Role of IRS-3 in the Insulin Signaling of IRS-1-deficient Brown Adipocytes*

Insulin receptor substrate-1 (IRS-1) plays an essential role in mediating the insulin signals that trigger mitogenesis, lipid synthesis, and uncoupling protein-1 gene expression in mouse brown adipocytes. Expression of IRS-3 is restricted mainly to white adipose tissue; expression of this IRS protein is virtually absent in brown adipocytes. We have tested the capacity of IRS-3 to mediate insulin actions in IRS-1-deficient brown adipocytes. Thus, we expressed exogenous IRS-3 in immortalized IRS-1 (cid:1) / (cid:1) brown adipocytes at a level comparable with that of endogenous IRS-3 in white adipose tissue. Under these conditions, IRS-3 signaling in response to insulin was observed, as revealed by tyrosine phosphorylation of IRS-3, and the activation of phosphatidylinositol (PI) 3-kinase associated with this recombinant protein. However, although insulin promoted the association of Grb-2 with recombinant IRS-3 in IRS-1 (cid:1) / (cid:1) cells, the exogenous expression of this IRS family member failed to activate p42/44 MAPK and mitogenesis in brown adipocytes lacking IRS-1. Downstream of PI 3-kinase, IRS-3 expression restored in-sulin-induced Akt phosphorylation, which is impaired by the 32 P-labeled cDNA (10 6 cpm/ml) for 24 h at 42 °C. cDNA labeling was carried out with [ (cid:2) - 32 P]dCTP by using a multiprimer DNA-labeling system. Blots were hybridized with probes for fatty acid synthase (FAS) (27) and for ADD-1/SREBP-1c (28). Membranes were subjected to au-toradiography, and the relative densities of the hybridization signals were determined by densitometric scanning of the autoradiograms. Transfections were performed in duplicate from three independent experiments. Determination of [ 3 H]Thymidine Incorporation into Acid-insoluble Material— DNA synthesis was determined after 24 h of cell culture in the absence or presence of various doses of insulin by [ 3 H]thymidine incorporation (0.2 (cid:4) Ci/ml) into acid-insoluble material over the last 4 h of culture (19). Results are expressed as disintegrations per min/dish.

Insulin receptor substrate-1 (IRS-1) plays an essential role in mediating the insulin signals that trigger mitogenesis, lipid synthesis, and uncoupling protein-1 gene expression in mouse brown adipocytes. Expression of IRS-3 is restricted mainly to white adipose tissue; expression of this IRS protein is virtually absent in brown adipocytes.

We have tested the capacity of IRS-3 to mediate insulin actions in IRS-1-deficient brown adipocytes. Thus, we expressed exogenous IRS-3 in immortalized IRS-1 ؊/؊ brown adipocytes at a level comparable with that of endogenous IRS-3 in white adipose tissue. Under these conditions, IRS-3 signaling in response to insulin was observed, as revealed by tyrosine phosphorylation of IRS-3, and the activation of phosphatidylinositol (PI) 3-kinase associated with this recombinant protein. However, although insulin promoted the association of Grb-2 with recombinant IRS-3 in IRS-1 ؊/؊ cells, the exogenous expression of this IRS family member failed to activate p42/44 MAPK and mitogenesis in brown adipocytes lacking IRS-1. Downstream of PI 3-kinase, IRS-3 expression restored in-
sulin-induced Akt phosphorylation, which is impaired by the lack of IRS-1 signaling. Whereas the generation of IRS-3 signals enhanced adipocyte determination and differentiation-dependent factor 1/sterol regulatory element-binding protein (ADD-1/SREBP-1c) and fatty acid synthase mRNA and protein expression, activation of this pathway was unable to reconstitute CCAAT/enhancerbinding protein ␣ and uncoupling protein-1 transactivation and gene expression in response to insulin. Similar results were obtained following insulin-like growth factor-I stimulation. In brown adipocytes expressing the IRS-3F4 mutant, the association of the p85␣ regulatory subunit via Src homology 2 binding was lost, but insulin nevertheless induced PI 3-kinase activity and Akt phosphorylation in a wortmannin-dependent manner. In contrast, activation of IRS-3F4 signaling failed to restore the induction of ADD-1/SREBP-1c and fatty acid synthase gene expression in IRS-1-deficient brown adipocytes. These studies demonstrate that recombinant IRS-3 may reconstitute some, but not all, of the signals required for insulin action in brown adipocytes. Thus, our data further implicate a unique role for IRS-1 in triggering insulin action in brown adipocytes.
Brown adipose tissue (BAT) 1 is a unique site for non-shivering thermogenesis in mammals characterized by the expression of the uncoupling protein-1 (UCP-1) in the mitochondria (1). UCP-1 uncouples fatty acid oxidation from ATP synthesis, which allows dissipation of energy from substrate oxidation as heat (2). In addition, BAT is a major site for lipid metabolism, using fatty acids as the main fuel to maintain its thermogenic capacity (for review see Ref. 3). Rodent brown adipocytes differentiate at the end of the fetal life via two programs, an adipogenic program of lipid synthesis and expression of lipogenic enzymes, which generates the multi-locular fat droplets phenotype (4 -6), and a thermogenic program related to heat production and UCP-1 expression (7). Regarding thermogenic differentiation, the main pathway involved in the regulation of UCP-1 gene expression is noradrenergic (8 -11). However, because the noradrenergic stimulus induced by hypothermia after birth is not yet fully developed in brown adipose tissue during late fetal development (5,9), other potential modulators of differentiation-related gene expression have been considered. Recently (12,13), regulatory elements for triiodothyronine and retinoic acid have been identified in the UCP-1 promoter, suggesting alternative pathways for brown fat thermogenesis. Moreover, during recent years (6) our laboratory has found that fetal brown adipocytes display high affinity binding sites for insulin. This classic hormone appears to be the primary mediator of fetal brown adipogenesis based on its ability to induce expression of metabolic genes (6,14); insulin may additionally exert a role in thermogenesis by inducing UCP-1 expression (15).
Previous work from our laboratory has demonstrated that primary fetal brown adipocytes express high levels of both IRS-1 and IRS-2. Interestingly, IRS-1, but not IRS-2, associates with Grb-2 in these cells to trigger ras/MAPK activation, thereby mediating the mitogenic response to insulin. The PI 3-kinase cascade is the main pathway implicated in the expression of both adipogenic-and thermogenic-related genes in primary fetal brown adipocytes (16). In addition, ras/MAPK signaling has been shown to down-regulate the UCP-1 gene expression in response to insulin (17). Recently (18), we have generated immortalized fetal brown adipocyte cell lines from IRS-1 knockout. This cell culture model has enabled us to identify the IRS-mediated signaling pathways responsible for the biological effects of insulin on brown adipose tissue during late fetal development. Thus, in brown adipocyte cell lines displaying the typical multilocular lipid droplets phenotype, IRS-1, but not IRS-2, is essential for insulin-stimulated mitogenesis (19), lipid synthesis (18), and also UCP-1 gene expression (20). No information is currently available regarding the role of IRS-3 in the insulin signaling cascades of brown adipocytes. IRS-3 was identified based on its specific expression in rat white adipose tissue (21) where it plays a major role in GLUT4 translocation and glucose transport in primary adipocytes (22,23). IRS-3 mediates PI 3-kinase activation rather than Grb-2/ras signaling pathway, suggesting a specific role for IRS-3 in triggering the metabolic rather than the mitogenic actions of insulin (24).
In the current study, we have addressed the potential of IRS-3 to mediate insulin signaling in brown adipocytes deficient for IRS-1. Our results reveal that expression of recombinant IRS-3 in our brown adipocyte model restores insulinstimulated Akt activation and fatty acid synthase gene expression but not p42/p44 MAPK activation and UCP-1 gene expression, critical pathways that are impaired in brown adipocytes lacking IRS-1. Thus, these observations further confirm a unique role for IRS-1 in triggering insulin action in brown adipose tissue.
Cell Culture and Retroviral Infections-Immortalized fetal brown adipocyte cell lines derived from wild-type and IRS-1-deficient mice have been generated as described previously (18). Viral Bosc-23 packaging cells were transfected at 70% confluence by calcium phosphate coprecipitation with 3 g/6-cm dish of the hygromycin-resistance retroviral vector pBabe encoding wild-type IRS-3 (kindly provided by C. R. Kahn, Joslin Diabetes Center, Boston, MA). Then, IRS-1-deficient brown adipocytes (clone 4) were infected at 60% confluence with polybrene (4 g/ml)-supplemented virus. Selection with 200 g/ml hygromycin was started 72 h after infection to generate stable cell lines, the expression of IRS-3 being assessed by Western blot.
Transfections-IRS-1-deficient brown adipocytes (clone 4) were cultured for 24 h in the presence of 10% FCS and then, when 60 -70% confluence was reached, cells were co-transfected with 10 g of pCIS2 IRS-3F4 and pBABE hygro constructs according to the calcium phosphate-mediated protocol. After 4 -6 h of incubation, cells were shocked with 3 ml of 15% glycerol for 2 min, washed, and then fed with Dulbecco's modified Eagle's medium-10% FCS. 24 h after transfection, hygromycin (200 g/ml) was added to select stable transfectants. The expression of the IRS-3F4 mutant was assessed by Western blot.
Immunoprecipitations-Quiescent cells were treated without or with several doses of insulin as indicated and lysed at 4°C in 1 ml of lysis buffer containing 10 mM Tris/HCl, 5 mM EDTA, 50 mM NaCl, 30 mM sodium pyrophosphate, 50 mM NaF, 100 M Na 3 VO 4 , 1% Triton X-100, and 1 mM phenylmethylsulfonyl fluoride, pH 7.6. Lysates were clarified by centrifugation at 15,000 ϫ g for 10 min. After protein content determination, equal amounts of protein (500 -600 g) were immunoprecipitated with the corresponding antibodies at 4°C. The immune complexes were collected on protein A-agarose or anti-mouse IgG-agarose beads. Immunoprecipitates were washed with lysis buffer and extracted for 5 min at 95°C in 2ϫ SDS-PAGE sample buffer (200 mM Tris/HCl, 6% SDS, 2 mM EDTA, 4% 2-mercaptoethanol, 10% glycerol, pH 6.8) and analyzed by SDS-PAGE.
Western Blotting-After SDS-PAGE, proteins were transferred to Immobilon membranes and were blocked using 5% non-fat dried milk or 3% bovine serum albumin in 10 mM Tris/HCl, 150 mM NaCl pH 7.5, and incubated overnight with several antibodies as indicated in 0.05% Tween 20, 10 mM Tris/HCl, 150 mM NaCl, pH 7.5. Immunoreactive bands were visualized using the ECL Western blotting protocol (Amersham Biosciences).
PI 3-kinase Activity-PI 3-kinase activity was measured in the anti-IRS-3 or anti-Tyr(P) immunoprecipitates by in vitro phosphorylation of phosphatidylinositol as described previously (18).
Protein Determination-Protein determination was performed by the Bradford dye method (25) using the Bio-Rad reagent and bovine serum albumin as the standard.
RNA Extraction and Northern Blot Analysis-Cells were cultured as described under "Results" and in the figure legends. At the end of the culture time, cells were washed twice in ice-cold phosphate-buffered saline, and RNA was isolated as described (26). Total cellular RNA (10 g) was submitted to Northern blot analysis, i.e. electrophoresed on 0.9% agarose gels containing 0.66 M formaldehyde, transferred to Ge-neScreen membranes (PerkinElmer Life Sciences), and cross-linked to the membranes by ultraviolet light. Hybridization was performed in 0.25 mM NaHPO 4 , pH 7.2, 0.25 M NaCl, 100 g/ml denatured salmon sperm DNA, 7% SDS, and 50% deionized formamide, containing denatured 32 P-labeled cDNA (10 6 cpm/ml) for 24 h at 42°C. cDNA labeling was carried out with [␣-32 P]dCTP by using a multiprimer DNA-labeling system. Blots were hybridized with probes for fatty acid synthase (FAS) (27) and for ADD-1/SREBP-1c (28). Membranes were subjected to autoradiography, and the relative densities of the hybridization signals were determined by densitometric scanning of the autoradiograms.
Extraction of Nuclear Proteins-At the end of the culture time, cells were resuspended at 4°C in Buffer A (10 mM HEPES-KOH, pH 7.9, 1.5 mM MgCl 2 , 10 mM KCl, 0.5 mM dithiothreitol, 0.2 mM phenylmethylsulfonyl fluoride, 0.75 g/ml leupeptin, 0.75 g/ml aprotinin), allowed to swell on ice for 10 min, and then vortexed for 10 s. Samples were centrifuged, and the pellet was resuspended in cold Buffer C (20 mM HEPES-KOH, pH 7.9, 25% glycerol, 420 mM NaCl, 1.5 mM MgCl 2 , 0.2 mM EDTA, 0.5 mM dithiothreitol, 0.2 mM phenylmethylsulfonyl fluoride, 0.75 g/ml leupeptin, 0.75 g/ml aprotinin) and incubated on ice for 20 min for high salt extraction. Cellular debris was removed by centrifugation for 2 min at 4°C, and the supernatant fraction was stored at Ϫ70°C.
Isolation of Mitochondrial Protein-Cells were cultured as described under "Results" and in the figure legends. At the end of the culture time, cells were scrapped off in isotonic isolation buffer (1 mM EDTA, 10 mM HEPES, 250 mM sucrose, pH 7.6), collected by centrifugation at 2,500 ϫ g for 5 min at 4°C, and resuspended in hypotonic isolation buffer (1 mM EDTA, 10 mM HEPES, 50 mM sucrose, pH 7.6). Then, cells were incubated at 37°C for 5 min and homogenized under a Teflon pestle (Overhead Stirrer; Wheaton Instruments, Milville, NJ). Hypertonic isolation buffer (1 mM EDTA, 10 mM HEPES, 450 mM sucrose, pH 7.6) was added to balance the tonicity of the buffer. Samples were centrifuged at 10,000 ϫ g for 10 min, and the pellets, containing the mitochondrial fraction, were resuspended in isotonic isolation buffer, and mitochondrial protein content was determined.
CAT Assays-Cells growing in the presence of 10% FCS were transiently transfected according to the calcium phosphate-mediated protocol as described previously (15). The plasmid constructs used were UCP-1-CAT (where the CAT reporter gene is under the control of a 4551-bp full-length 5Ј flanking region of the UCP-1 promoter) (29) and pCMV ␤-galactosidase (gal) (a viral promoter driving expression of the reporter gene ␤-gal). Ten micrograms of DNA-CAT, together with 2 g of DNA-␤-gal (to monitor transfection efficiency), were added to each 10-cm dish. After 4 h of incubation, cells were fed with serum-free medium for 15 h and stimulated with various doses of insulin or dibutyryl cyclic AMP (Bt 2 cAMP) for a further 24 h. Then, cells were harvested, and lysates were prepared for CAT and ␤-gal activity assays. CAT activity was determined by incubating 70 l of cell extracts with 0.25 Ci of [ 14 C]chloramphenicol and 0.5 mM acetyl coenzyme A in 0.25 mM Tris, pH 7.8, at 37°C for 15 h, and then samples were submitted to thin layer chromatography. The amount of substrate acetylated was directly quantified with a radioimaging device (Fujifilm BAS-1000). CAT enzyme activity was expressed in arbitrary units normalized to the internal control ␤-gal (assayed according to the Stratagene protocol).
Transfections were performed in duplicate from three independent experiments.

Determination of [ 3 H]Thymidine Incorporation into Acid-insoluble
Material-DNA synthesis was determined after 24 h of cell culture in the absence or presence of various doses of insulin by [ 3 H]thymidine incorporation (0.2 Ci/ml) into acid-insoluble material over the last 4 h of culture (19). Results are expressed as disintegrations per min/dish.

Expression of IRS-3 and Its Tyrosine Phosphorylation in IRS-1-deficient Brown Adipocytes-Previous experiments per-
formed in IRS-1-deficient fetal brown adipocytes revealed an essential role of IRS-1 in mediating insulin-induced lipid synthesis (18) and UCP-1 expression (20), IRS-2 being unable to compensate these responses. IRS-3 is a member of IRS family mainly expressed in white adipose cells (21). IRS-3 expression is virtually absent in BAT extracts, as shown in the Western blot depicted in Fig. 1A. To investigate a possible role of IRS-3 in triggering insulin actions in IRS-1-deficient brown adipocytes, we overexpressed IRS-3 in immortalized IRS-1 Ϫ/Ϫ brown adipocytes by retroviral gene transfer. By using this approach, we generated immortalized fetal brown adipocyte IRS-1-deficient cell lines expressing similar IRS-3 levels than those found in white adipose tissue extracts (see Fig. 1A). Our next step was to investigate whether overexpressed IRS-3 in IRS-1-deficient brown adipocytes undergoes tyrosine phosphorylation upon insulin stimulation. Confluent immortalized brown adipocytes (wild-type, IRS-1 Ϫ/Ϫ , IRS-1 Ϫ/Ϫ (ϩIRS-3)) were serum-deprived for 20 h and then stimulated with insulin (10 -100 nM) for 5 min. Total protein was immunoprecipitated with anti-IRS-3 antibody and submitted to Western blot analysis with the anti-Tyr(P) antibody. As shown in Fig. 1B, insulin induced a marked increased in IRS-3 tyrosine phosphorylation in IRS-1 Ϫ/Ϫ brown adipocytes overexpressing IRS-3, no tyrosine phosphorylation of IRS-3 being detected either in immortalized wild-type or IRS-1 Ϫ/Ϫ brown adipocytes.
Insulin Induced the Association of IRS-3 with p85a and Activated PI 3-kinase in IRS-1-deficient Brown Adipocytes-IRS-3 contains phosphotyrosine residues in a consensus sequence for binding to p85␣ regulatory subunit of PI 3-kinase following insulin stimulation (23). Accordingly, we investigated whether overexpressed IRS-3 associated p85␣ in immortalized IRS-1 Ϫ/Ϫ (ϩIRS-3) fetal brown adipocytes. Serum-deprived cells were further stimulated with insulin (10 -100 nM) for 5 min. Total protein was immunoprecipitated with anti-IRS-3 antibody and submitted to SDS-PAGE followed by Western blot with anti-p85␣ antibody. In the representative experiment shown in Fig. 2A, we did not detect association of p85␣ with IRS-3 in wild-type brown adipocytes stimulated with insulin. Surprisingly, insulin stimulation of IRS-1 Ϫ/Ϫ brown adipocytes resulted in a slight increase in p85␣ associated to IRS-3, despite of the absence of tyrosine-phosphorylated IRS-3 in these cells (see Fig. 1B). As expected, overexpressed IRS-3 in IRS-1 Ϫ/Ϫ brown adipocytes associated p85␣ in parallel to the levels of its tyrosine phosphorylation. Next, we checked whether IRS-3 was able to activate PI 3-kinase in brown adipocytes.

FIG. 1. Exogenous expression of IRS-3 in IRS-1-deficient brown adipocytes.
A, IRS-1 Ϫ/Ϫ brown adipocytes were cultured in the presence of 10% FCS to 80% confluence. Then, cells were infected with the retroviral expression vector pBabe-IRS-3-hygro for 5 h. After infection, brown adipocytes were maintained in culture medium for 72 h before selection with hygromycin (200 g/ml). Stable resistant cells were expanded and cultured under growing conditions. Cells were lysed, and 100 g of total protein was submitted to SDS-PAGE and analyzed by Western blot with the anti-IRS-3 antibody. Lanes 1 and 2 are controls of freshly isolated white and brown adipose tissue extracts (WAT and BAT, respectively). An arrowhead indicates the position of IRS-3. The position of molecular weight markers (ϫ 10 Ϫ3 ) is shown on the left. A representative experiment is shown. Scanning densitometry of the mean of three independent experiments is shown on the right. B, IRS-1 Ϫ/Ϫ and IRS-1 Ϫ/Ϫ (ϩIRS-3) brown adipocytes were serum-starved for 15 h, stimulated for 5 min with insulin (10 -100 nM), and lysed. 600 g of total protein was immunoprecipitated with the anti-IRS-3 antibody. The immune complexes were submitted to Western blot analysis with anti-Tyr(P) and anti-IRS-3 antibodies, respectively. A representative experiment of three is shown. Cells were stimulated as described above, and PI 3-kinase activity was measured in anti-IRS-3 immune complexes as described under "Experimental Procedures." As shown in Fig.  2B (left panel), IRS-3-associated PI 3-kinase enzymatic activity in parallel to IRS-3/p85␣ association. Although no IRS-3-associated PI 3-kinase activity was detected in wild-type cells, in IRS-1 Ϫ/Ϫ brown adipocytes insulin slightly induced IRS-3-associated PI 3-kinase activity in a dose-dependent manner, this effect being much higher when IRS-3 was overexpressed in these cells. Our laboratory has reported recently (18,20) that the lack of IRS-1 resulted in a 30% decrease in total (anti-Tyr(P)-associated) PI 3-kinase activity in insulin-stimulated brown adipocytes. To test whether IRS-3 can replace IRS-1 in stimulating total PI 3-kinase activity in fetal brown adipocytes, cells were treated with insulin (10 -100 nM) for 5 min and further assayed for PI 3-kinase enzymatic activity in anti-Tyr(P) immune complexes. Fig. 2B (right panel) showed that IRS-3 expression in IRS-1 Ϫ/Ϫ brown adipocytes turned out in a recovery of total PI 3-kinase activity, this effect being truly significant at 100 nM insulin concentration.
IRS-3 Expression Restored the Insulin Effect on Akt Phosphorylation, but Not MAPK Activation, in IRS-1 Ϫ/Ϫ Brown Adipocytes-In fetal brown adipocytes, the lack of IRS-1-associated PI 3-kinase activity cannot be compensated by IRS-2 and results in a failure of insulin to stimulate downstream targets such as Akt phosphorylation (18,20). The fact that IRS-3 binds to and activates PI 3-kinase in IRS-1 Ϫ/Ϫ cells prompted us to investigate a possible mechanism of recovery of the insulin response in activating Akt in the absence of IRS-1. Cells were stimulated with various doses of insulin and subsequently lysed. Akt Ser-473 phosphorylation was determined by direct Western blot analysis in wild-type, IRS-1 Ϫ/Ϫ , and IRS-1 Ϫ/Ϫ (ϩIRS-3) cells (Fig. 3A). In fetal brown adipocytes, the lack of IRS-1 resulted in a loss of Akt Ser-473 phosphorylation; only a remnant Akt Ser-473 phosphorylation was detected at 100 nM insulin concentration. The exogenous expression of IRS-3 in IRS-1 Ϫ/Ϫ brown adipocytes resulted in a total recovery of Akt Ser-473 phosphorylation in response to insulin.
It has been extensively reported that the association of IRS proteins with Grb-2 results in the activation of MAPK (30). In fact, we have shown recently (19) that in brown adipocytes the effect of insulin on MAPK activation is IRS-1-dependent. Accordingly, we investigated whether IRS-3 led to activation of MAPK in brown adipocytes. First, we checked IRS-3/Grb-2 association in IRS-1 Ϫ/Ϫ brown adipocytes expressing IRS-3. Cells were stimulated with various doses of insulin and then lysed. Total protein (600 g) was immunoprecipitated with anti-Grb-2 antibody and analyzed by Western blot with anti-IRS-3 antibody. As shown in Fig. 3B, there was no IRS-3/Grb-2 association either in wild-type or IRS-1 Ϫ/Ϫ brown adipocytes. However, insulin-stimulated IRS-1 Ϫ/Ϫ cells expressing exogenous IRS-3 showed a marked IRS-3/Grb-2 association (Fig. 3B,  upper panel). Next, we measured MAPK phosphorylation in IRS-1 Ϫ/Ϫ brown adipocytes in the presence of IRS-3. As shown in a representative Western blot depicted in Fig. 3B (lower  panel), insulin increased MAPK phosphorylation in wild-type cells but not in IRS-1 Ϫ/Ϫ , as reported previously (19). The expression of IRS-3 in IRS-1 Ϫ/Ϫ brown adipocytes did not result in the activation of MAPK in response to insulin. Thymidine incorporation into DNA revealed that insulin did not stimulate DNA synthesis in IRS-1 Ϫ/Ϫ cells despite the presence of p42 phosphorylation background, and as compared with wild-type cells, the expression of IRS-3 failed to restore mitogenesis.
IRS-3 Restored Insulin-induced Adipogenic-gene Expression in IRS-1 Ϫ/Ϫ Brown Adipocytes-Insulin-induced adipogenic differentiation program in brown adipocytes can be monitored by the expression of a set of lipogenic enzymes including FAS (5). In addition, insulin-induced FAS mRNA expression was inhibited in the presence of PI 3-kinase inhibitors in primary cultures of fetal brown adipocytes (15,16) or in immortalized fetal brown adipocytes lacking IRS-1 (20). Accordingly, we investigated whether the presence of IRS-3 in IRS-1 Ϫ/Ϫ cells could positively affect lipogenic gene expression. Cells were cultured for 12 and 24 h in the presence of various doses of insulin, and FAS mRNA and protein content were assessed by Northern and Western blot analysis, respectively. As shown in Fig. 4A, insulin treatment for 12 h induced FAS mRNA in wild-type brown adipocytes in a dose-dependent manner, maximal effect being elicited at 100 nM insulin concentration. This effect was abolished in IRS-1 Ϫ/Ϫ cells, as reported previously (20). However, the expression of IRS-3 in IRS-1-deficient cells resulted in a substantial up-regulation of FAS mRNA following insulin stimulation, reaching similar levels than that of the wild-type. At the protein level, IRS-3 led to a recovery of insulin-induced FAS protein content. Because FAS expression has been shown to be dependent on ADD1/SREBP-1c transcription factor (31), we analyzed ADD1/SREBP-1c expression in the presence of insulin in the three cell types. As shown in Fig. 4B, both ADD-1/SREBP-1c mRNA and protein content (the nuclear mature form of ADD-1/SREBP-1c) were increased after 12 h (mRNA) or 24 h (protein content) of insulin treatment in wildtype and IRS-1 Ϫ/Ϫ (ϩIRS-3) cells but not in IRS-1 Ϫ/Ϫ cells. These results indicate that, in brown adipocytes, IRS-3 and its signaling through PI 3-kinase up-regulated FAS expression as a result of the increase in ADD1/SREBP-1c expression.
IRS-3 Failed to Restore UCP-1 Gene Expression in Response to Insulin in IRS-1 Ϫ/Ϫ Brown Adipocytes-We have shown previously (15,16) that insulin is a novel thermogenic factor in brown adipocyte primary cultures involved in UCP-1 expression during late fetal development in a PI 3-kinase-dependent manner. Furthermore, the lack of IRS-1 in brown adipocytes resulted in a loss of the insulin response in inducing UCP-1, IRS-2 being unable to compensate this response (20). Based on that, our next purpose was to define whether IRS-3 can replace IRS-1 in mediating thermogenic-related gene expression induced by insulin in brown adipocytes. UCP-1 protein content was analyzed by direct Western blot analysis in mitochondrial protein extracts of IRS-1 ϩ/ϩ , IRS-1 Ϫ/Ϫ , and IRS-1 Ϫ/Ϫ (ϩIRS-3)

FIG. 3. Exogenous expression of IRS-3 restores Akt phosphorylation, but not MAPK activation, in IRS-1 ؊/؊ brown adipocytes following insulin stimulation. A, quiescent cells (15 h serum-starved)
were stimulated with insulin (10 -100 nM) for 5 min. Cells were lysed, and 50 g of total protein was submitted to SDS-PAGE and analyzed by immunoblotting with the corresponding antibodies against phospho-Akt (Ser-473) and total Akt. The autoradiograms corresponding to four experiments were quantitated by scanning densitometry. Results are expressed as arbitrary units of Akt phosphorylation and are means Ϯ S.E. B, upper panel, quiescent cells were stimulated with insulin (10 -100 nM) for 5 min. Cell lysates were immunoprecipitated with the anti-Grb-2 antibody, and the resulting immune complexes were analyzed by Western blot with anti-IRS-3 and anti-Grb-2 antibodies, respectively. A representative experiment of three is shown. Lower panel, serum-starved cells were incubated without or with insulin (10 -100 nM) for 5 min. Then, cells were lysed, and equal amounts of protein were submitted to SDS-PAGE followed by Western blot analysis with anti-phospho-MAPK and anti-MAPK antibodies. A representative experiment of three is shown. Cells were cultured for 24 h in serum-free medium either in the absence or presence of insulin (1-100 nM). DNA synthesis was determined by [ 3 H]thymidine incorporation (0.2 Ci/ml) over the last 4 h of culture. After two washes with ice-cold phosphate-buffered saline, cells were lysed, and trichloroacetate-precipitable DNA was then counted for incorporated radioactivity. Results are expressed as disintegrations/min/dish and are means Ϯ S.E. from six independent experiments, each one performed in triplicate. cell types upon insulin stimulation. As shown in Fig. 5A (left  panel), UCP-1 protein content was up-regulated by 24 h of insulin treatment in wild-type brown adipocytes whereas in IRS-1-deficient cells this response was completely blunted. Interestingly, the exogenous expression of IRS-3 in IRS-1-deficient cells did not restore the insulin effect on UCP-1 protein content. Another approach was to study the effect of insulin on transactivating the UCP-1 promoter in the three cell types. Cells were transiently transfected with the 4551-bp full-length UCP-1 promoter driving the expression of the CAT reporter gene (29). Upon transfection, cells were cultured for 24 h in a serum-free medium either in the absence or presence of 100 nM insulin and then CAT activity was determined. As shown in Fig. 5A (right panel), the 4551-CAT fusion gene was weakly transcribed in unstimulated wild-type brown adipocytes, insulin treatment resulting in 2.5-fold increase in CAT activity. IRS-1 Ϫ/Ϫ brown adipocytes lacked insulin stimulation of UCP-1 CAT activity, exogenous expression of IRS-3 being unable to restore insulin-induced UCP-1 CAT activity. Regarding the transcriptional machinery involved in the regulation of UCP-1, the 5Ј-flanking region of this gene contains C/EBP-regulated sites (32). In fact, recent data from our laboratory (20) indicated that C/EBP␣ protein content was positively regulated by insulin. Therefore, we performed anti-C/EBP␣ Western blot analysis in nuclear extracts from control and insulin-stimulated brown adipocytes. As shown in Fig. 5B, the expression of C/EBP␣ was significantly up-regulated by 24 h of insulin treatment in wild-type cells but not in IRS-1 Ϫ/Ϫ cells, regardless the exogenous expression of IRS-3.

IRS-3 Restored FAS Expression, but Failed to Restore UCP-1 Gene Expression, in IRS-1 Ϫ/Ϫ Brown Adipocytes following
IGF-I Stimulation-Our next purpose was to investigate the response of brown adipocyte cell lines to IGF-I regarding differentiation-related gene expression. Cells were cultured for 12 and 24 h in the presence of 10 nM IGF-I, and FAS mRNA and protein content were assessed by Northern and Western blot analysis, respectively. As shown in Fig. 6A (upper panel), IGF-I treatment for 12 h induced FAS mRNA in wild-type brown adipocytes. This effect was blunted in IRS-1 Ϫ/Ϫ cells. However, the expression of IRS-3 in IRS-1-deficient cells resulted in a substantial up-regulation of FAS mRNA, reaching similar levels than that of the wild-type. At the protein level, IRS-3 restored IGF-I-induced FAS protein content (Fig. 6A, middle  panel). As shown in Fig. 6A (lower panel), UCP-1 protein con-

FIG. 4. Overexpression of IRS-3 restores the insulin effect on FAS and ADD1/SREBP-1c expression in IRS-1 ؊/؊ brown adipocytes.
A, upper panel, quiescent wild-type, IRS-1 Ϫ/Ϫ (ϩIRS-3), and IRS-1 Ϫ/Ϫ brown adipocytes were serum-starved for 15 h and stimulated with insulin (10 -100 nM) for a further 12 h. Control cells were cultured in the absence of the hormone. Total RNA was isolated, submitted to Northern blot analysis, and hybridized with FAS cDNA probe. A representative experiment is shown. Scanning densitometry of the mean of five independent experiments is shown on the right. Brown adipocytes were serum-starved for 15 h and stimulated with insulin (10 -100 nM) for 24 h. Lower panel, at the end of the culture time, cells were lysed, and 50 g of total protein were submitted to SDS-PAGE followed by Western blot analysis with anti-FAS and anti-␤-actin antibodies. Representative autoradiograms are shown. Scanning densitometry of the mean of three-four independent experiments is shown on the right. B, upper panel, quiescent wild-type, IRS-1 Ϫ/Ϫ (ϩIRS-3), and IRS-1 Ϫ/Ϫ brown adipocytes were serum-starved for 15 h and stimulated with insulin (10 -100 nM) for a further 12 h. Control cells were cultured in the absence of the hormone. Total RNA was isolated, submitted to Northern blot analysis, and hybridized with ADD-1/SREBP-1c cDNA probe. A representative experiment is shown. Scanning densitometry of the mean of five independent experiments is shown on the right. Lower panel, brown adipocytes were serum-starved for 15 h and stimulated with insulin (10 -100 nM) for 24 h. At the end of the culture time cells were lysed, and 50 g of nuclear protein were submitted to SDS-PAGE followed by Western blot analysis with anti-ADD-1/SREBP-1c and anti-␤-actin antibodies. Representative autoradiograms are shown. Scanning densitometry of the mean of three-four independent experiments is shown on the right. tent was up-regulated upon 24 h of IGF-I treatment in wildtype brown adipocytes, whereas in IRS-1-deficient cells this response was completely blunted. In addition, the exogenous expression of IRS-3 in IRS-1-deficient cells did not restore the IGF-I effect on UCP-1 protein content. Because of the negative response of insulin or IGF-I in triggering UCP-1 expression and to rule out the possibility of alterations in these cells, we addressed the point of IRS-independent alternative signals. As shown in Fig. 6B, TGF-␤ induced FAS expression in wild-type and IRS-1 Ϫ/Ϫ cells. More importantly, Bt 2 cAMP, a second messenger of the ␤-adrenergic pathway, induced UCP-1 protein expression and UCP1-CAT transactivation at a similar extent in both in wild-type and IRS-1 Ϫ/Ϫ cells.

Expression of the IRS-3F4 Mutant Failed to Associate p85␣ in Insulin-stimulated IRS-1 Ϫ/Ϫ Brown Adipocytes-Results
shown above indicate that IRS-3 can replace IRS-1 in maintaining FAS and ADD1/SREBP-1c expression as a result of the activation of PI 3-kinase/Akt signaling pathway. A step further, we expressed an IRS-3 mutant (denoted IRS-3F4) in which Phe substituted to the four Tyr residues in Tyr-Xaa-Xaa-Met motifs (Tyr-341, Tyr-350, Tyr-361, and Tyr-390), which bind to Src homology 2 domains of p85␣ regulatory subunit (23). Importantly, the expression of IRS-3F4 was similar to that of wild-type IRS-3 in IRS-1 Ϫ/Ϫ brown adipocytes (Fig. 7A). Furthermore, mutation of these four tyrosines abolished the association of p85␣ with IRS-3F4 as revealed by anti-p85␣ Western blot analysis of anti-IRS-3 immunoprecipitates following insulin stimulation (Fig. 7A). Next, we checked whether IRS-1 Ϫ/Ϫ brown adipocytes expressing IRS-3F4 were able to induce PI 3-kinase activity and Akt phosphorylation. Quiescent cells were stimulated with insulin (10 -100 nM) for 5 min, and PI 3-kinase activity was assayed in the anti-IRS-3 immune complexes. Akt phosphorylation was determined by direct Western blot with the anti-phospho-Ser-473 antibody. Insulin induced both PI 3-kinase activity and Akt phosphorylation in IRS-1 Ϫ/Ϫ brown adipocytes expressing IRS-3F4 to a similar extend as that of the wild-type IRS-3 (Fig. 7B). Then, we investigated whether IRS-3-and IRS-3F4-associated PI 3-kinase activity was wortmannin-dependent. Fig. 7C shows that both IRS-3-and IRS-3F4-associated PI 3-kinase activity and Akt phosphorylation were prevented when cells were pretreated with wortmannin.
IRS-3F4 Mutant Failed to Restore Adipogenic Gene Expression in Insulin-stimulated IRS-1 Ϫ/Ϫ Brown Adipocytes-Finally, we explored the effect of overexpressing IRS-3F4 mutant on IRS-1-deficient brown adipocytes regarding the adipogenic phenotype. Cells (IRS-1 Ϫ/Ϫ (ϩIRS-3) and IRS-1 Ϫ/Ϫ (ϩIRS-3F4)) were stimulated with insulin for 12 and 24 h. Then, FAS and ADD-1/SREBP-1c expression was analyzed by Northern and Western blot, respectively. Fig. 8A shows that the expression of the IRS-3F4 mutant did not restore the insulin effect regarding FAS and ADD-1/SREBP-1c mRNA, as compared with the positive effect induced by wild-type IRS-3. These data were reinforced by those depicted in Fig. 8B, where no insulin effect was observed on FAS and ADD-1/SREBP-1c protein content in IRS-1-deficient brown adipocytes expressing IRS-3F4 mutant, as compared with wild-type IRS-3.

FIG. 5. IRS-3 fails to restore C/EBP␣ and UCP-1 gene expression in IRS-1 ؊/؊ brown adipocytes following insulin stimulation.
A, left panel, serum-starved brown adipocytes were stimulated with insulin (10 -100 nM) for 24 h. Mitochondrial proteins were extracted, and equal amounts of protein were submitted to SDS-PAGE and analyzed by Western blot with anti-UCP-1 and anti-cytochrome c antibodies. A representative experiment of three is shown. Right panel, wild-type, IRS-1 Ϫ/Ϫ (ϩIRS-3), and IRS-1 Ϫ/Ϫ brown adipocytes were transiently transfected with 10 g of full promoter UCP-1-CAT fusion gene. Upon transfection, cells were cultured for 24 h in serum-free medium in either the absence or presence of insulin (10 -100 nM). At the end of the culture period, cells were collected and assayed for CAT activity. The relative CAT activity normalized to ␤-gal activity is represented in the histograms. Results are means Ϯ S.E. from three independent experiments. B, quiescent brown adipocytes were serum-starved for 15 h and further stimulated with insulin (10 -100 nM) for 24 h. At the end of the culture, cells were lysed, and equal amounts of nuclear protein were submitted to SDS-PAGE followed by Western blot analysis with the anti-C/EBP␣ antibody. The autoradiograms corresponding to four experiments were quantitated by scanning densitometry. Results are expressed as arbitrary units and are means Ϯ S.E.

DISCUSSION
IRS-3 is the member of IRS family expressed mainly in white adipose tissue (21) where it is detected at levels comparable with IRS-1 (33). However, expression of IRS-3 in brown adipose tissue and brown adipocyte cell lines appears to be much lower than of IRS-1, as IRS-3 was barely detectable in control brown adipocytes and those lacking IRS-1 (Fig. 1). Previous experiments performed in IRS-1-deficient fetal brown adipocytes revealed a critical role for IRS-1 in mediating insulin-induced mitogenesis (19), lipid synthesis (18), and UCP-1 gene expression (20). In sharp contrast, IRS-2 was unable to trigger pathways responsible for the biological actions of insulin in these cells. To investigate whether IRS-3 signals might substitute the role of IRS-1 in mediating insulin action in IRS-1-deficient brown adipocytes, we overexpressed IRS-3 in immortalized IRS-1 Ϫ/Ϫ brown adipocytes by retroviral gene transfer. The level of expression of IRS-3 in IRS-1 Ϫ/Ϫ brown adipocytes was comparable with that detected in white adipose tissue (Fig. 1). Under these conditions, insulin-responsive IRS-3 signaling was established in our brown adipocyte model, as revealed by tyrosine phosphorylation of recombinant IRS-3 and the activation of associated PI 3-kinase. Interestingly, in brown adipocytes lacking IRS-1, we observed an enhanced association of p85␣ and corresponding activation of PI 3-kinase in immunoprecipi-tates of endogenous IRS-3, given the minimal expression of endogenous protein that we detected by Western blotting. The augmented capacity of endogenous IRS-3 to recruit PI 3-kinase may reflect its sub-cellular localization, which could be optimal for generation of lipid second messengers. Nevertheless, this IRS-3 signaling could not compensate for the absence of IRS-1 in mediating insulin-stimulated mitogenesis and lipid synthesis in brown adipocyte cell lines.
Two major insulin signaling pathways have been described in brown adipocytes. Insulin promotes IRS-1/Grb-2 association and p42/44 MAPK activation, producing a mitogenic response (19). Insulin also enhances IRS-1/IRS-2-dependent PI 3-kinase activation and Akt phosphorylation, resulting in the stimulation of lipid synthesis (18) and UCP-1 gene expression (20). Insulin stimulated dose-dependently the association of Grb2 with exogenous IRS-3 in IRS-1 Ϫ/Ϫ brown adipocytes (Fig. 3B). These data contrast the weak association between IRS-3 and Grb-2 observed in white adipocytes, where IRS-3 and IRS-1 are equally abundant (34). However, despite recruitment of Grb2, the exogenous expression of IRS-3 in brown adipocytes lacking IRS-1 failed to activate p42/44 MAPK and mitogenesis. These data differ from those reported in IRS-1-deficient mouse 3T3 embryo fibroblasts, where the impaired mitogenic response to IGF-I was reconstituted by supranormal levels of IRS-1 and was restored to almost normal by IRS-3 expression (35). These data strongly support the notion that IRS-1, but not IRS-3, is an essential component of the brown adipocyte mitogenic response to insulin, a pathway that is dependent on MAPK activation. Interestingly, IRS-3 overexpression reconstituted insulin-induced Akt phosphorylation, a response impaired in brown adipocytes lacking IRS-1 (Fig. 3A).
Diverging signals downstream of IRS-3 were observed in the absence of IRS-1. Whereas the introduction of IRS-3 signals in brown adipocytes led to an enhancement of insulin-mediated ADD-1/SREBP-1c and FAS mRNA and protein expression, exogenous IRS-3 failed to restore the C/EBP␣ and UCP-1 transactivation and gene expression in response to insulin (see Figs. 4 and 5, respectively). The inability of IRS-3 signaling to mediate thermogenic-related gene expression contrasts with results obtained by the exogenous expression of either C/EBP␣ or peroxisome proliferator-activated receptor ␥ in IRS-1-deficient brown adipocytes; both transcription factors induced UCP-1 expression in an insulin-independent manner (20). As insulininduced adipogenic differentiation program in brown adipocytes can be monitored by the expression of a set of lipogenic enzymes including FAS, our data suggest that IRS-3 may exert a role in the induction of adipogenic-related gene expression by insulin, with the regulation of ADD-1 expression representing a critical step. Similarly, given that UCP-1 gene expression reflects induction of thermogenic genes, our data demonstrate that IRS-3 signaling is unable to mediate the effects of insulin upon thermogenic-related gene expression in cells lacking IRS-1 because of its failure to reconstitute expression of C/EBP␣. The phenotypes of IRS-1 knockouts and IRS-1/IRS-3 double knockout mice have revealed dramatic differences in the development and/or maintenance of white fat; IRS-1-deficient mice develop a more or less normal white adipose tissue whereas the double knockouts display profound lipoatrophy. However, no differences in brown adipose tissue were observed between these knockout models, which both develop brown adipose tissue normally other than the fact that cells contain smaller lipid depots (36). More importantly, brown adipose tissue-specific deletion of the insulin receptor in mice (BAT-IRKO) causes profound brown fat lipoatrophy because of a significant reduction in lipid content, which appears to reflect smaller cell size and enhanced eosinophilia. However, the expression of UCP-1 was significantly increased in BATIRKO mice (37). Our current data are consistent with these later observations, emphasizing the role of IRS-3 in generating the insulin signals required for lipid synthesis but not for induction FIG. 7. IRS-3F4 mutant does not associate with p85␣ in response to insulin but activates PI 3-kinase and Akt phosphorylation. A, left, IRS-1 Ϫ/Ϫ brown adipocytes were transiently co-transfected with a plasmid encoding IRS-3F4 mutant, together with a hygromycin-resistant retroviral vector. IRS-1 Ϫ/Ϫ (ϩIRS-3) and IRS-1 Ϫ/Ϫ (ϩIRS-3F4) brown adipocytes were cultured to confluence under growing conditions. Cells were lysed, and equal amounts of protein (500 g) were immunoprecipitated with the anti-IRS-3 antibody. The resulting immune complexes were submitted to SDS-PAGE and analyzed by Western blot with the anti-IRS-3 antibody. The autoradiograms corresponding to three independent experiments were quantitated by scanning densitometry. Results are represented in the histograms. Middle panel, IRS-1 Ϫ/Ϫ (ϩIRS-3) and IRS-1 Ϫ/Ϫ (ϩIRS-3F4) brown adipocytes were serum-starved for 15 h and incubated with insulin (100 nM) for 5 min. Cells were lysed, and 600 g of total protein was immunoprecipitated with the anti-IRS-3 antibody. The immune complexes were washed and subjected to SDS-PAGE followed by immunoblotting with anti-p85␣ and anti-IRS-3 antibodies, respectively. A representative experiment is shown. On the right the results are depicted as arbitrary units of p85␣ association normalized to the amount of total IRS-3. B, serum-deprived brown adipocytes were stimulated with insulin (10 -100 nM) for 5 min. Total protein was used for an in vitro IRS-3-associated PI 3-kinase activity assay or analyzed by Western blot with anti-phospho-Akt and anti-Akt antibodies. C, serum-deprived brown adipocytes were treated with various doses of wortmannin for 30 min before the stimulation with insulin (100 nM) for a further 5 min. Cells were lysed, and total protein was used for an in vitro IRS-3-associated PI 3-kinase activity or analyzed by Western blot with anti-phospho-Akt and anti-Akt antibodies. A representative experiment of four is depicted.
of thermogenic-related genes in brown adipocytes. A key issue arises about the nature of additional signals involved in the UCP-1 expression and thermogenesis in the absence of IRS-1/ IRS-3-mediated insulin signaling. Among candidates, IGF-I or TGF-␤ play the same role as described for insulin (4,38). In addition, our data support the essential role played by the FIG. 8. IRS-3F4 mutant fails to restore FAS and ADD1/SREBP-1c gene expression in IRS-1 ؊/؊ brown adipocytes. A, quiescent wild-type, IRS-1 Ϫ/Ϫ (ϩIRS-3), and IRS-1 Ϫ/Ϫ (ϩIRS-3F4) brown adipocytes were serum-starved for 15 h and stimulated with insulin (10 -100 nM) for a further 12 h. Total RNA was isolated, submitted to Northern blot analysis, and hybridized with FAS and ADD1/SREBBP-1c probes. Representative autoradiograms are shown. Scanning densitometry of the mean of three-four independent experiments is shown on the right. B, quiescent brown adipocytes were serum-starved for 15 h and stimulated with insulin (10 -100 nM) for a further 24 h. Then cells were lysed, and equal amounts of protein were submitted to SDS-PAGE followed by Western blot analysis with anti-FAS, anti-ADD1/SREBP-1c, and anti-␤-actin antibodies. Representative autoradiograms are shown. Scanning densitometry of the mean of three-four independent experiments is shown on the right.
FIG. 9. Specific role of IRS in insulin signaling in brown adipocytes. Insulin signaling in brown adipocytes is regulated by the IRS family. IRS-1 (white arrows) directs insulin signaling to proliferation, lipid synthesis, and UCP-1 expression. However, in IRS-1-deficient brown adipocytes, exogenously expressed IRS-3 (black arrows) can modulate insulin-induced lipid synthesis through PI 3-kinase, Akt, and ADD-1 but cannot reconstitute insulin-induced UCP-1 expression. In addition IRS-2 (gray arrows) maintains the insulin signaling through PI 3-kinase and protein kinase C (PKC), which results in an increase of glucose uptake and Glut-4 translocation. sympathetic nervous enervation through the ␤-adrenergic receptors in inducing UCP-1 gene expression in the absence of IRS-dependent signaling.
Previous data have documented the essential contributions of IRS-1/PI 3-kinase in mediating insulin regulation of FAS and UCP-1 gene expression. This signaling pathway is not required for triggering UCP-2 or UCP-3 gene expression (20). Here we have investigated the role of an IRS-3F4 mutant defective in binding sites for the Src homology 2 domains of p85␣ regulatory subunit of PI 3-kinase (23). As expected, the IRS-3F4 in IRS-1 Ϫ/Ϫ brown adipocytes failed to recruit p85␣ in response to insulin (Fig. 7A). Surprisingly, insulin induced both PI 3-kinase activity and Akt phosphorylation in IRS-1 Ϫ/Ϫ brown adipocytes expressing IRS-3F4 comparable with the activation of these pathways in wild-type IRS-3 (Fig. 7B). Moreover, activation of PI 3-kinase in the IRS-3F4 cells was wortmannin-sensitive (Fig. 7, B and C, respectively). One likely explanation for these observations is that in brown adipocytes lacking IRS-1, insulin induces a weak association between p85 isoforms and the mutant IRS-3 sufficiently to promote PI 3-kinase activation but not strongly enough to survive immunoprecipitation. These results should also be considered in light of previous reports (33) demonstrating that IRS-3, but not IRS-1, is constitutively localized in the plasma membrane in COS-1 cells. Thus, the IRS-34F mutant in response to insulin may recruit and activate PI 3-kinase without an strong association to p85 isoforms. Downstream of PI 3-kinase, IRS-3F4 signaling activates Akt in response to insulin. However, IRS-3F4 signaling failed to up-regulate ADD-1/SREBP-1c and FAS gene expression, which are impaired in the absence of IRS-1 (Fig. 8).
Our results suggest that Akt activation in response to insulin can be dissociated from ADD-1/SREBP-1c expression in cells expressing the IRS-3 mutant lacking canonical p85␣ binding sites, which again may reflect differences in sub-cellular compartmentalization as compared with wild-type IRS-3. The current data are incongruous with results reported from COS-7 cells where exogenous IRS-3 localized to the nucleus and activated transcription (39). Our data strongly suggest that insulin requires IRS-3/p85␣/PI 3-kinase signaling to achieve transcriptional activation of adipogenic-related genes in brown adipocytes deficient for IRS-1.
In conclusion, introduction of IRS-3 signaling restores the PI 3-kinase/Akt pathway but fails to restore MAPK activation and mitogenesis in brown adipocytes lacking IRS-1. Downstream from PI 3-kinase, IRS-3 signaling restores ADD-1/FAS expression but not C/EBP␣/UCP-1. Although an IRS-3F4 mutant does not recruit p85␣, insulin still activates PI 3-kinase/Akt in cells expressing this recombinant molecule. However, IRS-3F4 signaling fails to up-regulate ADD-1/FAS gene expression. Thus, our data suggest that IRS-3 is capable of triggering the adipogenic-related gene expression in brown adipocytes that requires IRS-3/p85␣/PI 3-kinase signaling to achieve transcriptional activation. Finally, given that neither IRS-2 nor IRS-3 can fully reconstitute insulin action in IRS-1-deficient brown adipocytes, our studies thoroughly implicate a unique function for IRS-1 in brown adipose tissue (Fig. 9).