LRb-STAT3 Signaling Is Required for the Neuroendocrine Regulation of Energy Expenditure by Leptin

All animal experimentation was conducted in accordance with mandated standards of humane care approved by the institution animal care and use committee (Joslin Diabetes Center). All mice were housed in the Joslin Diabetes Center’s accredited mouse barrier facility on a 12-h light/dark cycle. All mice had ad libitum access to standard chow (Purina autoclavable laboratory mouse breeder chow) and water. C57BL/6 (backcrossed n = 6 generations) mice heterozygous for the lepr^s1138 mutation were bred to generate s/s and wild-type littermates for study. The db/db mice were obtained by breeding C57BL/6 db/+ heterozygotes (the lepr^db mutation used in this study is a splice mutation that specifically abrogates the expression of LRb) (30–32) obtained from The Jackson Laboratory (Bar Harbor, ME).

The 6-week-old male ad libitum–fed mice (wild type, s/s, and db/db) were briefly anesthetized with Avertin (tribromoethanol:tert-amyl alcohol, 0.015 ml/g i.p.) and subjected to dual-energy X-ray absorptiometry (DEXA; Lunar PIXImus2 densitometer; GE Medical Systems, Madison, WI) for the analysis of fat and lean body mass.

The 6-week-old male mice were housed individually with ad libitum access to food and evaluated for ambulatory activity using an OPTO-M3 sensor system (comprehensive lab animal monitoring system, or CLAMS; Columbus Instruments, Columbus, OH). A score of ambulatory activity was determined by consecutive photo-beam breaks occurring in adjacent beams. Beam breaks were counted every hour during a full light/dark cycle.

Indirect calorimetry was measured in 6-week-old male mice using an open-circuit Oxymax system (CLAMS; Columbus Instruments). Mice were individually housed in plexiglass cages through which air of known O₂ concentration was passed at a constant flow rate. After a 48-h acclimation period, exhaust air was sampled for 24 h in the fed state for the determination of O₂ and CO₂.

The core body temperature of 8-week-old ad libitum–fed male mice was measured at room temperature at 10:00 a.m. using a rodent telethermometer (Thermolyne; Fisher) equipped with a rectal probe, which was inserted 1 cm into the rectum and read after stabilizing for 5 s. Mice were then subjected to cold by maintenance in the bottom of a 2-l glass beaker submerged in ice for 75 min without food. Body temperature was assessed by removing the animals from the beaker (for <1 min) to measure rectal temperature at 30, 45, 60, and 75 min after the initiation of the study, and mice were killed at the end of the study or if the core body temperature decreased to <25°C.

For measurement of basal UCP1, 8-week-old male ad libitum–fed mice of genotypes +/+, s/s, and db/db (that had been housed at 25°C) were killed by CO₂ inhalation at 10:00 a.m. and interscapular BAT pads were removed and immediately snap-frozen. The frozen fat pads were homogenized in lysis buffer (20 mmol/l Tris, pH 7.4, containing 137 mmol/l NaCl, 2 mmol/l EDTA, 10% glycerol, 50 mmol/l β-glycerophosphate, 50 mmol/l NaF, 1% Nonidet P40, 2 mmol/l phenylmethylsulphonyl fluoride, and 2 mmol/l sodium orthovanadate) using a polytron homogenizer, and they were allowed to solubilize on ice for 1 h. Insoluble material was removed by centrifugation at 16,000g at 4°C for 15 min. Protein concentrations of the resulting lysates were determined, and equivalent amounts of protein were subjected to SDS-PAGE and immunoblot analysis with anti-UCP1 antibody (Santa Cruz Biotechnology, Santa Cruz, CA) with detection by ¹²⁵I-labeled protein A essentially as described (28).

Free T4 was determined in 8-week-old male ad libitum–fed mice killed by CO₂ inhalation at 10:00 a.m. Trunk blood was collected and centrifuged at 16,000g at 4°C. Serum was collected and frozen at −20°C. Free T4 was assessed by radioimmunoassay (radioimmunoassay-coated tube assay; ICN, Costa Mesa, CA).

Disruption of the LRb-STAT3 signal in s/s mice causes dramatic hyperphagia and obesity similar to that of db/db animals (29). Although the obese phenotype of db/db mice is known to be secondary to decreased energy expenditure as well as increased feeding, the relative contributions of hyperphagia and decreased energy expenditure to the phenotype of s/s animals have not been examined. While displaying dramatically increased body weight and body fat compared with wild-type (+/+) animals, s/s mice weigh slightly less and have a slightly lower body fat percentage by DEXA compared with db/db mice, which are devoid of all LRb signals (Fig. 1A and B). The slightly increased lean body mass of db/db compared with +/+ animals is consistent with the addition of lean (muscle, etc.) mass required to support and move the large adipose mass of the db/db animal (Fig. 1C). The increased lean body mass of s/s animals compared with both +/+ animals and the slightly heavier db/db animals is consistent with activation of the neuroendocrine growth axis (29). The lean mass data presented include bone mineral mass, but mineral constitutes ∼1.5% of total lean mass and varies only ∼20% among the various genotypes, thus the inclusion or exclusion of bone mineral in the lean mass calculation did not alter the observed differences in lean mass among the genotypes (data not shown). The slightly decreased body weight and percent body fat of s/s compared with db/db animals could result from increased metabolic rates/less efficient energy metabolism in s/s compared with db/db animals; we thus tried to determine the role of the LRb→STAT3 signal in the regulation of energy expenditure and its determinants in s/s and db/db animals.

To quantitatively examine parameters of energy utilization in these animals, we subjected +/+, s/s, and db/db animals to analysis in a comprehensive lab animal monitoring system, in which activity is continuously monitored by the breaking of light beams, and the energy utilization/metabolic rate is monitored by the consumption of oxygen (O₂) and the production of carbon dioxide (CO₂) in a closed cage system. Although this apparatus can also be used for the continuous monitoring of food intake, s/s and db/db mice are too obese to enable use of the monitored feeding system in this device; we were thus unable to assess feeding in this manner. We found that beam-break activity was decreased by ∼60–70% in s/s and db/db compared with +/+ mice (Fig. 2); s/s mice displayed slightly increased activity compared with db/db animals, however, suggesting that voluntary locomotor activity is less impaired in s/s compared with db/db animals.

Analysis of O₂ consumption and CO₂ production in these animals revealed no alterations in respiratory quotient among the three genotypes (data not shown), presumably because of the continuous presence of the same chow diet in each case. O₂ consumption data are often normalized per animal or per unit of body mass, but each of these methods skews comparisons among animals with different masses of metabolically active (e.g., lean tissue such as muscle and kidney) and metabolically quiet (e.g., adipose, bone) tissues; we have thus normalized O₂ consumption for each animal to lean tissue mass, derived by DEXA analysis as in Fig. 1C. It should be noted, however, that normalization by total body mass or animal did not substantially alter the statistical significance of differences among the various genotypes (data not shown). Resting (Fig. 3A) and total 24-h (Fig. 3B) O₂ consumption were reduced by 30–40% in s/s and db/db animals compared with +/+ mice, consistent with increased metabolic efficiency in these animals. Neither total 24-h nor resting O₂ consumption were different between s/s and db/db animals, suggesting that the LRb→STAT3 signal is central to the regulation of metabolic rate and energy expenditure by leptin.

The preceding results suggest that the components of the neuroendocrine regulation of energy expenditure should be similarly impaired in s/s and db/db animals. To determine whether this is the case, we investigated thermogenesis by measuring tolerance to acute cold exposure, the expression of BAT UCP1 protein (a mediator of nonshivering thermogenesis that is regulated by the SNS), and circulating thyroid hormone in s/s and db/db animals.

Although prolonged cold exposure studies more accurately gauge nonshivering thermogenesis than acute cold exposure studies, neither s/s nor db/db mice survive prolonged exposure to cold. We thus examined tolerance to acute cold exposure: animals accustomed to a 25°C environment were placed singly in the cold for 75 min in the absence of food, and core body temperature was measured periodically. In this experimental protocol, the ability to maintain core body temperature reflects primarily shivering thermogenesis, with a lesser component of nonshivering thermogenesis. Differences in initial core body temperature among genotypes at the outset of the experiment were not statistically significant. Whereas +/+ mice showed little variation in core body temperature during this experiment, the temperature of db/db animals dropped precipitously, such that the experiment had to be terminated early for each of these animals. This result is consistent with impairment of both shivering and nonshivering thermogenesis in the db/db animals. By contrast, although the core body temperature of s/s animals decreased during the experiment, they were able to maintain body temperatures of ∼33°C over the duration of the cold exposure, demonstrating improved total thermogenesis compared with db/db animals (Fig. 4).

UCP1-mediated thermogenesis in BAT is a major contributor to nonshivering thermogenesis and contributes to energy expenditure by uncoupling mitochondrial respiration from ATP synthesis; the synthesis and activity of BAT UCP1 are regulated by SNS stimulation. Impaired activity of the SNS in db/db mice decreases UCP1 expression and promotes the development of obesity. We examined UCP1 expression in +/+, s/s, and db/db animals acclimated to a 25°C environment by immunoblotting for UCP-1 protein in BAT extracts from these animals. As previously reported, BAT UCP1 expression was reduced 16 ± 2% in db/db animals compared with +/+ mice. Similarly impaired UCP1 expression was observed in BAT from s/s animals (decrease of 16 ± 2% compared with +/+), suggesting that LRb→STAT3 signaling is required for leptin-mediated BAT UCP1 expression (Fig. 5). Because 75 min of cold exposure (as in Fig. 4) is insufficient to increase BAT UCP1 from the baseline (shown in Fig. 5), differences in cold tolerance between s/s and db/db animals are not likely to result from differences in UCP1. These observations suggest that while nonshivering thermogenesis mediated by BAT UCP1 may be similarly impaired in s/s and db/db animals, the increased lean mass of s/s animals may enable compensatory increases in shivering thermogenesis in s/s compared with db/db animals during acute cold exposure (as in Fig. 4), although it is also possible that differences in the ability to control cold loss through vasoconstriction could play a role. Furthermore, because BAT UCP1 expression primarily reflects SNS activity (20–22), these data suggest similarly impaired SNS tone in s/s and db/db animals.

Thyroid hormone represents a major leptin-regulated endocrine mediator of metabolic rate; alterations in thyroid function often result in changes in body weight caused by dysregulation of metabolic rate and thermogenesis. To assess the role of LRb→STAT3 signaling in the control of thyroid function, free thyroxine (T4) levels were measured in 8-week-old s/s mice. Compared with +/+ mice, s/s animals displayed a 31 ± 7% decrease in serum free T4 concentrations (Fig. 6), which was similar to the 39 ± 9% decrease observed in db/db mice compared with +/+ animals. These data are consistent with the hypothesis that leptin mediates control of the thyroid axis via the LRb→STAT3 pathway, and they suggest that differences in energy expenditure between s/s and db/db animals are not attributable to differences in thyroid function. Although we were unable to examine levels of active (T3) thyroid hormone in these animals, the finding of decreased T4 is consistent with the previous description by others of hypothyroidism in rodents and humans that are deficient in leptin (9,33–36).

In aggregate, the data presented here are consistent with modest changes in energy balance in s/s compared with db/db mice that are attributable to increased lean body mass, whereas the neuroendocrine regulation of metabolic rate is similarly impaired in s/s and db/db animals, suggesting that leptin signals mediated via the LRb-STAT3 pathway are crucial for leptin regulation of energy expenditure in addition to leptin regulation of appetite. The obesity of s/s mice is caused not only by hyperphagia but also by disrupted neuroendocrine regulation of energy expenditure.

Investigation of ambulatory activity in s/s mice revealed that although movement is dramatically reduced in these animals compared with wild-type mice, it is slightly increased in s/s compared with db/db mice. Although the mechanism by which lack of leptin signaling decreases activity is not clear, possibilities include lack of motivation or the presence of physical impediments. We speculate that the physical impediments to movement may be slightly reduced in s/s compared with db/db mice because s/s animals have greater lean mass and distribute their adipose mass over a greater body length than db/db mice (29).

Certainly, this modest increase in activity in s/s compared with db/db animals did not significantly alter overall energy expenditure because neither the total nor the resting metabolic rate of s/s mice, as assessed by Vo₂, were different from those observed in db/db animals; the finding of similar respiratory quotient for each genotype suggests that these Vo₂ measurements reflect relative energy expenditure with reasonable accuracy. Because the activity levels of s/s and db/db animals are dramatically reduced compared with +/+ animals, the modest differences in activity between s/s and db/db animals are likely overwhelmed by the dramatic alterations in basal metabolic rate. The trend toward increased feeding in db/db compared with s/s animals, though not statistically significant (29), may also result in increased feeding thermogenesis in db/db animals that could counter the effect of the modestly increased activity of s/s mice on 24-h O₂ consumption. The finding that resting metabolic rate is similarly decreased in both s/s and db/db mouse models suggests that although signals independent of STAT3 may contribute to activity (and possibly also to shivering thermogenesis), the LRb-STAT3 signal is required for the majority of the neuroendocrine regulation of the metabolic rate. By extension of our previous findings of impaired hypothalamic melanocortin action with preserved neuropeptide Y (NPY) repression in s/s mice (29), our present results also suggest the importance of melanocortin action in the regulation of energy expenditure by leptin.

Whereas the ability to maintain body temperature in response to acute cold exposure is impaired in s/s compared with control mice, it is also improved compared with db/db animals. This intermediate acute cold tolerance phenotype of s/s animals is consistent with decreased neuroendocrine-regulated energy expenditure coupled with improved shivering thermogenesis compared with db/db animals. The hypothesis of increased shivering thermogenesis in s/s compared with db/db mice is commensurate with the increased muscle mass of s/s animals and their ability to support increased activity compared with db/db mice.

The inability of ob/ob and db/db mice to tolerate cold exposure is caused in part by decreased BAT UCP1 expression (20–22,37). The finding of equivalently decreased BAT UCP1 expression in s/s and db/db animals is consistent with the impaired cold tolerance of s/s animals and suggests that the LRb-STAT3 signal is critical to the regulation of UCP1 expression, and, by extension, of SNS tone. Although the mechanisms by which leptin regulates sympathetic tone and BAT UCP1 expression are not entirely clear, a great deal of evidence favors a primary role of melanocortin action. For example, mice null for the melanocortin receptor MC4R are resistant to leptin-induced UCP1 expression (38). Similarly, leptin-induced UCP1 expression can be blocked by melanocortin antagonists (18), and the melanocortin agonist melanotan II increases BAT UCP1 expression. These findings are consistent with the impairment of hypothalamic melanocortin action in s/s animals (29). Other evidence suggests a role for signaling by phosphatidylinositol 3′-kinase in hypothalamic melanocortin action, leptin-regulated SNS activity, and BAT UCP1 expression (13). These data and our findings are not mutually exclusive because it is possible that STAT3 and phosphatidylinositol 3′-kinase are each critical for the regulation of these events by leptin.

One of the primary indicators that leptin regulates the hypothalamic-pituitary-thyroid axis was the finding that db/db and ob/ob mice have significantly decreased serum free thyroxin levels, because it has been demonstrated that leptin depletion in mice caused by chronic fasting results in a fall in thyroid hormones that can be blunted if leptin is administered during starvation (9,14). Pro-TRH is highly expressed in the parvocellular region of the paraventricular nuclei of the hypothalamus; during starvation, pro-TRH expression in the paraventricular nuclei is suppressed, depressing thyroid-stimulating hormone secretion from the anterior pituitary and resulting in a fall in circulating levels of the thyroid hormones T3 and T4 (39). A number of mechanisms have been proposed to account for the leptin-mediated regulation of pro-TRH mRNA in the hypothalamus and thus regulation of circulating thyroid hormone levels. Indeed, several authors have suggested that LRb-STAT3 signaling may directly activate pro-TRH expression in pro-TRH–expressing hypothalamic neurons (16,17); this mechanism is consistent with our finding of depressed thyroid function in s/s animals. Leptin may also regulate pro-TRH indirectly via projections from LRb-expressing proopiomelanocortin and/or NPY-expressing neurons from the hypothalamic arcuate nucleus (40–42). However, the finding that s/s mice do not possess elevated hypothalamic NPY mRNA (unlike db/db mice) (29), but do display similar repression of the thyroid axis compared with db/db mice, suggests that leptin control of TRH message is predominantly mediated via direct LRb-STAT3 signaling in the TRH neuron or via projections from proopiomelanocortin/neurons, as opposed to via NPY neurons.

In the absence of leptin signaling, as in ob/ob or db/db mice, the reproductive, growth, and thyroid axes are inhibited, along with the SNS (9). In contrast, whereas s/s mice display intact reproductive and growth function (29), overall energy expenditure (with the exception of that attributable to increased lean mass) is similar to that of db/db animals, as is the repression of thyroid function and the SNS. Thus, the LRb-STAT3 pathway not only controls the majority of chronic feeding behavior but is also the primary determinant of overall energy expenditure. Thus, although it is possible that STAT3-independent LRb signals may also be important in mediating these and other leptin actions, the LRb-STAT3 signal represents a reasonable candidate as a site of intracellular leptin resistance in obesity.

This work was supported by National Institutes of Health Grants DK56731 and DK57768 (to M.G.M.), grants from the American Diabetes Association (to M.G.M.), National Institutes of Health Grant DK53978 (to E.M.-F.), and an American Diabetes Association/European Association for the Study of Diabetes transatlantic fellowship (to S.H.B.).

We thank Francis Marino for assistance.