Effect of selective expression of dominant-negative PPARγ in pro-opiomelanocortin neurons on the control of energy balance

Abstract

Peroxisome proliferator-activated receptor-γ (PPARγ), a master regulator of adipogenesis, was recently shown to affect energy homeostasis through its actions in the brain. Deletion of PPARγ in mouse brain, and specifically in the pro-opiomelanocortin (POMC) neurons, results in resistance to diet-induced obesity. To study the mechanisms by which PPARγ in POMC neurons controls energy balance, we constructed a Cre-recombinase-dependent conditionally activatable transgene expressing either wild-type (WT) or dominant-negative (P467L) PPARγ and the tdTomato reporter. Inducible expression of both forms of PPARγ was validated in cells in culture, in liver of mice infected with an adenovirus expressing Cre-recombinase (AdCre), and in the brain of mice expressing Cre-recombinase either in all neurons (NES^Cre/PPARγ-P467L) or selectively in POMC neurons (POMC^Cre/PPARγ-P467L). Whereas POMC^Cre/PPARγ-P467L mice exhibited a normal pattern of weight gain when fed 60% high-fat diet, they exhibited increased weight gain and fat mass accumulation in response to a 10% fat isocaloric-matched control diet. POMC^Cre/PPARγ-P467L mice were leptin sensitive on control diet but became leptin resistant when fed 60% high-fat diet. There was no difference in body weight between POMC^Cre/PPARγ-WT mice and controls in response to 60% high-fat diet. However, POMC^Cre/PPARγ-WT, but not POMC^Cre/PPARγ-P467L, mice increased body weight in response to rosiglitazone, a PPARγ agonist. These observations support the concept that alterations in PPARγ-driven mechanisms in POMC neurons can play a role in the regulation of metabolic homeostasis under certain dietary conditions.

the increasing prevalence of obesity is strongly linked to the consumption of high-fat diet (HFD) and is associated with resistance to insulin, leptin, and other feedback regulatory signals (11, 22, 29). The nuclear receptor peroxisome proliferator-activated receptor-γ (PPARγ) is a transcription factor highly expressed in adipose tissue, where it is essential for normal adipogenesis and regulation of lipid and glucose metabolism (6, 18, 24). Recently, it was discovered that PPARγ affects feeding behavior, energy balance, and leptin sensitivity through its actions in the brain rather than in the adipose tissue alone (21, 25). Injecting thiazolidinediones (TZD), synthetic agonists of PPARγ, or overexpressing PPARγ in the hypothalamus of rats increases food intake and body weight (25). Conversely, blocking endogenous PPARγ by either shRNA, pharmacological inhibitors, or genetic ablation of PPARγ from brain neurons decreases feeding and increases energy expenditure (21). A variety of mechanisms have been proposed to mediate the effects of PPARγ blockade including increased activation of intracellular leptin receptor signaling. This is evidenced by the enhanced phosphorylation of one of the key leptin receptor's downstream mediator, the signal transducer and activator of transcription-3 in the hypothalamus following PPARγ deletion (21).

The pro-opiomelanocortin (POMC)-expressing neurons are among several arcuate nucleus substrates of the central melanocortin system and play a critical role in the maintenance of energy balance (8–10, 28, 33). Activation of PPARγ via intracerebroventricular injection of the PPARγ agonist rosiglitazone was recently shown to decrease the formation of reactive oxygen species in the arcuate nucleus of mice (12). Importantly, and perhaps surprisingly, this leads to suppression of POMC and promotion of AgRP/NPY firing rate, implicating PPARγ as a critical regulator of the activity of these two subtypes of neurons. Most recently, a study by Long and colleagues (20) established that selective ablation of PPARγ in POMC neurons protects mice against effects of HFD and leads to improved leptin sensitivity. While the current evidence suggests that POMC may be the neuronal subtypes critical in mediating PPARγ's effect on energy balance, the molecular mechanisms of these effects remain to be fully elucidated.

To study the mechanisms by which PPARγ in POMC neurons controls energy balance, we generated transgenic mouse models in which a dominant-negative (DN) mutant (P467L) or a wild-type (WT) form of PPARγ is conditionally expressed in POMC neurons. The PPARγ-P467L is a rare mutation in the ligand-binding domain of the receptor and in humans causes early-onset severe hypertension, insulin resistance, along with lipodystrophy, and metabolic syndrome (2, 27). The experimental approach detailed in this study capitalizes on the novel use of the DN-PPARγ, which has been previously shown by our laboratory to be deficient in transcriptional activity and to effectively interfere with the transcriptional activity of endogenous PPARγ (3, 14–17, 23). For example, we previously showed that expression of DN-PPARγ results in a transcriptome in blood vessels, which is opposite to that of PPARγ-mediated activation by rosiglitazone (15). In other words, expression of PPARγ target genes that are induced by rosiglitazone are repressed by DN-PPARγ.

MATERIALS AND METHODS

Cell culture validation of constructs.

The function of the CAG-PPAR-WT and CAG-PPAR-P467L constructs was tested in either HEK293T or 3T3-L1 cell lines. The cells were seeded to 80% confluence 12–16 h prior to transfection. The 3T3-L1 cells were starved for 5 additional hours prior to transfection. Transfection was carried using Lipofectamine LTX (Life Technologies) according to manufacturers' protocol. Total protein or RNA lysates were collected and stored at −80°C until further analysis.

Mice and diets.

All animal procedures described below were approved by the University of Iowa Institutional Animal Care and Use Committee. All mice were fed either HFD, isocaloric-match control diet, or regular lab chow diet starting at 5 wk until 30 wk of age. The HFD was either 60% (D12492, Research Diets, New Brunswick, NJ) or 45% HFD (D12451). The isocaloric-match control diet was 10% fat (D12450J, Research Diets). Regular diet was 7013 (Teklad Premier Laboratory Diets). Body weights from group-housed mice were measured weekly at 8–9 AM. All experiments were performed at 25–27 wk of HFD, unless otherwise reported. Daily food intake and body weight were assessed in individually housed mice. The sex and number of mice in each experimental group are defined in each figure legend.

Conditional transgenic mice expressing either a human WT or a DN mutant (P467L) form of PPARγ were generated at the University of Iowa Gene Editing Facility. The CAG-PPAR-WT and CAG-PPAR-P467L mice carry a transgene designed to express both human PPARγ (P467L or WT) and the tdTomato reporter gene after its selective activation by Cre-recombinase. The mice were generated by injecting purified transgene DNA into the pronuclei of 1 cell fertilized mouse embryos derived B6SJL mice. Mice were continuously backcrossed bred to C57BL/6. The transgenic mice were crossed with POMC-Cre [Tg(Pomc1-Cre16Lowl/J), 006421] mice purchased from The Jackson Laboratory and bred at the University of Iowa. The following mice were used for the experiments described herein: POMC^Cre/PPARγ-P467L and their littermate controls (POMC^Cre−/PPARγ-P467L⁺, POMC^Cre+/PPARγ-P467L⁻, or POMC^Cre−/PPARγ-P467L⁻); POMC^Cre/PPARγ-WT and their littermate controls (POMC^Cre−/PPARγ-WT⁺, POMC^Cre+/PPARγ-WT⁻, or POMC^Cre−/PPARγ-WT⁻). Male and female mice of each genotype were tested as detailed in each figure legend.

Intracerebral activation of transgene.

Six- to eight-week-old, PPARγ-P467L F1 transgenic and nontransgenic littermate controls received Ad5-CMV-Cre or Ad5-CMV-eGFP (6 × 10⁷ plaque-forming unit/mouse) into the cerebral cortex. Mice were anesthetized using 5% isoflurane in O₂ and maintained with 2 ± 1% isoflurane in O₂. Mice were placed in a stereotaxic apparatus, the skin covering the skull was shaved and sterilized with betadine, and a 1.0–2.0 cm incision was made to uncover the skull. An area of the cortex just anterior to the lateral ventricle was targeted using the following coordinates: anterior-posterior, 0.2 mm; medial-lateral, 1.0 mm; dorsal-ventral, 3.0 mm. The skull was drilled to expose the brain surface. One microliter adenovirus was injected via a Hamilton syringe over a 5 min period. The skin was sewed shut, and the animal allowed to recover in home cage on a warming pad. Once ambulatory the mice were transferred to a designated room in the vivarium for shedding of the virus. Mice were euthanized 7 days postviral delivery. The brains were dissected and immersion-fixed in 4% paraformaldehyde solution for 24–48 h.

Immunohistochemistry.

Brains of PPARγ-P467L mice injected with Ad5-CMV-Cre or POMC^Cre/PPARγ-P467L double transgenic mice and littermate control mice were sectioned to 30 μm via vibratome. For antigen retrieval the sections were incubated sequentially in the following solutions: 1% NaOH and 1% H₂O₂ in distilled water for 20 min, 0.3% glycine in PBS for 10 min, 0.03% SDS in PBS for 10 min, and finally blocked in in PBS for 30 min. All incubations were performed at room temperature. Following the incubations the sections were rinsed with PBS three times for 10 min. PPARγ expression was determined using monoclonal anti-PPARγ (C26H12, Cell Signaling Technology) diluted in 3% normal goat serum with 0.1% Triton X-100 to a final concentration of 1:100 and incubated at 4°C overnight. ACTH expression was determined using antiserum to rabbit ACTH (A. F. Parlow, National Hormone and Peptide Program) diluted in 3% goat serum, 0.3% Tween 20 in Tris-buffered saline (TBS) to a final concentration of 1:1,000 and incubated at 4°C overnight. The sections were then rinsed in TBS-T (TBS with 0.3% Tween 20) three times for 10 min. The secondary antibody Alexa488 (Abcam) was diluted in 5% goat serum, 0.1% Triton X-100 in TBS to a final concentration of 1:200 and incubated at room temperature for 1 h. The sections were as described above. All incubation and rinsing steps were performed under constant agitation. The sections were mounted on glass slides with Vectashield (Vector Laboratories) and imaged for the presence of red fluorescence on Zeiss LSM710 confocal microscope. Single-plane images were collected. When comparing detection of transgene expression between samples, we kept the microscope settings, including laser power, gain, and offset, constant throughout image collection. Final images were processed using ImageJ software (version 1.48, National Institutes of Health; Java 1.6.0–20, 64-bit) to make adjustments to image size or linear parameters such as brightness and contrast. All adjustments were kept consistent across samples.

Western blotting.

HEK293T transfected cells, liver, brain tissue were homogenized in a lysis buffer containing: 50 mmol/l Tris Cl buffer, 0.1 mmol/l EDTA, 0.1 mmol/l EGTA (pH 7.5), and 0.1% vol/vol SDS, with protease inhibitors (Thermo Scientific). Protein concentration was determined using the BCA assay (Pierce Protein Research Products). Protein lysates were separated by SDS-PAGE gel and transferred to PVDF membranes (Immobilon-P, Millipore). Membranes were blocked for 30 min at room temperature with 5% BSA in TBS-T (TBS, 0.1% Tween 20). PPARγ expression was determined using monoclonal anti-PPARγ (C26H12, Cell Signaling Technology). Rabbit anti-β-actin was used as a loading control (ab8229, AbCam). Primary antibodies were diluted in 1% BSA/TBS-T to a final concentration of 1:1,000 for anti-PPARγ and 1:10,000 for anti-β-actin, and incubated overnight at 4°C. Secondary antibodies were diluted in 5% milk/TBS to final concentration of 1:10,000 and applied for 1 h at room temperature. Blots were treated with Pierce SuperSignal Western Pico chemiluminescence reagent for visualization.

Quantitative real-time PCR analysis.

Mediobasal hypothalami or 100 mg of liver tissue, kidney, heart, spleen, subcutaneous adipose tissue from POMC^Cre/PPARγ-P467L mice was suspended in 1.0 ml of ice-cold Trizol (Invitrogen). 3T3-L1 cells transfected with either CAG-PPAR-WT or CAG-PPAR-P467L constructs were suspended in 300 ml Trizol. To evaluate transgene expression in POMC neurons, punches were taken at the region of the arcuate nucleus from POMC^Cre/PPARγ-P467L and littermate control mice. Total RNA was isolated using RNeasy spin columns (RNeasy Mini Plus Kit or RNeasy Micro Kit, QIAGEN) following the manufacturer's instructions. The RNA concentration was determined using a NanoDrop ND-1000. cDNA was generated using SuperScript (Invitrogen). Quantitative real-time PCR (qRT-PCR) was performed using the TaqMan (applied Biosystems) gene expression assay from ∼1.0 ng cDNA in a total volume of 10.0 μl following the manufacturer's recommendations. The assay numbers for TaqMan (Applied Biosystems) probes were the following: Mm99999915_g1 (mouse GAPDH); Hs01115513_m1 (human PPARγ) and Mm 00445878_m1 (mouse FABP4 or aP2). The expression of tdTomato (forward: 5′-CAC CAT CGT GGA ACA GTA CG-3′ and reverse: 5′-GCG CAT GAA CTC TTT GAT GA-3) was determined using SYBR green (Bio-Rad) according to the manufacturer's protocol. Mouse GAPDH was used as an internal control.

Measurement of body composition and energy expenditure.

Body composition was measured at the end of diet period (25 wk) in vivo by nuclear magnetic resonance (Bruker Minispec LF-90). A second cohort of male POMC^Cre/PPARγ-P467L and control mice, 18 wk of age, on control diet for 13 wk were individually housed 1 wk before measurements. Animals were then transferred to CLAMS (Comprehensive Lab Animal Monitoring System, Columbus Instruments, Columbus, OH). Metabolic parameters were analyzed over a 3-day period using the Oxymax Comprehensive Lab Animal Monitoring System (Columbus Instruments). In brief, mice were acclimatized to the cages for 12 h before measurements began. The volumes of oxygen consumption (V̇o₂) and carbon dioxide consumption (V̇co₂) were measured through indirect calorimetry. Respiratory exchange ratio (RER) was calculated as the ratio of V̇o₂ into V̇co₂. Energy expenditure was calculated using the Lusk equation: 3.815 + 1.232*RER. Activity levels were counted as infrared beam breaks along the x-axis of the cage.

Fasting glucose measurements.

Mice were individually housed and fasted overnight (16–18 h). Fasting blood glucose levels were measured via Accu-Chek Aviva blood glucose meter (Roche) using tail blood.

Leptin sensitivity.

Male POMC^Cre/PPARγ-P467L and their littermate controls post-25 wk of low or high fat diet feeding were individually housed. Vehicle (PBS) was injected ip at 8 AM and 4 PM for 4 days (1 μl/g body wt) prior to leptin treatment. Recombinant murine leptin (1.0 μg/g body wt, Peprotech) was injected ip twice daily (at 8 AM and 4 PM) for 4 days. Body weight and food intake were measured daily at 8 AM and upon death 1 wk later.

Peripheral injection of rosiglitazone.

We used 14 to 16 wk old female POMC^Cre/PPARγ-P467L and POMC^Cre/PPARγ-WT and their respective littermate controls (n = 7–11/group) for this experiment as previous described (20). The mice were single-housed 1 day prior to the experiment, switched from regular laboratory chow to 45% HFD, and were injected ip with vehicle (10% DMSO in saline) for 5 days at 3:00 PM. Rosiglitazone (Cayman Chemicals, Ann Arbor, MI) was dissolved in 10% DMSO/saline and injected ip at a dose of 28 mg/kg for 5 days at 3:00 PM. In addition, the mice were fasted for 6 h (12:00–6:00 PM) prior to the onset of the dark cycle. Food intake was measured daily at 8:00 AM. Body weight was measured at the start and the end of the treatment.

Statistics.

All data are expressed as means ± SE. The means between two groups were analyzed by two-tailed Student's t-test. The means of two or more groups and two genotypes were analyzed by a two-way ANOVA with Bonferroni post hoc tests unless otherwise stated. ANCOVA was used to analyze the effect of both genotype and body weight on energy expenditure, and RER. Significance was recorded at P < 0.05.

RESULTS

We generated transgenic mice carrying conditionally activatable transgenes designed to express wild-type PPARγ (CAG-PPARγ-WT) or DN-PPARγ (CAG-PPARγ-P467L), each with the tdTomato reporter (Fig. 1A). Cotransfection of HEK293 cells with these plasmids and Cre-recombinase resulted in robust expression of PPARγ protein (Fig. 1B). Expression of aP2, a canonical PPARγ target gene was strongly induced by PPARγ-WT but not PPARγ-P467L in transfected 3T3L1 preadipocytes despite equivalent expression of both PPARγ isoforms (Fig. 1C). This is consistent with the P467L mutation in PPARγ being transcriptionally inactive (2), and our data showing that PPARγ-P467L acts dominant-negatively by effectively interfering with the transcriptional activity of PPARγ-WT (15). Transgenic mice expressing either CAG-PPARγ-WT or CAG-PPARγ-P467L exhibited liver-specific Cre-inducible PPARγ expression in response to intravenous AdCre but not AdeGFP (example provided in Fig. 1D). When AdCre was injected directly into the cerebral cortex of CAG-PPARγ-WT or CAG-PPARγ-P467L mice, neurons coexpressing PPARγ and tdTomato were observed along the needle tract (see example in Fig. 1E). Robust PPARγ expression was also observed in the brain when the CAG-PPARγ-P467L mice were bred with mice expressing Cre-recombinase in the nervous system (Nestin-Cre, Fig. 1F). These data clearly illustrate robust Cre-recombinase dependence of PPARγ expression in these novel transgenic mouse models.

Fig. 1.

Validation of transgenic models. A: schematic representation of the inducible transgene expression system. B: PPARγ Western blot of total cellular protein from HEK293 cells transfected with empty vector (Con), CAG-PPARγ-P467L, or CAG-PPARγ-WT in the presence or absence of Cre-recombinase. C: expression of transgenic PPARγ mRNA and aP2 mRNA in 3T3-L1 preadipocytes transfected with Cre-recombinase activated constructs expressing either empty vector (control), PPARγ-P467L, or PPARγ-WT (*P < 0.05, n = 4 per group). D: expression of PPARγ in the liver of CAG-PPARγ-P467L transgenic and control littermate mice in response to intravenous injection of AdCre or AdeGFP. Shown is a representative blot of many experiments performed in independent lines of mice. E: expression of tdTomato (red) and PPARγ (green) in neurons (nuclei stained by DAPI, blue) along the needle tract (white dashed arrow) in the cerebral cortex of a CAG-PPARγ-WT transgenic mouse injected with AdCre. Similar results were obtained in CAG-PPARγ-P467L transgenic mice (data not shown). F: expression of PPARγ in the brain of Nes^CRE/PPARγ-P467L transgenic and control littermate mice. The presence of the transgene and Nestin-Cre is indicated. Each lane represents brain protein isolated from an individual mouse (n = 8).

To evaluate the effect of interference with PPARγ function or overexpression of PPARγ specifically in POMC neurons, CAG-PPAR-P467L or CAG-PPAR-WT mice were crossed with mice expressing Cre-recombinase under the POMC promoter. We then performed immunocytochemistry at the level of the arcuate nucleus to confirm expression of DN PPARγ in POMC-expressing neurons of POMC^Cre/PPARγ-P467L. First, PPARγ-P467L staining and tdTomato fluorescence were evident only in brains of POMC^Cre/PPARγ-P467L but not in littermate control mice (Fig. 2A). Second, tdTomato (a reported for the transgene) was colocalized with ACTH, a product of POMC cleavage, in the arcuate nucleus (Fig. 2B). RT-qPCR analysis of punches from POMC^Cre/PPARγ-P467L mice showed increased expression of transgenic PPARγ in the arcuate nucleus (>10 CT value), but to a much lesser extent in cortex, compared with littermate control mice (Table 1). The pattern of expression of transgenic tdTomato mRNA correlated strongly with human PPARγ mRNA levels in all of the samples evaluated (data not shown). Expression of the transgene was evaluated in other peripheral tissues from POMC^Cre/PPARγ-P467L and control mice. Compared with arcuate nucleus there was very little or no expression of human PPARγ (Table 1) or tdTomato (data not shown) mRNA in liver, heart, kidney, spleen, subcutaneous white adipose tissue (WAT), perigenital WAT, and interscapular brown adipose tissue (BAT) in double transgenic mice. This provides evidence for selectivity of PPARγ expression in POMC neurons in our model.

Fig. 2.

Transgene expression in POMC^Cre/PPAR-P467L mice. A: immunofluorescence detection of PPARγ (green) and endogenous fluorescent detection of tdTomato (red) on coronal sections at the level of the mediobasal hypothalamus. Arrows indicate cells coexpressing PPARγ and tdTomato. B: POMC neurons immuno-labeled with antibody against ACTH compared with cells expressing tdTomato.

Table 1.

Tissue-specific expression of human PPARγ

Tissue	POMCCre/PPARγ-P467L Mice, CT ± SE	Littermate Controls, CT ± SE
Arcuate Nucleus	23.9 ± 1.1	33.8 ± 2.4
Cortex	26.7 ± 0.5	31.2 ± 1.9
Liver	31.2 ± 0.7	33.1 ± 0.1
Heart	29.4 ± 1.2	34.0 ± 2.1
Kidney	28.5 ± 0.2	34.5 ± 3.0
Spleen	27.8 ± 0.8	28.3 ± 2.9
SQ Adipose	25.3 ± 1.4	27.9 ± 2.8
PG Adipose	26.9 ± 0.9	27.5 ± 2.8
Brown Adipose	27.4 ± 2.7	30.0 ± 2.3

SQ, subcutaneous; PG, perigenital. Average CT value (n = 3) is shown.

The final body weight of POMC^Cre/PPARγ-P467L male mice was comparable to that of littermate controls (55.7 ± 1.0 g vs. 57.1 ± 1.2 g in controls, P = 0.14) 25 wk after 60% HFD feeding (Fig. 3A). There was no difference in body weight in female POMC^Cre/PPARγ-P467L mice (48.9 ± 0.5 g vs. 47.6 ± 0.4 g in controls, P = 0.30, Fig. 3B). To assess if a 60% HFD caused a ceiling effect, we placed a cohort of male POMC^Cre/PPARγ-P467L mice on 45% HFD. Similar to 60% HFD, there was no difference in body weight in POMC^Cre/PPARγ-P467L male mice compared with controls (48.1 ± 0.8 g vs. 48.0 ± 1.0 g in controls, P = 1.0). Consistent with the body weight data, there were no differences in body composition (Fig. 3, C and D), fasting glucose (Fig. 3, E and F), or in organ and fat pads weights (Table 2) between genotypes and within sexes in POMC^Cre/PPARγ-P467L mice.

Fig. 3.

Body weight, composition, and glucose during 25 wk of 60% high-fat diet. Body weight in male (A) and female (B) POMC^Cre/PPARγ-P467L (n = 11 male, n = 15 female) mice compared with littermate controls (n = 15 males, n = 22 females). Body composition is shown in male (C) and female (D) POMC^Cre/PPARγ-P467L (n = 11 male, n = 8 female) and control (n = 15 male, n = 12 female) mice. Fasting glucose is shown in male (E) and female (F) POMC^Cre/PPARγ-P467L (n = 11 male, n = 12 female) and control (n = 15 male, n = 17 female) mice. Data are expressed as means ± SE.

Table 2.

Organ weights

	Males			Females
	P467L	Control	P	P467L	Control	P
Liver	3.22 ± 0.19	3.17 ± 0.19	0.85	1.8 ± 0.25	1.47 ± 0.13	0.21
Heart	0.22 ± 0.01	0.21 ± 0.01	0.38	0.16 ± 0.01	0.17 ± 0.01	0.37
Kidney	0.49 ± 0.02	0.47 ± 0.01	0.42	0.35 ± 0.02	0.36 ± 0.01	0.70
Spleen	0.13 ± 0.01	0.14 ± 0.01	0.60	0.13 ± 0.02	0.12 ± 0.01	0.60
SQ WAT	3.17 ± 0.18	3.37 ± 0.16	0.40	3.21 ± 0.28	2.77 ± 0.20	0.21
PG WAT	1.48 ± 0.14	1.46 ± 0.09	0.90	3.38 ± 0.32	2.89 ± 0.32	0.32
BAT	0.30 ± 0.03	0.27 ± 0.02	0.37	0.27 ± 0.05	0.24 ± 0.31	0.57

Organ weights from male and female POMC^Cre/PPARγ-P467L mice are shown after 25 wk of 60% high-fat diet. SQ WAT, subcutaneous white adipose tissue; PG WAT, perigenital white adipose tissue; BAT, brown adipose tissue.

Next, we examined POMC^Cre/PPARγ-P467L mice fed a 10% fat isocaloric-matched control diet. Interestingly, male POMC^Cre/PPARγ-P467L mice gained significantly more weight 36.0 ± 0.9 g than the controls 31.7 ± 0.5 g (P < 0.0001, Fig. 4A). A similar trend was observed when POMC^Cre/PPARγ-P467L males were placed on a regular laboratory chow diet (7013 NIH-31, Harlan) containing 18% fat (41.1 ± 2.6 g vs. 39.2 ± 2.3 g in controls). The difference in body weight in the male POMC^Cre/PPARγ-P467L mice was strongly correlated with significantly increased fat mass, a small reduction in lean mass, and no changes in fluid content (Fig. 4B). There was no difference in the rate of weight gain (Fig. 4C), nor a change in fat or lean mass in response to low-fat control diet in female POMC^Cre/PPARγ-P467L mice (Fig. 4D). No differences in fasting glucose (male: 80.9 ± 6.7 mg/dl in POMC^Cre/PPARγ-P467L vs. 78.1 ± 2.5 mg/dl in controls, n = 10/group, P = 0.7; female: 77.3 ± 6.4 mg/dl in POMC^Cre/PPARγ-P467L vs. 72.6 ± 3.6 mg/dl in controls, n = 6–12/group, P = 0.5) and food intake (male: 2.6 ± 0.1 g/day in POMC^Cre/PPARγ-P467L vs. 2.6 ± 0.1 g/day in controls, n = 6–11/group, P = 0.74; female: 2.0 ± 0.3 g/day in POMC^Cre/PPARγ-P467L female mice vs. 2.2 ± 0.1 g/day in controls, n = 6–12/group, P = 0.33) were found comparing POMC^Cre/PPARγ-P467L to littermate control mice regardless of sex.

Fig. 4.

Body weight and composition during 25 wk of 10% fat, isocaloric-match diet. Body weight response to low-fat isocaloric control diet in male (A) and female (C) POMC^Cre/PPARγ-P467L (n = 10 male, n = 7 female) and control (n = 10 male, n = 13 female) mice. B and D: body composition in male POMC^Cre/PPARγ-P467L (n = 10) and control (n = 10) mice as measured by nuclear magnetic resonance. Data are expressed as means ± SE. *P < 0.0001; **P < 0.05.

Deletion of PPARγ throughout the entire brain (21) or antagonist-mediated inhibition of hypothalamic PPARγ was reported to increase leptin sensitivity (25), and ablation of PPARγ selectively in POMC neurons is sufficient to increase sensitivity to leptin (20). We therefore hypothesized that interference with PPARγ function in POMC neurons might alleviate leptin resistance caused by HFD. To test this, POMC^Cre/PPARγ-P467L mice fed either 10% fat isocaloric control diet or 60% HFD for 25 wk were subjected to twice daily ip injections of vehicle (PBS) or leptin (1 mg/kg) for 4 days. Control and POMC^Cre/PPARγ-P467L mice fed a low-fat diet exhibited an equivalent decrease in body weight (Fig. 5A) concomitant with a decrease in daily food intake (Fig. 5B), although this was only statistically significant in littermate controls. However, 1 wk following leptin treatment, the same cohort of POMC^Cre/PPARγ-P467L males had experienced a trend (P = 0.064) for a greater cumulative body weight loss (3.5 ± 0.58 g) compared with controls (1.9 ± 0.58 g). Despite this, a trend for increased mass in the subcutaneous adipose (0.64 ± 0.08 g vs. 0.48 ± 0.05 g in controls, P = 0.09) and BAT (0.17 ± 0.18 g vs. 0.13 ± 0.01 g in controls, P = 0.08) depots, but not in perigenital adipose depots was observed in POMC^Cre/PPARγ-P467L mice. Unlike the mice maintained on a low-fat diet, there was no cumulative weight loss (Fig. 5C) nor decrease in food intake (Fig. 5D) in control littermates or POMC^Cre/PPARγ-P467L mice fed a 60% HFD, suggesting that POMC^Cre/PPARγ-P467L mice develop leptin resistance when fed HFD just like control mice. There were no differences in female POMC^Cre/PPARγ-P467L mice between genotypes in response to leptin (data not shown).

Fig. 5.

Leptin sensitivity. Change in body weight (BW, A and C) or food intake (B and D) in response to vehicle or leptin (1.0 mg/kg) administration in male POMC^Cre/PPARγ-P467L (n = 9) and control (n = 10) mice fed 10% fat diet (A and B) or male POMC^Cre/PPARγ-P467L (n = 11) and control (n = 15) mice fed 60% high-fat diet (C and D). Data are expressed as means ± SE. *P < 0.01 vs vehicle.

To evaluate whether the increased body mass observed in POMC^Cre/PPARγ-P467L males is due to changes in energy expenditure and/or locomotor activity, a new cohort of male POMC^Cre/PPARγ-P467L (n = 7) and littermate control (n = 12) mice was fed the isocaloric control diet for 13 wk, a time point prior to the divergence in body weight. We hypothesized that if interference with PPARγ in POMC neurons leads to changes in metabolism or activity, which in turn induce the weight gain, such changes would be observable around the time body weight begins to diverge. However, there was no difference in energy expenditure compared with littermate controls in this cohort (0.34 ± 0.01 kcal/h vs. 0.34 ± 0.01 kcal/h in controls, P = 0.73, ANCOVA). Similarly, there was no difference in locomotor activity between genotypes (450.5 ± 88.0 beam breaks per 24 h in POMC^Cre/PPARγ-P467L vs. 440.9 ± 57.1 beam breaks in controls, P = 0.93). It is notable that when these mice were left to complete the diet period, body weight gain at 25 wk was not different in POMC^Cre/PPARγ-P467L (32.1 ± 1.0 g) relative to littermate controls (32.6 ± 0.61 g, n ≥ 6/group). This is interesting when one considers the differences in experimental cohorts. The first cohort was undisturbed throughout the entire feeding regimen except for weekly measurements of body weight. All other metabolic studies (Figs. 3 and 4) were performed at the end of the 30 wk test period. On the contrary, the second cohort underwent extensive metabolic phenotyping: 1) Respirometry was performed at 13 wk of age, 2) metabolic rate and other parameters were measured for 1 wk in a CLAMS starting at 14 wk of age, and 3) mice were housed singly in metabolism cages at 16 wk of age before being returned to their home cages for weeks 17–25. It is likely that the stress encountered in the second cohort could have interfered with the sensitivity of POMC^Cre/PPARγ-P467L mice to weight gain. Indeed, stress has been shown to strongly affect body weight measures in models of obesity (1).

Similar to POMC^Cre/PPARγ-P467L mice, male mice expressing wild-type PPARγ specifically in POMC neurons (POMC^Cre/PPARγ-WT) gained similar amounts of weight as their littermate controls 25 wk postfeeding with 60% HFD (Fig. 6A). A similar result was observed in female POMC^Cre/PPARγ-WT mice (Fig. 6B). Consistent with this, no differences were observed in body composition, fasting glucose, food intake, leptin sensitivity, and organ and fat pads weights between genotypes and within sexes in POMC^Cre/PPARγ-WT mice after 25 wk of HFD feeding (data not shown). However, unlike POMC^Cre/PPARγ-P467L mice, there was no difference in body weight in male POMC^Cre/PPARγ-WT mice in response to isocaloric control diet (data not shown). Female POMC^Cre/PPARγ-WTs were not placed on control diet because of low numbers.

Fig. 6.

Body weight during 25 wk of 60% high-fat diet in POMC^Cre/PPARγ-WT mice. Body weight in male (A) and female (B) POMC^Cre/PPARγ-WT (n = 9 male, n = 9 female) and control (n = 16 male, n = 14 female) mice during 60% high-fat diet. Data are expressed as means ± SE.

Administration of rosiglitazone into the 3rd ventricle or expression of constitutively active PPARγ specifically in the hypothalamus of rats causes hyperphagia and weight gain (25). As a control experiment to further validate the action of PPARγ in POMC neurons, we examined the body weight response to rosiglitazone in POMC^Cre/PPARγ-P467L mice employing the same protocol as Long et al. (20), who examined the response to peripheral administration of rosiglitazone in female POMC^Cre/PPARγ^Flox/Flox mice. Since POMC-specific deletion of PPARγ has been shown to lead to protection against the hyperphagic and body weight-increasing effects of rosiglitazone, we hypothesized that overexpression of PPARγ-WT in POMC neurons will have the opposite effects, i.e., will result in increased sensitivity to the drug. Intraperitoneal rosiglitazone administration to lean POMC^Cre/PPARγ-WT female mice exposed to 45% HFD resulted in a threefold increase in body weight compared with littermate control mice (Fig. 7A). In contrast to and consistent with transcriptional impairment of the PPARγ P467L, rosiglitazone failed to induce weight gain in female POMC^Cre/PPARγ-P467L mice compared with their littermate controls under the same condition as it did in POMC^Cre/PPARγ-WT mice (Fig. 7B). Neither transgenic mouse showed any significant changes in body weight gain in response to vehicle (saline) compared with nontransgenic control littermates (0.6 ± 0.58 g in POMC^Cre/PPARγ-WT vs. 0.6 ± 0.35 g in controls, P = 0.94; 0.5 ± 0.44 g in POMC^Cre/PPARγ-P467L vs. −0.5 ± 0.46 g in controls, P = 0.17).

Fig. 7.

Body weight response to rosiglitazone. A: the effect of rosiglitazone administration (28 mg/kg ip for 5 days) on body weight in lean POMC^Cre/PPARγ-WT (n = 8) and control (n = 11) female mice exposed to 45% high-fat diet. B: the effect of rosiglitazone administration (28 mg/kg ip for 5 days) on body weight in lean POMC^Cre/PPARγ-P467L (n = 7) and control (n = 7) female mice exposed to 45% high-fat diet. Data are expressed as means ± SE. *P < 0.01.

DISCUSSION

TZDs, high-affinity agonists of PPARγ, improve insulin sensitivity in Type II diabetes. This, along with the discovery of insulin-resistant patients with mutations in the ligand-binding domain of PPARγ, clearly demonstrated the importance of PPARγ for metabolic regulation (2). These mutations in PPARγ exhibit impaired transcriptional activity and act dominant-negatively by interfering with the activity of normal PPARγ (19). Data from our laboratory validated the DN activity of P467L PPARγ by showing that expression of P467L PPARγ causes repression of experimentally validated PPARγ target genes that are induced by PPARγ agonists (15, 16). Herein we utilized the same DN mutation occurring in humans (PPARγ-P467L) as a novel method to interfere with function of endogenous brain PPARγ. We generated transgenic mice overexpressing either the wild-type PPARγ (PPARγ-WT) or the mutant form of the receptor in specific neuronal populations and assessed the responses elicited by different dietary stressors. We suggest that expression of bona fide human disease-causing mutations in PPARγ is a better experimental system than genetic deletion because 1) null mutations do not occur in humans (because they are lethal), and 2) DN mutants prevent the loss of transcriptional repression caused by PPARγ in the unliganded state. For example, genetic ablation of PPARγ in smooth muscle leads to increased expression of β2-adrenergic receptor, a PPARγ target gene (5). Indeed, studies utilizing tissue-specific genetic deletion of PPARγ in mice failed to recapitulate hypertension (5, 31, 32), whereas mouse models with either a global (13, 30) or tissue-specific (3, 14, 17, 23) DN-PPARγ mutation have recapitulated the hypertensive phenotype observed in patients.

The results of our study demonstrate that overexpressing the wild-type PPARγ in POMC neurons acutely and significantly increases sensitivity to the body weight-accumulating effects of rosiglitazone. This is consistent with a previous study where administration of rosiglitazone in the 3rd ventricle of rats leads to acute induction of hyperphagia and weight gain (25). In addition, our findings corroborate those of Long et al. (20), where POMC-specific deletion of PPARγ renders the mice resistant to the combined effects of rosiglitazone and HFD. Interestingly, when we utilize a chronic model of HFD, mice overexpressing PPARγ-WT in POMC neurons show no difference in weight gain or body composition compared with littermate controls. The reason for this remains unclear. Perhaps saturating POMC neurons with additional PPARγ is not sufficient, and perhaps the presence of a ligand such as a TZD is required to provide the critical activation of the receptor. In the study by Lu et al. (21), the addition of rosiglitazone to the HFD led to significant body weight differences between global brain PPARγ knockouts and controls. This suggests that the body weight changes induced by manipulation of brain PPARγ are subtle and may require experimental paradigms manipulating not only the dietary variables but also the levels of agonists present.

The mice overexpressing PPARγ-P467L in POMC neurons did not gain weight when treated with a combination of rosiglitazone and 45% HFD, confirming that the mutant receptor is transcriptionally impaired and likely interferes with the ability of the endogenous wild-type form to be activated by the drug. Chronic exposure of the POMC^Cre/PPARγ-P467L to HFD, however, yielded no difference in weight gain compared with littermate controls. Saturating doses of a TZD agonist have been shown to overcome the transcriptional deficiencies in the mutant PPARγ-P467L (19). It remains unclear if chronic HFD leads to build up of PPARγ ligands in the brain that activate the mutant receptor resulting in similar phenotype as the littermate controls. On the contrary, HFD feeding leads to inactivation of PPARγ via phosphorylation of Ser273 (7). If this is occurring, HFD would emulate the DN mutant (which is transcriptionally impaired and inactivates endogenous) causing an equal level of impairment in littermate controls by phosphorylation and inactivation of endogenous PPARγ.

Interestingly, when the male POMC^Cre/PPARγ-P467L mice were subjected to low-fat isocaloric control diet, they gained significantly more weight compared with controls, suggesting that PPARγ expression in POMC neurons is a regulator of body weight and composition under certain dietary conditions. Furthermore, overexpression of PPARγ-P467L in the POMC neurons appears to promote adiposity in males under the conditions of control diet. Since the control diet is an isocaloric match to the HFD, carbohydrates are used to compensate for the decrease in fat calories. In this diet 70% of calories are derived from cornstarch. One possibility is that interference with PPARγ in POMC neurons sensitizes to the presence of increased glucose. One potential limitation of our study is that a second cohort of POMC^Cre/PPARγ-P467L treated with control diet for 25 wk did not replicate the enhanced weight gain of the first set of experimental animals. A similar discrepancy is evident in the Lu et al. (21) study when comparing two separate cohorts of HFD fed POMC^Cre/PPARγ^Flox/Flox mice (compare Fig. 2A to 3A in that paper). The reasons for this discrepancy in our study are not clear but may be due to the experimental manipulations related to energy expenditure measurements that the second cohort was subjected. This raises the possibility that weight gain in POMC^Cre/PPARγ-P467L mice may be sensitive to handling or stress, a finding consistent with other studies (1).

The effects of the isocaloric low-fat diet were sex dependent as interference with PPARγ in POMC neurons does not lead to changes in energy balance in female mice. The reason for the difference remains unclear. A previous study examining the effect of central PPARγ activation and inhibition in rats examined only males (25). Similarly, a study examining the effect of neuronal-specific deletion of PPARγ on food intake and body weight was performed solely in males (26). On the contrary, a POMC neuron-specific deletion of PPARγ resulted in a blunting of weight gain in both female and male mice in response to HFD, although the difference between wild-type and POMC-specific knockout males emerged a later time point (20). As described above, it is certainly possible that POMC neuron-specific deletion of PPARγ and POMC-specific expression of DN-PPARγ are not functionally equivalent at the molecular or physiological levels. Interestingly, a strong sexual dimorphism is observed in mice selectively expressing POMC in a 5-hydroxytryptamine 2c receptor (5-HT_2cR) expressing population of POMC neurons (4). Whereas male mice selectively expressing POMC in 5-HT_2cR exhibit a reversal of obesity and hyperinsulinemia (compared with mice lacking POMC in the hypothalamus), the females exhibit lower energy expenditure, impaired BAT activity, and obesity despite normalized food intake and insulin. This study is interesting in identifying a heterogeneity in POMC neurons in the hypothalamus and a sexual dimorphism in the function of the 5-HT_2cR/POMC neuron with regard to energy homeostasis. Consequently, it would be interesting to determine if there are differences in the function or expression of PPARγ in 5-HT_2cR-positive vs 5-HT_2cR-negative neurons and if our transgene was selectively activated in one population.

POMC-specific deletion of PPARγ leads to improved sensitivity to leptin in HFD-fed mice (20). Leptin treatment decreased food intake and body weight in low fat diet-fed mice and did so equivalently in POMC^Cre/PPARγ-P467L and controls. This suggests that under control conditions, interference with PPARγ in POMC neurons does not cause leptin resistance. The fact that there was a trend for a further decrease in body weight 1 wk after leptin suggests PPARγ interference in POMC neurons may facilitate leptin sensitivity, further suggesting that, chronically, PPARγ in POMC neurons may antagonize leptin. This would be consistent with the observation that brain-specific administration of TZDs increases appetite, an effect blunted by inhibition of PPARγ in the brain (25). However, there was no change in body weight or food intake in POMC^Cre/PPARγ-P467L and control mice after HFD, suggesting an equal degree of leptin resistance. It is likely that after 25 wk of HFD exposure the mice have developed significant hyperleptinemia and secondary leptin resistance such that a dose of 1.0 mg/g is not sufficient to induce a decrease in food intake. It is also possible that interference with PPARγ in POMC neurons may lead to altered response of these neurons to adipokines other than leptin, which may in turn contribute to the increased weight gain observed in these mice. Furthermore, since POMC neurons are only a subset of leptin-responsive neurons in the brain, it is possible that compensation by other neurons where the DN-PPARγ is not being expressed has offset those evoked by DN-PPARγ in POMC neurons. To further test the effect of DN-PPARγ on leptin signaling in the brain, one possibility is to use double transgenic mice in which the transgene is driven by the leptin receptor promoter.

GRANTS

This work was supported through research grants from the NIH to C. D. Sigmund (HL-084207, HL-048058, HL-062984, HL-125603) and K. Rahmouni (HL-084207), grants from the American Heart Association to C. D. Sigmund (15SFRN23480000) and K. Rahmouni (14EIA18860041), and the University of Iowa Fraternal Order of Eagles Diabetes Research Center to K. Rahmouni. The authors gratefully acknowledge the generous research support of the Roy J. Carver Trust.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the author(s).

AUTHOR CONTRIBUTIONS

M.S., K.R., and C.D.S. conception and design of research; M.S., D.-F.G., K.-T.L., M.M., and X.L. performed experiments; M.S., D.-F.G., K.-T.L., M.M., X.L., and C.D.S. analyzed data; M.S., D.-F.G., K.-T.L., M.M., X.L., K.R., and C.D.S. interpreted results of experiments; M.S., D.-F.G., K.-T.L., M.M., X.L., and C.D.S. prepared figures; M.S., K.R., and C.D.S. drafted manuscript; M.S., D.-F.G., K.-T.L., M.M., X.L., K.R., and C.D.S. edited and revised manuscript; M.S., D.-F.G., K.-T.L., M.M., X.L., K.R., and C.D.S. approved final version of manuscript.