R signaling is not required for the central anorexigenic actions of leptin, then why areAY/a mice resistant to an increase in either endogenous (2) or exogenous leptin (10)? To test the hypothesis that MC4-R signaling is not required for the weight-reducing action of leptin, we examined the effects of leptin administration in the four mouse genotypes. After a 4-day course of saline, the same adrenalectomized female mice used for analysis of weight gain were injected twice daily with human leptin (20). Wild-type and AY/amice were resistant to leptin relative to thelepob/lep obmice, as measured by the inability of leptin to induce weight loss (Fig. 2A), lower serum insulin (Fig. 2B), and decrease food intake (Fig. 2C). In contrast, absence of the leptin gene restored full leptin sensitivity to theAY/a lepob/lep ob mice, as demonstrated by use of all three measures (Fig. 2, A to C).

Figure 2

Effect of leptin administration on leptin-deficient lethal yellow obese(AY/a lepob/lep ob) mice. (A) Weight change from baseline (day −4) in adrenalectomized 8-month-old female mice from the colony in Fig. 1. Mice were given subcutaneous saline injections twice daily until day 0, and were then given leptin (2 mg/kg) twice daily. Symbols for the different phenotypes are as in Fig. 1. (B) Fasting serum insulin concentrations before and after leptin administration. (C) Average daily food intake before (days −4 to −1) and after (days 1 to 5) leptin administration. Data are reported as the mean ± SE. Weight change from baseline curves was compared by two-way ANOVA. All curves were significantly different (P< 0.001) with the exception of AY/a lepob/lep obversus a/a lepob/lep ob andAY/a +/+ versusa/a +/+ (P > 0.01). The effect of leptin on serum insulin and food intake was analyzed by the one-tailedt test (pre- versus postleptin administration: *P < 0.05, **P < 0.01, ***P ≤ 0.001). The number of mice used in these experiments is as follows:a/a +/+ (n = 3),AY/a +/+ (n= 1), a/a lepob/lep ob(n = 3), and AY/a lepob/lep ob (n = 3).

To examine the effect of circulating glucocorticoids, gender, and age on leptin sensitivity in the AY/a lepob/lep ob mouse, we administered leptin to young nonadrenalectomized mice (Fig.3A). Restoration of leptin responsiveness by deletion of the leptin gene in theAY/a mouse was independent of adrenal status. Whereas full leptin responsiveness was restored in adrenalectomized and normal female AY/a lepob/lep ob mice, leptin responsiveness was only partially restored in maleAY/a lepob/lep obanimals. In agreement with previously reported data (10), both male and female AY/a mice exhibited greater leptin resistance than C57BL/6J controls.

Figure 3

Effect of leptin administration in young male and female nonadrenalectomized leptin-deficient lethal yellow mice (AY/a lepob/lep ob). (A) Weight change from baseline (day −2) in 3-month-old male (open symbols) and female (closed symbols) mice. Symbols for the different phenotypes are as in Fig. 1. Mice were given subcutaneous saline injections twice daily until day 0, and were then given leptin (2 mg/kg) twice daily. (B) Fasting serum insulin concentrations in female mice before and after leptin administration. (C) Fasting serum insulin concentrations in male mice before and after leptin administration. (D) Serum corticosterone concentrations before and after leptin administration (26). Data are reported as the mean ± SE. Weight change from baseline curves was compared within gender by two-way ANOVA. All curves were significantly different (P < 0.005) with the exception of female A y /a lep ob/lep ob versus a/a lepob/lepob and maleA Y/a +/+ versus a/a +/+ curves (P > 0.01). The effect of leptin on serum insulin and corticosterone concentrations was analyzed by the one-tailedt test (pre- versus postleptin administration: *P < 0.05, **P < 0.01, ***P < 0.001). Postleptin serum insulin concentrations in the maleAY/a lepob/lep ob mice were compared with those in a/a lepob/lep ob mice by the two-tailed t test (P < 0.05). The number of mice used in all experiments except male serum insulin is as follows: a/a +/+ (n = 3),AY/a +/+ (n= 3), a/a lepob/lep ob(n = 3), and AY/a lepob/lep ob (n = 3). The number of mice used in the male serum insulin experiment is as follows: a/a +/+ (n = 6),AY/a +/+ (n= 6), a/a lepob/lep ob(n = 6), and AY/a lepob/lep ob (n = 6).

Leptin administration also produced a reduction in serum insulin concentrations in AY/a lepob/lep ob andlepob/lep obyoung females by a factor of 15 (Fig. 3B) and inAY/a lepob/lep ob andlepob/lep obmales by a factor of 5 and 15, respectively (Fig. 3C), indicating a significant restoration of leptin sensitivity in nonadrenalectomizedAY/a lepob/lep ob mice. However, the serum insulin concentration in leptin-treatedlepob/lep obmales was significantly lower than that in leptin-treatedAY/a lepob/lep ob males. This observation suggests there is a male-specific leptin-independent pathway for regulation of insulin by POMC neurons. Alternatively, theA Y allele may cause a minor defect in this specific leptin response in the male only.

AY/a mice have normal concentrations of serum corticosterone (19, 21), whereaslepob/lepobmice have increased basal corticosterone levels. The AYallele had no effect on basal serum corticosterone in thelepob/lepobbackground (Fig. 3D). Furthermore, leptin administration reduced corticosterone to normal levels in both thelepob/lepoband AY/a lepob/lepob mice. Adrenalectomy revealed that increased glucocorticoids due to the absence of leptin are responsible for much of the hyperinsulinemia in thelepob/lepobmice (compare Fig. 1B to Fig. 3, B and C). The elevated glucocorticoid levels also increased fat deposition, as 3-month-old nonadrenalectomizedlepob/lepobmice were ∼60% heavier than adrenalectomizedlepob/lepobmice (16).

These data demonstrate that the A Y gene, and by inference disruption of central POMC signaling, does not cause obesity by acting as a genetic roadblock to the central anorexigenic action of leptin. The ability of Agouti to induce weight gain irrespective of the leptin state of the animal argues strongly that the POMC neurons can act independently of leptin in their actions on energy homeostasis. It remains likely that some aspect of leptin function not tested here is mediated by POMC neurons, given that mRNA encoding the long form of the receptor is expressed in some POMC-containing arcuate nucleus neurons (22). Furthermore, a recent report shows that, within a narrow dose range, an antagonist of the MC3 and MC4 receptors is able to block the acute inhibition of feeding by leptin in the rat (23). However, the results of the genetic studies shown here argue that normal POMC signaling is not required for the long-term ability of leptin to reduce weight, serum insulin, or serum corticosterone. The reduced sensitivity to leptin-induced weight loss and reduction in serum insulin in the maleAY/a lepob/lep ob animal indicates the existence of a minor, sexually dimorphic defect in leptin signaling resulting from Agouti inhibition of MC4-R. Alternatively, the data could simply highlight a male-specific effect of POMC signaling downstream and independent of leptin action, reflected also by the observation that AY/a males gain weight faster and are more hyperinsulinemic thanAY/a females (19).

Apparent leptin resistance is a hallmark of obesity in multiple species (10, 11); however, our data suggest that it may be erroneous to assume that leptin resistance is indicative of genetic defects blocking leptin action. Rather, the data presented for theAY/a animal show that removal of leptin from this strain restores complete leptin sensitivity, strongly arguing that animals are leptin resistant as a consequence of desensitization to further leptin action. The absence of leptin in thelepob/lep obanimal is responsible for the deregulation of two genes expressed in the arcuate nucleus: neuropeptide Y (24) and Agouti-related transcript, a newly described brain homolog of Agouti (8,9). The expression levels of both these genes in theAY/a animal (9,25) further indicates a normal sensing of leptin in the arcuate nucleus of these mice. Given the apparent independence of the POMC and leptin pathways with regard to energy and insulin homeostasis, additional work will be required to determine which peripheral or central signals are dependent on the POMC neurons for their integration into the energy homeostasis equation.

  • * To whom correspondence should be addressed. E-mail: cone{at}ohsu.edu

REFERENCES AND NOTES

View Abstract

Stay Connected to Science

Navigate This Article