Inactivation of Signal Transducer and Activator of Transcription 3 in Proopiomelanocortin (Pomc) Neurons Causes Decreased Pomc Expression, Mild Obesity, and Defects in Compensatory Refeeding

Leptin is an adipocyte-derived hormone that signals body energy status to the brain by acting on multiple neuronal subgroups in the hypothalamus, including those that express proopiomelanocortin (Pomc) and agouti-related protein (Agrp). Signal transducer and activator of transcription 3 (Stat3) is an important intracellular signaling molecule activated by leptin, and previous studies have shown that mice carrying a mutated leptin receptor that abolished Stat3 binding are grossly obese. To determine the extent to which Stat3 signaling in Pomc neurons was responsible for these effects, we constructed Pomc-specific Stat3 mutants using a Cre recombinase transgene driven by the Pomc promoter. We find that Pomc expression is diminished in the mutant mice, suggesting that Stat3 is required for Pomc transcription. Pomc-specific Stat3 female mutant mice exhibit a 2-fold increase in fat pad mass but only a slight increase in total body weight. Mutant mice remain responsive to leptin-induced hypophagia and are not hypersensitive to a high-fat diet; however, mutant mice fail to mount a normal compensatory refeeding response. These results demonstrate a requirement for Stat3 in transcriptional regulation of Pomc but indicate that this circuit is only one of several components that underlie the neuronal response to leptin and the role of Stat3 in that response.

ENERGY BALANCE IS controlled by the coordinated action of hypothalamic neurons, which sense and integrate peripheral signals that report changes in body energy stores. Leptin, an adipocyte-derived hormone circulating at levels proportional to body fat mass, signals energy status to the brain by directly acting on target neurons in the hypothalamus (reviewed in Ref. 1). Located in the arcuate nucleus of the hypothalamus, two groups of neurons, agouti-related protein (Agrp) and proopiomelanocortin (Pomc) neurons, express leptin receptors and exert opposite effects on energy balance. Agrp stimulates feeding, and overexpression of Agrp causes obesity (2). In contrast, Pomc neurons secrete the neuropeptide α-MSH, which decreases food intake and increases energy expenditure. Mutations of the Pomc gene in mice or in humans causes profound obesity (3–7), underscoring the essential role of Pomc in regulation of energy balance.

Several observations indicate that Pomc and Pomc neurons are positively regulated by leptin and/or changes in energy balance. First, leptin causes depolarization of Pomc neurons (8); second, in leptin-deficient ob/ob mice, expression of Pomc is reduced and can be activated by leptin administration (9). Finally, in normal mice, Pomc mRNA levels are reduced during fasting when leptin levels are low and up-regulated during forced overfeeding (10).

One of the best characterized components in leptin signaling is signal transducer and activator of transcription 3 (Stat3). Phosphorylated Stat3 (pStat3) is a potent transcriptional activator of many target genes (11, 12), and leptin administration causes rapid phosphorylation of Stat3 in the hypothalamus (13). Previous studies in transfected cells indicate that leptin induced Pomc promoter activity via Stat3 activation (14). Most important, a neural-specific deletion of Stat3 using a Cre recombinase transgene driven by the Nestin promoter caused severe obesity and reduced expression of Pomc and recapitulated the phenotype of mice deficient in leptin or its receptor (15). Moreover, mice carrying a mutated leptin receptor that abolished Stat3 binding are grossly obese with decreased Pomc but not Agrp and neuropeptide Y (Npy) expression (16). These results have led to a view whereby Stat3-mediated regulation of Pomc transcription constitutes a major component of how the central nervous system responds to leptin. However, more recent studies have revealed a role for additional leptin-responsive neurons outside the arcuate nucleus of the hypothalamus (17, 18) and additional intracellular mediators of leptin besides Stat3 (19–21).

We have previously described the characterization of Cre recombinase transgenes under control of Pomc or Agrp regulatory sequences and the application of these transgenes to study specific signaling molecules or cell types (6, 20). Surprisingly, we found that removal of activatable Stat3 from Agrp neurons had no effect on Agrp mRNA levels (22). Here we have applied a similar strategy to remove activatable Stat3 specifically from the Pomc neurons. We find that mutant mice exhibit reduced levels of Pomc mRNA, indicating that Stat3 is essential for modulating Pomc expression. Mutant female mice exhibit mild obesity with a 2-fold increase in isolated fat pad mass. Moreover, mutant mice remain leptin responsive but exhibit defects in compensatory refeeding. Thus, Stat3 is required for transcriptional regulation of Pomc, but this circuit is only one of several components that underlie the neuronal response to leptin and the role of Stat3 in that response.

Materials and Methods

Mouse genetics

Animals carrying the R26R LacZ reporter allele were obtained from Dr. Phillipe Soriano (University of Washington) and are maintained in our laboratory as homozygotes on a mixed background that includes contributions from FVB/N, C57BL/6J, and 129 strains. The Tg.PomcCre was maintained as heterozygotes on the same mixed background. Construction and application of the Stat3 conditional (Stat3^flox/flox) has been described previously (23). In brief, the loxP sites flank exon 22, which encodes a phosphorylation site that is required for normal Stat3 function. Animals with Pomc-specific Stat3 deletion were generated by intercrossing Tg.PomcCre/+ and Stat3^flox/flox mice. To help control for differences in litter size and modifier genes, results were analyzed by two-way ANOVA in which sibship was included as a variable. Although the genetic background is heterogeneous, there is less variation (both environmental and genetic) within litters than between litters. Animals were housed in a room with a 12-h light (0700–1900 h), 12-h dark cycle. Mice were fed standard mouse chow (14 kcal% fat, 60 kcal% carbohydrates, 26 kcal% protein, ProLab RMH 3000; LabDiet, Richmond, IN). For high-fat feeding, mice at 10 wk of age were placed on a high-fat diet (45 kcal% fat, 35 kcal% carbohydrates, 20 kcal% protein; Research Diets, Inc., New Brunswick, NJ) for a total of 12 wk. All experiments were carried out under a protocol approved by the Stanford Administrative Panel on Laboratory Animal Care.

Body composition and percent adiposity

Body fat mass, lean mass, and bone density of 3.5-month-old mice were determined by dual-energy x-ray absorptiometry (DEXA) using a PIXImus2 instrument (Lunar Corp., Madison, WI). The machine was calibrated using the Phantom mouse supplied by the manufacturer, which has a Phantom value of 16.1% fat and 0.0624 g/cm² bone density. Quality control was performed to fit within ±2% of the standard value before each experiment. Percent adiposity was determined by adding the weights of the scapular, inguinal, retroperitoneal, and reproductive fat pads and dividing by body weight. Percent adiposity was determined from nine Pomc-Stat3 mutants and six control females (age, ∼4–6 months) and four males of each genotype (age, ∼6–8 months).

Measurement of serum circulating leptin levels

For leptin measurements, plasma was collected from mice fed ad libitum. The samples were centrifuged, plasma was removed, and levels were determined using a mouse leptin ELISA (Crystal Chem, Downers Grove, IL). Circulating leptin levels were determined from nine Pomc-Stat3 mutants and six control females (age, ∼4–6 months) and five Pomc-Stat3 knockout and four control males (age, ∼6–8 months).

Gene expression analysis

Gene expression analysis was performed as described previously (6). Measurements of mRNA levels were carried out by quantitative RT-PCR on RNA extracted from dissected hypothalamic tissue isolated from 3.5-month-old mice. Total RNA for each hypothalamus was quantified by spectrophotometry after purification using TRIzol reagent (Invitrogen, Carlsbad, CA) and an RNeasy mini kit (QIAGEN, Valencia, CA). Then, 250 ng of each total RNA sample was reverse transcribed and PCR amplified using a Lightcycler (Roche Applied Science, Indianapolis, IN) and SYBR green to measure relative cDNA levels. All neuropeptide expression was normalized by β-actin expression. Agrp primers were TGCTACTGCCGCTTCTTCAA and CTTTGCCCAAACAACATCCA; Npy primers were TAACAAGCGAATGGGGCTGT and ATCTGGCCATGTCCTCTGCT; Pomc primers were AGGCCTGACACGTGGAAGAT and AGGCACCAGCTCCACACAT; Cart primers were GTGCTTGTGAAGGGACGACA and AGCCAGGCTCCAGGGATAAT; and Actb primers were CTGCGTTTTACACCCTTTCTTTG and GCCATGCCAATGTTGTCTCTTAT. Efficiency for each primer set was estimated from standard curves made with serial cDNA dilutions.

X-gal staining and immunohistochemistry

Colocalization studies were performed as previously described (20). Briefly, weight-matched mice were injected ip with leptin (3.5 mg/kg), and the animals were perfused with 4% paraformaldehyde 45 min later. Brains were removed, placed in fixative overnight, and infiltrated with 30% sucrose in PBS at 4 C, and then 12-μm frozen sections were prepared and mounted on Superfrost/Plus slides. Sections were washed with 0.1% Triton X-100 in PBS and then incubated with X-gal staining buffer (5 mm potassium ferricyanide, 5 mm potassium ferrocyanide, 2 mm MgCl₂, and 1 mg/ml X-gal) overnight at 37 C. After X-gal staining, sections were washed and immunostained with a phospho-Y705 Stat3 antibody (Cell Signaling Technology, Beverly, MA) according to the manufacturer’s protocols.

Leptin sensitivity assay

The mice (5 months old) were singly housed for greater than 1 wk before the start of the experiment. One week before leptin injections, all animals were injected once a day at 1500 h with PBS vehicle. Body weight and food intake were measured at this time. On d 1 of the experiment, food was removed from the cages at 1300 h. At 1500 h, mice were injected ip with either 5 mg/kg leptin (R&D Systems, Minneapolis, MN) or vehicle (half of each genotype got vehicle, and half got leptin). The next morning, mice received another injection of either leptin or vehicle at 0800 h. At 0900 h, food was returned to the hopper and food intake measurements were taken 2, 4, 8, and 24 h later. Body weight was measured 24 h later. The experiment was repeated after 1 wk. The animals that received vehicle in the first round received leptin in the second round and vice versa. Mice were housed in a room with a 12-h light (0700–1900 h), 12-h dark cycle.

Results

Pomc-Stat3 knockout mice do not show phosphorylation of Stat3 in the Pomc neurons after leptin administration

Mice carrying Tg.PomcCre, previously employed and characterized in multiple studies (6, 20, 24, 25), were crossed with animals carrying Stat3^flox, a gene-targeted Stat3 allele with loxP sites that flank exon 22, which encodes a tyrosine residue whose phosphorylation is critical for Stat3 dimerization and nuclear translocation (23).

We first investigated the ability of leptin to induce phosphorylation of Stat3 in control and Pomc-specific Stat3 mutant mice. Control (Tg.PomcCre/+; Stat3^+/+) and mutant (Tg.PomcCre/+; Stat3^flox/flox) animals were crossed to mice harboring the Cre reporter allele Gt(Rosa)26Sor^tm1Sor (R26R). LacZ expression of the R26R allele is activated upon Cre-mediated recombination, and therefore serves as a cell-autonomous marker to mark the Pomc neurons in histological studies. Forty-five minutes after ip administration of leptin, animals were euthanized, and hypothalamic sections were double stained for β-galactosidase activity and immunohistochemical presence of pStat3.

In this preparation, a perinuclear dot of blue X-gal staining marks Pomc neurons, whereas brown staining of the entire nucleus signifies pStat3 immunostaining (Fig. 1, E and F). As expected, leptin-treated animals exhibited nuclear pStat3 accumulation in many neurons throughout the hypothalamus similar to what was reported before (14, 26); close inspection of the arcuate nucleus reveals that about 39% of the Pomc neurons are positive for pStat3 upon leptin treatment (Fig. 1, E and G). However, in Pomc-specific Stat3 mutant mice, Pomc neurons failed to stain with pStat3 (Fig. 1F). The pattern of pStat3 staining outside the arcuate nucleus was identical in control and Pomc-specific Stat3 mutant mice, demonstrating failure of leptin-induced pStat3 specifically in the Pomc neurons.

Weight gain, growth, and feeding abnormalities in Pomc-specific Stat3 mutant mice

To generate a large cohort of mutant and control mice for physiological studies, we first established a breeding colony of animals homozygous for the Stat3^flox allele among which segregation of the Tg.PomcCre allele determined assignment to control (+/+; Stat3^flox/flox) or mutant (Tg.PomcCre/+; Stat3^flox/flox) group. In separate studies (data not shown), we found that neither Tg.PomcCre nor the Stat3^floxby itself had any effect on body weight regulation, food consumption, linear growth, or body composition.

Body weight gain over a 6-month period was the same in male mutant compared with male control mice (Fig. 2A). Mutant females, however, exhibited a 5–10% increase in body weight compared with control females, which first became apparent at 2 months of age (Fig. 2B). Using DEXA to measure body composition, mutant females, but not males, showed a modest but significant increase in fat mass, with no apparent change in lean body mass (Fig. 2, C and D). To gain more information about fat distribution, scapular, inguinal, reproductive, and retroperitoneal fat pads were isolated. Male mutant mice did not exhibit a significant increase in any of the fat depots examined (Fig. 2E). However, inguinal, reproductive, and retroperitoneal fat pads showed a significant 2-fold increase in female Pomc-Stat3 mutants (Fig. 2F), with a nearly 2-fold increase in percent adiposity (Fig. 2G). Both females and males exhibited a trend in which mutant animals were slightly longer than control animals, but these changes were not significant (supplemental Fig. 1, published as supplemental data on The Endocrine Society’s Journals Online web site at http://endo.endojournals.org). In addition, bone density was identical between controls and mutants (supplemental Fig. 1).

To better understand the basis for mild obesity in female Pomc-specific Stat3 mutant mice, we measured both daily food consumption (averaged over 7–14 d) and ability to mount a compensatory refeeding response after a 48-h fast. Under ad libitum feeding, mutant females showed a mild but significant increase in food intake (Fig. 3A). Food intake was no different between mutant and control male mice (data not shown). When allowed to refeed after a 48-h fast, mutant and control female mice consumed the same amount of food in a 24-h period (Fig. 3B) but showed a slight decrease when compared with their normal food intake (Fig. 3C). However, control mice normalized their body weight after 24 h of refeeding, whereas mutant mice were unable to do so (Fig. 3D).

Neuropeptide gene expression in Pomc-specific Stat3 mutant mice

Next, we measured neuropeptide mRNA levels in hypothalamic biopsies of normal and Pomc-specific Stat3 mutant mice using real-time quantitative PCR. Similar to the sexual dimorphism in body weight described above, levels of hypothalamic Pomc mRNA were significantly reduced in mutant females but not in mutant males (Fig. 4). We also measured expression of Cart, Agrp, and Npy mRNA levels and observed no difference between control and mutant animals in either males or females (Fig. 4).

Sensitivity to leptin and high-fat diet in Pomc-specific Stat3 mutant mice

To examine whether Pomc-Stat3 mutant mice have reduced sensitivity to leptin, we first measured circulating leptin levels in these mice. Neither male nor female mutant mice showed significant increase in serum leptin levels, although there is a trend in the female mutants (Fig. 5A). To test directly whether or not Stat3 signaling in Pomc neurons was required for leptin-induced anorexia, we examined the time course of food consumption and weight gain in mutant and control female animals after leptin treatment. We used a paradigm and crossover study design in which the same animals received either leptin or vehicle 1 wk apart. Leptin treatment reduced food intake by approximately 20% in both control and mutant mice; this effect was detectable by 2 h and remained apparent up to 24 h after leptin treatment (Fig. 5B). Hence, Pomc-specific Stat3 mutant mice maintain their ability to respond to leptin.

Finally, we investigated whether disruption of Stat3 function in the Pomc neurons predisposes mice to diet-induced obesity. Mice were placed on a high-fat diet (45 kcal% fat) for a total of 12 wk. No significant difference in body weight increase was observed in male or female mutant mice (Fig. 5, C and D) compared with wild-type and heterozygous controls.

Discussion

Mutation of the Stat3 binding site in the leptin receptor produced mice that are grossly obese (16), suggesting that Stat3 function is essential in mediating leptin signaling directly in first-order leptin target neurons. Pomc and Agrp/Npy neurons in the arcuate nucleus of the hypothalamus are key first-order neurons that express leptin receptors. Pomc expression, but not Agrp/Npy expression is reduced in mice carrying the mutated leptin receptor with no Stat3 signaling, suggesting that the gross obesity phenotype could be attributed to lack of Stat3 function in the Pomc neurons (16). In this study, we showed that removal of activatable Stat3 from Pomc neurons causes reduced levels of Pomc mRNA and increase in adiposity in a sex-specific manner. In recent studies analogous to those described here where we removed activatable Stat3 from Agrp neurons, we observed no effect on Agrp mRNA levels (22). Thus, there exists an interesting dichotomy for the role of Stat3 in transcriptional regulation of arcuate nucleus neuropeptide, in that it is required for normal regulation of Pomc but not of Agrp or Npy. The reduction of Pomc mRNA levels in the mutant mice described here is likely to be an important contributor to their abnormal phenotype, because both the molecular and the physiological abnormalities were present in females but not in males. Hence, our result that Pomc, but not the Agrp expression, is regulated by leptin-mediated Janus kinase (Jak)-Stat3 signaling is consistent with what was observed in mice with a mutated leptin receptor that is incapable of Jak-Stat3 signaling (16). The differential requirement of Jak-Stat3 signaling by leptin in the regulation of gene expression suggests that different signaling mechanisms are engaged in these two neuronal subtypes, or that compensatory mechanisms develop in a cell-type-specific fashion.

The mechanism of sex-dimorphic regulation of Pomc expression by Stat3 is currently unknown. However, a gender-specific response to leptin has been described, in that females are more sensitive to the anorexic effects of leptin than males and that such differences are controlled by gonadal hormones (27, 28). So, it is possible that sex hormones may interact with leptin signaling to influence Pomc expression.

We have previously shown that Pomc neurons, but not the Agrp neurons, are required for normal compensatory refeeding (6). The Pomc-specific Stat3 mutant mice described here exhibit similar defects in compensatory refeeding, which suggests that modulation of Stat3 contributes to the mechanism by which Pomc expression is regulated in response to negative energy balance. It is interesting to note that Pomc-specific Stat3 mutant mice have a normal response to leptin, whereas deletion of suppressor of cytokine signaling-3 (Socs3) from the Pomc neurons results in increased leptin sensitivity (29). It is possible that Socs3 might inhibit other signaling pathways of leptin in addition to the Jak-Stat3 pathway. Socs3 has been shown to induce insulin resistance in peripheral tissue via inhibition of tyrosine phosphorylation of insulin receptor substrate (IRS)-1 and -2 (30). Because IRS-phosphatidyinositol-3-kinase signaling has been implicated in leptin signaling (19, 20), it is possible that deletion of Socs3 leads to increased leptin signaling via the Jak-Stat3 and IRS-phosphatidyinositol-3-kinase pathways. Deletion of Socs3 could potentially lead to increased insulin sensitivity as well.

Recent studies have investigated the involvement of Pomc neurons in the nucleus of the solitary tract (NTS) of the caudal brainstem. One study has shown that leptin does not stimulate pStat3 in Pomc neurons in the NTS (31), whereas a different study has reached the opposite conclusion (32). Our preliminary results show that Cre is expressed in the NTS of the Pomc-Cre mice (6), suggesting that Stat3 is also abolished in the Pomc neurons in the NTS. However, the mild phenotype in Pomc-Stat3 knockout mice (especially in the males) suggests that Stat3 in the NTS Pomc neurons does not play a major role in energy balance regulation.

In light of the dominant view in the field that Pomc neurons are primary first-order neurons that mediate leptin function, it is somewhat surprising that deletion of Stat3 function from the Pomc neurons results in only a mild phenotype, and even those are apparent only in female animals. These mild effects are in sharp contrast with the neural-specific Stat3 deletion (15), which develop severe obesity, hyperglycemia, and infertility. These results are consistent with a modified view of central control of energy homeostasis whereby leptin signals through multiple, possibly redundant, groups of neurons in several regions of the brain. Indeed, this view is consistent with recent work of Balthasar et al. (33) and Dhillon et al. (17), in which cell-type-specific removal of the leptin receptor in a subpopulation of neurons in the hypothalamus caused mild obesity.

An alternative hypothesis to account for the surprisingly mild phenotype we observed is to posit that Pomc neurons do, indeed, serve as a primary conduit for leptin signaling but that the signaling itself is transmitted via intracellular mediators in addition to Stat3. However, this seems less likely given the phenotype of animals with Pomc-specific deletion of the leptin receptor, in which one of the most surprising aspects is the relatively mild extent of their obesity (33). In this respect, an important puzzle is why removal of the Pomc gene itself causes severe obesity (4, 5, 7), much greater than that caused by removal of Stat3 or the leptin receptor from Pomc neurons. Recent studies have revealed several novel mechanisms involved in central nervous system regulation of energy homeostasis, including changes in electrical activity (8, 34), modulation of synaptic plasticity (35), and de novo neurogenesis (36). In addition, a simple explanation for the apparent discrepancy between the severe phenotype caused by removal of Pomc and the mild phenotypes caused by removal of Pomc-specific leptin signaling machinery could be that Pomc neurons receive leptin input indirectly from other leptin-responsive neurons in several brain regions. This latter view, in which non-Pomc leptin-responsive neurons might feed through Pomc neurons, is consistent with a role of Agrp/Npy neurons in providing inhibitory GABA input onto Pomc neurons (8, 34) and with the recent discovery of a circuit from the ventromedial hypothalamus to Pomc neurons (18). Neuron-specific disruption of additional signaling molecules will help dissect the unique function of these molecules in individual neuronal subtypes and will provide insight into the multitude of neuronal circuits in the regulation of energy homeostasis.

We thank Dr. Akira at Osaka University for providing the Stat3^flox/flox mice.

This work was supported by grants from the National Institutes of Health to G.S.B. (DK48506 and DK68384).

Disclosure Statement: The authors have nothing to disclose.

Abbreviations:

 
• Agrp,

  Agouti-related protein;

 
• DEXA,

  dual-energy x-ray absorptiometry;

 
• IRS,

  insulin receptor substrate;

 
• Jak,

  Janus kinase;

 
• Npy,

  neuropeptide Y;

 
• NTS,

  nucleus of the solitary tract;

 
• Pomc,

  proopiomelanocortin;

 
• pStat3,

  phosphorylated Stat3;

 
• Socs3,

  suppressor of cytokine signaling-3;

 
• Stat3,

  signal transducer and activator of transcription 3.