Differential Roles of Insulin Receptor Substrates in the Anti-apoptotic Function of Insulin-like Growth Factor-1 and Insulin*

Insulin-like growth factor-1 (IGF-1) and insulin are known to prevent apoptosis. The signaling network of IGF-1 and insulin occurs via multiple pathways involving different insulin receptor substrates (IRSs). To define their roles in the anti-apoptotic function of IGF-1 and insulin, we established brown pre-adipocyte cell lines from wild-type and IRS knockout (KO) animals. In response to 16 h of serum deprivation, IRS-1-deficient cells showed a significant decrease in response to IGF-1 protection from apoptosis, whereas no changes were observed in the IRS-2, IRS-3, or IRS-4 KO cells. Five hours after serum withdrawal, cells already began to undergo apoptosis. At this early time point, IGF-1 and insulin were able to protect both wild-type and IRS-1 KO cells from death by 85–90%. After a longer period of serum deprivation, the protective ability of insulin and IGF-1 was decreased, and this was especially reduced in the IRS-1 KO cells. Reconstitution of these cells with IRS-1, IRS-2, IRS-3, or IRS-1/IRS-2 chimeras restored the anti-apoptotic effects of IGF-1, whereas overexpression of IRS-4 had no effect at long time points and actually reduced the effect of IGF-1 at the short time point. The biochemical basis of the defect in anti-apoptosis was not dependent on phosphorylation of mitogen-activated protein kinase; whereas phosphoinositide 3-kinase activity was decreased by 30% in IRS-1 KO cells. Akt phosphorylation was slightly reduced in these cells. Phosphorylation of the transcription factors cAMP response element-binding protein and FKHR by IGF-1 and insulin was markedly reduced in IRS-1 KO cells. In addition, both IGF-1 and insulin prevented caspase-3 cleavage in the wild-type cells, and this effect was greatly reduced in the IRS-1-deficient cells. These findings suggest that the IRS proteins may play differential roles in the anti-apoptotic effects of IGF-1 and insulin in brown pre-adipocytes, with IRS-1 being predominant, possibly acting through caspase-3-, CREB-, and FKHR-dependent mechanisms.

The pleiotropic effects of IGF-1 1 and insulin on metabolism, mitogenesis, and cell survival are mediated by a complex net-work of intracellular signaling pathways (1). The biological effects of both of these peptides are mediated by the activation of their respective cell surface receptors. Activation of these receptors results in phosphorylation of several IRSs. These, in turn, interact with SH2 domain-containing proteins such as PI3K, Grb2, SHP2, and others. Activation of PI3K leads to activation of the main downstream effector Akt and stimulation of several biological responses, including glucose metabolism and cell proliferation and survival. Association of IRS proteins with Grb2 leads to recruitment of SOS and Ras and results in activation of the MAPK pathway, a major regulatory pathway for gene expression.
The IRS proteins are a growing family of proteins that are phosphorylated by the activated insulin and IGF-1 receptors as well as growth hormone and cytokine receptors. There are at least four members of this family that have been identified (IRS-1, IRS-2, IRS-3, and IRS-4). With the exception of IRS-3, which is about 60 kDa in size and not found in the humans, the other IRS proteins range in size between 160 and 185 kDa and are present in both humans and rodents.
To obtain a better understanding of the physiological role of the four different IRS proteins, we and others have established mice with a targeted knockout (KO) of each IRS gene. IRS-1deficient mice have a phenotype of growth retardation and insulin resistance (2,3), whereas IRS-2 KO mice are overtly diabetic due to reduced ␤-cell mass and insulin resistance (4). In contrast, mice with disruption of IRS-3 do not show any obvious defect in growth or glucose metabolism (5), and mice lacking IRS-4 exhibit very mild defects in growth, reproduction, and glucose homeostasis (6). Our laboratory has successfully established brown pre-adipocyte cell lines derived from different KO mice and begun to use these cells as a model system to study the roles of various IRS proteins in cell growth, survival, and differentiation. We have found that differentiation of IRS-1 KO cells is impaired, suggesting that IRS-1 plays an essential role in differentiation of brown adipocytes (7). In contrast, IRS-2 KO cells can be differentiated into matured adipocytes but have an impaired insulin-induced glucose uptake (8). Recently, we have found that IRS-3 and IRS-4 act as negative regulators of IGF-1 signaling pathway by suppressing the function of IRS-1 and IRS-2 in embryonic fibroblasts (9). However, the extent to which the various IRS proteins may play unique versus redundant or complementary roles in other insulin-and IGF-1-mediated responses is still poorly understood.
Apoptosis is a form of programmed cell death, which plays a critical role in controlling cell number and eliminating misplaced cells, and is characterized by chromatin condensation, cytoplasmic blebbing, and DNA fragmentation (10). At the molecular level, apoptosis involves at least three major components: the Bcl-2 protein family; the caspases, which belong to a family of cysteine proteases; and the Apaf-1/CED-4 protein that relays the signals integrated by Bcl-2 protein family to caspases (11). Depletion of growth factors is a common cause of apoptosis. Several peptide growth factors are considered to generate signals for growth and survival, including IGF-1 and insulin. Multiple mechanisms have been proposed by which these factors protect cells from apoptosis, including the PI3K/ Akt pathway (12)(13)(14), the Ras/MAPK pathway (13), Jun Nterminal kinase (15), and p38 MAPK (16). These pathways lead to phosphorylation of several downstream proteins involved in apoptosis, such as the Bcl-2 family member Bad (17), caspase-9 (18), the forkhead transcription factors (19), CREB (20), NF-B (21), and others. Although intensive studies have been done to elucidate the involvement of different downstream signaling molecules in the anti-apoptotic function of IGF-1 and insulin, little is know about the contribution of different IRS proteins in mediating these responses.
In the present study, we have investigated the role of the IRS proteins in the anti-apoptotic function of IGF-1 and insulin by using KO cell lines derived from all four IRS KO mice. IRS-1 deficiency causes a significant reduction in response to IGF-1 and insulin protection from serum withdrawal-induced apoptosis. Caspase-3 and the transcription factors CREB and FKHR appear to be downstream mediators of IRS-1, which may mediate the anti-apoptotic function of IGF-1 and insulin.
Cell Culture, Plasmids, Transfection, and Retroviral Infection-Brown pre-adipocytes were isolated and immortalized as previously described (7,23). Cells were maintained in Dulbecco's modified Eagle's medium containing 10% FBS at 37°C in a 5% CO 2 environment. Retroviral expression vectors of human IRS-1, mouse IRS-2, mouse IRS-3, and mouse IRS-4 were prepared as previously described (7)(8)(9). N1.C2 and N2.C1 chimeras were constructed by using the AflIII sites of IRS-1 and IRS-2, which located approximately half of each protein. The N1.C2 chimera had N-terminal half of IRS-1 and C-terminal region of IRS-2. Conversely, the N2.C1 chimera contained the N-terminal domain of IRS-2 and C-terminal half of IRS-1. 3 g of these plasmids was transfected into ⌽NX-packaging cells in 6-cm-diameter plates using the calcium phosphate method (24). The cells were refed 10 h later, and viral supernatants were harvested 48 h after transfection. KO cells were infected at 70% confluence in a 12-well plate with Polybrene (8 g/ml)-supplemented virus-containing supernatant for overnight. 48 h after infection, cells from each well were trypsinized and transferred to a 15-cm-diameter plate. Stable cell lines were established by selection in medium containing 250 g/ml bleomycin analogue, ZeocinC.
For transient transfection, 5 ϫ 10 5 cells were plated on a well of six-well plates and grown overnight at 37°C. 5 g of pEBG-mBad DNA (Cell Signaling Technology) was transfected into the cells using LF2000 reagent (Invitrogen, Carlsbad, CA) for 5 h. Then the cells were washed twice with medium and incubated in Dulbecco's modified Eagle's medium containing 10% FBS for overnight. 24 h after transfection, cells were serum-deprived for 4 h and stimulated with 10 nM IGF-1 or insulin for 10 min. Cell lysates were prepared as described below.
Apoptosis Assays-Cells were plated at 4000 cells per well in multiple 96-well plates and incubated at 37°C for 2 days. Then the cells were washed twice with medium containing no FBS and serum-deprived in medium supplemented with 0.1% bovine serum albumin alone or treated with 100 nM IGF-1 or insulin for indicated times. Cells cultured in serum-containing medium were used as negative controls. Apoptosis was measured by detection of DNA fragmentation using a cell death detection enzyme-linked immunosorbent assay kit (Roche Molecular Biochemicals, Indianapolis, IN). The enrichment of mono-and oligonucleosomes released into cytosol was calculated using the formula suggest by the manufacture: (absorbance of sample cells)/(absorbance of negative control cells). The protective effects of IGF-1 and insulin were calculated by using the following formula: (enrichment factors of serumdeprived cells subtracts enrichment factors of IGF-1-or insulin-treated cells)/(enrichment factors of serum-deprived cells). The numbers were then multiple by 100 to give the percentage of protection. Mean and standard error values from three or four independent experiments were calculated as percentage of wild-type control within each experiment.
Immunoblotting-Cells grown on a 100-mm dish were serum-deprived overnight in medium containing 0.1% bovine serum albumin and then treated for the indicated times with IGF-1 or insulin at a final concentration of 10 nM. After stimulation, cells were washed twice with ice-cold phosphate-buffered saline and scraped into 0.5 ml of lysis buffer as previously described (25). Protein concentrations were determined using the Bradford protein assay (Bio-Rad). Lysates (30 -50 g) were subjected to SDS-PAGE followed by immunoblotting using specific antisera and detection with chemiluminescence (ECL, Amersham Biosciences, Piscataway, NJ).
For detection of caspase-3 cleavage, cells were grown to confluence and serum-deprived for 4 or 7 h in the absence or presence of IGF-1 or insulin. Cells cultured in serum-containing medium were used as a negative control. Both floating and attached cells were collected and subjected to lysis and immunodetection as described above.
Phosphoinositide 3-Kinase Assay-Cell lysates were obtained as described above, and supernatants containing 500 g of protein were subjected to immunoprecipitation with anti-phosphotyrosine antibody (4G10) overnight at 4°C. Immune complexes were collected with protein A-Sepharose, washed three times with lysis buffer, washed twice with PI3K reaction buffer (20 mM Tris-HCl, pH 7.4, 100 mM NaCl, and 0.5 mM EDTA), and resuspended in 50 l of PI3K reaction buffer containing 0.1 mg/ml phosphatidyl inositol (Avanti Polar Lipids). The reactions were performed, and phosphorylated lipids were separated by thin-layer chromatography as described previously (26).
Statistical Analysis-Data are expressed as mean Ϯ S.E. Differences between two groups were evaluated by an unpaired Student t test. p Ͻ 0.05 was defined as indicating the presence of a statistically significant difference.

IRS-1 Plays a Critical Role in the Anti-apoptotic Function of IGF-1-
To determine if the brown pre-adipocyte cell lines were sensitive to serum withdrawal-induced apoptosis, we grew the wild-type cells in media deprived from serum in the absence or presence of IGF-1 or insulin for various times (Fig. 1A). Cells began to undergo apoptosis at as early as 5 h after serum deprivation as measured in the DNA fragmentation assay and reached a maximal level at 13 h. Both IGF-1 and insulin were able to protect these cells from apoptosis, with IGF-1 having the greater effect. At 13 h, IGF-1 was able to reduce cell death about 75%, whereas insulin could only protect 40% of the cells, and little or no apoptosis occurred when cells were cultured in the presence of serum. These results were confirmed by using other assays for apoptosis, such as propidium iodide staining coupled with flow cytometry analysis, and TUNEL assays (data not shown).
Using the quantitative real-time PCR, we estimated the relative amount of different IRS variants in wild-type cells. 2 Both IRS-1 and IRS-2 were abundant in these cells with IRS-1 expressed at a slightly higher level than IRS-2. IRS-3 was expressed at a low level in the pre-adipocytes, whereas IRS-4 was barely detectable. We next examined the ability of IGF-1 to protect cells from 16-h serum withdrawal-induced apoptosis in different IRS KO cells (Fig. 1B). There was a significant 30% decrease of IGF-1 protection in the IRS-1 KO cells as compared with the wild-type cells. In contrast, the anti-apoptotic effect of IGF-1 appeared to be normal in the IRS-2-, IRS-3-, or IRS-4deficient cells. Basal rates of apoptosis following serum withdrawal were similar in all KO cells to that of the wild-type cells (data not shown). These data suggested that IRS-1 played an important role in the anti-apoptotic function of IGF-1.
Biphasic Responses of IGF-1 and Insulin in Protection from Serum Withdrawal-induced Apoptosis-To further investigate the role of IRS-1 in mediating the anti-apoptotic function of IGF-1 and insulin, we assessed the time course of apoptosis and rescue in wild-type and IRS-1 KO cells (Fig. 2). In both cell types, there was a time-dependent decrease in the ability of IGF-1 and insulin to protect against cell death. Interestingly, during the first 5 h, both hormones were able to protect at least 85% of the cells from apoptosis in both cell lines. After 5 h, deficiency of IRS-1 caused a significant reduction in the antiapoptotic ability of IGF-1 and insulin. Based on these observations, we defined a biphasic response to IGF-1 and insulin protection of cells during serum withdrawal-induced apoptosis. At the early phase (e.g. 5 h), IRS-1 deficiency did not affect the anti-apoptotic function of both factors, whereas at the late phase (e.g. 13 h) IRS-1 played an important role.
We also examined other downstream effectors that may mediate the anti-apoptotic function of IGF-1 and insulin. Caspase-3 is one of the key executioners of apoptosis. Activation of caspase-3 requires proteolytic processing of its inactive zymogen into active p17 and p12 subunits. Using an antibody that specifically recognized the large fragment (17 kDa) of activated caspase-3, we detected significant accumulation of cleaved p17 subunit of caspase-3 at as early as 4 h of serum deprivation (Fig. 3A, lanes 1 and 5). In the wild-type cells, both IGF-1 and insulin were able to prevent the occurrence of caspase-3 cleavage, although with insulin the effect was not complete (Fig. 3A, lanes 2 and 3). In the IRS-1 KO cells, there was a slight increase of cleaved caspase-3, even in the presence of IGF-1 (Fig. 3A, lane 6), and this process was dramatically increased in the insulin-treated cells (Fig. 3A, lane 7). As a control, serum completely block caspase-3 cleavage in both wild-type and IRS-1 KO cells (Fig. 3A, lanes 4 and 8). After 7 h of serum deprivation, there was an increase in the amount of cleaved caspase-3 in both wild-type and IRS-1 KO cells treated with either IGF-1 or insulin, but a similar pattern was observed (Fig. 3B). Taken together, these data suggest that, in the early phase, although we could not detect any defect in the anti-apoptotic function of IGF-1 and insulin in the IRS-1 KO cells by DNA fragmentation assay, these cells already displayed the defect in response to IGF-1 and insulin protection as measured by other earlier apoptotic event, such as caspase-3 cleavage.
Re-expression of IRS-1 in IRS-1-deficient Cells Reconstitutes the Anti-apoptotic Function of IGF-1 at the Late Phase-To confirm that the lack of IRS-1 was responsible for impaired responses to IGF-1 and insulin protection from apoptosis, and to examine the possible complementary versus redundancy among IRS family members with respect to their ability to mediate the anti-apoptotic function of IGF-1 and insulin, the IRS-1 KO cells were infected with retroviruses expressing fulllength of IRS-1, IRS-2, IRS-3, or IRS-4. Western blot analysis revealed about 30 -80% (average 65%) IRS-1 re-expression of that seen in the wild-type cells (Fig. 4A). IRS-2 protein was present in both wild-type and IRS-1 KO cells (Fig. 4B). The IRS-1 KO cells exhibited a slight increase in IRS-2 protein expression compared with the wild-type cells. This is consistent with previous observation by Valverde et al. (27) in fetal brown adipocytes and was not altered by IRS-1 re-expression. Using retroviral-mediated gene transfer, we could increase the level of IRS-2 protein to about 2-fold over that in the parental cells. Both IRS-3 and IRS-4 proteins could be detected only in cells infected with retroviruses encoding the respective genes (Fig.  4, C and D). All IRS proteins became tyrosine-phosphorylated upon 5 min of IGF-1 stimulation (Fig. 4E). Consistent with our previous findings in embryonic fibroblasts (9) Cells were grown on 96-well plates and serum-deprived for indicated time in the absence or presence growth factors. Apoptosis assay were performed by detection of DNA fragmentation as described under "Experimental Procedures." A, effects of IGF-1, insulin, and serum on protection from apoptosis in wild-type cells. Data are presented as the degree of apoptosis as measured by a cell death detection enzyme-linked immunosorbent assay kit over a time course. B, IGF-1 protection from serum withdrawal-induced apoptosis in wild-type and different IRS KO cells. Data are from three independent experiments and are presented as the protective effects of IGF-1 from apoptosis caused by 16-h serum deprivation as described under "Experimental Procedures." either IRS-3 or IRS-4 in the IRS-1 KO cells showed a significant decrease in both IRS-2 protein expression and tyrosine phosphorylation of IRS-2 (Fig. 4, B and D). To estimate the relative "functional" concentration of each IRS protein in the reconstituted cells, we quantified the corresponding phosphorylated IRS protein (Fig. 4E). This revealed reconstitution of IRS tyrosine phosphorylation to about 90, 100, 110, and 140% of that of wild-type in cells reconstituted with IRS-1, -2, -3, and -4, respectively. This result indicated that all the isoforms were sufficiently and almost equally expressed in the reconstituted cells.
We next examined the ability of IGF-1 to protect these cell lines from 16-h serum withdrawal-induced apoptosis (Fig. 5A). IRS-1-reconstituted cells showed a very significant increase in response to IGF-1 protection from apoptosis as compared with the IRS-1-deficient cells (p Ͻ 0.001). Expression of IRS-2 or IRS-3 in these cells also increased the anti-apoptotic function of IGF-1 (p Ͻ 0.05). In contrast, IRS-4 overexpression did not improve the deficiency in response to IGF-1 protection from cell death.
Overexpression of IRS-4 in IRS-1-deficient Cells Inhibits the Anti-apoptotic Function of IGF-1 at the Early Phase-Brown pre-adipocytes are very sensitive to serum withdrawal-induced apoptosis. By 5 h of serum deprivation, we observed cell death as marked by DNA fragmentation (Figs. 1A and 2), caspase-3 cleavage (Fig. 3A), by microscopy (data not shown), and annexin staining coupled with flow cytometry analysis (data not shown). At the early phase (e.g. 5 h), IRS-1 was not required for the anti-apoptotic effect of IGF-1 and insulin, because both hormones produced almost equal protective effects in both the wild-type and IRS-1 KO cells. Interestingly, overexpression of IRS-4 in the IRS-1-deficient cells caused an almost 50% reduction of IGF-1 protection from 5-h serum withdrawal-induced apoptosis, while other IRS proteins did not show any effect at this time point (Fig. 5B). These data suggested that IRS-4 might act as a negative regulator in the anti-apoptotic function of IGF-1.
Deficiency of the Anti-apoptotic Effect of IGF-1 in IRS-1 KO Cells Can Be Restored by IRS-1/IRS-2 Chimeras-To further address the structure-function relationship of IRS proteins, we generated cell lines expressing IRS-1/IRS-2 chimeric proteins. Fortuitously, both IRS-1 and IRS-2 contained an AflIII site located approximately in the middle of each protein, and this site was in-frame with the coding sequence of IRS-1 and IRS-2 allowing for simple exchange of the two halves of each protein (Fig. 6A). This resulted in two different IRS-1/IRS-2 chimeras. The N1.C2 construct contained an N-terminal half of IRS-1 and C-terminal region of IRS-2. Conversely, the N2.C1 chimera had all the N-terminal domains of IRS-2 and C-terminal half of IRS-1. Both chimeras were expressed in the IRS-1 KO cells by retroviral-mediated gene transfer. Using two different antibodies, JD 287 and JD 288, which recognize amino acids 511-859 and 859 -1233 of IRS-1, respectively, we found that the N1.C2 construct was expressed at a similar level to that seen in the wild-type cells, but the N2.C1 chimera was expressed only about 10% of that of the wild-type cells (Fig. 6B). We also noticed that the levels of IRS-1 re-expression varied from 30% to 80% of that seen in the wild-type cells, which was consistent with the findings in Fig. 4. Interestingly, both chimeras appeared to be able to enhance the protective effect of IGF-1 from apoptosis caused by 16-h serum deprivation (Fig. 6C). Expression of the N1.C2 construct in IRS-1-deficient cells displayed a greater effect than that of expression of IRS-1 or IRS-2 alone or the N2.C1 chimera in the IRS-1 KO cells. The enhancing effect of N1.C2 construct may be due to overexpression of this chimeric protein in the cells or to an important role of the N terminus in mediating the anti-apoptotic effects of IRS-1.
Impact of Expression of Different IRS Proteins on IGF-1induced PI3K Activity, Akt Phosphorylation, and MAPK Phosphorylation in IRS-1 KO Cells-To address the potential signaling pathways involved in the anti-apoptotic function of IGF-1, we determined PI3K activity, Akt phosphorylation, and MAPK phosphorylation in wild-type, IRS-1 KO, and IRS-1 KO cells reconstituted with different IRS proteins (Fig. 7). As all the IRS proteins became tyrosine-phosphorylated by IGF-1 stimulation (Fig. 4E), we measured phospho-tyrosine (pY)-associated PI3K activity in these cells. The IRS-1-deficient cells showed about 30 -40% decrease of pY-associated PI3K activity. It was slightly increased by IRS-1 re-expression. Overexpression of IRS-2 in these cells also increased PI3K activity, whereas IRS-3 overexpression showed no effect. Interestingly, overexpression of IRS-4 in the IRS-1-deficient cells increased pY-associated PI3K activity to the level above that in the wildtype cells, although it did not rescue the anti-apoptotic effect of the hormone (compare Figs. 7A and 5A).
Activation of PI3K leads to the recruit of Akt to membrane and promotes its phosphorylation by other membrane-associated phospholipid-depend kinase (28,29). Ser-473 phosphorylation plays a critical role in the activation of Akt. Activation of Akt by Western blotting with a phospho-specific antibody against Akt Ser-473 revealed a slight decrease in the IRS-1deficient cells (Fig. 7B). None of the IRS proteins significantly affected IGF-1-stimulated Akt phosphorylation. Interestingly, IRS-3 or IRS-4 overexpressing cells showed an increased basal level of Akt phosphorylation, and this is consistent with our previous finding in the embryonic fibroblast cells (9).
The MAPK cascade is another major signaling pathway activated by IGF-1 and insulin stimulation (1). As previously observed in tissues of IRS-1 KO mice (30), activation of MAPK detected by Western blotting with a phospho-specific p44/p42 MAPK antibody was not altered in the IRS-1-deficient cells (Fig. 7C). Overexpression of different IRS proteins in the IRS-1 KO cells had no influence on MAPK phosphorylation.

Phosphorylation of Transcription Factors CREB and FKHR Is Significantly Reduced in the IRS-1 KO Cells-To investigate
other downstream molecules that may involve in the antiapoptotic function of IGF-1 and insulin via IRS-1-dependent pathway, we investigated three molecules, which have been described to be involved in anti-apoptosis by IGF-1 or insulin in other studies. CREB is a transcription factor that binds to its specific sequence known as CRE. IGF-1 is known to prevent apoptosis in neuronal cells through a CREB-dependent pathway (31). IGF-1 induces Bcl-2 promoter activities through CREB and CRE (20). Phosphorylation of the serine 133 residue of CREB increases its transcriptional activities (32). ATF-1 is a CREB-related transcription factor that shares identical consensus phosphorylation sequence of CREB and is detected by the same phospho-specific antibody (31). We found that phosphorylation of CREB (Ser-133) was induced by 5 min of IGF-1 or insulin stimulation in the wild-type cells (Fig. 8A). This stimulation was markedly diminished in the IRS-1-dificient cells and fully restored by IRS-1 reconstitution. Expression of IRS-2 or IRS-3 in the IRS-1 KO cells also slightly restored CREB phosphorylation by IGF-1 stimulation. In general, the phosphorylation pattern of ATF-1 was paralleled that of CREB. These data suggest an important role of IRS-1 in IGF-1-and insulin-stimulated CREB/ATF-1 phosphorylation.
Forkhead transcription factors, such as FKHR, have been demonstrated to be phosphorylated by insulin (33) and IGF-1 (34) and negatively regulated by these survival factors through an Akt-dependent pathway, leading to anti-apoptosis (19,35,36). Phosphorylation of FKHR (Ser-256), but not FKHRL1 (Ser-253), was greatly induced by IGF-1 stimulation for 30 min in wild-type cells and diminished in the IRS-1 null cells (Fig. 8B). However, IRS-1 reconstitution was unable to restore this response.
Bad is an apoptotic member of the Bcl-2 family. Survival factors inhibit the apoptotic activity of Bad by phosphorylation at Ser-112 and Ser-136 (37). In brown pre-adipocytes, expression of Bad protein was barely detected with immunoprecipitation followed by immunoblotting (data not shown). Therefore, we transiently transfected wild-type and IRS-1 KO cells with a eukaryotic expression vector carrying the cDNA of mouse Bad. IGF-1 and insulin slightly increased Bad phosphorylation in both cell lines, however, no significant difference was observed between wild-type and IRS-1 null cells (Fig. 8C). DISCUSSION IGF-1 and insulin evoke diverse biological effects through receptor-mediated tyrosine phosphorylation of the IRS proteins. The four known IRS proteins share similar overall architecture. Disruption of each individual IRS gene causes distinct phenotypes in mice (2, 4 -6), suggesting that the four IRS proteins play different roles in regulation of metabolism, cell growth, survival, and differentiation. One of the best approaches in determining the degree to which the IRS proteins play unique versus redundant or complementary roles in insulin-and IGF-1-mediated responses is to examine the ability of these proteins to replace one another in mediating the actions of insulin and IGF-1. This approach, however, requires a cell system that lacks IRS proteins or in which an IRS protein is removed and replaced by other IRS proteins. Thus far most of work in the field has relied on 32D leukemia cells (38 -40) or cells derived from IRS-deficient mice (7-9, 27, 30). Using SV40 T antigen-immortalized brown pre-adipocytes, we previously demonstrated that IRS-1 plays a critical role in brown adipocyte differentiation (7), whereas IRS-2 is essential for insulinstimulated glucose uptake in cells (8).
Brown adipose tissue is specialized for non-shivering thermogenesis and regulated energy expenditure. It has been shown that age-related decline in thermogenesis and energy expenditure in humans and rodents is associated with a significant reduction in the mass of functional brown fat tissue (41). The exact factors contributing to this decrease in brown adipose tissue are largely unknown. In this study, we find that brown pre-adipocytes are very sensitive to apoptosis under conditions of serum withdrawal. In this tissue, both IGF-1 and insulin exhibit anti-apoptotic effects. These effects are impaired in the IRS-1-deficient cells and restored by re-expression of IRS-1, emphasizing a specific role of IRS-1 in the antiapoptotic effects. IRS-1 deficiency also defines a biphasic nature of the apoptosis. During the first 5 h, IRS-1 deficiency does not affect the protective effect of IGF-1 and insulin, whereas during the second phase of apoptosis (e.g. 13-16 h), Reconstitution studies show that IRS-1 re-expression not only fully restores but also enhances the anti-apoptotic function of IGF-1 in IRS-1-deficient cells at 16 h. The enhancing effect seen in the IRS-1-reconstituted cells may due to a combination of IRS-1 re-expression and the compensatory response of the cells to overcome IRS-1 deficiency, such as an increase in IRS-2 protein synthesis. Whereas this increase is not sufficient to completely overcome the IRS-1 deficiency, retrovirally induced overexpression of IRS-2 or IRS-3 in IRS-1 KO cells is also able to restore the defect of these cells in response to IGF-1 protection from apoptosis caused by 16-h serum deprivation, suggesting that these three IRS proteins may be in part interchangeable in mediating the anti-apoptotic effect of IGF-1. This finding is further supported by the experiments using IRS-1/

FIG. 7. Changes of IGF-1 induced PI3K activity, Akt phosphorylation, and p44/42 MAPK phosphorylation in wild-type cells, IRS-1 KO cells, and IRS-1 KO cells reconstituted with different IRS proteins.
Cells were stimulated with 10 nM IGF-1 for 5 min, and cell lysates were prepared as described under "Experimental Procedures." A, 500 g of proteins was immunoprecipitated with anti-phosphotyrosine antibody and subjected to PI3K assay. B and C, 50 g of whole cell lysates was separated by SDS-PAGE and immunoblotted with antiphospho-Akt (Ser-473) antibody (B) or anti-p44/42 MAPK (Tyr-204) antibody (C). Blots were exposed to Kodak BioMax films, and the films were scanned by densitometer and quantified by ImageQuaNT software. Data were obtained from three or four independent experiments. IRS-2 chimeras, N1.C2 and N2.C1, both of which are able to effectively mediate the protective effect of IGF-1, indicating that at least the anti-apoptotic function of IGF-1 is mediating through common features shared by IRS-1 and IRS-2, such as the PH and PTB domains and/or the conserved tyrosine phosphorylation sites.
Interestingly, overexpression of IRS-4 in the IRS-1 KO cells inhibits the anti-apoptotic function of IGF-1 at the early phase, suggesting that IRS-4 may act as a negative regulator. This is not due to insufficient expression of IRS-4 in the IRS-1 KO cell, because, in fact, tyrosine phosphorylation of the reconstituted IRS-4 reached the level of 140% of combined tyrosine phosphorylation of IRS-1 and IRS-2 seen in wild-type cells. Moreover, using two IRS-1 KO cell lines expressing different levels of IRS-4, we find a dose-responsive effect of IRS-4 to inhibit the anti-apoptotic function of IGF-1 (data not shown). One possible explanation for the negative role of IRS-4 in mediating the anti-apoptotic function of IGF-1 is that IRS-4 may attenuate the IRS-2 signaling. Support for this possibility stems from our previously studies using IRS-1-deficient embryonic fibroblasts (9). In that study, we have found that overexpression of IRS-3 and IRS-4 impair IRS-2-mediated signaling through the mechanisms of decreasing IRS-2 mRNA and protein levels and IGF-1-stimulated tyrosine phosphorylation of IRS-2. Similarly, in this study, we observe a significant decrease of IRS-2 protein expression and a diminished IGF-1-induced tyrosine phosphorylation of IRS-2 in IRS-1 KO cells overexpressing IRS-4. When overexpressed in these cells, IRS-3 also slightly reduces tyrosine phosphorylation of IRS-2 stimulated by IGF-1 but does not affect IRS-2 protein synthesis. IRS-4 most likely decreases IGF-1-stimulated IRS-2 tyrosine phosphorylation by competing for receptor interaction. Because IRS-4 is clearly tyrosine-phosphorylated, it may bind to the NPXY motif in the juxtamembrane domain of the IGF-1 receptor that binds the IRS proteins (42). Alternatively, different subcellular localizations of the IRS proteins (43) may also contribute to the efficiency of interaction between the IGF-1 receptor and the various IRS proteins. Whether or not IRS-4 can attenuate IRS-1-mediated signaling in brown pre-adipocytes is currently under investigation and beyond the scope of the current study.
Multiple signaling pathways have been suggested to mediate the anti-apoptotic function of IGF-1 and insulin; however, there is still uncertainty as to the "essential" components for this response, which may due to the usage of different model systems in which expression of receptors and signaling molecules vary from one system to the other. The most frequent paradigm involves signaling through IRS proteins, PI3K and Akt (14). In the IRS-1-deficient cells, we find a 30 -40% decrease in IGF-1stimulated pY-associated PI3K activity as compared with that of the wild-type cells. This decrease could contribute to the defect in anti-apoptotic function of IGF-1, however, IRS-1 reconstitution in these cells only partially restores PI3K activity while completely restoring anti-apoptosis, and overexpression of IRS-4 causes an enhanced increase of PI3K activity without rescue of anti-apoptosis. Likewise, there is a very slight decrease in Ser phosphorylation of Akt in the IRS-1 null cells, and none of the IRS proteins significantly affect this pathway in these cells. Furthermore, treatment of the wild-type cells with the PI3K inhibitor, LY294002, only slightly inhibited the protective effects of IGF-1 and insulin from apoptosis caused by 5and 16-h serum deprivation (data not shown). Thus, changes in PI3K activity and Akt phosphorylation do not correlate well with the IGF-1-induced anti-apoptotic phenotypes observed in different cell lines, suggesting that some alternative pathways may also involved. This is consistent with the work of Kulik and Weber (44) suggesting IGF-1 may utilize PI3K-and Aktdependent and -independent pathways to regulate cell survival depending on the backgrounds of the cells.
Two additional potential anti-apoptotic pathways activated by the IGF-1 receptor are activation of MAPK and mitochondrial translocation of Raf (17). Phosphorylation of MAPK is unchanged in the IRS-1 KO cells, and treatment of the wildtype cells with MEK inhibitors (PD98059 and UO126) had no effect on the anti-apoptotic function of IGF-1 (data not shown), indicating that, at least in the brown pre-adipocytes, MAPK activation is not required for the anti-apoptotic function of IGF-1. Although the exact signaling pathway or pathways utilized by IGF to exert its anti-apoptotic function in these cells are still not clear, it appears that multiple different pathways may be utilized to achieve the full protective effect. When one major pathway is not available, alternative pathways take over part of these functions (17). Nevertheless, our data suggest among the several signaling pathways elicited by IGF-1 and insulin, the IRS-1-dependent pathway plays a more important role in anti-apoptotic function of these growth factors in the brown pre-adipocytes. In addition, significant apoptosis is detected in the wild-type cells even with hormone treatment, suggesting the existence of IRS-independent pathways or the requirement of other receptor pathway for full protection of the cells from apoptosis. Growth factors, such as epidermal growth factor or platelet-derived growth factor, may be potential candidates for this action, because they are known to utilize pathways independent of IRS to protect cells from apoptosis in many other systems (56). Moreover, the presence of ␤-adrenergic receptors in these cells (23) and the roles of these receptors in anti-apoptosis (57) suggest that ␤-adrenergic stimulation may also contribute to protection of these cells from death.
In searching for other downstream molecules involved in the anti-apoptotic function of IGF-1, we find that cleavage of caspase-3 caused by growth factor withdrawal is strongly inhibited by IGF-1 and insulin. Since caspase activation is required for complete apoptotic phenotype, and caspase-3 is one of the executioners in this process (11), prevention of caspase-3 cleavage is certainly one of the most powerful sites by which IGF-1 and insulin protect cells from apoptosis. In addition, this preventive effect of IGF-1 and insulin is impaired in the IRS-1 KO cells, suggesting that caspase-3 is involved in the IRS-1-dependent pathway by which these growth factors protect cells from death.
Transcription factor CREB is known to regulate IGF-1-dependent anti-apoptosis in a phosphorylation-dependent manner (Ser-133), presumably via increasing the transcription of Bcl-2 (20). IGF-1-and insulin-induced CREB phosphorylation is dramatically reduced in the IRS-1-deficient cells and can be fully restored by reconstitution of these cells with IRS-1, but not other IRS proteins, suggesting that IRS-1 is an essential factor in CREB phosphorylation by IGF-1 and insulin stimulation. Many kinases have been suggested to activate CREB, including protein kinase A (32), protein kinase C (45), PI3K/ Akt(46),ERK1/2(47),p38MAPK (31,48),andcalcium/calmodulindependent protein kinase (49). We recently found that IGF-1induced CREB phosphorylation in wild-type brown preadipocytes is blocked by inhibitors of MEK and calcium/ calmodulin. 2 Because MAPK phosphorylation remains unchanged in the IRS-1 KO cells, IGF-1 may utilize pathways coupling IRS-1 and calcium/calmodulin to regulate CREB activity. Previously we have identified an insulin-dependent interaction between IRS proteins and the calcium ATPases of the sarcoplasmic and endoplasmic reticulum, SERCA 1 and SERCA2, in muscle and heart (50). Whether or not that the coupling of IRS-1 and calcium/calmodulin-dependent pathways is involved in the activation of CREB by IGF and insulin or plays a role in the anti-apoptotic effect of these factors is currently under investigation.
Forkhead transcription factors are other potential targets that may be involved in IGF-1-or insulin-dependent antiapoptosis. Recently, several lines of evidence have suggested a critical role of these factors in cell death (19,35,36,51); both IGF-1 and insulin are able to regulate FKHR activity through a phosphorylation-dependent mechanism (33,34,52,53). Phosphorylation of FKHR by IGF-1 and insulin stimulation is markedly reduced in the IRS-1 KO cells, but IRS-1 reconstitution failed to completely restore this deficiency. This may reflect the fact that a normal concentration of IRS-1 is required for restoration of FKHR phosphorylation, and the 65% reconstitution of IRS-1 that occurs for retroviral infection of IRS-1-deficient cells may not be sufficient to recover FKHR phosphorylation.
A third potential target for the protective effect of IGF-1 and insulin is Bad (54). Although many studies have emphasized an important role of Bad in regulation of cell death (17, 37), we did not find a significant change in Bad phosphorylation by IGF-1 or insulin stimulation between wild-type and IRS-1 KO cells. Constitutive activation of Akt has also been shown to regulate cell survival in human macrophages through a mechanism independent of Bad (55). Thus, Bad may not be the critical factor in the anti-apoptotic function of IGF-1 and insulin in brown pre-adipocytes.
In summary, using the brown pre-adipocytes derived from different IRS KO animals, we have shown that IRS-1 plays a critical role in the anti-apoptotic effect of IGF-1 and insulin. This defect can be restored by reconstitution of these cells with IRS-1, IRS-2, or IRS-3. IRS-4 may act as a negative regulator in this function. The anti-apoptotic pathways of IGF-1 and insulin are very complex and may involve caspase-3, CREB, and FKHR. These data suggest that the IRS proteins may play unique as well as complementary roles in IGF-1-and insulinmediated anti-apoptosis, which may be due to activation of different signaling pathways.