p66Shc-generated Oxidative Signal Promotes Fat Accumulation*

Reactive oxygen species (ROS) and insulin signaling in the adipose tissue are critical determinants of aging and age-associated diseases. It is not clear, however, if they represent independent factors or they are mechanistically linked. We investigated the effects of ROS on insulin signaling using as model system the p66Shc-null mice. p66Shc is a redox enzyme that generates mitochondrial ROS and promotes aging in mammals. We report that insulin activates the redox enzyme activity of p66Shc specifically in adipocytes and that p66Shc-generated ROS regulate insulin signaling through multiple mechanisms, including AKT phosphorylation, Foxo localization, and regulation of selected insulin target genes. Deletion of p66Shc resulted in increased mitochondrial uncoupling and reduced triglyceride accumulation in adipocytes and in vivo increased metabolic rate and decreased fat mass and resistance to diet-induced obesity. In addition, p66Shc-/- mice showed impaired thermo-insulation. These findings demonstrate that p66Shc-generated ROS regulate the effect of insulin on the energetic metabolism in mice and suggest that intracellular oxidative stress might accelerate aging by favoring fat deposition and fat-related disorders.

Genetic evidence suggests that insulin/IGF1 signaling in the adipose tissue plays a critical role in the regulation of lifespan in both invertebrates and mammals; (i) overexpression in the fat tissue of Foxo, a key target of insulin/IGF-1 signaling that is inactivated by insulin/IGF-1, prolongs lifespan in Drosophila and Caenorhabditis elegans (1-3) (ii) fatspecific disruption of the insulin receptor gene decreases body fat and increases lifespan in mice (FIRKO mice) (4,5). Furthermore, sirtuin1 (SIRT1), the mammalian ortholog of the life-extending yeast gene silent information regulator 2 (SIR2), inhibits adipogenesis of precursor cells and reduces fat storage in differentiated adipocytes by repressing the peroxisome proliferator-activated receptor-␥ nuclear receptor, a master regulator of fat cell development in the insulin/ IGF1 signaling pathway (6).
At the molecular level evidence suggests that the most important mechanisms of aging involve damage to cellular macromolecules. Reactive oxygen species (ROS) 2 , as generated by mitochondria during respiration, induce oxidative stress, which accumulates over life and is considered the proximal mechanism of aging and a major determinant of lifespan (free radical or mitochondrial theory of aging) (7). Indeed, mutations of p66 Shc , a gene that increases production of mitochondrial ROS, or overexpression of catalase, which increases their scavenging, prolong lifespan in mice (8,9). Whether these lifespan determinants (insulin/IGF1 signaling in the fat tissue and oxidative stress) are mechanistically related is unknown.
Recent data showed that one specific ROS molecule, hydrogen peroxide (H 2 O 2 ), is directly implicated in the physiological regulation of different signal transduction pathways, including the insulin/IGF1 pathway (10). Induction of cell proliferation by insulin or IGF1 as well as several other growth factors (e.g. epidermal growth factor or platelet-derived growth factor) correlates with a transient increase of intracellular H 2 O 2 , whereas anti-oxidant treatments prevent insulin/IGF-1-induced DNA synthesis (10,11). Exogenously added H 2 O 2 leads to functional inactivation of tyrosine and serine/threonine phosphatases and activation of tyrosine kinases and various transcription factors (see the following supplemental references: Ross et al. The property of H 2 O 2 to function as signaling molecule is based on its ability to induce fully reversible protein modifications. It has been demonstrated that H 2 O 2 oxidizes directly cysteinyl thiol, inducing disulfide bonds and sulfenic acids formation, and induces glutathionylation of cysteine residues or sulfoxidation of methionine residues in a variety of molecular targets (see the following supplemental references: Song et al.  )). In the case of the PTP1B and PTEN phosphatases, which are directly involved in the attenuation of the insulin/IGF1 signaling, it has been shown that H 2 O 2 -induced protein modifications are functionally relevant (12,13). It appears, therefore, that H 2 O 2 regulates directly the intensity of insulin/IGF1 signaling within cells.
Indirect evidence suggests that ROS are also involved in the regulation of fat development. Treatment with antioxidants prevents in vitro differentiation of pre-adipocytes and, in mice, diet-induced obesity (14). Caloric restriction in laboratory rodents reduces their adipose tissue mass, extends lifespan, and is associated with decreased oxygen metabolism and, consequently, mitochondrial H 2 O 2 production (15,16). It is not clear, however, whether these effects of H 2 O 2 or antioxidants on fat development are mediated by modifications of the insulin/ IGF1 signaling pathway in the adipose tissue.
A role for ROS as signaling molecules is further supported by recent findings that the generation of H 2 O 2 by mitochondria is not just the byproduct of respiration but can also be the result of specific enzymatic systems, such as p66 Shc . p66 Shc functions as an inducible redox enzyme, which is activated by stress and generates H 2 O 2 to trigger apoptosis (17). For this function, p66 Shc uses reducing equivalents of the mitochondrial electron transfer chain through the direct oxidation of cytochrome c (17). The H 2 O 2 generated by p66 Shc is ϳ30% that of the total pool of intracellular H 2 O 2 and is biologically relevant, as shown by p66 Shc ability in vitro and in vivo to induce mitochondrial permeability transition and by the findings that cells and tissues derived from p66 ShcϪ/Ϫ mice accumulate significantly less oxidative stress (17)(18)(19)(20)(21)(22)(23)(24)(25). It is not known, however, whether p66 Shcgenerated H 2 O 2 is also involved in the regulation of receptoractivated signal transduction pathways. We report here that p66 Shc -generated H 2 O 2 regulates insulin signaling in adipocytes and fat development and propose that the insulin-p66 Shc oxidative signaling pathway regulates energetic metabolism.

EXPERIMENTAL PROCEDURES
Reagents-Antibodies against AKT, pAKT, pMAPK, FOXO1, and pFoxo1 were from Cell Signaling; anti-Shc and anti-pShc were from BD Transduction Laboratories and Alexis; anti-PTEN, anti-IRS-1, anti-p85a, and anti-GLUT4 were from Santa Cruz; the anti-phosphotyrosine was from Upstate, and human recombinant insulin was from Roche Applied Science.
Mice-This study was performed according to national law and following authority guidelines. Experiments were carried out on p66 ShcϪ/Ϫ and WT mice (Sv/129 and C57Bl/6), maintained in a temperature-controlled room with a 12-h light/12-h dark cycle. A high fat (HF) diet (D12492 with 60% kcal from fat) was purchased from Research Diets Inc. Eight-week-old mice were fed with standard diet or D12492 diet. C57Bl/6 and 129Sv mice were fed HF diet and tap water or water supplemented with 40 mM N-acetylcysteine.
Blood glucose levels were measured using Accu-Chek active; triglycerides, cholesterol, and free fatty acid levels were measured using the Reflotron Plus system from Roche Applied Science. Body weight was measured weekly. Total body fat mass was evaluated after chemical extraction as described (26). Measurements of energy expenditure were made using a closed-circuit dual-gas respirometer (Micro-Oxymax System, Columbus Instruments).
Adipocyte Preparation and Treatments-Primary brown pre-adipocytes were isolated from brown adipose tissue (BAT) of intrascapular fat pads from 2-day-old newborns upon digestion with 1.5 mg/ml collagenase type IV buffer (125 mM NaCl, 5 mM KCl, 1.3 mM CaCl 2 , 5 mM glucose, 100 mM Hepes, 4% BSA) for 40 min at 37°C. Cells were then plated in growth media (Dulbecco's modified Eagle's medium supplemented with 20% fetal bovine serum, 20 mM HEPES). The following day cultures were washed twice with phosphate-buffered saline, then after confluence, 2 ϫ 10 5 cells were plated in 12-well dishes in Dulbecco's modified Eagle's medium plus 10% fetal bovine serum. The day after, media were supplemented with insulin at different dosages as indicated and in the presence of 1 nM T3 (27). Lipids accumulation was evaluated after 6 days. Differentiation for some experiments was induced with 40 ng/ml insulin plus 250 M ␣-lipoic acid or 0.1 M antimycin A or 1 M dexamethasone and 1 mM isobutylmethylxanthine.
Primary white adipocytes were isolated from abdominal white fat pads from 1-month-old mice. Fat pieces were minced and digested with 1.5 mg/ml collagenase type IV at 37°C for 35 min. Cells were filtered through 250-m cell strainer, centrifuged at 250 g, and plated in the same media as for BAT-derived cells. At day 2 upon confluence, cells were stimulated with 1 g/ml insulin, and lipid accumulation was evaluated after 10 days with Oil red-O staining.
For retroviral infections, pre-adipocytes were infected on day 2 of culture with empty pBABE vectors or vectors expressing WT p66 Shc or the p66 Shc qq or p66 Shc a mutants and selected with 2 g/ml puromycin.
For ROS analysis, levels of intracellular H 2 O 2 were determined using 6-carboxy-2Ј,7Ј-dichlorodihydrofluorescein (DCF) diacetate (H2-DCFDA). Briefly, cells were treated with 20 M H2-DCFDA for 45 min in dark (20). Intracellular DCF fluorescence was detected and quantified by fluorescence-activated cell sorter. Direct fluorescence imaging of attached cells stained with H2-DCFDA was also performed to confirm fluorescence-activated cell sorter analysis data. For siRNA down-regulation of p66 Shc expression, pre-adipocytes were transfected with a p66 Shc -specific 5Ј-GTACAACCCACT-TCGGAATG (28) and a scrambled nucleotide sequence as a negative control, siRNAs.
Biochemistry and Immunofluorescent Analysis-For insulin stimulation experiments cells were serum-starved for 24 h and then treated with 1 g/ml insulin for 90 min (for FOXO1 and p66 Shc activation) or for 5 min (for AKT activation). For immunofluorescence cells were fixed in paraformaldehyde, permeabilized for 10 min with 0.4% Triton X-100 in phosphate-buffered saline supplemented with 0.2% BSA, blocked for 30 min in 2% BSA, and incubated with primary anti-FOXO antibodies produced in rabbit and then with secondary anti-rabbit IgC conjugated with Cy3. For biochemical analysis, cells were washed and lysed in a buffer containing 50 mM HEPES, pH 7.5, 150 mM NaCl, 10% glycerol, 1% Triton X-100, 1.5 mM MgCl 2 , 1 mM EGTA, 100 mM NaF, 10 mM sodium pyrophosphate, and 500 M sodium orthovanadate supplemented with a protease inhibitors mixture from Calbiochem. Immunoprecipitations were performed on a platform rocker at 4°C. Equal amounts of proteins were incu-bated with anti-Shc or anti-IRS-1 antibodies and then with 50 l of protein A-agarose followed by 3 washes with lysis buffer. Beads were resuspended in 2ϫ Laemmli loading buffer, and proteins were separated by 8% SDS-PAGE. After SDS-PAGE, proteins were transferred to a polyvinylidene difluoride membrane. The immunoprecipitates were analyzed by Western blotting using indicated antibodies. All blotting steps were performed according to the manufacturer's instructions. Chemiluminescence was detected with ECL Western blotting detection reagent from Amersham Biosciences.
Fatty Acid Intake and Oxidation Rate in Cell Cultures-To determine FA intake and oxidation rate, isolated BAT adipocytes were grown to confluence, washed with phosphate-buffered saline, 0.5% BSA, incubated at 37°C for 90 min with 1 Ci/ml [ 14 C]oleate (Amersham Biosciences) in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum and 0.5% BSA, washed twice with phosphate-buffered saline, 0.5% BSA, and lysed in 200 l of 0.5 N NaOH. After neutralization, radioactivity was counted with a liquid scintillation analyzer and normalized for protein content. Fatty acid ␤-oxidation rate was assessed by the release of 14 CO 2 after [ 14 C]oleate uptake as described (29).
Gene Expression Analysis-Primary BAT pre-adipocytes were grown to confluence and incubated with or without 100 nM insulin for 4 h. Cells then were lysed in a guanidine-isothiocyanate containing buffer, and total RNA was extracted using RNeasy kit (Qiagen) following the manufacturers instructions. For each experiment, total RNA was isolated and pooled from triplicate culture dishes and used for microarrays Affymetrix GeneChip hybridization. Biotin-labeled cRNA targets were obtained from 3 g of total RNA derived from samples as described above. cDNA synthesis was performed with Invitrogen SuperScript Custom cDNA synthesis kit, and biotin-labeled antisense RNA was transcribed using the in vitro MEGAscript High Yield Transcription kit (Ambion Inc., Austin, TX) and included Bio-11-UTP and Bio-11-CTP (PerkinElmer Life Sciences) in the reaction. GeneChip hybridization, washing, staining, and scanning were performed according to Affymetrix (Santa Clara, CA) protocols. A copy of the Mouse Genome 430 2.0 GeneChip Array was hybridized with each target. Absolute and comparative analyses were performed with Affymetrix GeneChip Operating Software (GCOSv1.4), scaling all images to a TGT value of 500. Results were further exported to GeneSpring GX software Version 7.3 (Agilent Technology). For quantitative PCR (QPCR) validation of gene-chip data, total RNA from three independent experiments was reverse-transcribed using Superscript II reverse transcriptase (Invitrogen), and the obtained cDNA was analyzed by quantitative real-time PCR using specific primers for selected genes and Taqman chemistry. Results were expressed as -fold change relative to the controls after normalization using GAPDH gene expression levels.
Statistical Analysis-Data are presented as the mean Ϯ S.D. and analyzed by Student's t test. Differences between means were assessed by one-way analysis of variance. The minimum level of significance was set at p Ͻ 0.05.

RESULTS
Insulin Activates the Redox Activity of p66 Shc in Adipocytes-p66 Shc is expressed in primary adipocytes (32) and supplemental Fig. 1A) where it is rapidly phosphorylated on Ser-36 after insulin treatment (Fig. 1A). After phosphorylation in the cytosol, p66 Shc binds to the peptidyl-prolyl isomerase Pin-1 and undergoes a major conformational change that favors translocation of p66 Shc within mitochondria and activation of its redox enzymatic activity (33).
Phosphorylation-defective mutants of p66 Shc are unable to maintain proper levels of intracellular H 2 O 2 and to mediate pro-apoptotic stress signals (17,33). We investigated whether insulin treatment induces H 2 O 2 accumulation in BAT primary pre-adipocytes, as reported for other cell types (34) and whether this effect is mediated by phosphorylation of p66 Shc . Intracellular levels of H 2 O 2 were determined by fluorescenceactivated cell sorter analysis of cells stained with H2-DCFDA, a molecular probe that becomes fluorescent upon oxidation by H 2 O 2 . Steady-state levels of H 2 O 2 in p66 ShcϪ/Ϫ adipocytes were reduced by ϳ35% as compared with control WT cells (Fig. 1B). Insulin treatment induced a 2-3-fold increase of H 2 O 2 in WT BAT pre-adipocytes, whereas it had no effect on p66 ShcϪ/Ϫ cells (Fig. 1C). Reintroduction of p66 Shc but not of the phosphorylation-defective p66 Shc Ser36Ala (p66 Shc a) mutant into p66 ShcϪ/Ϫ pre-adipocytes restored insulin-induced H 2 O 2 upregulation ( Fig. 1C), demonstrating that p66 Shc phosphorylation is critical to the ability of insulin to increase intracellular H 2 O 2 . To demonstrate that p66 Shc mediates insulin-dependent H 2 O 2 up-regulation through its redox activity, we used the redox-defective mutant p66 Shc E132Q,E133Q (p66 Shc qq), which is unable to oxidize cytochrome c and to produce mitochondrial ROS (17). Expression of the p66 Shc qq mutant in p66 ShcϪ/Ϫ BAT pre-adipocytes failed to mediate H 2 O 2 up-regulation after insulin stimulation (Fig. 1C). Together, these findings demonstrate that insulin activates the redox activity of p66 Shc and that p66 Shc mediates insulin redox signaling.
p66 Shc -generated H 2 O 2 Regulates Insulin-dependent Activation of AKT-FOXO1 in Adipocytes-Insulin-stimulated H 2 O 2 modulates proximal and distal insulin signaling at least in part through the oxidative inhibition of phosphatases that negatively regulate the insulin pathway (PP2A, PTEN) (35).
Tyrosine phosphorylation of IRS-1, the intracellular target of activated insulin-receptor, did not differ significantly in the WT and p66 ShcϪ/Ϫ samples ( Fig. 2A; the slight decrease of IRS-1 phosphorylation at later time points after insulin treatment was not seen in other separated experiments). Once phosphorylated, IRS-1 binds various adapter molecules, such as p85 and Grb-2, which activate, respectively, the phosphatidylinositol 3-kinase-AKT and the Ras-MAPK pathways (36). Analysis of the association of IRS-1 with the p85 subunit of phosphatidylinositol 3-kinase in WT and p66 ShcϪ/Ϫ BAT pre-adipocytes revealed no differences ( Fig.  2A). On the contrary, in the p66 ShcϪ/Ϫ pre-adipocytes, the extent of insulin-induced AKT phosphorylation was markedly reduced as compared with WT adipocytes, whereas MAPK phosphorylation proceeded normally (Fig. 2, A and  B), indicating that loss of p66 Shc affects the phosphatidylinositol 3-kinase-AKT branch in particular.
Among the several targets of activated AKT, the Forkhead transcription factor FOXO1 couples insulin signaling to adipocyte differentiation (37). AKT-dependent FOXO1 phosphorylation leads to its cytoplasmic sequestration and inhibition of transcription from FOXO1-target genes (37). Immunofluorescence analysis of WT BAT pre-adipocytes showed massive relocalization of FOXO1 in the cytoplasm after insulin treatment (from 80 to 5% of cells with nuclear FOXO-1), which was markedly reduced in p66 ShcϪ/Ϫ cultures (from 80 to 75%) (Fig. 3, A  and B). It appears, therefore, that p66 Shc expression is indispensable for the activation of the AKT-FOXO1 pathway after insulin treatment. Notably, the effect of p66 Shc on FOXO1 appeared more pronounced than the effect on AKT, suggesting that p66 Shc regulates insulin signaling at multiple levels.
To determine whether the effect of p66 Shc on insulin-mediated AKT-FOXO1 activation was dependent on its redox activity, we analyzed the signaling functions of phosphorylation-and redox-defective p66 Shc mutants and the effect of p66 Shc on oxidation of PTEN, which dephosphorylates and inactivates PIP3, a lipid second messenger produced by phosphatidylinositol 3-kinase (35). Re-expression of WT p66 Shc into p66 ShcϪ/Ϫ BAT pre-adipocytes restored a normal response of FOXO1 to insulin treatment (from 70 to 5% cells with nuclear FOXO-1), whereas the p66 Shc a or p66 Shc qq mutants had no effects ( Fig. 3B and supplemental Fig. 1B). PTEN is inactivated by endogenous H 2 O 2 that catalyzes the formation of a disulfide bond between the PTEN active sites Cys-124 and Cys-71. This event correlates with a shift in its electrophoretic mobility in non-reducing PAGE (12). In p66 ShcϪ/Ϫ pre-adipocytes, the ratio between reduced PTEN (the slowly migrating form) and oxidized PTEN (the fast migrating form) was increased with respect to the same ratio observed in WT pre-adipocytes both in basal conditions (100% of increase) and after insulin treatment (30% of increase) (Fig. 3C, left panel), whereas total PTEN levels,  measured by Western blotting (WB) upon reducing SDS-PAGE, was comparable between WT and p66 ShcϪ/Ϫ pre-adipocytes (Fig. 3C, right panel). It appears, therefore, that the ability of p66 Shc to regulate FOXO1 activation depends on the integrity of its redox center, suggesting that p66 Shc regulates insulin signaling by generating H 2 O 2 .
p66 Shc Regulates Lipogenesis in Adipocytes by Inhibiting Fatty Acid Oxidation-We then investigated the biological contribution of p66 Shc to lypogenesis in adipocytes. Because intracellular ROS are critical mediators of insulin-dependent lipogenesis (38), we measured insulin-induced triglyceride (TG) accumulation in WT and p66 ShcϪ/Ϫ adipocytes. Treatment of WT BAT (Fig. 4A) and white adipose tissue (WAT) (Fig. 4B)-derived pre-adipocytes with 40 mg/ml or 1 g/ml of insulin, respectively, in the presence of 1 nM triiodothyronine (T3) induced TG accumulation in virtually all cells, as seen by Oil red staining 6 and 10 days after treatment (Fig. 4,  A and B). Strikingly, p66 ShcϪ/Ϫ adipocytes from both BAT and WAT were almost completely Oil red-negative after the same insulin treatment (Fig. 4, A and B). Dose-response analysis of insulin-induced TG accumulation in BAT adipocytes revealed a reduction of insulin sensitivity in the absence of p66 Shc expression of about 30-fold (the half-maximal differentiating doses of insulin were 10 -20 and 600 ng/ml in WT and p66 ShcϪ/Ϫ adipocytes, respectively; Fig.  4C). Notably, the effect of p66 Shc loss was comparable with that of   6 days (b, f, d, h, l, and n). B, micrographs of plates of Oil red-stained WT and p66 ShcϪ/Ϫ WAT adipocytes (ADPC) untreated and after insulin treatment. C, optical density (OD) values of Oil red extracts from stained WT, antimycin A (AntA)-, or lipoic acid (LipAc)-treated WT and p66 ShcϪ/Ϫ BAT adipocytes (average of eight independent experiments for each condition). D, white light microscopy pictures of Oil red-stained p66 ShcϪ/Ϫ BAT adipocytes infected with empty (Vect), p66 Shc , p66 Shc a, or p66 Shc qq retroviruses and treated with insulin and their corresponding optical density (OD; 490 nm) values. DECEMBER 5, 2008 • VOLUME 283 • NUMBER 49 antioxidants, such as lipoic acid (Figs. 1B and 4C) or N-acetylcysteine (not shown). Conversely, levels of ROS and insulininduced lipid accumulation were increased by antimycin A, an inhibitor of respiration that increases mitochondrial ROS generation (Figs. 1B and 4C). Re-expression of p66 Shc , but not of the p66 Shc a or p66 Shc qq mutants, in p66 ShcϪ/Ϫ BAT pre-adipocytes restored insulin-induced TG accumulation (Fig. 4D). Adipocyte differentiation by other stimuli (3-isobutyl-1-methylxanthine and dexamethasone (supplemental Fig. 1C) proceeded normally in the p66 ShcϪ/Ϫ BAT pre-adipocytes, suggesting that the effect of p66 Shc is specific for insulin-induced lipogenesis.

Role of p66 Shc in Adipogenesis
In adipocytes, TG accumulation stimulated by insulin is determined by increased uptake of FA or glucose and inhibition of FA oxidation. To investigate FA metabolism, we measured 14 C incorporation (FA uptake) and 14 CO 2 release (FA ␤-oxidation) in cells preincubated with [ 14 C]oleate. Although the rate of [ 14 C]oleate uptake was similar in basal and insulin-treated WT and p66 ShcϪ/Ϫ BAT pre-adipocytes (Fig. 5A), the release of 14 CO 2 was higher in p66 ShcϪ/Ϫ cells both at the steady state and after insulin treatment (Fig. 5B). Together, these results indicate that p66 Shc controls the lipogenic function of insulin in adipocytes by inhibiting FA oxidation.
To investigate whether p66 Shc regulates selectively downstream adipogenic effectors of the insulin signaling pathway, we performed global gene expression analysis of WT and p66 ShcϪ/Ϫ BAT pre-adipocytes treated or not with insulin (4 h). Biotinylated cRNA targets were synthesized from RNA pools of three independent experiments and hybridized to Affymetrix oligonucleotide chips. Analysis of microarray data identified two overlapping lists of 598 and 610 genes regulated by insulin in WT and p66 ShcϪ/Ϫ preadipocytes, respectively (supplemental Fig. 2).
QPCR validation of 30 gene regulations revealed full concordance (not shown). Cross-comparisons of WT and p66 ShcϪ/Ϫ gene sets distinguished three groups of insulin targets; (i) 282 genes regulated by insulin in both WT and p66 ShcϪ/Ϫ pre-adipocytes (p66 Shcindependent insulin targets); (ii) 316 genes regulated by insulin in WT but not p66 ShcϪ/Ϫ pre-adipocytes (p66 Shc -dependent targets), and (iii) 328 genes regulated by insulin in p66 ShcϪ/Ϫ but not WT pre-adipocytes (p66 Shc -repressed insulin targets) (supplemental Fig. 2). Among the 282 p66 Shc -independent insulin targets, several genes linked to adipogenesis have been found, some of which are mechanistically implicated in the process of adipogenesis. Notably, the impaired response to insulin in p66 Shc Ϫ/Ϫ BAT pre-adipocytes is accompanied by a lower induction of these adipogenesis-linked genes (Egr1, Egr3, Ptgs2, Pai-1, Fabp3, CD36, App; Fig. 5C and supplementary references). QPCR validation of these targets in p66 ShcϪ/Ϫ or in WT BAT pre-adipocytes after depletion of p66 Shc expression by siRNA is shown in Fig. 5D. This correlates with the observation that insulin response is impaired in p66 Shc -deficient BAT pre-adipocytes, although we can not draw any conclusion regarding the contribution of the genes considered to the observed phenotype. Expression of peroxisome proliferator-activated receptor-␣, ERR␣, PGC1, which are critical regulators of adipocyte differentiation, was instead normal in p66 ShcϪ/Ϫ cells (not shown). Together, these data indicate that the effect of p66 Shc on insulin-dependent gene expression is restricted to specific transcriptional targets, which might explain selectivity of the effect of p66 Shc on lipid accumulation.
p66 Shc Improves Mitochondrial Coupling in Adipocytes-We then investigated mitochondrial energetic functions in WT and p66 ShcϪ/Ϫ BAT mitochondria and tissue. Oxygraph analysis of energized mitochondria (using either pyruvate/ malate or glycerol 3-phosphate) and in the presence of fatty acid-free albumin (to sequester free fatty-acids) revealed  (Table 1). In the absence of albumin, instead, p66 ShcϪ/Ϫ mitochondria showed higher levels of oxygen consumption (Table 1). Consistently, the addition of palmitate increased respiration more potently in the p66 ShcϪ/Ϫ mitochondria ( Table 1), suggesting that respiration is higher in the p66 ShcϪ/Ϫ mitochondria when fatty acids are available. Because fatty acids stimulate proton backflow to the mitochondrial matrix (uncoupled respira-tion), we treated BAT mitochondria with the artificial protonophores dinitrophenol or FCCP. Strikingly, dinitrophenol (not shown) or FCCP (Fig. 6A) increased respiration of WT but not p66 ShcϪ/Ϫ mitochondria ( Table 1), suggesting that p66 ShcϪ/Ϫ mitochondria are uncoupled. Treatment with the inhibitor of uncoupling proteins GDP, instead, reduced respiration of WT and p66 Shc Ϫ/Ϫ BAT mitochondria to equal levels ( Table 1). As seen for the isolated BAT mitochondria, minced BAT from p66 ShcϪ/Ϫ mice showed increased respiration upon the addition of succinate (p Ͻ 0.01), which was not further enhanced by treatment with FCCP. Notably, also basal respiration was higher in p66 ShcϪ/Ϫ BAT (Fig. 6A).
To characterize directly coupling efficiency, we measured the protonmotive force of WT and p66 ShcϪ/Ϫ BAT mitochondria using the potentiometric tetramethylrhodamine methyl ester (TMRM). Results showed a slight, yet statistically significant, reduction of the membrane potential of p66 ShcϪ/Ϫ energized mitochondria (Fig. 6B). Treatment with GDP abolished this difference, further supporting our conclusion that p66 ShcϪ/Ϫ mitochondria are uncoupled. Together, these findings suggest that the increased oxygen consumption of p66 ShcϪ/Ϫ mitochondria is attributable to increased uncoupled state.
Uncoupled respiration in BAT mitochondria is due to expression of the anion-carrier protein UCP1, which mediates a regulated proton-leak across the mitochondrial membranes (39). Inspection of our microarray data showed that UCP1 is up-regulated in p66 ShcϪ/Ϫ adipocytes. QPCR analysis confirmed the increased expression of UCP1 mRNA in the p66 ShcϪ/Ϫ adipocytes both under basal conditions and after insulin treatment (Fig. 6C). Western blotting showed increased levels of the UCP1 protein in BAT adipose tissue of newborn and adult p66 ShcϪ/Ϫ mice as compared with WT controls (Fig. 6D). These data indicate that p66 Shc inhibits basal and insulin-induced UCP1 expression in adipocytes. Thus, high levels of UCP1 might be responsible for the observed uncoupling state of mitochondria in p66 ShcϪ/Ϫ adipocytes.
p66 Shc Induces Fat Development in Vivo-We then investigated the effect of p66 Shc on fat accumulation in vivo. The body weight of p66 ShcϪ/Ϫ mice was normal from birth to 4 weeks of age (not shown). Starting at 2 months of age, p66 ShcϪ/Ϫ mice maintained 5-10% lower body weights throughout the rest of their lives as compared  with WT mice (Fig. 7A) despite normal locomotory activity, 3 food intake (Fig. 8C), and overall caloric absorption (as determined by measuring the residual caloric content of feces by calorimetric bomb; not shown). Differences of total body weight between WT and p66 ShcϪ/Ϫ mice were mainly due to a ϳ35% reduction in fat mass in the p66 ShcϪ/Ϫ mice (total-body TG content; not shown). Perigonadal, abdominal, and interscapular fat pads were all reduced in p66 ShcϪ/Ϫ mice (Fig. 7A). Differences were more marked in the abdominal district (75% reduction in the p66 ShcϪ/Ϫ males) and were observed in both 129Sv and C57Bl6 p66 ShcϪ/Ϫ strains (Fig. 7A). No differences between WT and p66 ShcϪ/Ϫ mice were observed in the weight of other organs (Fig. 7A).
Histological analysis of the same samples showed high cell size heterogeneity, with a larger frequency of adipocytes below 100 m diameter (Fig. 7B). Although the functional significance of this heterogeneity in p66 ShcϪ/Ϫ adipose tissue is unclear, it resembles that observed in mice with a null mutation in the insulin receptor gene (FIRKO), which also show reduced fat mass and increased lifespan (5). Electron microscopy examination of BAT (interscapular) in adult p66 ShcϪ/Ϫ mice revealed smaller sized lipid deposits in the p66 ShcϪ/Ϫ BAT (Fig. 7C).
To test whether decreased fat mass in p66 ShcϪ/Ϫ mice was due to a cell-autonomous defect of adipocytes, we transplanted WT and p66 ShcϪ/Ϫ primary BAT pre-adipocytes into WT nude mice. One month post-transplant, WT adipocytes transplants showed a dramatic increase in size and a decrease in cell heter-3 A. Berry and F. Cirulli, manuscript in preparation.  ogeneity, whereas p66 ShcϪ/Ϫ adipocyte transplants grew significantly less, and the cell population remained heterogeneous ( Fig. 7D and not shown), thus suggesting that reduced fat accumulation in p66 ShcϪ/Ϫ mice is the consequence at least in part of an abnormal intrinsic function of adipocytes.
Diet-induced obesity is mediated by insulin-induced lipid accumulation in the fat tissue (5). To determine the role of p66 Shc in diet-induced obesity, we studied body weight and fat mass in WT and p66 ShcϪ/Ϫ mice fed HF diet for 14 weeks. The profile of circulating lipids after administration of the high fat diet was similar in WT and p66 ShcϪ/Ϫ mice (supplemental Fig.  3A). As shown in Fig. 8A, WT mice gained significantly more body weight than the p66 ShcϪ/Ϫ mice (144 versus 119% maximal body weight increase, respectively; p Ͻ 0.001). Notably, the effects of the HF diet on p66 ShcϪ/Ϫ mice body weight were comparable with those observed in WT mice supplemented with the anti-oxidant N-acetylcysteine in drinking water (supplemental Fig. 3B and (40). Gross appearance at sacrifice after 14 weeks of HF diet revealed markedly reduced fat mass in the p66 ShcϪ/Ϫ mice as compared with WT animals (Fig. 8B). Energetic intake (Fig. 8C) and residual fecal caloric content (not shown) were comparable in the two groups. Together these findings demonstrate that p66 ShcϪ/Ϫ mice are protected from age-related and diet-induced obesity and suggest that p66 Shc is a genetic determinant of fat development in adult mice.
Increased Basal Metabolic Rate and Impaired Cold Adaptation in the p66 ShcϪ/Ϫ Mice-We then investigated the functional consequences on energy metabolism of the reduced WAT and BAT in p66 ShcϪ/Ϫ mice. WAT is the site where energy is stored after feeding in the form of TG and from where energy is released during periods of fasting in the form of FA. Plasma concentrations of TG, FA, cholesterol, and low density lipoproteins were comparable in WT and p66 ShcϪ/Ϫ animals (measured in both the fed and fasted states in age-matched animals of 3, 6, 9, 12, and 18 months; supplemental Fig.  3A). Oxygen consumption normalized per body weight was slightly, yet significantly increased in p66 ShcϪ/Ϫ males as compared with matched controls (differences did not reach statistical significance in the female animals; Fig. 9A). Consistently, energetic expenditure measured in a 24-h interval was moderately higher in the p66 ShcϪ/Ϫ male mice (Fig. 9B).
WAT has the additional function of protecting from body heat loss (thermo-insulation), whereas BAT generates heat for the maintenance of body temperature when newborn animals are exposed to cold (thermogenesis). The effects of p66 Shc loss on thermoregulation were studied by exposing WT and p66 ShcϪ/Ϫ mice to cold (5°C) for 6 h. Basal body temperature was slightly increased in the p66 ShcϪ/Ϫ mice (average of 0.6°C; Fig. 9C). The maximal loss of body heat after cold exposure occurred after 4 h in the WT mice and was in the range of 3°C of body temperature. In the p66 ShcϪ/Ϫ mice, instead, maximal loss was around 6°C and occurred after 3 h (Fig. 9D). Notably, both WT and p66 ShcϪ/Ϫ mice returned to their basal body temperature after 6 h, suggesting that thermogenesis is not impaired in p66 ShcϪ/Ϫ mice (Fig. 9D). Thus, abnormal cold adaptation in the p66 ShcϪ/Ϫ mice is the consequence of accelerated heat loss, likely due to the reduced thermal insulation effect of fat pads. Notably, levels of UCP1 in p66 ShcϪ/Ϫ mice kept at 22°C were as high as those of WT mice after cold exposure (Fig. 6C, right panel). All together, these findings suggest that the functional alterations observed in the p66 ShcϪ/Ϫ adipocytes translate into critical modifications of the energetic metabolism in p66 ShcϪ/Ϫ mice, e.g. increased basal body temperature and increased energy expenditure, due to increased BAT metabolism, and accelerated heat loss, due to reduced WAT mass.

DISCUSSION
ROS are implicated in the physiological regulation of the insulin/IGF1 signal-transduction pathway and in the regulation of fat development (10,11,14). It is not known, however, whether regulation of fat development by ROS occurs through a direct effect on the insulin/IGF1signaling pathway in the adipose tissue. We reported here that insulin activates the redox enzyme-activity of p66 Shc in adipocytes and that p66 Shc -generated H 2 O 2 , through its effect on the insulin-signaling cascade, reduces mitochondrial oxygen consumption and favors TG accumulation. In vivo, mice lacking p66 Shc showed increased basal metabolism, reduced fat development, and increased insulin sensitivity of peripheral tissues. We propose that this newly identified insulin-p66 Shc -H 2 O 2 signaling pathway is a critical determinant of energetic metabolism in mammals.
Our findings demonstrate that p66 Shc is a downstream target of the insulin signaling pathway and a critical mediator of insulindependent ROS-up-regulation in adipocytes. Furthermore, we showed that phosphorylation-and redox-defective mutants of p66 Shc are unable to restore redox signaling by insulin, demonstrating that p66 Shc phosphorylation and p66 Shc redox activity are both integral to insulin signaling. Based on previous characterization of these two mutants (17), we propose a model whereby insulin stimulation induces phosphorylation of p66 Shc in the cytosol, its translocation into the mitochondrial intermembrane space, activation of its redox enzymatic activity, and generation of H 2 O 2 (supplemental Fig. 4).
p66 Shc -generated H 2 O 2 appears to regulate insulin signaling selectively and at multiple levels. In p66 ShcϪ/Ϫ adipocytes, activation of AKT by insulin is attenuated, whereas MAPK activation is normal. Selective activation of AKT by p66 Shc might be due to a direct effect of p66 Shc -generated H 2 O 2 on phosphatases that regulate phosphatidylinositol 3-kinase, such as PTEN (35). In this respect we noticed that the effect of p66 Shc is much more marked on FOXO, a critical transcription factor downstream of AKT (37), than on AKT itself and that the effect of p66 Shc on insulin-dependent gene expression is markedly selective. Some of the gene regulations induced by insulin in WT adipocytes were maintained in the absence of p66 Shc ; others were attenuated or lost, and "new" transcriptional effects of insulin became evident. Thus, the selective effects of p66 Shc on the cellular responses to insulin might be achieved through multiple redox-dependent mechanisms that selectively regulate insulin signaling, including AKT phosphorylation, Foxo1 localization, and selected insulin target genes.
Regulation of insulin signaling by p66 Shc results into critical modifications of energy metabolism in adipocytes. By studying the effect of p66 Shc deletion in BAT adipocytes, we discovered two functions of p66 Shc ; they are inhibition of insulin-induced FA oxidation and inhibition of basal and insulin-induced UCP1 expression. These activities of p66 Shc led, respectively, to increased TG accumulation in WAT and BAT adipocytes and reduced oxygen consumption in BAT adipocytes, thus contributing to switch the energetic balance of adipocytes toward energy conservation.
One implication of these findings is that p66 Shc functions to connect mitochondrial respiration to insulin signaling. Because the mitochondrial electron transfer chain is the source of reducing equivalents for the generation of H 2 O 2 by p66 Shc , accumulation of energetic substrates in mitochondria is critical for the redox activity of p66 Shc . On the other hand, p66 Shcgenerated H 2 O 2 sets the threshold of cellular sensitivity to specific insulin effects, including inhibition of mitochondrial FA consumption and increase of TG accumulation. In this model p66 Shc functions as a specific signal transducer that integrates in adipocytes insulin signaling and the mitochondrial energetic state (supplemental Fig. 4). Notably, the consequences of p66 Shc -generated redox signals are not equivalent to generic oxidative damage. In fact, ROS are expected to increase lipid peroxidation in mitochondria and, accordingly, decrease mito-chondrial coupling (41). p66 Shc , instead, generates specifically H 2 O 2 but has no effect on other oxygen radicals such as O 2 . p66 Shc -generated H 2 O 2 has marginal effects on mitochondrial lipid peroxidation (not shown) and, on the contrary, signals to increase mitochondrial coupling.
The signaling activity of p66 Shc in adipocytes has important in vivo consequences. We demonstrated that p66 ShcϪ/Ϫ mice have reduced body weight due to reduced mass of both WAT and BAT. In addition, p66 ShcϪ/Ϫ mice are protected from dietinduced obesity, suggesting that p66 Shc is a genetic determinant of fat development in adult mice. Leanness of the p66 ShcϪ/Ϫ mice is not explained by changes in food intake, intestinal absorption of nutrients, or locomotor activity. Rather, it may reflect defective lipogenesis in adipocytes, as suggested by the reduced lipid accumulation of p66 ShcϪ/Ϫ adipocytes transplanted into WT recipient mice. This interpretation of the mechanisms leading to decreased fat mass in the p66 ShcϪ/Ϫ mice, however, poses the question of how the energy balance is maintained in the absence of p66 Shc and why energy storage is reduced. Notably, p66 ShcϪ/Ϫ mice showed increased basal body temperature and increased basal metabolic rate, suggesting that the increased uncoupled respiration in the BAT mitochondria of p66 ShcϪ/Ϫ mice leads to increased energy expenditure, thus contributing to resistance to body weight gain.
The findings of reduced adiposity in p66 ShcϪ/Ϫ mice might have important implications for the effect of p66 Shc on lifespan.
In conclusion, our findings demonstrate that p66 Shc -generated oxidative signal directly regulates the threshold of sensitivity to insulin in adipocytes and the development of fat tissue in vivo, suggesting that regulated generation of H 2 O 2 has evolved to control energy conservation. Attenuation of p66 Shc or insulin signaling in the fat tissue, as by genetic mutations, reduces adiposity and increases lifespan, posing the question of why the threshold of insulin sensitivity is apparently "so low" in the fat tissue under physiological conditions. One could hypothesize that the physiological size of the fat tissue is set to serve other functions, which are evolutionarily more critical than longevity. Adaptation to cold, which is altered in the lean p66 ShcϪ/Ϫ mice, could be one of such functions or, alternatively, optimization of energy storage when food is available.