Beta-Carotene Reduces Body Adiposity of Mice via BCMO1

Evidence from cell culture studies indicates that β-carotene-(BC)-derived apocarotenoid signaling molecules can modulate the activities of nuclear receptors that regulate many aspects of adipocyte physiology. Two BC metabolizing enzymes, the BC-15,15′-oxygenase (Bcmo1) and the BC-9′,10′-oxygenase (Bcdo2) are expressed in adipocytes. Bcmo1 catalyzes the conversion of BC into retinaldehyde and Bcdo2 into β-10′-apocarotenal and β-ionone. Here we analyzed the impact of BC on body adiposity of mice. To genetically dissect the roles of Bcmo1 and Bcdo2 in this process, we used wild-type and Bcmo1 -/- mice for this study. In wild-type mice, BC was converted into retinoids. In contrast, Bcmo1-/- mice showed increased expression of Bcdo2 in adipocytes and β-10′-apocarotenol accumulated as the major BC derivative. In wild-type mice, BC significantly reduced body adiposity (by 28%), leptinemia and adipocyte size. Genome wide microarray analysis of inguinal white adipose tissue revealed a generalized decrease of mRNA expression of peroxisome proliferator-activated receptor γ (PPARγ) target genes. Consistently, the expression of this key transcription factor for lipogenesis was significantly reduced both on the mRNA and protein levels. Despite β-10′-apocarotenoid production, this effect of BC was absent in Bcmo1-/- mice, demonstrating that it was dependent on the Bcmo1-mediated production of retinoids. Our study evidences an important role of BC for the control of body adiposity in mice and identifies Bcmo1 as critical molecular player for the regulation of PPARγ activity in adipocytes

Increasing evidence has been provided that BC-derived apocarotenoid signaling molecule can influence adipocyte physiology. Bcmo1 knockout mice are highly susceptible to high fat diet-induced obesity and show increased expression of peroxisome proliferator-activated receptor c (PPARc) regulated genes in fat depots [7]. PPARs control the expression of genes for lipid and glucose metabolism [10;11] and PPARc is pivotal for adipocyte differentiation and lipogenesis in mature adipocytes [12]. Bcmo1 gene expression is under the control of PPARc [13;14] and is induced during adipocyte differentiation [15]. The primary BC cleavage product retinaldehyde has been shown to inhibit PPARc activity both in adipocyte cell cultures and mouse models [16]. Furthermore, evidence has been provided that Bcmo1 plays an important role for retinoic acid production and retinoic acid receptor (RAR) signaling in adipocytes [15]. Retinoic acid influences adipocyte differentiation [17;18] and fat deposition [19], mitochondrial uncoupling [20;21], oxidative metabolism [22;23] and the expression of adipokines such as leptin, resistin, and serum retinol binding protein [24][25][26][27]. Part of these effects are mediated via RARs, which upon retinoic acid binding regulate the expression of direct target genes [28] and interfere with the activity of other transcription factors, including early adipogenic transcription factors such as PPARc [17;18]. In addition, retinoic acid may influence PPAR-mediated responses by activating the retinoid X receptor (RXR) moiety of permissive PPAR:RXR heterodimers [29] and, possibly, by serving as an agonist of PPARb/d [30;31]. Finally, BC-derived long chain apocarotenoids such as b-apo-149-carotenal can inhibit PPARc activity and adipogenesis in cell culture [32]. Moreover, b-13-apocarotenone has been shown to inhibit RXRa activity [33].
These findings implicate that a tissue-specific conversion of BC via carotenoid-oxygenases can influence the activities of key transcription factor that control adipocyte physiology. However, this concept lacks experimental testing in animal models. Here we wished to systematically analyze the effects of BC on the adipose phenotype in mice. To genetically dissect the contributions of Bcmo1 and Bcdo2 to this process, we used wild-type (WT) and Bcmo1 -/mice in this study. Our analysis included a genome wide expression analysis of inguinal white adipose tissue (iWAT) to identify the target genes of BC action in this tissue.

Effects of dietary BC supplementation on BC and retinoid levels in serum and adipose tissue
Five-week-old WT and Bcmo1 -/female mice were fed a defined, pelletized diet containing 1500 IU vitamin A/kg with (BC diet) or without (control diet) 150 mg BC/kg diet (n = 6 animals per group). After 14 weeks, we first studied expression of Bcmo1 in iWAT of different dietary groups and genotypes. Bcmo1 was expressed in iWAT of WT but not Bcmo1 -/mice ( Figure 1A). We also analyzed the mRNA expression levels of Bcdo2, which encodes a second carotenoid metabolizing enzyme [5]. This analysis revealed a marked up-regulation of Bcdo2 expression in iWAT of Bcmo1-null mice, both with control and BC diet ( Figure 1B). Similar results were obtained for Bcmo1 and Bcdo2 expression in gonadal WAT ( Figure S1).
Serum and iWAT BC levels were significantly increased after the 14 weeks of dietary intervention in all animals subjected to BC supplementation ( Figures 1C and 1E). This increase was much more pronounced in the Bcmo1 -/mice, in which BC accumulation was readily evidenced by an orange coloring of the fat depots (not shown). Serum retinol levels were similar regardless of genotype or diet ( Figure 1D). iWAT total retinol levels (free retinol plus retinyl ester) were increased upon the BC diet in WT but not in Bcmo1 -/mice ( Figure 1F). In Bcmo1 -/mice supplemented with BC, a compound that migrated shortly after all-trans-retinol with an absorption maximum of 402 nm was present (Fig. 1G). Spectral characteristics and retention time indicated that this compound is b-109-apocarotenol, the corresponding alcohol of the primary Bcdo2 cleavage product b-109apo-carotenal. To confirm this hypothesis, we used a b-109apocarotenal standard and reduced it to the corresponding alcohol by BH 4 treatment. In fact, this b-109-apocarotenol standard had the same retention time and spectral characteristics of the compound that accumulated in Bcmo1 -/mice upon BC supplementation (Fig 2A and 2B). Finally, we confirmed the identity of this compound by tandem mass spectrometry as compared b-109-apocarotenol standard substance (Fig. 2C-F). Thus, we concluded that in WT mice supplemented BC is largely converted to retinoids. When BC accumulated in Bcmo1 -/mice, it is converted in part to b-109-apocarotenal that is further reduced to the corresponding alcohol.

Dietary BC reduces adiposity in WT mice
Bcmo1 -/mice had a lower average body weight than WT mice at the beginning of the study (17.360.4 g versus 18.560.2 g; p,0.05 Student's t test) and showed a tendency of higher body weight gain over the 14 week experimental period ( Figure 3A). Body weight at the end of the study was comparable in all groups (data not shown). Cumulative energy intake over the 14 week experimental period was significantly higher in mice fed the BC diet than in those fed the control diet, and in Bcmo1 -/mice as compared to WT mice ( Figure 3B). Energy efficiency (body weight gained per energy eaten) was similar among groups ( Figure 3C). Even though BC supplementation did not affect final body weight in either genotype, it significantly decreased the percentage of body fat (by 28%, p,0.01) in WT but not in Bcmo1 -/mice ( Figure 3D). In WT mice, reduction of depot mass after BC diet was evident for all three WAT depots examined (inguinal, gonadal and retroperitoneal) ( Figure 3E); no such effect of BC on fat depots was observed in the Bcmo1 -/mice. Leptin levels in serum reflected changes in body adiposity, being reduced following dietary BC supplementation in the WT mice but not in the Bcmo1 -/mice (Table 1). There was a direct, positive correlation between serum leptin levels and iWAT mass (r = 0.617, p,0.001). Morphometric analysis showed that the mean sectional area of unilocular iWAT adipocytes was reduced after dietary BC supplementation in the WT mice ( Figure 3F and 3G). Mean iWAT adipocyte area was similar in the two genotype groups after the control diet, but significantly higher in the Bcmo1 -/mice after the BC diet. Circulating levels of insulin, glucose and triacylglycerol were unaffected by diet or genotype (Table 1).

Dietary BC results in a general down-regulation of gene expression in adipose tissue of WT mice
To analyze the effects of BC supplementation on transcriptional activities in adipose tissue in different genotypes we performed genome-wide expression analysis of iWAT of all individual animals. Principal component analysis of microarray results revealed a different gene expression signature in WT and Bcmo1-null mice, and a clear effect of dietary BC supplementation that changed the overall gene expression signature in the WT mice, but not in the Bcmo1-null mice ( Figure 4A). From the 25139 spots with signal intensities above two times the background, 1597 positive spots were differentially expressed due to BC supplementation in the WT mice and only 84 in the Bcmo1-null mice (p,0.01) ( Figure 4B). These 84 positive spots correspond to 81 genes, that are given in Table S1 and are likely regulated by BC or b-109-apocarotenoid accumulation. None of the genes affected by BC supplementation in the WT mice (p,0.01) were affected in the Bcmo1-null mice ( Figure 4B, 0 genes overlapping). In the WT mice, genes significantly regulated (p,0.01) by BC supplementation appeared to differ most in expression as compared to the other three groups; WT control diet, Bcmo1 -/control diet and Bcmo1 -/-BC diet ( Figure 4D). Remarkably, upon BC supplementation, 87% of the genes affected in WT mice were found to be down-regulated ( Figures 4C and 4D). Pathway analysis using Metacore showed that genes encoding components of lipid and glucose metabolism-related pathways (including cholesterol biosynthesis, fatty acid metabolism, glycerophospholipid metabolism, pentose phosphate pathway, pyruvate metabolism and glycolysis/ gluconeogenesis) as well as ATP synthesis and oxidative phosphorylation were among the most affected (Table S2). Because of the close connections between neovascularisation and adipose tissue expansion [34][35], we specifically examined changes in the expression of angiogenesis-related genes. As expected, BC supplementation triggered a generalized down-regulation of these genes in the WT mice, but not in the Bcmo1 -/mice (Table S3). Dietary BC down-regulates PPARc and PPARc target genes in adipose tissue of WT mice A general effect of BC supplementation in reducing global transcriptional activity in iWAT of WT mice suggested that the activity of a master control transcription factor of adipose tissue biology was affected. In this context, a batch promoter analysis with Eldorado coupled to matinspector (Genomatix Suite) revealed that 54% of the top 50 down-regulated genes in iWAT of BC supplemented WT mice harbored a putative PPAR response element in their promoter region (p value: 1.33e 26 ) with at least 85% homology with the matrix similarity defined by Genomatix (Table 2). Interestingly, the same kind of analysis in iWAT of BC supplemented Bcmo1 -/mice revealed that the presence of putative PPAR responsive element was not significant (p value: 0.81). Microarray results indicated that iWAT gene expression of PPARc and its heterodimeric partner RXRa were significantly decreased following BC supplementation in the WT mice (-1.75 and 21.57 fold-change, respectively, p,0.05), but not in the Bcmo1 -/mice. Q-PCR analysis confirmed the microarray results and thus down-regulation of the mRNA expression levels of PPARc and RXRa after BC supplementation in iWAT of WT mice, but not of Bcmo1 -/mice ( Figure 5A and 5B). Additionally, PPARc protein expression levels in iWAT followed the same pattern as the mRNA levels, as indicated by immunoblotting results ( Figure 5D). In contrast, iWAT mRNA levels of CYP26a1, a retinoic acid hydroxylase that is transcriptionally induced by retinoic acid in a strictly RAR-dependent manner [36;37], were increased by 138% after BC supplementation in the WT mice, but not in the Bcmo1 -/mice ( Figure 5E), suggesting a Bcmo1-dependent enhancement of retinoic acid signalling in iWAT following chronic BC supplementation. Finally, to provide direct evidence for down-regulation of PPARc activity after BC supplementation in the WT mice, we determined iWAT mRNA levels of LPL, a well-known target of PPARc:RXRa, by Q-RT-PCR. As shown in Fig. 5C, LPL mRNA levels were significantly decreased in WT mice upon BC supplementation. Changes in gene expression of PPARc, RXRa and LPL in the gonadal fat depot followed the same pattern as in the inguinal depot ( Figure S1).

Discussion
Here we analyzed the effects of dietary BC on whole-body adiposity and adipose tissue physiology. We supplemented WT and Bcmo1 -/mice with BC and compared them to nonsupplemented littermates. Our results revealed a Bcmo1-dependent, adiposity reducing effect of BC supplementation, as percent body fat, adipose depot mass, iWAT adipocyte size and leptin serum levels were all significantly reduced after BC supplementation in the WT mice, but not in the Bcmo1 -/mice ( Figure 3). Reduced adiposity in WT mice after BC supplementation was not associated with adverse effects on blood parameters measured such as increased circulating triacylglycerol levels. Moreover, no harmful effects of BC supplementation on parameters such as body weight gain, food intake, or lung tissue histology were observed in the animals (this work and [38]). Although relatively low dietary vitamin A was fed, blood retinol levels were well maintained and within the expected range for adult mice in the different genotypes and supplementation groups ( Figure 1D). Additionally, stored retinol was found in the adipose tissue ( Figure 1F) and in the liver (data not shown), indicating no gross changes in vitamin A homeostasis in animals of different supplementation groups.
Microarray analyses revealed that BC supplementation altered the overall iWAT gene expression signature in WT mice but not in Bcmo1 -/mice. Strikingly, none of the genes altered in its expression levels upon BC supplementation in the WT mice (p,0.01) was changed in Bcmo1 knockout mice ( Figure 4). BC supplementation led to a general down-regulation of gene expression in iWAT of WT mice. This finding indicated that the activity of major transcriptional regulator of iWAT was affected by BC supplementation. Microarray, Q-PCR and immunoblotting analysis consistently confirmed a reduction of PPARc expression Table 1. Fed blood parameters in wild-type and Bcmo1 -/mice following control and b-carotene enriched diet.    Figure 5A and 5D). In silico analysis of the promoter regions of the top 50 BC down-regulated genes in wild-type mice showed that more than 50% contained a PPAR response element ( Table 2). The p value associated (1.33e 26 ) strongly supported the importance of PPARc within this gene set, since it represent the probability to obtain an equal or greater number of sequences with a match in a randomly list of genes. Interestingly, such result was not reproduced in the Bcmo1 -/mice (p value 0.81; not significant) underlying the key role of this enzyme in the repression of PPARc activity by BC. Thus, the reduction of PPARc activity and the down-regulation of its target genes likely explain the reduced adiposity of WT mice upon BC supplementation.   Centric cleavage of BC by Bcmo1 leads to retinaldehyde production which can be converted to retinol and retinyl ester for storage and to retinoic acid. Enzymes catalyzing these reactions are expressed in adipocytes [7;16]. Retinoids such as retinoic acid and retinaldehyde are ligands for nuclear receptors including RARs and PPARs that help control adipogenesis and adipocyte metabolism. Treatment with retinoic acid has been shown to inhibit differentiation of pre-adipose cells in culture by blocking PPARc induction in an RAR-mediated manner [18;39], and to result in vivo in mice in reduced PPARc expression levels in WAT [19] and reduced whole body adiposity [19;31]. Moreover, a recent study showed that BC, but not retinol, decreases PPARc and reduces lipid storage capacity in mature white adipocytes in culture [15]. This effect involves production of retinoic acid from BC, is blocked by a Bcmo1 chemical inhibitor (fenretinide), and is dependent on RARs [15]. Importantly, BC was accumulated following BC diet in serum and tissues of WT mice ( Figure 1C and 1E), suggesting that not all BC undergoes intestinal cleavage in WT mice. Thus, an adipose tissue-specific, Bcmo1-dependent production of retinoids, such as retinoic acid, from BC might explain the down-regulation of iWAT PPARc expression levels and the overall changes in iWAT gene expression patterns and adiposity-related end-points observed in this work in WT mice following chronic exposure to a BC-enriched diet. The involvement of RAR-signalling is sustained by the CYP26a1 geneexpression data showing increased expression of this retinoic acid target gene following BC supplementation selectively in the iWAT of WT mice ( Figure 5E). Additionally, iWAT mRNA levels of some other genes for which there is evidence of transcriptional upregulation or down-regulation in response to retinoic acid were significantly increased and decreased, respectively, following BC supplementation in the WT mice, but not in the Bcmo1-null mice (microarray data shown in Table S4), including leptin. Leptin expression has been shown to be decreased by retinoic acid [24][25][26][27] and blood levels of this adipokine were significantly reduced in WT mice supplemented with BC.
Interestingly, expression of the Bcdo2 gene was up-regulated in iWAT of Bcmo1-null mice, both under control and BC diet ( Figure 1B). In fact, relatively large amounts of b-109-apocarotenol became detectable in WAT of Bcmo1 -/mice upon BC supplementation. This compound derives by the reduction of the primary BC cleavage product b-109-apo-carotenal. It has been reported that apocarotenoids such as b-14-apo-carotenal can repress PPARc -and RXR-mediated responses in cell culture [32]. However, we found not much evidence that apocarotenoid formation in Bcmo1 -/mice influenced adipocyte physiology. Our results indicate that the effects of BC on adipocyte biology are clearly dependent on Bcmo1, as those effects were not found in Bcmo1-deficient animals that express Bcdo2.
In summary, this study revealed a Bcmo1-dependent, adiposity reducing effect of long-term BC supplementation in mice. In this context, we provide evidence that BC-derived retinoids influence adipocyte physiology. Bcmo1 is a PPARc target gene [13], and our analysis indicates that the Bcmo1-dependent, adiposity-reducing effect of BC is mediated by a reduction of PPARc activity in adipocytes. Thus, our findings establish Bcmo1 as key component for the crosstalk between RAR-and PPARc-signaling pathways in the control of body adiposity.
Our results may also have implications in the context of the current obesity pandemics. There is evidence of an impact of BC and vitamin A status on parameters related to adiposity in humans [40]. Circulating BC levels are inversely correlated with risk of human type 2 diabetes, a pathology associated with obesity [41][42][43][44] . Additionally, reduced plasma levels of carotenoids, including BC, are commonly found in obese children [45]. Genetic polymorphisms in the Bcmo1 gene have been described that alter the activity of the encoded human protein [46;47]. Our work suggests that BC intake may have different consequences on body adiposity in subjects carrying different Bcmo1 functional gene variants. Thus, the involvement of BC and Bcmo1 in human obesity deserves further research.

Animals and diets
The animal experiment was conducted according to accepted standards of humane care and use of laboratory animals, and was approved by the bioethical committee of the University of Freiburg (Az.: 35/9185.81/G-07/30). C57/BL6;129Svj-Bcmo1 tm1Dnp (Bcmo1 -/-) mice previously described by Hessel et al. [7] and mixed background F1 generation of C57/BL6;129Svj mice as WT controls were used. During the breeding and weaning periods and up to 5 weeks of age all mice were maintained on breeder chow containing 14000 IU vitamin A/kg diet (Provimi Kliba AG, Kaiseraugst, Switzerland). Five-week-old WT and Bcmo1 -/female mice were fed a defined, pelletized diet containing 1500 IU vitamin A/kg and 10% energy as fat, with (BC diet) or without (control diet) 150 mg BC/kg diet (n = 6 animals per group) for 14 weeks, after which they were euthanized. Detailed composition of these diets is shown in Table 3. A relatively low dose of dietary vitamin A was used to maximize intestinal BC absorption, which can be repressed by preformed vitamin A [48]. The dose used was previously shown to be sufficient to maintain vitamin A homeostasis while allowing tissue BC accumulation both in WT and Bcmo1 -/mice [9]. The experimental diets were prepared by Research Diets, Inc (New Brunswick, NJ, USA) by cold extrusion so as to protect BC from heat treatment and incorporating a water-soluble formulation in the form of beadlets containing BC, DL-a-tocopherol, ascorbyl palmitate, corn oil, fish gelatin, sucrose and corn starch. The control diet contained control beadlets devoid of BC, but with all other ingredients. All beadlets were kindly provided by DSM Nutritional Products Ltd (Basel, Switzerland). Mice were maintained under environmentally controlled conditions (24uC, 12 h/ 12 h light/dark cycle) in groups of 3 animals per cage and with ad libitum access to feed and water. Body weight and food intake were determined twice per week (3 and 4 days intervals). Energy intake was estimated on a per-cage basis from the actual amount of food consumed by the animals and its caloric equivalence (16 MJ/kg for both control and BC diet). On the day of the necropsy, the animals were sacrificed according to the animal protocol of the University of Freiburg (Germany). First, animals were subjected to inhalation of isofluran. Then, 100-150 mg/kg KGW Ketamin (Ketamin 10%, Essex) and 5-10 mg/kg KGW Xylazin (Rompun 2 %, Bayer) were injected intraperitoneally. Blood was drawn from the vena cava, and animals were then sacrificed by cervical dislocation. Inguinal (iWAT), gonadal (periovarian) and retroperitoneal white adipose tissue (WAT) depots were excised in their entirety according to Cinti [49], weighed, snap-frozen in liquid nitrogen and stored at -80uC until analysis. The continuous subcutaneous fat depot comprising the dorsolumbar and inguinal areas was taken as iWAT. The combined mass of all WAT pads taken expressed as a percentage of body weight was used as adiposity index. For three animals in each group, one entire iWAT lobule was fixed for histological studies.

Histology and immunohistochemistry
Entire iWAT lobules were fixed by immersion in the perfusion fixative (4% paraformaldehyde in 0.1 M sodium phosphate buffer, pH 7.4) overnight at 4uC, dehydrated, cleared and then paraffin-embedded so that the plane of section corresponded with the one of the wider surface. 3 mm-thick sections at the same level were obtained and stained with hematoxylin/eosin for morphometric analysis, performed by digital acquisition of adipose tissue areas (Digital Still Camera DXM 1200 and NIKON 6000 Eclipse Microscope). Two histological sections were counted per mouse.

Serum parameters
Commercial enzymatic colorimetric kits were used for the determination of circulating levels of triacylglycerol (Triglyceride (INT) 20, Sigma Diagnostics, St Louis, MO, USA) and glucose (D-Glucose UV-method, Roche Biopharm GmbH, Darmstadt, Germany). Serum levels of leptin and insulin were measured as described before [50] with some minor differences. Briefly, sera were diluted 4x in High Performance ELISA buffer (Sanquin, Amsterdam, The Netherlands) before adipokine analyses using the mouse serum adipokine Lincoplex Kit (Linco Research, Nuclilab, Ede, The Netherlands). The assays were conducted according to the manufacturer's protocol and measured using the Bio-PLex (Bio-Rad, Venendaal, The Netherlands) system with Bio-Plex software. All individual samples were analyzed in duplicate and averaged.

Analysis of retinoids, apocarotenoids, and carotenoids
Retinoids and carotenoids were extracted from serum under dim red safety light (600 nm) as described previously [7]. Adipose tissue (,40 mg) was incubated with 100 mL of 12% pyrogallol in ethanol, 200 mL of 30% KOH in water and 1 ml ethanol for 2 h at 37uC. After saponification, 1 ml H 2 O was added. The samples were extracted with 3 ml of ether/hexane (2:1, stabilized with 1% ethanol) twice, and diluted with 2 ml H 2 O and 2 ml ethanol. After centrifugation for 5 min at 8006 g, the organic layer was collected. The extract was evaporated in a Speedvac and dissolved in HPLC solvent. HPLC separation of carotenoids and retinoids and quantification of peak integrals was performed as described previously [6]. Solvents for HPLC and extraction were purchased in HPLC grade from Merck (Darmstadt, Germany). b-109-apocarotenal was a gift of Dr. Hansgeorg Ernst (BASF, Germany). For b-109-apocarotenol production, the standard was dissolved in 1 ml ethanol and reduced with BH 4 on ice. After 20 min, reaction products were extracted upon addition of 1 ml H 2 O with 1 ml hexanes. The organic phase was dried with a speedvak and redissolved in hexanes and subjected to HPLC analysis. The identity of b-109-apocarotenol was further confirmed by LC/MS. The saponified lipophilic extract from iWAT was injected onto a Zorbax Sil column (Agilent Technologies, Santa Clara, CA) equilibrated with 90 % hexane and 10 % acetylacetate (v:v). The eluant was directed into LXQ mass spectrometer through an atmospheric pressure chemical ionization source (Thermo Scientific Waltham, MA) working in a positive mode. b-109-apocarotenol was identified based on its molecular mass (m/z = 361. 42 [MH] + ) and a fragmentation pattern of the parent ion identical with that of the b-109apocarotenol standard.

RNA isolation
WAT was homogenized in liquid nitrogen using a liquid nitrogen cooled mortar and pestle. Total RNA was isolated using TRIzol reagent (Invitrogen, Carlsbad, CA, USA) followed by purification using RNeasy columns (Qiagen, Venlo, The Netherlands) according to the instructions of the manufacturer. RNA concentration and purity were measured using the Nanodrop spectrophotometer (ND-1000, Isogen Life Science, Maarssen, The Netherlands). Approximately 20 mg of total RNA was isolated from each WAT depot. A260/A280 ratios were above 2 and A260/A230 ratios were above 1.9 for all samples, indicating good RNA purity. RNA degradation was checked using StdSense Chips (Biorad) and no apparent RNA degradation was seen.

Western blot analysis
For PPARc determination in iWAT, 100 mg of total protein from each animal tissue was fractionated in SDS-polyacrylamide (12% acrylamide) and then electroblotted to a PVDF membrane. Membranes were blocked in fat-free milk (5%) dissolved in Tris buffered saline-Tween, washed and incubated overnight at 4uC with anti-mouse PPARc (Abcam, Cambridge, MA) diluted 1:500. As a protein loading control, b-actin (Cell Signalling, Boston, MA) diluted 1:1000 was used. Secondary antibody (1:5000) was horseradish peroxidase-conjugated anti-rabbit IgG (Promega, Madison, WI). Immunoblots were developed with the ECL system (Amersham Biosciences) and scanned. Quantification of bands was performed by the ImageJ (Java).

Microarray hybridization procedure
The 4 x 44 k Agilent whole mouse genome microarrays (G4122F, Agilent Technologies, Inc. Santa Clara, CA, USA) were used and microarray hybridization was performed essentially according to the manufacturer's protocol. In brief, cDNA was synthesized out of 1 mg iWAT RNA using the Agilent Low RNA Input Fluorescent Linear Amplification Kit for each animal without addition of spikes. Thereafter samples were split in 2 equal amounts, to synthesize Cyanine 3-CTP (Cy3) and Cyanine 5-CTP (Cy5) labeled cRNA using half the amounts as indicated by the manufacturer per dye (Agilent Technologies) and as has been assessed previously [50]. Labeled cRNA was purified using RNeasy columns (Qiagen). Yield, A260/A280 ratio and Cy3 or Cy5 activity were examined for every sample on the nanodrop. All samples met the criteria of a cRNA yield higher than 825 ng and a specific activity of at least 8.0 pmol Cy3 or Cy5. 1200 ng of every Cy3 labeled cRNA sample was pooled and used as a common reference pool. Individual 825 ng Cy5-labeled cRNA and 825 ng pooled Cy3-labeled cRNA were fragmented in 1x fragmentation and 1x blocking agent (Agilent Technologies) at 60uC for 30 minutes and thereafter mixed with GEx Hybridization Buffer HI-RPM (Agilent Technologies) and hybridized in a 1:1 ratio at 65uC for 17 h in an Agilent Microarray hybridization chamber rotating at 4 rpm. After hybridization slides were washed according to the wash protocol with Stabilization and Drying solution (Agilent Technologies). Arrays were scanned at an Agilent scanner with 10% and 100% laser power intensities (Agilent Technologies).

Data analyses and statistical methods
Signal intensities for each microarray spot were quantified using Feature Extraction 9.1 (Agilent Technologies). Median density values and background values of each spot were extracted for both the experimental samples (Cy5) and the reference samples (Cy3). Quality control was performed visually, by using Quality control graphs from Feature extraction and M-A plots and boxplots were made using limmaGUI in R (Bioconductor) [52] for every microarray. Data were imported into GeneMaths XT (Applied Maths, Sint-Martens-Latem, Belgium). Spots with a Cy5 and Cy3 signal twice above background were selected and log transformed. 31295 spots had at least a 2-fold signal to noise ratio. The Cy5 signal was normalized against the Cy3 intensity as described before [53]. P-values for differential expressions were calculated between 2 groups using t-test statistics on log intensity values. Changes were considered statistically different at p,0.05. Vulcano plots were made using Graphpad Prism 5.0.2 (La Jolla, CA) with all spots or a selection of spots. Heatmapping and hierarchical clustering was performed in GeneMathsXT (Applied Maths, St.Martens Lathem, Belgium). Pathway analysis was performed using MetaCore (GeneGo Inc, St. Joseph, MI, USA). Promoter analysis and the literature based network were performed with the Genomatix suite (Genomatix Software GmbH, Germany). Biochemical and biometric parameters in Figures 1, 2, and 4 are presented as means 6 SEM. Statistical significance of effects on these parameters was assessed by two-way ANOVA and two-tailed Student's t test using SPSS 14.0 for windows (SPSS, Chicago, IL, USA), with threshold of significance set at p,0.05. Microarray data has been submitted in a MIAME compliant form to Gene Expression Omnibus (GEO) data base of the National Center of Biotechnology Information (NCBI),, and is accessible through GEO series accession number GSE27271 (http:// www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc = GSE27271).