A role for central nervous system PPAR-γ in the regulation of energy balance.

The peroxisome proliferator-activated receptor-γ (PPAR-γ) is a nuclear receptor that is activated by lipids to induce the expression of genes involved in lipid and glucose metabolism, thereby converting nutritional signals into metabolic consequences. PPAR-γ is the target of the thiazolidinedione (TZD) class of insulin-sensitizing drugs, which have been widely prescribed to treat type 2 diabetes mellitus. A common side effect of treatment with TZDs is weight gain. Here we report a previously unknown role for central nervous system (CNS) PPAR-γ in the regulation of energy balance. We found that both acute and chronic activation of CNS PPAR-γ, by either TZDs or hypothalamic overexpression of a fusion protein consisting of PPAR-γ and the viral transcriptional activator VP16 (VP16-PPAR-γ), led to positive energy balance in rats. Blocking the endogenous activation of CNS PPAR-γ with pharmacological antagonists or reducing its expression with shRNA led to negative energy balance, restored leptin sensitivity in high-fat-diet (HFD)-fed rats and blocked the hyperphagic response to oral TZD treatment. These findings have implications for the widespread clinical use of TZD drugs and for understanding the etiology of diet-induced obesity.

Despite intense investigation, PPARγ remains an orphan nuclear receptor; i.e., a high-affinity endogenous ligand has not yet been identified. Rather, several dietary fats and their metabolites bind to PPARγ, but with moderate affinity, leading to the suggestion that the physiological role of PPARγ is to act as a sensor for the integrated flux of multiple fatty acids1. Consistent with this possibility, PPARγ is highly expressed in white adipose tissue (WAT) where it is a key regulator of adipogenesis3,4 and where PPARγ activation promotes increased lipid storage5,6. Chronic peripheral administration of exogenous PPARγ agonists, including the TZD Rosiglitazone (RSG), improves glycemic control at the expense of increased caloric intake, body weight and body-fat gain2,7,8. Chronic peripheral administration of PPARγ antagonists also confers protection from diet-induced obesity9.

The traditional view has been that these changes in energy balance are mediated primarily by the actions of PPARγ to induce adipogenesis in WAT. However we would emphasize the point, made elsewhere by Rosen and Spiegelman10, that simply “having more fat cells does not make an animal fatter. In the absence of altered energy balance, an increase in adipogenesis will result in smaller fat cells with no change in total adiposity.” Pertinent to this, PPARγ is also expressed in regions of the hypothalamus important for the central regulation of energy balance11-13. We therefore hypothesized that: 1) activation of CNS PPARγ by RSG contributes to its effect on energy balance, and 2) activation of CNS PPARγ by its endogenous lipid agonists provides a direct mechanism underlying HFD-induced hyperphagia and leptin resistance.

We hypothesized that direct activation of CNS PPARγ would result in positive energy balance. To test this, we administered small doses of RSG or its vehicle directly into the 3rd-cerebral ventricle (i3vt) of male Long-Evans rats in the area of the ventral hypothalamus. Acute i3vt RSG resulted in a 50% higher caloric intake over 24 h, with a corresponding higher body weight change (Fig. 1a,b) compared to i3vt vehicle alone. Furthermore, a single bolus of i3vt RSG led to significantly greater food intake for as many as 3 d (Fig. 1c) and body fat gain was still higher 7 d following the single injection (Fig. 1d), compared to i3vt vehicle alone. We found no differences in chow intake following an oral dose of RSG (0 vs. 0.1 mg kg·bw–1) roughly 30 times greater than our central dose (VEH: 26.34 g ± 0.58, RSG: 27.64 g ± 0.71), ruling out that our i3vt treatments have peripheral orexigenic effects. To determine whether RSG could activate neuronal populations involved in the regulation of energy balance, we measured c-Fos immunoreactivity in rat hypothalamus 1 h following an acute i3vt injection of RSG. There was a significantly greater induction of c-Fos in the paraventricular (PVH, Fig. 1e,f) but not in the arcuate (ARH) or dorsomedial nucleus of the hypothalamus (DMH) (Supplementary Fig. 1a,b) among rats injected with RSG compared to those injected with vehicle alone.

Figure 1

Activation of hypothalamic PPARγ leads to positive energy balance

a,b) 24 h caloric intake (a) and weight change (b) following i3vt RSG or vehicle (Kruskal-Wallis, Dunn's posthoc) c,d) Cumulative food intake (c) and body fat gain (d) following the bolus infusion of RSG or vehicle on day 0 (RM ANOVA with Tukey posthoc) e) Representative sections (top = vehicle, bottom = RSG; left = 10X, right = 20X) showing c-Fos immunoreactivity in the PVH at 1 h following i3vt RSG or vehicle. Scale bar = 100 μm f) Quantification of c-Fos response to 1 μg RSG or its vehicle i3vt (Mann-Whitney test) g,h,i) Caloric intake (g), body weight change (h), and body fat gain (i) 4 weeks following over-expression of a constitutively active form of PPARγ (VP16-PPARγ) or its scrambled control in the medial hypothalamus (RM ANOVA with Tukey posthoc, T-tests) * = p < 0.05, ** = p < 0.01, *** = p < 0.001. The mean for each group is represented ± S.E., n = 3–6 animals per group for the c-Fos experiment, for all other experiments n = 7–12 animals per group.

To investigate the effects of chronic activation of CNS PPARγ, we used a lentiviral vector to over-express VP16-PPARγ fusion protein locally in the hypothalamus of male rats (Supplementary Fig. 2a,b). Fusion of the viral transcriptional activator, VP16, to PPARγ potently and constitutively activates PPARγ in the absence of ligand14, and has been used to explore the effects of chronic tissue-specific PPARγ activation15,16. Over the course of 4 weeks following the lentivirus infusions, rats that had received the VP16-PPARγ virus consumed more calories, gained more weight, and gained twice as much body fat as control-treated rats (Fig. 1g–i). Collectively these data suggest a potentially important role for hypothalamic PPARγ in both the acute and chronic regulation of food intake and adiposity.

These findings additionally suggest that peripheral administration of TZD drugs might increase weight gain by activating CNS PPARγ. To test this hypothesis explicitly, we administered RSG or its vehicle via oral gavage and simultaneously infused the PPARγ antagonist GW9662 or its vehicle i3vt. It is well-established that treatment with oral TZDs over the course of weeks or months induces weight gain in humans7,8,17 and in rodents2,18. We observed that a single oral dose of RSG induced hyperphagia and greater body weight gain within 24 h (Fig. 2a,b) compared to vehicle alone. To rule out that this weight gain was associated with greater plasma volume19, we measured hematocrits from rats following oral RSG or its vehicle (Fig. 2c,d). Further, when CNS PPARγ receptors were blocked by administering 1 μg of GW9662 i3vt, the acute orexigenic effect of oral RSG was completely attenuated (Fig. 2e). To test the hypothesis in a second way, we administered an shRNA against PPARγ locally into the hypothalamus using a lentiviral vector (Supplementary Fig. 3). Acute oral RSG again elicited greater food intake, compared to its vehicle, among rats that had received the scrambled control shRNA, but this effect was blunted in rats receiving the shRNA against PPARγ (Fig. 2f). These data suggest that peripheral administration of TZDs results in activation of hypothalamic PPARγ that can drive higher food intake.

Figure 2

Activation of CNS PPARγ is required for the hyperphagic effect of oral RSG

a,b) Caloric intake (a) and body weight change (b) 24 h following 30 mg kg–1 RSG or its vehicle by oral gavage (RM ANOVA with Tukey posthoc) c,d) Body weight gain (c) and hematocrits (d) 24 h following 10 mg kg–1 RSG or its vehicle by oral gavage (T-tests) e) Caloric intake 24 h following 10 mg kg–1 RSG or its vehicle by oral gavage and 1 μg GW9662 or its vehicle i3vt (ANOVA with Tukey posthoc) f) Caloric intake 24 h following 10 mg kg–1 RSG or its vehicle by oral gavage in rats previously infected with a lentivirus expressing an shRNA targeted against PPARγ in the medial hypothalamus (RM ANOVA with Tukey posthoc). * = p < 0.05. The mean for each group is represented ± S.E., n = 5–12 animals per group.

Consistent with these findings, two groups have reported that a modest amount of systemically-delivered RSG crosses the blood-brain-barrier18,20. However, those groups draw opposing conclusions regarding the implications of these findings. One study18 concluded that the modest levels of RSG found in whole mouse brain following its intravenous administration were insufficient to account for blunted thyroid status and sympathetic tone observed following peripheral RSG. This conclusion was based primarily on the failure to observe any changes in energy balance following a 14 d continuous i.c.v. infusion of RSG via osmotic pump in a small number of rats (n = 4 group–1). The discrepancy between these findings and ours might be explained if RSG was unstable in an aqueous solution at body temperature over this time frame. Alternatively, they may result from other differences in experimental design. Conversely, another study20 concluded that the modest levels of RSG present in mouse cortex following oral delivery of a therapeutic dose (a comparable dose to ours, 10 mg kg·bw–1 d–1) exceeded the EC50 for RSG at PPARγ and could account directly for changes they observed in the CNS following an oral dose.

Next we investigated the physiological role of CNS PPARγ in the regulation of energy balance. If hypothalamic PPARγ has a physiologically-important role, blocking its endogenous activation locally within the CNS should lead to negative energy balance. We found that fasted, but not ad-libitum fed, rats maintained on a low-fat chow diet and treated with acute i3vt GW9662 exhibited lower 24 h food intake compared to those treated with vehicle alone (Fig. 3a,b). When GW9662 or its vehicle was administered i3vt to HFD-induced obese rats and their age-matched chow-fed controls, the antagonist-treated HFD-fed rats consumed 40% fewer calories compared to vehicle controls, but the same dose had no effect in the chow-fed animals (Fig. 3c). This difference in caloric intake was reflected by a lower body weight change in HFD-fed, but not chowfed, rats (Fig. 3d). Consistent with this, a previous study12 reported that neuron-specific PPARγ knockout mice exhibit no metabolic phenotype when maintained on a chow diet; no data were presented from HFD-fed mice. Together these data suggest that the physiological role of CNS PPARγ is greater during fasting or in rats fed a high-fat diet (HFD), both physiological states characterized by an increased flux of fatty acids.

Figure 3

Blocking the activation of CNS PPARγ with GW9662 leads to negative energy balance

a,b) 24 h chow intake following 1 μg GW9662 or vehicle i3vt to ad lib-fed (a), or 24 h fasted (b) rats (T-tests) c) Caloric intake 24 h following 1μg GW9662 (grey bars) or vehicle (black bars) i3vt, to ad lib-fed rats maintained on either chow or HFD (RM ANOVA with Tukey posthoc). d) Weight change 24 h following 1μg GW9662 (grey bars) or vehicle (black bars) i3vt, to ad lib-fed rats maintained on either chow or HFD (T-tests). * = p < 0.05, ** = p < 0.01. Means are depicted ± S.E., n = 9–12 rats per group.

Notably, we also found that plasma thyroid-stimulating hormone (TSH) was more than 5-fold higher among DIO rats acutely-treated with i3vt GW9662 compared to those treated with vehicle alone (Supplementary Fig. 4a). These data agree with previous reports of cross-talk between PPARγ and the HPT axis 21,22. Moreover, they are consistent with reduced thyroid status previously observed among adult rats following oral RSG18, and they fit with an overall catabolic effect of the antagonist. In contrast, a recent study reported that RSG increases thyroid status in newborn mice22, suggesting that development may be a critical factor modulating this interaction. We did not observe differences in thyrotropin-releasing hormone (Trh) mRNA (Supplementary Fig. 4), and there was a trend for higher plasma tri-iodothyronine (T3) (Supplementary Fig. 4b). Given the relative sluggishness of the HPT axis, we suspect that T3 had not yet reached its peak response to increased TSH at this time point. Thus the role of the HPT axis to mediate effects of CNS PPARγ signaling must therefore remain speculative, and is a priority for future research.

A hallmark of HFD-induced obesity is leptin resistance. Leptin is a peptide hormone secreted by WAT in proportion to total adiposity, thereby providing an indicator of the levels of stored energy in adipose tissue23. Leptin activates its receptors in various hypothalamic nuclei; its downstream effectors are integrated in the PVH, where Trh is also expressed24. Leptin's actions in the hypothalamus and in other areas of the CNS reduce food intake and deplete energy storage in WAT25. Leptin signaling in the hypothalamus is blunted in animals fed a HFD26, and this leptin resistance is thought to contribute to the continued accumulation of body fat. Because the endogenous ligands of PPARγ are thought to be fatty acids, and given the results of our previous experiments, we hypothesized that hypothalamic PPARγ may contribute to the development of HFD-induced leptin resistance. We predicted that chronic antagonism of CNS PPARγ with GW9662 would restore leptin sensitivity in HFD-induced leptin-resistant rats. To test this, we chronically delivered GW9662 or its vehicle into the lateral ventricle of HFD rats, using a sub-threshold dose that has no effect on body weight but that results in significantly lower hypothalamic expression of the PPARγ target gene lipoprotein lipase (Lpl) (Fig. 4). After 14 d, we challenged these rats with exogenous i.p. leptin at a dose that reduces caloric intake and body weight in lean animals27. Among rats concurrently receiving i.c.v. GW9662, those also receiving i.p. leptin exhibited lower 24 h food intake (Fig. 4c) and body weight change (Fig. 4d) compared to those also receiving i.p. vehicle, whereas the weight-matched vehicle-treated rats remained leptin-resistant. These findings implicate PPARγ as a potential molecular link between fatty acids and leptin action to regulate food intake.

Figure 4

Blocking the activation of CNS PPARγ with GW9662 leads to improved leptin sensitivity

a,b) 24 h food intake (a) and body weight change (b), among ad lib HFD-fed rats receiving a chronic, sub-threshold i.c.v. dose of GW9662 (3 μg kg·bw–1) or its vehicle, 24 h following a 1 mg kg·bw–1 acute i.p. dose of leptin (white bars) or its vehicle (black bars) (ANOVA with Tukey posthoc) c) Body weights of rats in panels (a) and (b) (ANOVA) d) Hypothalamic expression of lipoprotein lipase (Lpl) relative to the housekeeping gene L32 (T-test), among ad lib HFD-fed rats receiving chronic sub-threshold i.c.v. dose of GW9662 (3 μg kg·bw–1) or its vehicle. * = p < 0.05, ** = p < 0.01, *** = p < 0.001. The mean for each group is depicted ± S.E., n = 4–7 rats per group.

The present data collectively reveal a novel role of hypothalamic PPARγ in the central regulation of energy balance, and they imply that CNS mechanisms may underlie at least some of the weight gain observed with PPARγ-modulating drugs. In further support of this possibility, Sugii et al.15 recently reported that constitutive PPARγ activation specifically in adipocytes improves the glucose tolerance of DIO mice but does not lead to increased weight gain. An additional implication, therefore, is that differential access to the CNS by different PPARγ ligands and their formulations may result in different therapeutic outcomes. Notably, these data also identify a physiological pathway by which increased consumption of lipids can promote leptin resistance and obesity via hypothalamic PPARγ.