Hepatic Gene Expression Profiling Reveals Key Pathways Involved in Leptin-Mediated Weight Loss in ob/ob Mice Authors: Ashok Sharma, Shoshana M. Bartell, Clifton A. Baile, Bo Chen, Robert H. Podolsky, Richard A. McIndoe, Jin-Xiong She Published: 2010, PLOS ONE DOI: 10.1371/journal.pone.0012147 ABSTRACT Leptin, a cytokine-like protein, plays an important role in the regulation of body weight through inhibition of food intake and stimulation of energy expenditure. Leptin circulates in blood and acts on the brain, which sends downstream signals to regulate body weight. Leptin therapy has been successful in treating leptin deficient obese patients. However, high levels of leptin have been observed in more common forms of obesity indicating a state of leptin resistance which limits the application of leptin in the treatment of obesity. If the central effect of leptin could be by-passed and genes which respond to leptin treatment could be regulated directly, new therapeutic targets for the treatment of obesity may be possible. The purpose of this study was to identify genes and subsequent pathways correlated with leptin-mediated weight loss. We utilized microarray technology to compare hepatic gene expression changes after two types of leptin administration: one involving a direct stimulatory effect when administered peripherally (subcutaneous: SQ) and another that is indirect, involving a hypothalamic relay that suppresses food intake when leptin is administered centrally (intracerebroventricular: ICV). We identified 214 genes that correlate with leptin mediated weight loss. Several biological processes such as mitochondrial metabolic pathways, lipid metabolic and catabolic processes, lipid biosynthetic processes, carboxylic acid metabolic processes, iron ion binding and glutathione S-transferases were downregulated after leptin administration. In contrast, genes involved in the immune system inflammatory response and lysosomal activity were found to be upregulated. Among the cellular compartments mitochondrion (32 genes), endoplasmic reticulum (22 genes) and vacuole (8 genes) were significantly over represented. In this study we have identified key molecular pathways and downstream genes which respond to leptin treatment and are involved in leptin-mediated weight loss. Many of these genes have previously been shown to be associated with obesity; however, we have also identified a number of other novel target genes. Further investigation will be required to assess the possible use of these genes and their associated protein products as therapeutic targets for the treatment of obesity. INTRODUCTION Leptin, a cytokine-like protein containing 167 amino acids that is predominantly expressed in adipocytes, was discovered in 1994 by positional cloning of the obese ( ob ) gene [1] . Leptin plays a very important role in regulation of body weight and fat deposition through the inhibition of food intake and stimulation of energy expenditure [2] . Leptin-deficient ( ob/ob ) mice and leptin receptor-deficient ( db/db ) mice are severely obese, infertile and insulin resistant, have increased food intake, reduced body temperature and decreased energy expenditure, immune function and bone formation [3] – [5] . The liver has an integral role in lipid metabolism and also has significant energetic demands; therefore, studies have suggested that leptin's effect on metabolism may be mediated by the liver. Leptin-deficient mice have significant abnormalities in macronutrient metabolism, which can be corrected by leptin administration. A recent study has shown that acute leptin infusion rapidly reverses hepatic steatosis and plasma dyslipidemia induced by a high sucrose diet in rats [6] , and the preservation of hepatic leptin action after a high sucrose diet is associated with the maintenance of low adiposity and plasma leptin concentrations. Leptin therapy has been used to treat selected cases of both congenital and acquired leptin deficiencies in humans. Leptin usually works as a negative-feedback signal regulating the mass of the adipose tissue. However, high levels of leptin have been observed in obese humans and rodents, suggesting the development of leptin resistance in these cases [7] – [10] . An important component in leptin resistance appears to be the development of defective leptin uptake into the brain, downregulation of leptin receptors and decreased leptin sensitivity. The molecular mechanism by which leptin reduces body weight has been partially elucidated. There is a need to have a precise understanding of the metabolic pathways affected by leptin deficiency or leptin resistance. Therefore, understanding the mechanisms by which leptin and its targets influence energy balance could lead to new therapeutic targets for obesity. The purpose of this study was to identify the genes which respond to leptin treatment and subsequent pathways correlated with leptin-mediated weight loss. We utilized microarray technology to compare two types of leptin administration: one involving a direct stimulatory effect when administered peripherally (subcutaneous: SQ) and another that is indirect, involving a hypothalamic relay that suppresses food intake when leptin is administered centrally (intracerebroventricular: ICV). We report here the impact of central and peripheral administration of leptin on food intake, body weight and body fat composition in ob/ob mice. We also report hepatic gene expression changes caused by central versus peripheral leptin administration. MATERIALS AND METHODS Leptin deficient ( ob/ob ) female mice (11 wk old) on the C57BL6 background were obtained from Jackson Laboratory (Bar Harbor, ME). These mice were housed 2 per cage in shoebox cages with food and water available ad libitum . Their health, body weight and food intake were monitored and recorded weekly for 4 wk as they adapted to the pellet diet (LabDiet 5010, PMI Nutrition International). At 15 wk of age, the mice were divided into four treatment groups such that the mean body weight between treatments was not statistically different. Two groups of ten mice each were subjected to the subcutaneous (SQ) treatment (vehicle and leptin) and two groups of ten mice each were subjected to the intracerebroventricular (ICV) treatment (vehicle and leptin). Surprisingly, we observed negligible weight loss (ranging from −5.42% to 1.49%) in four animals of leptin-ICV treatment group, in which the cannula may not have been in place or treatment did not work. Therefore, these animals were removed from the analyses while evaluating effects of leptin administration on body weight, food intake and insulin levels. However, we thought that it would be interesting to determine if the gene expression profile of these animals were different from the animals showing weight loss. Therefore, for microarray analyses, we sub divided the LEP-ICV group into two groups: (a) LEP_ICV group: treatment worked (weight loss range: 25%–38%) and (b) LEP_ICV*: treatment did not work (weight loss −5.42% to 1.49%). This resulted in total five groups: VEH-SQ, LEP-SQ, VEH-ICV, LEP-ICV, and LEP-ICV*. To minimize the cost of the experiment, we randomly selected mice (n = 4) from these five groups for microarray profiling. LEPTIN ADMINISTRATION Mice assigned to ICV treatments were surgically prepared with unilateral cannulas directed towards the right lateral cerebral ventricle as previously described [11] . Osmotic mini-pumps were surgically placed SQ to provide continuous infusion of control and test solutions for 12 days. For ICV treatments, polyethylene tubing filled with 6µl artificial cerebrospinal fluid (aCSF) was connected from the mini-pump to the ICV cannula. All mice were allowed to recover for 2 days after surgery; during this time 6µl of saline or 6µl of aCSF were infused prior to the treatment infusion. Leptin was continuously delivered subcutaneously (SQ; Control: 6 µl/d saline; Leptin group: 10.0 µg/d leptin) or into the lateral cerebral ventricle (ICV) (Control: 6 µl/d aCSF; Leptin group: 1.5 µg/d leptin) via osmotic pumps for 12 days. Body weight and food intakes were measured daily. Water was available ad libitum . The mice were sedated with CO 2 , sacrificed by guillotine prior to the collection of blood and tissues. Serum samples were stored at −80°C until assayed. Insulin, was determined using the Lumine×100™ instrumentation and a multiplex assay kit (Mouse Endocrine Immunoassay Panel catalog # MENDO-75K, protocol - http://www.lincoresearch.com/protocols/mendo-75k.html ) manufactured by LINCO Research, Inc (St. Charles, MO). Brains of mice assigned to ICV treatments were sectioned to confirm proper cannula placement. All surgical and experimental procedures in this study were conducted in accordance with the NIH Guidelines and were approved by the Animal Care and Use Committee for The University of Georgia prior to initiating the studies under the Animal Use Proposal # A2007-10224-0. MICROARRAY HYBRIDIZATION Total RNA was isolated using TRI REAGENT (Sigma), according to the manufacturer's protocol. Purified RNA (200 ng) was further used for cRNA amplification and labeling with biotin using Target Amp cDNA synthesis kit (Epicenter catalog no. TAB1R6924). Approximately 750ng of labeled cRNA was hybridized to the MouseRef-8 v2.0 Expression BeadChips as per manufacturer's instructions. MouseRef-8 BeadChip targets 25,600 RefSeq transcripts and enables the interrogation of eight samples in parallel. After hybridization, arrays were washed, stained with Cy3-conjugated streptavidin, and scanned. Image data were analyzed with BeadStudio V2.0. Raw intensity data from BeadStudio were exported for further statistical analyses. The microarray data are MIAME compliant and have been deposited in NCBI Gene Expression Omnibus and are accessible through GEO Series accession number GSE20878. DIFFERENTIAL GENE EXPRESSION ANALYSIS Microarray data were normalized using the lumi package in R, using the variance stabilizing transformation (VST) of the package and robust spline normalization (RSN). Differential expression analyses were conducted using the LIMMA (Linear Models for Microarray Analysis) package from the Bioconductor project [12] . LIMMA uses an empirical Bayes approach that uses the variability in all genes for testing for signficiant differences, with this approach resulting in more stable inferences for a relatively small number of arrays [12] , [13] . We used LIMMA to test for a linear association between gene expression levels and amount of weight loss (grams), using the amount of weight loss in a linear model and testing for the significance of the regression coefficient for weight loss. We used the false discovery rate (FDR) adjust for multiple testing [14] . B-statistics (the log of the odds of a gene showing either any association with weight loss) were calculated for each gene. Duplicate genes, when present, were removed and their expression levels averaged across the duplicates. We used a combination of “B statistic” and absolute value of slope (logFC) for selecting the differentially expressed genes. B statistic is the log odds of a gene showing either positive or negative correlation with weight loss (also known as a lod score) and logFC determines the magnitude of change in expression with weight loss. We used a cut-off of B>0 and logFC>0.02 (which corresponds to ∼2-fold or greater change in expression from lowest weight loss to highest weight loss point in graph). A total of 221 probes representing 214 unique genes were found significantly correlated with weight loss. CLUSTER ANALYSIS Cluster analysis was performed for grouping differentially expressed genes exhibiting similar expression patterns using the HPCluster program [15] . HPCluster is a two-stage algorithm: the first stage is based on BIRCH (Balanced Iterative Reducing and Clustering using Hierarchies), while the second stage is a conventional k-Means. With BIRCH, a tree of cluster features defining the partitioning of high dimensional space is generated, followed by a conventional k-Means clustering of each cluster feature obtained with BIRCH. Two hundred and fourteen genes could be grouped into four clusters based on the expression data. GENE ONTOLOGY ANALYSIS Associations of the 214 differentially expressed genes with biological processes, molecular functions and cellular compartments were annotated using the GOTree Machine [16] . The GOTree Machine uses the hypergeometric test to evaluate the significance of enrichment for each category by determining if the observed number of gene counts exceeded the expected counts. The “FDR” method was used for adjusting the p-values for multiple testing. The results were visualized as a directed acyclic graph (tree) in order to show the relationship among the enriched GO categories. RESULTS The present study was designed to identify hepatic genes and pathways correlated with leptin mediated weight loss. Leptin deficient ob/ob mice were continuously administered with leptin via ICV (1.5 µg/d) or SQ (10 µg/d) over the 12-day treatment period. Liver RNA was extracted and cDNA microarray technology was used to explore the change in global gene expression profiles after two types of leptin administration. The four animals with negligible weight loss (ranging from −5.42% to 1.49%) after leptin-ICV treatment animals were removed from the analyses while evaluating effects of leptin administration on body weight, food intake and insulin levels. However we profiled these animals for gene expression, and the data were included for Identification of genes correlated with weight loss. Both central and peripheral leptin administrations caused a significant reduction in food intake (FI) in leptin treated ob/ob mice as compared to vehicle treated ob/ob mice (SQ: 3.66±0.49 g (leptin) vs 5.80±0.79 g (vehile); ICV: 2.45±0.65 g (leptin) vs 5.57±0.65 g (vehile); Figure 1A ). Similarly, there was a significant reduction in body weight (BW) in both ICV & SQ leptin groups as compared to their respective controls. Interestingly, percent weight loss in the ICV leptin-treated group was significantly greater than the weight loss in SQ leptin-treated group (SQ: 14.89±1.15 g (leptin) vs −4.50±0.76 (vehile); ICV: 30.51±2.87 g (leptin) vs −4.23±0.78 g (vehile); Figure 1B ). Values are mean ± SEM. Leptin deficient, ob/ob mice were continuously administered with leptin over 12-days using central or peripheral route of administration. Both types of leptin infusion significantly decrease food intake, weight, and serum insulin levels, as compared to their respective controls. The weight loss caused by central treatment is significantly greater than the peripheral treatment. Also, the decrease in the insulin level was more in the central group as compared to the peripheral group. VEH_SQ: vehicle subcutaneous treatment, LEP_SQ: leptin subcutaneous treatment, VEH_ICV: vehicle intracerebroventricular treatment, LEP_ICV: leptin intracerebroventricular treatment. Values are mean ± SEM; n = 9 Veh_SQ, n = 10 Lep_SQ, n = 9 Veh_ICV and n = 4 Lep_ICV; *P<0.05 leptin vs respective vehicle, † p<0.05 Lep_ICV vs Lep_SQ. LEPTIN ADMINISTRATION DECREASES SERUM INSULIN LEVELS The 12 days of leptin infusion in ob/ob mice, resulted in a significant decrease in serum insulin levels in both central and peripheral leptin treatment groups, as compared to their respective controls (SQ: 91.79±8.91 (leptin) vs 787.7±70.74 pM (vehile); ICV: 29.25±6.44 (leptin) vs 815.2±98.15 pM (vehile), Figure 1C ). There was also a significant difference between the leptin-treated groups; the decrease in the insulin levels being greater in the ICV group as compared to the SQ group ( Figure 1C ). Values are mean ± SEM. IDENTIFICATION OF GENES CORRELATED WITH LEPTIN-MEDIATED WEIGHT LOSS We adopted microarray technology to compare the global gene expression profiles after the central (ICV) and peripheral (SQ) leptin treatments in ob/ob mice. C57BL6 mice were used for the baseline gene expression. Two hundred and twenty-one probes representing 214 unique genes were found to be significantly correlated with leptin mediated weight loss. We have listed the top 15 upregulated genes (with a two-fold or greater change in expression for each treatment) in Table 1 and 37 downregulated genes in Table 2 . The plots of the regression analyses (expression vs weigh loss graphs) of the top six upregulated and top six downregulated genes mentioned above are shown in Figure 2 . The phenotype data and weight loss of four ICV leptin treated animals (represented by red crossed circle) were very similar to the vehicle treated animals. Sectioning of the brain could not confirm placement of the cannula in two of these animals. As seen in the heatmap and correlation plots, the expression of these animals is very similar to the vehicle treated animals. Correlation plots of all 214 differentially expressed genes are available in supplementary material ( File S1 : Top-214 Correlation plots.pdf). The regression analysis was performed using “ LIMMA ” to identify genes whose expression is significantly correlated with leptin mediated weight loss. Blue open circle is vehicle subcutaneous treatment; Blue filled circle is leptin subcutaneous treatment; Red open circle is vehicle intracerebroventricular treatment; Red filled circle is leptin intracerebroventricular treatment; Red crossed circle is leptin intracerebroventricular treatment but no weight loss. FC_SQ: fold change in gene expression after peripheral (subcutaneous) leptin administration as compared to vehicle subcutaneous treatment. FC_ICV: fold change in gene expression after central (intracerebroventricular) leptin administration as compared to vehicle intracerebroventricular treatment. FC_ob/B6: fold change in gene expression in vehicle treated ob/ob mice as compared to the expression in B6 mice. adj.P.Val: adj. p-value of regression analysis performed to test for a linear association between gene expression levels and amount of weight loss. FC_SQ: fold change in gene expression after peripheral (subcutaneous) leptin administration as compared to vehicle subcutaneous treatment. FC_ICV: fold change in gene expression after central (intracerebroventricular) leptin administration as compared to vehicle intracerebroventricular treatment. FC_ob/B6: fold change in gene expression in vehicle treated ob/ob mice as compared to the expression in B6 mice. adj.P.Val: adj. p-value of regression analysis performed to test for a linear association between gene expression levels and amount of weight loss. TABLE: Top 15 upregulated genes involved in leptin mediated weight loss. Symbol FC _ SQ a FC _ ICV b FC _ ob/B6 c adj.P.Val d Definition Igfbp2 6.54 16.10 0.06 2.10E-08 insulin-like growth factor binding protein 2 Acot1 3.12 5.60 0.65 0.000739 acyl-CoA thioesterase 1 Lyz 2.33 4.93 0.41 0.000107 Lysozyme Mat2a 2.55 4.10 0.37 0.000703 methionine adenosyltransferase II, alpha Lpl 2.32 4.03 0.86 0.000211 lipoprotein lipase (Lpl) Pla2g12a 2.04 4.57 0.66 0.000125 phospholipase A2, group XIIA Wbp5 3.01 3.44 0.35 0.000535 WW domain binding protein 5 C1qb 2.32 3.22 0.28 0.000584 complement component 1, q subcomponent, beta polypeptide H2-Ab1 2.52 2.88 0.50 0.001534 histocompatibility 2, class II antigen A, beta 1 Hadhb 2.04 3.17 0.99 0.000344 hydroxyacyl-Coenzyme A dehydrogenase Egfr 2.58 2.48 0.16 0.001144 epidermal growth factor receptor Slc25a33 2.60 2.43 0.38 0.000271 solute carrier family 25, member 33 Slc40a1 2.24 2.77 0.68 0.001166 solute carrier family 40 (iron-regulated transporter), member 1 LOC100046120 2.30 2.65 0.33 0.001791 s musculus similar to clusterin Rps11 2.12 2.52 0.38 0.001243 ribosomal protein S11 TABLE: Top 37 downregulated genes involved in leptin mediated weight loss. Symbol FC _ SQ a FC _ ICV b FC _ ob/B6 c adj.P.Val d Definition Scd1 0.05 0.01 4.60 3.33E-07 stearoyl-Coenzyme A desaturase 1 Gsta1 0.09 0.11 41.28 0.000125 glutathione S-transferase, alpha 1 Hamp2 0.32 0.03 2.16 8.76E-07 hepcidin antimicrobial peptide 2 Cyp17a1 0.24 0.17 2.59 6.85E-06 cytochrome P450, family 17, subfamily a, polypeptide 1 Ahcy 0.25 0.18 4.50 7.94E-05 S-adenosylhomocysteine hydrolase Raet1b 0.41 0.12 7.26 8.76E-07 retinoic acid early transcript beta Sardh 0.27 0.17 3.86 0.001766 sarcosine dehydrogenase Gsta2 0.23 0.22 8.24 0.000131 glutathione S-transferase, alpha 2 Rdh16 0.29 0.20 5.93 5.65E-06 retinol dehydrogenase 16 Aass 0.29 0.22 4.63 0.000942 aminoadipate-semialdehyde synthase Pla2g4f 0.23 0.28 4.14 0.000178 phospholipase A2, group IVF Lrtm1 0.32 0.20 2.05 6.44E-05 leucine-rich repeats and transmembrane domains 1 Gstt3 0.41 0.12 2.67 2.13E-07 glutathione S-transferase, theta 3 Thrsp 0.35 0.21 0.90 0.000931 thyroid hormone responsive SPOT14 homolog Bhmt 0.35 0.23 5.59 0.000696 betaine-homocysteine methyltransferase Selenbp2 0.32 0.30 1.76 0.000285 selenium binding protein 2 Rcan2 0.30 0.33 3.01 8.09E-05 regulator of calcineurin 2 Mmd2 0.29 0.34 2.79 0.000309 monocyte to macrophage differentiation-associated 2 Gamt 0.36 0.29 1.85 0.000344 guanidinoacetate methyltransferase Dmgdh 0.31 0.34 3.82 0.000703 dimethylglycine dehydrogenase precursor Olig1 0.41 0.25 6.69 2.86E-06 oligodendrocyte transcription factor 1 Tmie 0.48 0.18 4.48 0.000109 transmembrane inner ear Snhg11 0.35 0.31 3.98 0.000314 small nucleolar RNA host gene 11 Kcnk5 0.45 0.22 1.99 0.000682 potassium channel, subfamily K, member 5 Aes 0.36 0.31 2.66 0.000315 amino-terminal enhancer of split Nnmt 0.48 0.21 1.29 0.002523 nicotinamide N-methyltransferase 9130409I23Rik 0.35 0.35 3.35 0.001156 RIKEN cDNA 9130409I23 gene Slc38a4 0.40 0.32 5.60 0.002091 solute carrier family 38, member 4 Slc25a1 0.44 0.30 1.81 2.96E-05 solute carrier family 25 (mitochondrial carrier, citrate transporter), member 1 Uroc1 0.48 0.28 2.84 6.63E-05 urocanase domain containing 1 Hsd17b10 0.42 0.36 3.47 0.000125 hydroxysteroid (17-beta) dehydrogenase 10 Pls3 0.46 0.33 4.45 0.002206 plastin 3 Fam158a 0.44 0.39 2.84 0.00198 family with sequence similarity 158, member A C530044N13Rik 0.47 0.39 3.14 0.002232 RIKEN cDNA C530044N13 gene Slc44a1 0.48 0.40 2.09 5.28E-05 solute carrier family 44, member 1 Oprs1 0.47 0.43 2.03 0.001192 opioid receptor, sigma 1 Mfsd2 0.45 0.48 2.59 0.001226 major facilitator superfamily domain containing 2 CLUSTER ANALYSIS OF CORRELATED GENES Next, we performed cluster analyses on the 214 genes significantly correlated with leptin mediated weight loss and found 4 clusters, each of which defined a unique pattern of expression ( Figure 3 ). The four animals in which the cannula may not have been in place or treatment did not work are represented on the heatmap as LEP-ICV*. The expression pattern of these animals is similar to the vehicle groups. Seventy genes represented by cluster-1 are overexpressed in ob/ob mice as compared to B6 mice. After leptin replacement, the expression of these genes is normalized towards B6 mice. Similarly, 46 genes in cluster-2 are underexpressed in ob/ob as compared to B6 and leptin treatment could correct the expression back to normal. 56 genes of cluster-3 and 42 genes of cluster-4 have similar expression in B6 and ob/ob but these genes are downregulated (Cluster-3) or upregulated (Cluster-4) after leptin treatment. The complete spreadsheet with gene symbols and cluster assignment is available as supplementary material ( File S2 : 214_Gene_matrix.xls). Each row represents one gene and each column represents one sample. Red indicates higher expression and green indicates lower expression. The global gene expression profiles were compared after the central and peripheral leptin treatments in ob/ob mice, while C57BL6 mice were used for the baseline gene expression. Cluster analysis on the gene expression data was performed to group the genes with similar expression patterns across experimental conditions. VEH-SQ: vehicle subcutaneous treatment, LEP-SQ: leptin subcutaneous treatment, VEH-ICV: vehicle intracerebroventricular treatment, LEP-ICV: leptin intracerebroventricular treatment, LEP-ICV * : represents four animals in which the cannula may not have been in place or treatment failed. While analyzing the phenotype data, we found that there was no weight loss in these four animals. Sectioning of the brain could not confirm placement of the cannula in these animals. The expression of these animals is very similar to the vehicle treated animals. BIOLOGICAL PROCESSES OF GENES REGULATED BY LEPTIN TREATMENT Using the GOTree Machine web application [16] , the 214 differentially expressed genes were categorized as being associated with specific biological processes, molecular functions and cellular compartments. Statistically significant over represented biological processes were determined using the hypergeometric test. Among the cellular compartments, mitochondrion (32 genes), endoplasmic reticulum (22 genes) and vacuole (8 genes) were significantly over represented. To determine if there was a difference in enrichment of upregulated versus downregulated genes, the two groups (126 downregulated and 88 upregulated genes) were also mapped separately to GO categories. The results were visualized as a gene ontology tree in order to show the relationship among the enriched GO categories ( Figure 4 ). Several biological processes such as cellular metabolic processes (101 genes), catabolic processes (36 genes), biosynthetic processes (28 genes), oxidation reduction (27 genes), response to chemical stimulus (35 genes), and inflammatory response (10 genes) were found to be significantly over represented. Among the molecular functions, oxidoreductase activity (26 genes), cofactor binding (14 genes), catalytic activity (89 genes), electron carrier activity (10 genes) and iron ion binding (12 genes) were significantly over represented. The complete gene ontology analyses results are available as supplementary material ( File S3 : Gene Ontology.pdf). Gene ontology analysis was performed using the GOTree Machine software which performs the hypergeometric test to evaluate the significance of enrichment for each category by determining if the observed number of gene counts exceeded the expected counts. Among the cellular compartments mitochondrion and endoplasmic reticulum are enriched by downregulated genes and vacuole, vesicle, lysosome are significantly enriched by upregulated genes. DOWNREGULATED BIOLOGICAL PATHWAYS AFTER LEPTIN ADMINISTRATION IN OB/OB MICE The gene ontology analyses revealed that most significantly enriched cellular compartments among downregulated genes are mitochondrion and endoplasmic reticulum. Thirty-two genes could be mapped to mitochondrion, 26 genes were downregulated and 6 genes were upregulated after leptin treatment ( Table 3 ). The most affected mitochondrial pathways are, ketone metabolic process, organic acid metabolic process, oxidation reduction, amino acid metabolic process, and nitrogen compound metabolic process. Among the molecular functions, electron carrier activity, catalytic activity, and coenzyme binding were most affected after leptin treatment. We found that overexpression of genes related to cellular metabolism is a characteristic of ob/ob mice, while leptin replacement normalizes this expression back towards B6. The livers of ob/ob mice are overloaded with large amounts of free fatty acids either through increased fatty acid biosynthesis or through decreased fatty acid oxidation. Leptin administration promotes a hypometabolic state and downregulates cellular metabolic activity. Surprisingly, this condition is very similar to hibernation [17] . The key to the survival under hibernation lays in an inherent ability to downregulate cellular metabolic rate to new hypometabolic steady states. There was a downregulation of most genes involved in amino acid metabolism, lipid, fatty acid and steroid metabolism and oxidation/reduction after leptin treatment in ob/ob mice. Mitochondrial metabolic pathways are major regulators of leptin-mediated weight loss. FC_SQ: fold change in gene expression after peripheral (subcutaneous) leptin administration as compared to vehicle subcutaneous treatment. FC_ICV: fold change in gene expression after central (intracerebroventricular) leptin administration as compared to vehicle intracerebroventricular treatment. FC_ob/B6: fold change in gene expression in vehicle treated ob/ob mice as compared to the expression in B6 mice. adj.P.Val: adj. p-value of regression analysis performed to test for a linear association between gene expression levels and amount of weight loss. TABLE: 32 mitochondrial genes involved in leptin signaling. Symbol FC _ SQ a FC _ ICV b FC _ ob/B6 c adj.P.Val d Definition Sardh 0.27 0.17 3.86 0.001766 sarcosine dehydrogenase Aass 0.29 0.22 4.63 0.000942 aminoadipate-semialdehyde synthase Gck 0.58 0.21 2.96 0.000114 Glucokinase Abat 0.53 0.25 2.30 7.01E-05 4-aminobutyrate aminotransferase Dmgdh 0.31 0.34 3.82 0.000703 dimethylglycine dehydrogenase precursor Slc25a1 0.44 0.30 1.81 2.96E-05 solute carrier family 25 (mitochondrial carrier, citrate transporter), member 1 Hsd17b10 0.42 0.36 3.47 0.000125 hydroxysteroid (17-beta) dehydrogenase 10 Otc 0.82 0.32 1.71 0.002029 ornithine transcarbamylase Cisd1 0.53 0.38 2.35 0.001166 CDGSH iron sulfur domain 1 Kynu 0.62 0.46 1.98 3.95E-05 kynureninase (L-kynurenine hydrolase) Gcdh 0.57 0.52 1.67 0.000211 glutaryl-Coenzyme A dehydrogenase Nipsnap1 0.69 0.48 1.61 1.01E-06 4-nitrophenylphosphatase domain and non-neuronal SNAP25-like protein homolog 1 (C. elegans) Ndufs8 0.62 0.51 1.47 0.001217 NADH dehydrogenase (ubiquinone) Fe-S protein 8 Sfxn1 0.76 0.55 1.77 1.61E-05 sideroflexin 1 Dbi 0.60 0.61 2.05 0.000292 diazepam binding inhibitor Cyp17a1 0.24 0.17 2.59 6.85E-06 cytochrome P450, family 17, subfamily a, polypeptide 1 Haao 0.57 0.20 1.26 0.000502 3-hydroxyanthranilate 3,4-dioxygenase Prodh 0.61 0.21 1.35 2.57E-05 proline dehydrogenase Gls2 0.73 0.25 0.84 1.45E-03 glutaminase 2 (liver, mitochondrial) Ass1 0.80 0.34 0.94 0.001664 argininosuccinate synthetase 1 Mmab 0.75 0.47 0.95 2.96E-05 methylmalonic aciduria (cobalamin deficiency) type B homolog (human) Macrod1 0.95 0.50 0.64 0.001609 MACRO domain containing 1 Cyb5r3 0.60 0.56 1.04 0.000137 cytochrome b5 reductase 3 Hsd3b2 0.83 0.49 1.13 0.000502 hydroxy-delta-5-steroid dehydrogenase, 3 beta- and steroid delta-isomerase 2 Ndufa3 0.74 0.60 1.08 0.000687 NADH dehydrogenase (ubiquinone) 1 alpha subcomplex, 3 Uqcrc1 0.58 0.64 1.24 0.000902 ubiquinol-cytochrome c reductase core protein 1 Hadhb 2.04 3.17 0.99 0.000344 hydroxyacyl-Coenzyme A dehydrogenase/3-ketoacyl-Coenzyme A thiolase/enoyl-Coenzyme A hydratase (trifunctional protein), beta subunit Ung 1.27 2.95 0.76 0.00021 uracil DNA glycosylase Mrps6 1.20 1.91 0.92 0.002234 mitochondrial ribosomal protein S6 Slc25a33 2.60 2.43 0.38 0.000271 solute carrier family 25, member 33 Hsp90ab1 1.69 2.32 0.47 0.001625 heat shock protein 90kDa alpha (cytosolic), class B member 1 Cyba 1.20 2.00 0.68 7.33E-05 cytochrome b-245, alpha polypeptide LIPID METABOLIC PROCESS We found 26 genes differentially expressed under lipid metabolic processes, out of which 19 genes were downregulated. Also, 12 of these genes could be mapped to lipid biosynthetic process, out of which 10 genes were downregulated. Leptin influences the rates of synthesis and degradation of lipids through its autocrine and paracrine actions. The molecular mechanism of leptin action in fatty acid metabolism and lipid biosynthesis has not yet been fully elucidated. We found that Scd1, Faah, Acot1, Rdh16 and Lss are downregulated by leptin treatment. The microsomal enzyme, stearoyl-CoA desaturase-1 (Scd1) catalyzes the biosynthesis of monounsaturated fatty acids from saturated fatty acids. In our study we found that leptin-deficient ob/ob mice have very high levels (∼5-fold) of Scd1 gene expression as compared to B6 control mice and after leptin replacement in ob/ob mice, the Scd1 gene expression is extremely downregulated (∼20-fold in SQ and ∼100-fold in ICV) as compared to their respective controls. Therefore, Scd1 appears to be a major player in the leptin mediated lipid biosynthesis and fatty acid metabolism. This finding is consistent with other studies which have shown that ob/ob mice with mutations in Scd1 gene are significantly less obese than ob/ob controls and have markedly increased energy expenditure. A significant proportion of metabolic effects of leptin are the result of inhibition of this enzyme [18] . On the other hand the Scd1 −/− mice in SV129 background are lean, resistant to diet-induced obesity, have increased insulin sensitivity, and increased metabolic rate [19] . CARBOXYLIC ACID METABOLIC PROCESS Twenty-four downregulated genes including Gck, Faah, Otc, Arg1, Gls2, Ass1, Prodh and Aass, could be mapped to carboxylic acid metabolic pathways. Glucokinase (GcK) regulates rate-limiting reactions in glycolysis. Expression of Gck is regulated in the liver in response to fasting and feeding. We found that Gck is 2-fold and 5-fold downregulated after leptin administration. Hepatocyte nuclear factor 4 alpha (HNF4alpha) plays an important role in transcriptional regulation of GcK gene [20] and many genes expressed in the liver. We found that HNF-4 has a downward trend after leptin administration, but could not reach to statistical significance. HNF4alpha is also critical for urea homeostasis by direct regulation of the Otc gene [21] . Otc is a key enzyme in the urea cycle to detoxify ammonium produced from amino acid catabolism. Here, we found that Otc was also significantly downregulated after leptin replacement. IRON ION BINDING In this study, we found that 12 genes could be annotated to the iron ion binding group. The 9 downregulated genes were Cisd1, Haao, Cyp17a1, Dpyd, Hpd, Scd1, Cyp2c29, Ndufs8, and Sfxn1 and 3 upregulated genes of this group were Cyba, Hpx and, Slc40a1. There is a well established link between obesity and iron metabolism. Ob/ob mice have higher iron absorption as compared to lean mice [22] . There is evidence in the literature that insulin resistance is associated with hepatic iron overload. A recent study found a role of Hepacidin expression in metabolic syndrome and hepatic iron overload associated with insulin resistance [23] . Interestingly, in our study, we also found that expression of hepcidin antimicrobial peptide 2 (Hamp2) was 3-fold and 28-fold downregulated after leptin treatment. GLUTATHIONE S TRANSFERASES Glutathione is a major endogenous anti-oxidant for the cell. We found that three of the glutathione S-transferases were significantly overexpressed in ob/ob mice as compared to B6 (Gsta1: 41-fold, Gsta2: 8-fold, Gstt3: 3-fold). Also, leptin replacement caused the downregulation of Gsta1, Gsta2, Gsta3, Gstt1 and Gstt3 in ob/ob mice. Reactive oxygen species (ROS) production by mitochondria plays a critical role in many physiological processes and therefore over quenching of ROS in ob/ob mice may result in pathological conditions. A recent study has shown that mice lacking one of the key enzymes involved in the elimination of physiological ROS, glutathione peroxidase 1 (Gpx1), were protected from high-fat-diet-induced insulin resistance [24] . GENES INVOLVED IN ADIPOCYTE PROLIFERATION AND DIFFERENTIATION It is well known that adipocyte proliferation and differentiation plays a critical role in obesity. Leptin causes alterations in the expression of many genes which inhibit adipocyte differentiation. The expression of Igfbp2 gene, which negatively regulates the biological activity of IGF1, was 6× (SQ) and 16× (ICV) upregulated after leptin replacement. Igfbp2 is almost 17× downregulated in ob/ob mice as compared to B6. Another mitochondrial gene, Cisd1 was 2.4× upregulated in ob/ob mice as compared to B6 and corrected back to normal by leptin. This protein encodes for MitoNEET protein which exists in low levels in preadipocytes, and its expression increases exponentially in differentiated adipocytes. Leptin has a mitogenic potential in liver cancer cells, which is mediated through the induction of Mat2a and Mat2b genes [25] . In this study, we found that expression of Mat2a was induced after both types of leptin treatment. UPREGULATED BIOLOGICAL PROCESSES AFTER LEPTIN ADMINISTRATION IN OB/OB MICE We found that the cellular compartments enriched by upregulated genes were lysosome (5 genes), vacuole (6 genes), and vesicle (9 genes). Lysosomes contain enzymes which destroy organelles that have been damaged. Also, when adequate amount of food is not available for the cell, the organelle degradation and lysosomal activity is increased. A recent study has identified the previously unknown role of autophagy in regulating intracellular lipid stores [26] . Lysosomes continuously use portions of lipid droplets and process them for energy production during nutrient deprivation [26] . IMMUNE SYSTEM AND INFLAMMATORY RESPONSE Leptin has structural similarities with the family of long-chain helical cytokines, including IL-6 and IL-12 and also has been found to play a critical regulatory role within the immune system that affects the course of inflammation [27] . Various studies have shown that immune cells respond to pharmacological doses of leptin. In this study we found that the immune system process and inflammatory response were enriched in the upregulated group of genes. The genes which were upregulated in this group are; C1qb, Cd44, Ccl4, Nupr1, Cyba, Fcer1g, Rmcs2, Fcer1g, Rps19, Plscr1, Ednrb, Cd44, Ung, Ctnnb1, Lmo2, and Hpx. A recent study has shown the function of leptin within T-cell polarization [28] . Utilizing the model of oxazolone-induced colitis, authors found that ob/ob mice were protected, whereas wild-type and leptin-reconstituted ob/ob mice developed colitis. This protection of Th1- as well as in Th2-dependent inflammation was associated with decreased expression of T-bet and GATA-3, in naive ob/ob T cells [28] . In this study, we also found that epidermal growth factor receptor (EGFR) expression was upregulated almost 2.5 times after leptin replacement in ob/ob mice. It is known that EGFR mediates both chemotaxis and proliferation in monocytes and macrophages and treatment with EGFR inhibitor decreases the protein expression of TNF-alpha and IL-6, and reduces subclinical inflammation in HFD-fed mice [29] . COMPARISON OF TWO TYPES OF GENE EXPRESSION CHANGES AFTER LEPTIN TREATMENT IN OB/OB MICE Cluster analysis revealed that 214 differentially expressed genes could be sub-divided into two major groups: (A) 116 genes of Cluster-1 and Cluster-2, which have different expression in ob/ob when compared to B6, and leptin treatment normalizes the expression in ob/ob towards B6; (B) 98 genes of Cluster-3 and Cluster-4, which have similar expression in ob/ob and B6 but are downregulated or upregulated after leptin treatment. These two groups of genes were separately analyzed and were mapped to cellular compartments and biological functions ( Table S1 ). The enriched cellular compartments in both groups were mitochondrion (group A: 19 genes; group B: 14 genes), cytosol (group A: 10 genes; group B: 10 genes) and endoplasmic reticulum (group A: 11 genes; group B: 11 genes). However, the plasma membrane (6 genes) and vacuole/lysosome (5 genes) were only enriched in group B. We have listed the comparison of enriched biological processes of each group in Table S1 . Both groups are enriched in many metabolic processes whereas major catabolic processes are mainly enriched in group A. We also found that steroid metabolic process is only enriched in Group B and the genes involved in this process are Gpsn2, Cyb5r3, Lss, Cyp17a1, Pcsk9, Mvd, Hsd3b2. Differentially expressed genes were further analyzed using Ingenuity Pathway Analysis software (IPA; Ingenuity Systems, Mountain View, CA; www.ingenuity.com ) to construct and visualize molecular interaction networks. Three high scoring networks could be identified. Network-1 ( Figure 5 ) is enriched in lipid metabolism and cell-cell signaling. EGFR (which is upregulated after leptin treatment) is one of the central hub genes of this network. Also a lot of genes involved in this network are connected through important kinases. Expression of these kinases does not change after leptin replacement, but these might be mediators for the signal transduction between the genes which respond to leptin treatment. The transcriptional regulators in the network are Ets2, Xbp1, Sap30, Nupr1, Hmga1, Zfp36, Maged1 and Cttnb1. Network-2 and 3 are available in supplementary material ( File S4 : Networks.pdf). Network-2 is enriched in lipid metabolism and cell cycle and involves many cytokines. Network-3 is enriched in cell morphology and most of the genes in this network are connected through Hnf4a, Ifng and Htt. Ingenuity Pathway Analysis software was used to construct and visualize molecular interaction networks. DISCUSSION Despite the fact that leptin has shown exciting results in the mouse models, the success rate of leptin for the reduction of body weight in humans is very low. This is due mainly to the fact that a very small fraction of the obese human population is leptin deficient, and the most common forms of obesity are associated with high leptin levels [9] . In such cases, most obese patients are leptin resistant and fail to respond effectively to exogenous leptin [8] . Several defects may contribute to the leptin resistant state, including defective leptin transport across the blood-brain barrier, reduced leptin receptor expression, defects in leptin signal transduction, or the induction of feedback inhibitors [10] [7] . However, if leptin resistance could be by-passed and genes/pathways which respond to leptin treatment could be regulated directly, new therapeutic strategies for the treatment of obesity may be possible. In this study, we determined the global hepatic gene expression profiles after peripheral and central leptin treatment. We found that the effect of central treatment on weight loss was significantly better than peripheral treatment; which confirms that leptin action on weight loss is primarily controlled by a central pathway via hypothalamic relay. We found 214 genes significantly correlated with leptin-mediated weight loss. Obesity results from the imbalance of energy intake and energy utilization. When energy intake exceeds energy expenditure, excess energy is stored as triglycerides in lipid droplets. Conversely during nutrient deprivation, stored triglycerides are hydrolyzed into fatty acids and converted back to energy. Therefore, metabolic pathways are key regulators of energy balance. In this study, we found that genes which belong to metabolic, catabolic and biosynthetic pathways in mitochondria are highly expressed in ob/ob mice when compared with the B6 baseline expression. However, leptin replacement results in the downregulation of these pathways back to normal. The Stearoyl-CoA Desaturase 1 (Scd1), which is a lipogenic enzyme involved in fatty acid synthesis and oxidation, is overexpressed in the livers of ob/ob mice as compared to B6 and is downregulated after leptin replacement. This is consistent with the results of previous studies which have shown that Scd1 deficient mice are protected against obesity and insulin resistance [18] , [19] . Also in humans, genetic variations in the Scd1 gene were found to be associated with body fat deposition and insulin sensitivity [30] . Some recent studies have used pharmacological inhibition of Scd1 for treatment of obesity and insulin resistance [31] , [32] . Glucokinase (GcK) phosphorylates glucose to produce glucose-6-phosphate and therefore regulates glucose metabolism. We found that expression of this gene is negatively correlated with leptin-mediated weight loss in ob/ob mice. Long-term overexpression of Gck increases hepatic lipogenesis and circulating lipids, which leads to insulin resistance [33] . Weight loss also has been linked to GH-IGF1 axis in the literature, which has a critical role in the proliferation and differentiation of adipocytes [34] . IGF-binding proteins (IGFBPs) modulate bioavailability of IGF. In this study we found that Igfbp2 was highly underexpressed in livers of ob/ob mice but was normalized after leptin replacement. Recent studies have shown that overexpression of Igfbp2 protects against the development of obesity and improves insulin sensitivity [35] . Treatment with recombinant Igfbp2 impairs 3T3-L1 differentiation and hence adipogenesis [35] . Elevated insulin and body fat have been shown to be associated with decreased Igfbp1 and Igfbp2 levels cross-sectionally [36] . A recent study observed markedly lower expression of Igfbp2 in liver from morbidly obese women [37] . Such studies provide an impetus for investigating the effects of Igfbp2 for treatment of obesity. However, studies have also shown that Igfbp2 is overexpressed in a wide spectrum of cancers and IGF-independent effects of Igfbp2 are emerging [38] . Igfbp2 expression was significantly higher in breast cancer tissue compared with benign breast tissue and Igfbp2 inhibition attenuated the associated aggressive phenotype of breast cancer cells both in vitro and in vivo [39] . Overexpression of Igfbp2 has also been correlated with glioblastoma [40] , [41] and lymph node metastasis in patients with invasive breast carcinomas [42] . These findings suggest that further studies are required to evaluate IGF-dependent and -independent functions of Igfbp2 before choosing Igfbp2 as a therapeutic target for treatment of obesity. We found that Cisd1, the mitochondrial gene which encodes for the MitoNEET protein, is upregulated in livers of ob/ob mice as compared to B6 and could be normalized by leptin. MitoNEET is an iron-containing outer mitochondrial membrane protein that regulates oxidative capacity and is involved in the control of maximal mitochondrial respiratory rates [43] . This protein exists in low levels in preadipocytes, and its expression increases exponentially in differentiated adipocytes. Also, MitoNEET has been identified as a target for the thiazolidinedione class of diabetes drugs that may contribute to lipid lowering and/or antidiabetic actions [44] . Interestingly, the expression of five glutathione S-transferases (Gsta1, Gsta2, Gsta3, Gstt1, and Gstt3) were significantly downregulated after leptin administration. In an earlier study it was shown that Gsta3 expression is markedly induced during adipose conversion, which is virtually undetectable in confluent 3T3-L1 cells under basal conditions [45] . Inhibition of the 3T3-L1 adipogenic program demonstrated that Gsta3 expression is associated specifically with acquisition of the adipocytic phenotype. One important finding of this study was the upregulation of hepatic lysosomal pathways, including vesicles and vacuole, after leptin treatment. This finding is consistent with a recent study which has identified the previously unknown role of autophagy in regulating intracellular lipid stores [26] . Starvation causes the induction of autophagy, which delivers intracellular proteins and organelles sequestered in double-membrane vesicles (autophagosomes) to lysosomes for degradation and use as an energy source. It was found that inhibition of autophagy in cultured hepatocytes and mouse liver increased triglyceride storage in lipid droplets. In conclusion, this study identified key molecular pathways and downstream target genes involved in the leptin-mediated weight loss. Many of these genes have previously been shown to be associated with obesity; however, we also identified a number of novel target genes. Further detailed studies will be required to evaluate the possible use of these genes and the associated protein products as therapeutic targets for the treatment of obesity. SUPPORTING INFORMATION Top-214 Correlation plots.pdf. Regression plots of 214 genes significantly associated with leptin mediated weight loss. We used LIMMA to test for a linear association between gene expression levels and amount of weight loss (grams). (0.93 MB PDF) 214_Gene_matrix.xls. Normalized gene expression data of 214 genes significantly associated with leptin mediated weight loss. (0.52 MB XLS) Gene Ontology.pdf. Biological processes, molecular functions and cellular compartments of hepatic genes regulated by leptin treatment. (0.24 MB PDF) Networks.pdf. Top molecular networks and their biological functions affected by leptin treatment. Network analysis was performed using Ingenuity Pathway Analysis software. (1.17 MB PDF) Differentially expressed genes could be sub-divided into two major groups: 116 genes of Cluster-1 and Cluster-2 (group-A), and 98 genes of Cluster-3 and Cluster-4 (group-B). These two groups of genes were separately analyzed and were mapped to cellular compartments and biological functions. (0.08 MB DOC) REFERENCES [1] Zhang Y Proenca R Maffei M Barone M Leopold L 1994 Positional cloning of the mouse obese gene and its human homologue. Nature 372 425 432 [2] Halaas JL Boozer C Blair-West J Fidahusein N Denton DA 1997 Physiological response to long-term peripheral and central leptin infusion in lean and obese mice. Proc Natl Acad Sci U S A 94 8878 8883 [3] Chehab FF Lim ME Lu R 1996 Correction of the sterility defect in homozygous obese female mice by treatment with the human recombinant leptin. Nat Genet 12 318 320 [4] Ducy P Amling M Takeda S Priemel M Schilling AF 2000 Leptin inhibits bone formation through a hypothalamic relay: a central control of bone mass. Cell 100 197 207 [5] Friedman JM Halaas JL 1998 Leptin and the regulation of body weight in mammals. Nature 395 763 770 [6] Huang W Dedousis N O'Doherty RM 2007 Hepatic steatosis and plasma dyslipidemia induced by a high-sucrose diet are corrected by an acute leptin infusion. J Appl Physiol 102 2260 2265 [7] Munzberg H 2010 Leptin-signaling pathways and leptin resistance. Forum Nutr 63 123 132 [8] Friedman JM 2009 Leptin at 14 y of age: an ongoing story. Am J Clin Nutr 89 973S 979S [9] Maffei M Halaas J Ravussin E Pratley RE Lee GH 1995 Leptin levels in human and rodent: measurement of plasma leptin and ob RNA in obese and weight-reduced subjects. Nat Med 1 1155 1161 [10] Caro JF Kolaczynski JW Nyce MR Ohannesian JP Opentanova I 1996 Decreased cerebrospinal-fluid/serum leptin ratio in obesity: a possible mechanism for leptin resistance. Lancet 348 159 161 [11] Choi YH Li C Page K Westby A la-Fera MA 2003 Melanocortin receptors mediate leptin effects on feeding and body weight but not adipose apoptosis. Physiol Behav 79 795 801 [12] Smyth GK 2004 Linear models and empirical bayes methods for assessing differential expression in microarray experiments. Stat Appl Genet Mol Biol 3 Article3 [13] Kerr KF 2009 Comments on the analysis of unbalanced microarray data. Bioinformatics 25 2035 2041 [14] Benjamini Y Hochberg Y 1995 Controlling the False Discovery Rate - A Practical and Powerful Approach to Multiple Testing. Journal of the Royal Statistical Society Series B-Methodological 57 289 300 [15] Sharma A Podolsky R Zhao J McIndoe RA 2009 A modified hyperplane clustering algorithm allows for efficient and accurate clustering of extremely large datasets. Bioinformatics 25 1152 1157 [16] Zhang B Schmoyer D Kirov S Snoddy J 2004 GOTree Machine (GOTM): a web-based platform for interpreting sets of interesting genes using Gene Ontology hierarchies. Bmc Bioinformatics 5 [17] Boutilier RG 2001 Mechanisms of cell survival in hypoxia and hypothermia. J Exp Biol 204 3171 3181 [18] Cohen P Miyazaki M Socci ND Hagge-Greenberg A Liedtke W 2002 Role for stearoyl-CoA desaturase-1 in leptin-mediated weight loss. Science 297 240 243 [19] Ntambi JM Miyazaki M Stoehr JP Lan H Kendziorski CM 2002 Loss of stearoyl-CoA desaturase-1 function protects mice against adiposity. Proc Natl Acad Sci U S A 99 11482 11486 [20] Hirota K Sakamaki J Ishida J Shimamoto Y Nishihara S 2008 A combination of HNF-4 and Foxo1 is required for reciprocal transcriptional regulation of glucokinase and glucose-6-phosphatase genes in response to fasting and feeding. J Biol Chem 283 32432 32441 [21] Inoue Y Hayhurst GP Inoue J Mori M Gonzalez FJ 2002 Defective ureagenesis in mice carrying a liver-specific disruption of hepatocyte nuclear factor 4alpha (HNF4alpha). HNF4alpha regulates ornithine transcarbamylase in vivo. J Biol Chem 277 25257 25265 [22] Failla ML Kennedy ML Chen ML 1988 Iron metabolism in genetically obese (ob/ob) mice. J Nutr 118 46 51 [23] Le GG Chanseaume E Ruivard M Morio B Mazur A 2007 Study of iron metabolism disturbances in an animal model of insulin resistance. Diabetes Res Clin Pract 77 363 370 [24] Loh K Deng H Fukushima A Cai X Boivin B 2009 Reactive oxygen species enhance insulin sensitivity. Cell Metab 10 260 272 [25] Ramani K Yang H Xia M Ara AI Mato JM 2008 Leptin's mitogenic effect in human liver cancer cells requires induction of both methionine adenosyltransferase 2A and 2beta. Hepatology 47 521 531 [26] Singh R Kaushik S Wang Y Xiang Y Novak I 2009 Autophagy regulates lipid metabolism. Nature 458 1131 1135 [27] Lago F Dieguez C Gomez-Reino J Gualillo O 2007 Adipokines as emerging mediators of immune response and inflammation. Nat Clin Pract Rheumatol 3 716 724 [28] Batra A Okur B Glauben R Erben U Ihbe J 2010 Leptin: a critical regulator of CD4+ T-cell polarization in vitro and in vivo. Endocrinology 151 56 62 [29] Prada PO Ropelle ER Mourao RH de Souza CT Pauli JR 2009 EGFR tyrosine kinase inhibitor (PD153035) improves glucose tolerance and insulin action in high-fat diet-fed mice. Diabetes 58 2910 2919 [30] Warensjo E Ingelsson E Lundmark P Lannfelt L Syvanen AC 2007 Polymorphisms in the SCD1 gene: associations with body fat distribution and insulin sensitivity. Obesity (Silver Spring) 15 1732 1740 [31] Li CS Belair L Guay J Murgasva R Sturkenboom W 2009 Thiazole analog as stearoyl-CoA desaturase 1 inhibitor. Bioorg Med Chem Lett 19 5214 5217 [32] Issandou M Bouillot A Brusq JM Forest MC Grillot D 2009 Pharmacological inhibition of stearoyl-CoA desaturase 1 improves insulin sensitivity in insulin-resistant rat models. Eur J Pharmacol 618 28 36 [33] Ferre T Riu E Franckhauser S Agudo J Bosch F 2003 Long-term overexpression of glucokinase in the liver of transgenic mice leads to insulin resistance. Diabetologia 46 1662 1668 [34] Hausman DB DiGirolamo M Bartness TJ Hausman GJ Martin RJ 2001 The biology of white adipocyte proliferation. Obes Rev 2 239 254 [35] Wheatcroft SB Kearney MT Shah AM Ezzat VA Miell JR 2007 IGF-binding protein-2 protects against the development of obesity and insulin resistance. Diabetes 56 285 294 [36] Ahmed RL Thomas W Schmitz KH 2007 Interactions between insulin, body fat, and insulin-like growth factor axis proteins. Cancer Epidemiol Biomarkers Prev 16 593 597 [37] Elam MB Cowan GS Jr Rooney RJ Hiler ML 2009 Hepatic gene expression in morbidly obese women: implications for disease susceptibility. Obesity (Silver Spring) 17 1563 1573 [38] Wheatcroft SB Kearney MT 2009 IGF-dependent and IGF-independent actions of IGF-binding protein-1 and -2: implications for metabolic homeostasis. Trends Endocrinol Metab 20 153 162 [39] So AI Levitt RJ Eigl B Fazli L Muramaki M 2008 Insulin-like growth factor binding protein-2 is a novel therapeutic target associated with breast cancer. Clin Cancer Res 14 6944 6954 [40] Fukushima T Kataoka H 2007 Roles of insulin-like growth factor binding protein-2 (IGFBP-2) in glioblastoma. Anticancer Res 27 3685 3692 [41] Marucci G Morandi L Magrini E Farnedi A Franceschi E 2008 Gene expression profiling in glioblastoma and immunohistochemical evaluation of IGFBP-2 and CDC20. Virchows Arch 453 599 609 [42] Wang H Arun BK Wang H Fuller GN Zhang W 2008 IGFBP2 and IGFBP5 overexpression correlates with the lymph node metastasis in T1 breast carcinomas. Breast J 14 261 267 [43] Wiley SE Murphy AN Ross SA van der GP Dixon JE 2007 MitoNEET is an iron-containing outer mitochondrial membrane protein that regulates oxidative capacity. Proc Natl Acad Sci U S A 104 5318 5323 [44] Colca JR McDonald WG Waldon DJ Leone JW Lull JM 2004 Identification of a novel mitochondrial protein (“mitoNEET”) cross-linked specifically by a thiazolidinedione photoprobe. Am J Physiol Endocrinol Metab 286 E252 E260 [45] Jowsey IR Smith SA Hayes JD 2003 Expression of the murine glutathione S-transferase alpha3 (GSTA3) subunit is markedly induced during adipocyte differentiation: activation of the GSTA3 gene promoter by the pro-adipogenic eicosanoid 15-deoxy-Delta12,14-prostaglandin J2. Biochem Biophys Res Commun 312 1226 1235