PPARγ ablation sensitizes proopiomelanocortin neurons to leptin during high-fat feeding

Activation of central PPARγ promotes food intake and body weight gain; however, the identity of the neurons that express PPARγ and mediate the effect of this nuclear receptor on energy homeostasis is unknown. Here, we determined that selective ablation of PPARγ in murine proopiomelanocortin (POMC) neurons decreases peroxisome density, elevates reactive oxygen species, and induces leptin sensitivity in these neurons. Furthermore, ablation of PPARγ in POMC neurons preserved the interaction between mitochondria and the endoplasmic reticulum, which is dysregulated by HFD. Compared with control animals, mice lacking PPARγ in POMC neurons had increased energy expenditure and locomotor activity; reduced body weight, fat mass, and food intake; and improved glucose metabolism when exposed to high-fat diet (HFD). Finally, peripheral administration of either a PPARγ activator or inhibitor failed to affect food intake of mice with POMC-specific PPARγ ablation. Taken together, our data indicate that PPARγ mediates cellular, biological, and functional adaptations of POMC neurons to HFD, thereby regulating whole-body energy balance.

Pharmacological compounds that activate PPARs, such as thiazolidinediones, have been used as highly effective oral medications for type 2 diabetes due to their ability to increase systemic insulin sensitivity. An adverse effect of these compounds is weight gain (1, 2), which limits their potential applications for a variety of obesity-related metabolic diseases. Whether tissue-specific action of PPARs can explain this phenomenology is not known.

Several studies, including our own, have suggested a central role for PPARγ in whole-body energy homeostasis (3–7). Hypothalamic Pparg mRNA expression is several folds higher compared with Ppara or Ppard mRNA expression, and Pparg mRNA expression increases significantly in diet-induced obese (DIO) mice when compared with that of lean controls (7). While brain-specific deletion of PPARγ decreased food intake and increased energy expenditure (EE) in DIO mice (6), chronic hypothalamic-specific PPARγ activation induced increased food intake and body weight in DIO mice (5). In addition, deletion of PPARγ in the brain or in the hypothalamus conferred resistance to rosiglitazone-induced food intake and body weight gain (5, 6). Nevertheless, the identity of PPARγ-expressing neurons and the mode of action on cellular mechanisms in these cells that mediate the effect of PPAR agonists on energy homeostasis remains elusive.

We have shown that systemic administration of the PPARγ agonist, rosiglitazone, induced increased food intake on high-fat diet (HFD), a process that was associated with peroxisome proliferation, reduced reactive oxygen species (ROS) levels, and decreased activity of proopiomelanocortin (POMC) neurons in the arcuate nucleus of the hypothalamus (7). Whether PPARγ action within POMC neurons brings about these apparent crucial cellular biological adaptations in response to HFD is not known. To address this, we studied animals with POMC-specific deletion of PPARγ. We focused on cellular biological events and whole-body physiology of these animals and their control littermates.

Generation and validation of Pomc-Cre Pparg^fl/fl mice. To develop mice with selective ablation of PPARγ in POMC neurons, Cre transgenic mice, in which Cre recombinase is expressed specifically under the POMC promoter were crossed with Pparg^fl/fl mice (Jackson Lab). In situ hybridization for Pparg mRNA combined with POMC immunocytochemistry showed colocalization in control mice (Supplemental Figure 1; supplemental material available online with this article; doi:10.1172/JCI76220DS1), while no colocalization was observed in Pomc-Cre Pparg^fl/fl mice (Supplemental Figure 1), confirming PPARγ deletion in the hypothalamic POMC neurons. Because POMC promoter also drives Cre recombinase expression in the pituitary, we examined the pituitary-adrenal axis. The expression levels of Pomc mRNA (1.00 ± 0.11 in male controls vs. 1.04 ± 0.12 in male Pomc-Cre Pparg^fl/fl mice) as well as circulating corticosterone levels (46.6 ± 9.9 ng/ml in male controls vs. 47.1 ± 10.8 ng/ml in male Pomc-Cre Pparg^fl/fl mice) were unaltered between the 2 genotypes. These data suggest that deletion of PPARγ in the pituitary does not affect the pituitary-adrenal axis.

Deletion of PPARγ in POMC neurons does not alter energy balance on standard chow diet. On a standard chow diet, no difference in body weight was found among female (Supplemental Figure 2A) and male (Supplemental Figure 3A) Pomc-Cre Pparg^fl/fl, Pomc-Cre Pparg^fl/+, and Pparg^fl/fl (Cre negative defined as control) mice up to 16 weeks of age. The amount of adipose depot in both female (Supplemental Figure 2, B and C) and male (Supplemental Figure 3, B and C) Pomc-Cre Pparg^fl/fl mice was not statistically significant compared with that of Pomc-Cre Pparg^fl/+ and control mice. Leptin levels were also similar between Pomc-Cre Pparg^fl/fl and control groups (1.15 ± 0.09 ng/ml in female Pomc-Cre Pparg^fl/fl mice vs. 1.38 ± 0.05 ng/ml in controls; P > 0.05). EE and locomotor activity in both females and males were also similar between Pomc-Cre Pparg^fl/fl and control mice (Supplemental Figure 2 and Supplemental Figure 3, D–H). No differences in thyroid hormones levels were found between these experimental groups (data not shown). When food intake was measured in a 24-hour period, female and male Pomc-Cre Pparg^fl/fl mice showed no significant difference in feeding compared to that of control mice (females: 3.61 ± 0.16 g in Pomc-Cre Pparg^fl/fl mice vs. 3.78 ± 0.15 g in controls; males: 3.86 ± 0.06 g in Pomc-Cre Pparg^fl/fl mice vs. 4.02 ± 0.31 g in controls).

Loss of PPARγ in POMC neurons decreases peroxisome density and increases ROS levels in POMC neurons of HFD-exposed mice. Exposure to HFD significantly increases peroxisome density and decreases ROS levels in POMC neurons (7). To test the involvement of PPARγ, we analyzed peroxisome density and ROS levels in HFD-exposed Pomc-Cre Pparg^fl/fl and control mice. A significant reduction in peroxisome density was observed in POMC neurons of Pomc-Cre Pparg^fl/fl mice compared with in control mice (0.31 ± 0.05 peroxisomes/μm² POMC cytosols in Pomc-Cre Pparg^fl/fl mice vs. 0.58 ± 0.07 peroxisomes/μm² POMC cytosols in control littermates, P < 0.05; Figure 1, A and B). POMC neurons of Pomc-Cre Pparg^fl/fl mice also showed significantly greater ROS levels (6.26 ± 0.37 fluorescent particles in 10 μm² cytosol) than those of control mice (4.23 ± 0.38 fluorescent particles in 10 μm² cytosol; P < 0.05; Figure 1, C and D).

Figure 1

Lower peroxisome density and elevated ROS levels and neuronal activation in POMC neurons of Pomc-Cre Pparg^fl/fl mice on HFD. (A) Electron micrographs showing representative sections of POMC perikarya in the arcuate nuclei of control and Pomc-Cre Pparg^fl/fl mice. Red arrows point to peroxisomes. Images on bottom row are high-power magnification images of peroxisomes from the top row, respectively (numbers in images on the top row correspond with numbers in images on the bottom row). Scale bars: 500 nm. (B) Graph showing significantly lower peroxisome density in Pomc-Cre Pparg^fl/fl animals compared with controls (n = 4 per group). (C) Representative confocal micrographs from control and Pomc-Cre Pparg^fl/fl mice fed on HFD showing DHE (red) in POMC neurons (green). Scale bar: 20 μm. (D) Graph showing the density levels of DHE in POMC neurons of control (n = 4) and Pomc-Cre Pparg^fl/fl mice (n = 4). (E) Representative confocal micrographs of double labeling for c-fos (red nuclei) and POMC (green cytoplasm) showing colocalization (white arrows) in the arcuate nucleus of the hypothalamus of a control and a Pomc-Cre Pparg^fl/fl mouse fed on HFD. Scale bar: 50 μm. III, third ventricle. (F) Graph showing a significant increase in the percentage of c-fos/POMC double-labeled neurons in Pomc-Cre Pparg^fl/fl mice compared with control littermates on HFD (n = 4 per group). Data in all graphs are shown as mean ± SEM. *P < 0.05.

Loss of PPARγ in POMC neurons induces increased POMC neuronal activity of HFD-exposed mice. Increased ROS levels in POMC neurons have been associated with increased POMC neuronal activity (7, 8). In agreement with these studies, a greater percentage of POMC neurons in Pomc-Cre Pparg^fl/fl mice (19.20% ± 2.60%) showed c-fos staining in their nuclei compared with control mice (9.30% ± 1.81%; Figure 1, E and F).

Ablation of PPARγ in POMC neurons prevents hyperphagia and alters energy balance on HFD. The body weight of both female and male Pomc-Cre Pparg mice on HFD for 16 weeks was significantly lower than that of control mice. In female mice, a significant difference in body weight appeared as soon as 2 weeks of HFD feeding (Figure 2A). Female Pomc-Cre Pparg mice showed resistance to high-fat feeding, reaching an average body weight of 23.49 ± 0.99 g compared with that of control mice after 16 weeks of HFD exposure 27.44 ± 1.42 g (P = 0.03) (Figure 2A). On the other hand, the difference in body weight between males became evident later, at 9 weeks of HFD feeding (32.47 ± 0.48 g in controls and 30.22 ± 0.59 g in Pomc-Cre Pparg^fl/fl mice), and after 16 weeks of HFD feeding, the average body weight of male Pomc-Cre Pparg^fl/fl mice was 33.60 ± 1.98 g, while that for control mice was 39.62 ± 1.55 g (P < 0.05) (Figure 2B). Pomc-Cre Pparg^fl/fl mice were also significantly lighter than Pomc-Cre Pparg^fl/+ mice (Supplemental Figure 4) at 9 weeks of HFD, indicating that the effect on body weight between the controls (Pparg^fl/fl-Cre negative) and the Pomc-Cre Pparg^fl/fl mice was not due to the presence of the Pomc-Cre transgene. Consistent with the body weights, significant changes in body composition were observed (Figure 2, C–F). A decrease in fat mass (Figure 2, C and D) was observed in both female and male Pomc-Cre Pparg^fl/fl mice compared with control littermates. The significant difference in fat mass in both males and females was reached at around 4 weeks of HFD exposure and persisted for the duration of the experiment (female fat mass at 16 weeks of HFD: 3.67 ± 0.37 g in Pomc-Cre Pparg^fl/fl mice vs. 8.14 ± 1.45 g in control mice, Figure 2C; male fat mass at 16 weeks of HFD: 8.59± 1.25 g in Pomc-Cre Pparg^fl/fl mice vs. 13.44± 1.13 g in control mice, P < 0.05, Figure 2D). Consistently, leptin levels were significantly (P = 0.009) lower in Pomc-Cre Pparg^fl/fl mice (13.9 ± 1.6 ng/ml) compared with those in littermate controls (21.7 ± 1.2 ng/ml). No difference in lean mass was observed between groups (Figure 2, E and F).

Figure 2

Lower peroxisome density and elevated ROS levels and neuronal activation in POMC neurons of Pomc-Cre Pparg^fl/fl mice on HFD is associated with a leaner phenotype. Analysis of (A and B) body weight (n = 9–14 per group), (C and D) fat mass (n = 9–14 per group), and (E and F) lean mass (n = 9–14 per group) showed that both (A, C, and E) female and (B, D, and F) male Pomc-Cre Pparg^fl/fl mice were resistant to HFD compared with littermate controls. Data in all graphs are shown as mean ± SEM. *P < 0.05; **P < 0.01; ***P < 0.001.

To understand the cause of reduced body weight gain, we then measured food intake. Twenty-four–hour food intake analysis showed a slight but not significant decrease in food consumption in Pomc-Cre Pparg^fl/fl animals compared with that in the controls on HFD (2.75 ± 0.21 g in female controls vs. 2.38 ± 0.15 g female Pomc-Cre Pparg^fl/fl mice, P > 0.05, n = 5, Figure 3A; 2.87 ± 0.18 g in male controls vs. 2.68 ± 0.17 g in male Pomc-Cre Pparg^fl/fl mice, P > 0.05, Figure 3B). However, when food intake was analyzed during the dark-and-light cycles, a significant difference in food intake was found during the dark phase (1.48 ± 0.17 g in female controls vs. 1.13 ± 0.09 g in female Pomc-Cre Pparg^fl/fl mice, P < 0.05, Figure 3A; 1.40 ± 0.13 g in male Pomc-Cre Pparg^fl/fl mice vs. 2.03 ± 0.22 g in male controls, P < 0.05, Figure 3B). No difference was observed during the light phase in both female mice (Figure 3A) and male mice (Figure 3B).

Figure 3

Pomc-Cre Pparg^fl/fl mice on HFD have reduced feeding. Analysis of daily food intake in (A) female (n = 5–6 per group) and (B) male (n = 9–12 per group) mice on HFD shows that in both genders Pomc-Cre Pparg^fl/fl mice ate less than control littermates. This difference in food intake was significant during the dark cycle in both female and male mice. Data in all graphs are shown as mean ± SEM. *P < 0.05 compared with controls.

We then measured EE and locomotor activity in male and female Pomc-Cre Pparg^fl/fl mice and controls on HFD. Male Pomc-Cre Pparg^fl/fl mice showed no difference in EE compared with that of controls when EE was normalized by lean mass (Figure 4A). However, when EE was analyzed using analysis of covariance (ANCOVA), a significant increase was observed in male Pomc-Cre Pparg^fl/fl mice compared with that in controls (Figure 4B; 12.259 ± 0.136 kcal per day in Pomc-Cre Pparg^fl/fl mice vs. 11.374 ± 0.123 kcal per day in controls, P < 0.05 both for body weight and genotype effects). Furthermore, EE was significantly increased during the light cycle (11.754 ± 0.211 kcal per day in Pomc-Cre Pparg^fl/fl mice and 10.738 ± 0.190 kcal per day in controls, P < 0.05 both for body weight and genotype effects), but not during the dark cycle (12.482 ± 0.338 kcal per day in Pomc-Cre Pparg^fl/fl mice and 12.023 ± 0.304 kcal per day in controls; P < 0.05 for body weight effect; P = 0.38 for genotype effect). Elevated levels of VO₂ (2,369.675 ± 24.604 ml per day in controls and 2,545.735 ± 27.360 ml per day in Pomc-Cre Pparg^fl/fl mice, P < 0.05 both for body weight and genotype effects, Figure 4C) and VCO₂ (1,839.580 ± 29.576 ml per day in controls and 2,014.103 ± 32.888 ml per day in Pomc-Cre Pparg^fl/fl mice, P < 0.05 both for body weight and genotype effects; Figure 4D) were also observed in Pomc-Cre Pparg^fl/fl mice compared with control mice. No difference in the respiratory exchange rate (0.778 ± 0.010 in controls and 0.785 ± 0.013 in Pomc-Cre Pparg^fl/fl mice) was observed (Figure 4E).

Figure 4

Pomc-Cre Pparg^fl/fl mice on HFD have increased EE. (A and B) The results of EE analysis either after normalization by (A) lean mass or (B) ANCOVA analysis between male Pomc-Cre Pparg^fl/fl and control mice (n = 5–6 per group). (C and D) The results of (C) VO₂ and (D) VCO₂ analysis in male Pomc-Cre Pparg^fl/fl and control mice using ANCOVA (n = 5–6 per group). (E) No difference in respiratory quotient (RQ) was found between the 2 groups. (F) Increased locomotor activity during both dark and light cycles of male Pomc-Cre Pparg^fl/fl mice compared with male controls. (G) Elevated BAT Ucp1 mRNA expression levels in male Pomc-Cre Pparg^fl/fl mice (n = 7) compared with male controls (n = 4). Data in all graphs are shown as mean ± SEM. *P < 0.05 compared with controls.

The total locomotor activity of Pomc-Cre Pparg^fl/fl mice on HFD was significantly higher than that of controls (56,217 ± 2,902 beam breaks per 48 hours in controls and 63,796 ± 3,680 beam breaks per 48 hours in Pomc-Cre Pparg^fl/fl mice, P < 0.05; Figure 4F). This difference was due to an increased locomotor activity during both dark (41,379 ± 3,052 beam breaks per 48 hours in controls and 45,348 ± 2,342 beam breaks per 48 hours in Pomc-Cre Pparg^fl/fl mice, P < 0.05; Figure 4F) and light phases (14,837 ± 1,140 beam breaks per 48 hours in controls and 18,448 ± 1,890 beam breaks per 48 hours in Pomc-Cre Pparg^fl/fl mice, P < 0.05; Figure 4F).

To further analyze the increase in EE, we then assessed the expression of uncoupling protein 1 (UCP1) in the brown adipose tissues (BATs). A significant increase in Ucp1 mRNA was observed in Pomc-Cre Pparg^fl/fl mice compared with their littermate controls (Figure 4G). However, no differences in thyroid hormone levels were observed between the 2 groups (free T3: 3.1 ± 0.8 ng/dl, n = 4 male controls vs. 3.6 ± 1.9 ng/dl, n = 6 male Pomc-Cre Pparg^fl/fl mice; free T4: 1.9 ± 0.7 ng/dl, n = 4 male controls vs. 1.4 ± 0.3 ng/dl, n = 5 male Pomc-Cre Pparg^fl/fl mice).

Female Pomc-Cre Pparg^fl/fl mice showed increased EE compared with control mice on HFD both after normalization to lean mass (Supplemental Figure 5A) or after ANCOVA analysis (Supplemental Figure 5B; 11.279 ± 0.206 kcal per day in Pomc-Cre Pparg^fl/fl mice vs. 10.594 ± 0.186 kcal per day in controls, P < 0.05 both for body weight and genotype effects). In contrast to the males, EE was significantly greater during the dark cycle in females (12.078 ± 0.202 kcal per day in Pomc-Cre Pparg^fl/fl mice and 11.259 ± 0.0.183 kcal per day in controls, P < 0.05 for body weight and genotype effects) but not during the light cycle (10.499 ± 0.244 kcal per day in Pomc-Cre Pparg^fl/fl mice and 9.946 ± 0.221 kcal per day in controls, P < 0.05 both for body weight effect; P = 0.149 for genotype effect). Elevated levels of VO₂ (2,199.616 ± 37.778 ml per day in controls and 2,338.401 ± 41.809 ml per day in Pomc-Cre Pparg^fl/fl mice, P < 0.05 both for body weight and genotype effects; Supplemental Figure 5C) and VCO₂ (1,741.982 ± 34.032 ml per day in controls and 1,864.273 ± 37.663 ml per day in Pomc-Cre Pparg^fl/fl mice, P < 0.05 for body weight effects, P = 0.052 for genotype effect; Supplemental Figure 5D) were also observed in Pomc-Cre Pparg^fl/fl mice compared with control mice. No difference in the respiratory exchange rate (0.792 ± 0.005 in controls and 0.793 ± 0.004 in Pomc-Cre Pparg^fl/fl mice, Supplemental Figure 5E) was observed. The total locomotor activity of Pomc-Cre Pparg^fl/fl mice on HFD was significantly higher than that of controls (104,898 ± 8,851 beam breaks per 48 hours in controls and 157,906 ± 30,918 beam breaks per 48 hours in Pomc-Cre Pparg^fl/fl mice, P < 0.05, Supplemental Figure 5F). This difference was due to an increased locomotor activity during both the dark phase (77,675 ± 5,783 beam breaks per 48 hours in controls and 102,227 ± 6031 beam breaks per 48 hours in Pomc-Cre Pparg^fl/fl mice, P < 0.05, Supplemental Figure 5F) and the light phase (23,527 ± 1,271 beam breaks per 48 hours in controls and 35,361 ± 2,622 beam breaks per 48 hours in Pomc-Cre Pparg^fl/fl mice, P < 0.05, Supplemental Figure 5F).

Insulin sensitivity in Pomc-Cre Pparg^fl/fl mice. Analysis of glucose and insulin tolerance tests showed a greater glucose tolerance and insulin sensitivity of Pomc-Cre Pparg^fl/fl mice compared with control littermates (Figure 5, A and B). No difference in insulin levels was found between the 2 experimental groups in fasted (0.31 ± 0.07 ng/ml in female Pomc-Cre Pparg^fl/fl mice vs. 0.31 ± 0.02 ng/ml in control mice) or fed conditions (2.30 ± 0.57 ng/ml in female Pomc-Cre Pparg^fl/fl mice vs. 2.19 ± 0.88 ng/ml in control mice). Liver analysis showed a decrease in the mRNA levels of hepatic gluconeogenesis enzymes, such as phosphoenolpyruvate carboxykinase (Pepck) (0.53 ± 0.06 in male Pomc-Cre Pparg^fl/fl mice vs. 0.92 ± 0.15 in controls, P < 0.05, Figure 5C) and glucose-6-phosphatase (G6pase) (0.34 ± 0.07 in male Pomc-Cre Pparg^fl/fl mice vs. 1.00 ± 0.25 in controls, P < 0.05, Figure 5D) in Pomc-Cre Pparg^fl/fl mice compared with control littermates, indicating an increased hepatic insulin sensitivity in these mice.

Figure 5

Pomc-Cre Pparg^fl/fl mice on HFD have improved glucose metabolism. (A) Increased glucose and (B) insulin tolerance in female Pomc-Cre Pparg^fl/fl animals (n = 5) compared with controls (n = 8). (C and D) Graphs shown a significant decrease in the liver gluconeogenesis of Pomc-Cre Pparg^fl/fl mice compared with matched controls, as shown by the liver mRNA levels of (C) Pepck (n = 4 per group) and (D) G6pase (n = 4 per group). Data in all graphs are shown as mean ± SEM. *P < 0.05 compared with controls.

To further analyze glucose metabolism in Pomc-Cre Pparg^fl/fl mice, we performed hyperinsulinemic euglycemic clamp on mice on 9 weeks of HFD. To maintain blood glucose levels between 110 mg/dl and 130 mg/dl in both groups (Figure 6A), we needed to infuse approximately 2-fold glucose into Pomc-Cre Pparg^fl/fl mice compared with control mice (the average glucose infusion rate between t = 75 minutes and t = 115 minutes was 8.6 ± 2.2 mg/kg/min in controls, n = 7, and 16.8 ± 2.8 mg/kg/min in Pomc-Cre Pparg^fl/fl mice, n = 6, Figure 6B), suggesting that Pomc-Cre Pparg^fl/fl mice had higher insulin sensitivity than control mice. Consistent with this, Pomc-Cre Pparg^fl/fl mice had a significantly higher rate of disappearance (Rd), which represents whole-body glucose utilization during the clamp period, compared with control mice (16.6 ± 1.7 mg/kg/min in control mice and 21.6 ± 1.3 mg/kg/min in Pomc-Cre Pparg^fl/fl mice, Figure 6C). Simultaneously, Pomc-Cre Pparg^fl/fl mice also showed increased insulin-stimulated inhibition of endogenous glucose production, which mainly represents hepatic insulin sensitivity (40.2% ± 6.3% in controls and 68.2% ± 10.0% in Pomc-Cre Pparg^fl/fl mice, Figure 6D). In agreement with the result of Rd, Pomc-Cre Pparg^fl/fl mice had significantly greater 2-deoxy-d-glucose (2DG) uptake in epididymal white adipose tissue (EWAT) (Figure 6E; 3.4 ± 0.4 nmol/g/min in control mice and 5.9 ± 0.4 nmol/g/min in Pomc-Cre Pparg^fl/fl mice), soleus, and the red portion of gastrocnemius muscle (Gastro-R) compared with that of control mice during the clamp period (soleus: 100.3 ± 18.4 nmol/g/min in controls and 188.5 ± 37.0 nmol/g/min in Pomc-Cre Pparg^fl/fl mice; Gastro-R: 27.8 ± 4.0 nmol/g/min in controls and 58.6 ± 9.6 nmol/g/min in Pomc-Cre Pparg^fl/fl mice; Figure 6F). 2DG uptake in both spleens and brains, which are not insulin-sensitive tissue, were not significantly different between groups (Figure 6G). These results suggest that deletion of PPARγ in POMC neurons improves glucose metabolism due to the enhancement of insulin sensitivity in EWAT, muscle, and liver during HFD feeding.

Figure 6

Pomc-Cre Pparg^fl/fl mice on HFD have increased insulin sensitivity. (A) Glucose levels during the hyperinsulinemic-euglycemic clamp study in control (n = 7) and Pomc-Cre Pparg^fl/fl mice (n = 6). (B) Graph showing a significant greater glucose infusion rate (GIR) during the hyperinsulemic-euglycemic clamp study in Pomc-Cre Pparg^fl/fl mice compared with controls. (C) The increase in the rate of whole-body glucose disappearance (Rd) in Pomc-Cre Pparg^fl/fl mice compared with controls. (D) Graph showing increased percentage of inhibition of hepatic glucose production in Pomc-Cre Pparg^fl/fl mice compared with controls during the hyperinsulemic-euglycemic clamp study. (E and F) The increase in 2DG uptake in (E) EWAT and (F) muscles (soleus and Gastro-R) in Pomc-Cre Pparg^fl/fl mice compared with controls. (G) No difference in 2DG uptake was observed in the spleens and in the brains of Pomc-Cre Pparg^fl/fl and control mice. Data in all graphs are expressed as the mean ± SEM. *P < 0.05; **P < 0.01.

PPARγ deletion in POMC neurons increases leptin sensitivity on HFD. To test leptin sensitivity, HFD-fed Pomc-Cre Pparg^fl/fl and control mice were i.p. injected with leptin twice a day for 3 days, with a dose of leptin of 1.5 μg/g body weight. Compared with controls, Pomc-Cre Pparg^fl/fl mice showed a significant reduction in food intake (day 3: –30% ± 6.59% reduction of food intake) compared with control littermates (day 3: –12.42% ± 2.65% reduction food intake, P < 0.05 vs. WT mice, Figure 7A). Pomc-Cre Pparg^fl/fl mice also showed a significant reduction in body weight gain (day 3: 1.80% ± 0.72% in control vs. 5.41% ± 0.60% of Pomc-Cre Pparg^fl/fl mice; P < 0.05 vs. control mice, Figure 7B).

Figure 7

Pomc-Cre Pparg^fl/fl mice on HFD are sensitive to leptin. (A and B) Daily food intake and body weight of control and Pomc-Cre Pparg^fl/fl mice (n = 4–6 per group) treated with leptin for 3 days on HFD for 12 weeks. Arrows indicate leptin administration. (C) Representative confocal micrographs of p-STAT3 (red) and POMC (green) double immunolabeled hypothalamic sections (white arrows) from leptin-treated control and Pomc-Cre Pparg^fl/fl mice. Scale bar: 50 μm. (D) A higher percentage of POMC neurons with p-STAT3–labeled nuclei in the arcuate nuclei of Pomc-Cre Pparg^fl/fl mice was detected compared with controls (n = 4–6 per group). (E) Representative confocal photomicrographs of control and Pomc-Cre Pparg^fl/fl mice double immunostained (white arrows) for POMC (green) and c-fos (red). Scale bar: 50 μm. (F) Pomc-Cre Pparg^fl/fl mice showed an increased level of c-fos/POMC immunostaining in the arcuate nucleus of the hypothalamus compared with controls (n = 4–6 per group). (G) Electron micrographs showing representative sections of POMC perikarya in the arcuate nuclei from control and Pomc-Cre Pparg^fl/fl mice. Red asterisks show mitochondria in close association with ER. Scale bar: 1 μm. (H) Quantification of mitochondria-ER interaction revealed higher numbers of mitochondria-ER contacts in POMC neurons of Pomc-Cre Pparg^fl/fl animals compared with control animals (n = 4 per group). Data in all graphs are expressed as the mean ± SEM. *P < 0.05; **P < 0.01; ***P < 0.001.

To assess activation of the leptin-associated signal pathway, p-STAT3 immunoreactivity was analyzed after leptin treatment. After i.p. leptin injection, p-STAT3 expression in POMC neurons was significantly increased in Pomc-Cre Pparg^fl/fl mice (45.63% ± 4.91%) compared with control mice on HFD (19.42% ± 4.04%, Figure 7, C and D). Similarly, c-fos immunostaining in POMC neurons was significantly higher in Pomc-Cre Pparg^fl/fl mice (24.28% ± 3.90%) compared with that in control littermates (6.00% ± 2.09%, Figure 7, E and F).

Endoplasmic reticulum (ER) stress plays a key role in the development of obesity and leptin resistance (9, 10). Recently, close interaction between mitochondria and ER has been shown to play an important role in the onset of obesity and leptin resistance (11). Analysis of mitochondria-ER interaction in our study showed that the number of mitochondria-ER contacts in POMC neurons of Pomc-Cre Pparg^fl/fl mice was significantly greater (53.05 ± 2.71) than those in POMC neurons of control mice (23.35 ± 3.61, P < 0.001, Figure 7, G and H)

PPARγ in POMC neurons mediates rosiglitazone and GW9662 effects on food intake. To assess whether the effects of PPARγ activation or inhibition are mediated by PPARγ expression in POMC neurons, we then assessed food intake of Pomc-Cre Pparg^fl/fl and control mice treated i.p. with either rosiglitazone, a PPARγ agonist, or GW9662, a PPARγ antagonist, for 5 days.

Rosiglitazone administration to lean control mice exposed to HFD caused a significant increase in food intake (23.3% ± 3.03%) and body weight gain (4.47% ± 0.63%) compared with Pomc-Cre Pparg^fl/fl mice (–0.2% ± 8.4% food intake and 1.08% ± 1.16% body weight, P < 0.05, Figure 8, A and B, respectively).

Figure 8

Peripheral administration of either rosiglitazone or GW9662 fails to affect food intake in Pomc-Cre Pparg^fl/fl mice. (A) The effect of rosiglitazone administration on the food intake of lean control mice exposed to HFD (n = 5–6 per group). The percentage change in food intake before (vehicle treated) and after rosiglitazone treatment shows that rosiglitazone did not significantly alter food intake in Pomc-Cre Pparg^fl/fl mice. (B) The effect of rosiglitazone administration on the body weight of lean control mice exposed to HFD. The percentage change in body weight before (vehicle treated) and after rosiglitazone treatment shows that rosiglitazone did not significantly alter body weight in Pomc-Cre Pparg^fl/fl mice. (C) The effect of GW9662 administration on HFD-exposed mice for 16 weeks (n = 6 per group). The percentage change in food intake before (vehicle treated) and after GW9662 treatment shows that GW9662 failed to significantly decrease food intake in Pomc-Cre Pparg^fl/fl mice. (D) The effect of GW9662 administration on the body weight of HFD-exposed mice for 16 weeks. The percentage change in body weight before (vehicle treated) and after GW9662 treatment shows that GW9662 failed to significantly decrease body weight in Pomc-Cre Pparg^fl/fl mice. Data in the graphs are expressed as the mean ± SEM. *P < 0.05; **P < 0.01.

Similarly, GW9662 administration to HFD-exposed mice for 16 weeks significantly reduced food intake (–11.87% ± 3.86%) and body weight change (–4.95% ± 1.16%) in control mice compared with Pomc-Cre Pparg^fl/fl mice (1.42% ± 4.22% food intake and –0.41% ± 0.28% body weight, P < 0.05, Figure 8, C and D, respectively).

The results of this study unmasked a crucial role for PPARγ in POMC neurons to bring about cellular biological adaptations of these cells to the changing metabolic environment. We showed that the arcuate nucleus POMC neurons are important sites of action of PPARγ during high-fat feeding. We showed that deletion of PPARγ in POMC neurons attenuates hyperphagia, increases EE, and protects the mice from DIO and the development of leptin resistance. The cellular mechanisms responsible for these effects involve proliferation of peroxisomes, changes in ROS levels, and mitochondrial ER interactions. Ablation of PPARγ in POMC neurons prevented the increase of peroxisome density and increased ROS levels and mitochondrial ER in association with higher activity of POMC neurons. Considering that PPARγ is widely expressed within the central nervous system, our work also highlights PPARγ as a potential general metabolic switch of neurons.

Pomc-Cre Pparg^fl/fl mice exposed to HFD showed a decrease in body weight gain compared with control littermates. The reduced body weight in Pomc-Cre Pparg^fl/fl mice was due to a marked decrease in adipose mass, induced by a significant decrease in food intake and increase in EE, locomotor activity, and BAT activity in Pomc-Cre Pparg^fl/fl mice. In agreement with these data, Pomc-Cre Pparg^fl/fl mice showed an improved glucose metabolism compared with controls. Interestingly, a previous report using a mouse model in which PPARγ was deleted in the whole brain, BPPARγKO (synapsin I-Cre mice; ref. 6), showed that these mice, while having a reduced body weight gain when exposed to HFD, did not show improvements in their glucose metabolism (6). Our study shows that Pomc-Cre Pparg^fl/fl mice have a significantly greater insulin sensitivity in several tissues, including muscle and white adipose tissue, compared with controls, which was associated with a downregulation of hepatic enzymes involved in the gluconeogenesis and an increased insulin-induced inhibition of endogenous glucose production in the liver. However, no difference in glucose tolerance or insulin-induced suppression of hepatic glucose production was found in BPPARγKO mice compared with controls, despite the differences in body weight and body composition (6). Thus, this observation suggests that the improved glucose tolerance and insulin sensitivity may be independent of changes in body weight and body composition.

By binding its receptors on cell membrane, leptin activates POMC neurons and exerts its anorexigenic effect and increases EE (12). Increased activity of POMC neurons has been associated with elevated intracellular ROS levels (7, 8). Thus, during DIO, sustained higher levels of leptin would predict higher intracellular levels of ROS and activity of POMC neurons. However, during DIO leptin resistance (13), a phenomenon characterized by the inability of exogenously administered leptin to induce Stat3 phosphorylation and simultaneously by an increase in suppressor of cytokine signaling 3 (SOC3) (14, 15), develops, and the positive correlation between circulating leptin levels and intracellular ROS levels in POMC cells is lost (7). Interestingly, an increase in peroxisome density in POMC neurons occurs in DIO mice (7). Peroxisomes are cellular organelles involved in the catabolism of long-chain fatty acids through β-oxidation (16, 17). Peroxisomes also contain several enzymes, including catalase, which, by using hydrogen peroxidase to oxidize several substrates, detoxify the cell from deleterious ROS (18). Thus, the increase of peroxisome density in POMC neurons of DIO mice, caused by decreasing ROS levels, may affect POMC activity and sensitivity to leptin. In support of this, our data show that preventing the increase of peroxisome density in POMC neurons by selective deletion of PPARγ induces an elevation in ROS levels together with increased POMC activity and leptin sensitivity.

Our study shows that mice with deletion of PPARγ in POMC neurons have a significant increase in mitochondria-ER interaction compared with controls. ER stress and mitochondrial dysfunction are implicated in the pathogenesis of obesity and leptin resistance (9–11, 19). Disruption of mitochondria-ER interaction was shown recently to be a key element in leptin resistance of POMC neurons specifically (11). Thus, our observation that PPARγ deletion in POMC neurons reversed the loss of mitochondria-ER connection in response to HFD highlights the importance of PPARγ in the etiology of ER stress.

Our study assessed the effect of constitutive loss of PPARγ during development in POMC neurons. Gene deletion during development may induce compensatory mechanisms that may affect the overall metabolic phenotype of an organism and undermine the significance of that gene in the regulation of homeostasis. This is the case for AgRP neuron deletion, which has been shown to induce different metabolic outcomes according to the timing of neuronal ablation (11, 20). Furthermore, recent data showed that multiple lineages of hypothalamic neurons express POMC during development, including a subpopulation of arcuate AgRP neurons that does not express POMC in adult mice (21). However, a very recent study using a new model of POMC-Cre mice showed that gene deletion in POMC neurons either prenatally or postnatally did not change the metabolic outcomes (21). Nevertheless, we cannot exclude the possibility that PPARγ deletion in a subpopulation of AgRP neurons may have contributed to the metabolic phenotype of our Pomc-Cre Pparg^fl/fl mice, as changes in ROS levels in AgRP neurons have been shown to inactivate these neurons, thus reinforcing the anorexigenic tone of POMC neurons (7, 8).

In summary, our data showed that PPARγ deletion in POMC neurons dramatically affected cellular biological processes and activity of POMC neurons on HFD. These observations give further support to the notion that appropriate cellular biological adaptations in hypothalamic neurons, including ROS control, represent crucial components in leptin sensitivity and resistance and related metabolic and glucose control.

View Supplemental data

This work was supported by NIH grant DK 097566 (to S. Diano).

Address correspondence to: Sabrina Diano, Department of Obstetrics, Gynecology, and Reproductive Sciences, Yale University School of Medicine, PO Box 208063, New Haven, Connecticut 06520-208063, USA. Phone: 203.737.1216; E-mail: sabrina.diano@yale.edu.

Conflict of interest: The authors have declared that no conflict of interest exists.

Reference information: J Clin Invest. 2014;124(9):4017–4027. doi:10.1172/JCI76220.

1. Lehrke M, Lazar MA. The many faces of PPARγ. Cell. 2005;123(6):993–999.
   View this article via: PubMed CrossRef Google Scholar
2. Shimizu H, Tsuchiya T, Sato N, Shimomura Y, Kobayashi I, Mori M. Troglitazone reduces plasma leptin concentration but increases hunger in NIDDM patients. Diabetes Care. 1998;21(9):1470–1474.
   View this article via: PubMed CrossRef Google Scholar
3. Larsen PJ, et al. Differential influences of peroxisome proliferator-activated receptors γ and -α on food intake and energy homeostasis. Diabetes. 2003;52(9):2249–2259.
   View this article via: PubMed CrossRef Google Scholar
4. Sarruf DA, et al. Expression of peroxisome proliferator-activated receptor-gamma in key neuronal subsets regulating glucose metabolism and energy homeostasis. Endocrinology. 2009;150(2):707–712.
   View this article via: PubMed CrossRef Google Scholar
5. Ryan KK, Li B, Grayson BE, Matter EK, Woods SC, Seeley RJ. A role for central nervous system PPAR-γ in the regulation of energy balance. Nat Med. 2011;17(5):623–626.
   View this article via: PubMed CrossRef Google Scholar
6. Lu M, et al. Brain PPAR-γ promotes obesity and is required for the insulin-sensitizing effect of thiazolidinediones. Nat Med. 2011;17(5):618–622.
   View this article via: PubMed Google Scholar
7. Diano S, et al. Peroxisome proliferation-associated control of reactive oxygen species sets melanocortin tone and feeding in diet-induced obesity. Nat Med. 2011;17(9):1121–1127.
   View this article via: PubMed CrossRef Google Scholar
8. Andrews ZB, et al. UCP2 mediates ghrelin’s action on NPY/AgRP neurons by lowering free radicals. Nature. 2008;454(7206):846–851.
   View this article via: PubMed CrossRef Google Scholar
9. Zhang X, Zhang G, Zhang H, Karin M, Bai H, Cai D. Hypothalamic IKKβ/NF-κB and ER stress link overnutrition to energy imbalance and obesity. Cell. 2008;135(1):61–73.
   View this article via: PubMed CrossRef Google Scholar
10. Ozcan L, et al. Endoplasmic reticulum stress plays a central role in development of leptin resistance. Cell Metab. 2009;9(1):35–51.
    View this article via: PubMed CrossRef Google Scholar
11. Schneeberger M, et al. Mitofusin 2 in POMC neurons connects ER stress with leptin resistance and energy imbalance. Cell. 2013;155(1):172–187.
    View this article via: PubMed CrossRef Google Scholar
12. Pinto S, et al. Rapid rewiring of arcuate nucleus feeding circuits by leptin. Science. 2004;304(5667):110–115.
    View this article via: PubMed CrossRef Google Scholar
13. Myers MG, Myers MG Jr, Leibel RL, Seeley RJ, Schwartz MW. Obesity and leptin resistance: distinguishing cause from effect. Trends Endocrinol Metab. 2010;21(11):643–651.
    View this article via: PubMed Google Scholar
14. Munzberg H, Flier JS, Bjorbaek C. Region-specific leptin resistance within the hypothalamus of diet-induced obese mice. Endocrinology. 2004;145(11):4880–4889.
    View this article via: PubMed CrossRef Google Scholar
15. Mori H, et al. Socs3 deficiency in the brain elevates leptin sensitivity and confers resistance to diet-induced obesity. Nat Med. 2004;10(7):739–743.
    View this article via: PubMed CrossRef Google Scholar
16. Dreyer C, Krey H, Givel F, Helftenbein G, Wahli W. Control of the peroxisomal β-oxidation pathway by a novel family of nuclear hormone receptors. Cell. 1992;68(5):879–887.
    View this article via: PubMed Google Scholar
17. I Green S. Activation of a member of the steroid hormone receptor superfamily by peroxisome proliferators. Nature. 1990;347(6294):645–650.
    View this article via: PubMed Google Scholar
18. del Rio LA, Sandalio LM, Palma JM, Bueno P, Corpas FJ. Metabolism of oxygen radicals in peroxisomes and cellular implications. Free Radic Biol Med. 1992;13(5):557–580.
    View this article via: PubMed CrossRef Google Scholar
19. Dietrich MO, Liu ZW, Horvath TL. Mitochondrial dynamics controlled by mitofusins regulate agrp neuronal activity and diet-induced obesity. Cell. 2013;155(1):188–199.
    View this article via: PubMed CrossRef Google Scholar
20. Gropp E, et al. Agouti-related peptide-expressing neurons are mandatory for feeding. Nat Neurosci. 2005;8(10):1289–1291.
    View this article via: PubMed CrossRef Google Scholar
21. Padilla SL, Carmody JS, Zeltser LM. Pomc-expressing progenitors give rise to antagonistic neuronal populations in hypothalamic feeding circuits. Nat Med. 2010;16(4):403–405.
    View this article via: PubMed CrossRef Google Scholar
22. Kwon Jeong J, Dae Kim J, Diano S. Ghrelin regulates hypothalamic prolyl carboxypeptidase expression in mice. Mol Metab. 2013;2(1):23–30.
    View this article via: PubMed CrossRef Google Scholar
23. Jeong JK, Szabo G, Kelly K, Diano S. Prolyl carboxypeptidase regulates energy expenditure and the thyroid axis. Endocrinology. 2012;153(2):683–689.
    View this article via: PubMed CrossRef Google Scholar
24. Toda C, et al. Extracellular signal-regulated kinase in the ventromedial hypothalamus mediates leptin-induced glucose uptake in red-type skeletal muscle. Diabetes. 2013;62(7):2295–2307.
    View this article via: PubMed CrossRef Google Scholar
25. Ayala JE, Bracy DP, McGuinness OP, Wasserman DH. Considerations in the design of hyperinsulinemic-euglycemic clamps in the conscious mouse. Diabetes. 2006;55(2):390–397.
    View this article via: PubMed Google Scholar