To assess the role of insulin action and inaction in the liver, immortalized hepatocyte cell lines have been generated from insulin receptor substrate (IRS)-2−/− and wild-type mice. Using this model, we have recently demonstrated that the lack of IRS-2 in neonatal hepatocytes resulted in insulin resistance. In the current study, we show that immortalized neonatal hepatocytes undergo apoptosis on serum withdrawal, with caspase-3 activation and DNA laddering occurring earlier in the absence of IRS-2. Insulin rescued wild-type hepatocytes from serum withdrawal-induced caspase-3 activation and DNA fragmentation in a dose-dependent manner, but it failed to rescue hepatocytes lacking IRS-2. In IRS-2−/− cells, insulin failed to phosphorylate Bad. Furthermore, in these cells, insulin was unable to translocate Foxo1 from the nucleus to the cytosol. Adenoviral infection of wild-type cells with constitutively active Foxo1 (ADA) induced caspase-8 and caspase-3 activities, proapoptotic gene expression, DNA laddering and apoptosis. Dominant negative Foxo1 regulated the whole pathway in an opposite manner. Prolonged insulin treatment (24 hours) increased expression of antiapoptotic genes (Bcl-xL), downregulated proapoptotic genes (Bim and nuclear Foxo1), and decreased caspase-3 activity in wild-type hepatocytes but not in IRS-2−/− cells. Infection of IRS-2−/− hepatocytes with adenovirus encoding IRS-2 reconstituted phosphatidylinositol 3-kinase (PI 3-kinase)/Akt/Foxo1 signaling, restored pro- and antiapoptotic gene expression, and decreased caspase-3 activity in response to insulin, thereby blocking apoptosis. In conclusion, IRS-2 signaling is specifically required through PIP3 generation to mediate the survival effects of insulin. Epidermal growth factor, via PIP3/Akt/Foxo1 phosphorylation, was able to rescue IRS-2−/− hepatocytes from serum withdrawal-induced apoptosis, modulating pro- and anti-apoptotic gene expression and downregulating caspase-3 activity. Supplementary material for this article can be found on the HEPATOLOGY website (http://interscience.wiley.com/jpages/0270-9139/suppmat/index.html). (HEPATOLOGY 2004;40:1285-1294.)
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
In the liver, regulation of apoptosis is essential for development and hepatic homeostatic mechanisms.1 Thus, liver hyperplasia during development or regeneration may be aided by inhibition of apoptosis. Conversely, liver atrophy occurs by apoptosis of liver cells in the absence of regeneration (for review, see Patel et al.2). Progression of the apoptotic program can be inhibited at different levels depending on the origin of the death stimulus. In this regard, insulin promotes survival in a number of cell types. This effect is mediated by a complex network of intracellular signaling pathways.3 The cascade begins when the activated insulin receptor (IR) phosphorylates several docking proteins, including the insulin receptor substrate (IRS) family4 and Shc. Then, phosphorylated IRSs (IRS-1, -2, -3, and -4) connect with the phosphatidylinositol 3-kinase (PI 3-kinase) signaling pathway,5 which plays a central role in the metabolic actions elicited by insulin, and the Ras/MAPK pathway, a major regulatory pathway for gene expression.6 Among the members of the IRS family, IRS-2 triggers insulin actions in the liver. In fact, mice deficient in IRS-2 develop hepatic insulin resistance and fatty liver along with β-cell failure,7–10 producing a severe type-2 diabetic phenotype. At the molecular level, a defect in PI 3-kinase/Akt signaling occurs in immortalized hepatocytes derived from IRS-2–deficient mice.11
Apoptosis is characterized by chromatin condensation, cytoplasmic blebbing, and DNA fragmentation.12 Mitochondria are deeply involved in the regulation of apoptosis.13, 14 The family of Bcl-2–related proteins function like checkpoints through which survival and death signals must pass before they determine the cell fate. Indeed, a major site of activity of the Bcl-2 proteins is the mitochondrial membrane, promoting or preventing cytochrome c release. It is well known that depletion of growth factors is a common cause of apoptosis in mammalian cells. During the last years, the role of insulin-like growth factor-I as a survival factor has been extensively studied.15–20 Moreover, insulin protects from apoptosis in its target cells such as brown adipocytes by activating the PI 3-kinase/Akt and the Ras/MAPK signaling pathways,15 with IRS-1 playing a predominant role.20 These pathways lead to such acute effects as phosphorylation of downstream proteins involved in the apoptotic process including Bad, caspase-9, the Foxo family of transcription factors (previously known as FKHR), NF-κB, and CREB.20–24 However, little is known about the contribution of the distinct members of the IRS family in mediating the antiapoptotic effect of insulin in the different tissues that regulate glucose homeostasis.
In this study, we have investigated whether IRS-2, a key molecule for the metabolic actions of insulin in the liver, is essential for the signals that induce cell death as well as survival in response to insulin. As a model system, we have used immortalized neonatal hepatocyte cell lines recently generated in our laboratory.11 These include wild-type and IRS-2–deficient cells. Interestingly, we have found that insulin fails to rescue IRS-2–deficient neonatal hepatocytes from apoptosis induced by serum withdrawal because of multiple defects in the signaling pathways downstream from PIP3/Akt.
Anti-Tyr(P) (clone 4G10) (05-321), anti–IRS-1 (06-248) and anti-MnSOD (Ref. 06–984) antibodies were from Upstate Biotechnology (Lake Placid, NY). Anti-phospho Akt (Ser473 #9271), anti-Akt (#9272), anti-phospho MAPK (Thr202/Tyr204 #9101), anti-MAPK (#9102), anti-phospho Foxo1 (Ser256 #9461), anti-Foxo1 (#9462), anti-phospho Bad (Ser136 #9295), and anti-active caspase-3 (#9661) antibodies were purchased from Cell Signaling Technology (Beverly, MA). Anti-Bim (Ref. 556499), anti–Bcl-X (Ref. 559685) and the agonist anti-Fas Jo2 (Ref. 15401D) antibodies were from BD Biosciences PharMingen (San Diego, CA). Anti-Bid antibody (Ref. AF860) was from R&D Systems (Minneapolis, MN). Anti-p27Kip antibody (Ab-1, Ref. 256) was from NeoMarkers (Fremont, CA). Anti–PIP3-FIIC antibody (Ref. Z-6345) was from Echelon Research Laboratories Inc. (Salt Lake City, UT). Cy3-conjugated goat anti-rabbit antibody was from Amersham (Aylesbury, United Kingdom).
Cell Culture and Retroviral Infections.
Primary hepatocytes were obtained from livers of 3.5- to 4-day–old neonatal mice from 2 to 3 pregnant IRS-2+/− mice mated with IRS-2+/− males and were further submitted to primary culture as described.25 Immortalized neonatal hepatocyte cell lines (IRS-2+/+ and IRS-2−/−) were obtained as previously described.11
Immunofluorescence and Confocal Imaging.
Cells were grown in glass coverslips, fixed in methanol (−20°C) for 2 minutes, and processed to immunofluorescence. Anti-Foxo1 primary antibody (diluted 1:500) was applied for 1 hour at 37°C in phosphate-buffered saline (PBS) 1% bovine serum albumin. The secondary antibody (Cy3-conjugated goat anti-rabbit diluted 1:500) was applied for 30 minutes. Immunofluorescence was examined in an MRC-1024 (Bio-Rad, Hempstead, United Kingdom) confocal microscope adapted to an inverted Nikon Eclipse TE 300 microscope (Nikon Instruments, Japan). Images were taken with 514-nm laser excitation, and fluorescence emissions were detected through a 605/15-nm bandpass filter. Cells were stimulated with insulin or epidermal growth factor (EGF) and then fixed. The generation of PIP3 was measured by immunofluorescence with the anti–PIP3–FITC antibody.
Transduction of IRS-2–Deficient Hepatocytes by Adenoviral Infection.
Neonatal hepatocytes were infected with adenoviruses encoding wild-type IRS-2, constitutively active Foxo1 (ADA), or Foxo 1 Δ256 (dominant negative). Cells (80%-90% confluence) were routinely infected for 24 to 48 hours as previously described.11
Quiescent cells were treated as indicated and lysed at 4°C as previously described.26 After protein content determination, equal amounts of protein were used for immunoprecipitation at 4°C with the corresponding antibodies. Immunoprecipitates were separated by sodium dodecyl sulfate–polyacrylamide gel electrophoresis and analyzed by Western blot.
Extraction of Total Cell Lysates.
Cells from supernatants were collected by centrifugation at 2,000g for 5 minutes at 4°C. Attached cells were scraped off in ice-cold PBS, pelleted by centrifugation at 4,000g for 10 minutes at 4°C, and resuspended in lysis buffer as described.27 Samples were sonicated for 30 seconds at 1.5 mÅ, and lysates were clarified by centrifugation at 12,000g for 10 minutes.
Extraction of Nuclear Proteins.
Nuclear protein extracts were prepared as previously described.27
Isolation of Mitochondrial and Cytosolic Proteins.
Mitochondrial and cytosolic proteins were isolated as previously described.28
PI 3-Kinase Activity.
PI 3-kinase activity was measured in the anti-phosphotyrosine immunoprecipitates as previously described.26
Protein determination was performed by the Bradford dye (Bio-Rad).
To assess the fragmentation of extranuclear DNA, the method of Lyons et al.29 was used.
Analysis of Cell Viability.
At the end of the culture time, nonadherent and adherent cells were collected, washed once, and resuspended in PBS. Trypan blue (0.1% vol/vol) was added, and cells were counted immediately.
Analysis of Caspase-3 and Caspase-8 Activities.
Caspase-3 activity was measured in a luminescence spectrophotometer (Perkin Elmer LS-50, Norwalk, CT) (λ excitation, 380 nm; λ emission, 440 nm) using Ac-DEVD-AMC (Ref. 66081; BD Biosciences PharMingen) as a substrate.28 Caspase-8 activity was measured with the ApoAlert Caspase-8 fluorescent assay kit (Clontech, Ref. K2028) using IETD-AFC as a substrate (λ excitation, 380 nm; λ; emission, 505 nm). Protein concentration of cell lysates was determined, and final expression of the results is presented as arbitrary units of caspase activity/μg total protein.
Statistically significant differences between mean values were determined using paired Student t tests.
Differential Sensitivity of Wild-Type and IRS-2−/− Neonatal Hepatocytes to Undergo Apoptosis After Serum Withdrawal.
To determine whether immortalized neonatal hepatocyte cell lines derived from wild-type (IRS-2+/+) and IRS-2–deficient (IRS-2−/−) mice showed a differential sensitivity to serum withdrawal–induced apoptosis, cells were deprived of trophic factors for various periods, and DNA cleavage was analyzed. In wild-type cells, DNA fragmentation was slightly visible at 12 hours of serum withdrawal, and reached the maximal level after 24 hours. By contrast, IRS-2–deficient hepatocytes displayed maximal DNA fragmentation after 12 hours (Fig. 1 A). Moreover, IRS-2–deficient hepatocytes showed a substantial reduction of cellular viability during the time course of serum deprivation compared with the wild type (Fig. 1B). In wild-type neonatal hepatocytes, serum deprivation triggered a gradual increase in the accumulation of the large fragment (17 kd) active caspase-3 and also caspase-3 activity up to the maximal value observed after 8 hours (Supplementary Fig. 1). However, absence of trophic factors triggered an earlier increase in active caspase-3 content and caspase-3 activity in cells lacking IRS-2, reaching a maximum after 4 hours.
Insulin Fails to Rescue From Apoptosis in IRS-2–Deficient Hepatocytes: Involvement of Akt Targets.
In wild-type cells, insulin prevented caspase-3 activity in a dose-dependent manner, half maximal effect being elicited at 10 nM and maximal effect at 100 nM (Fig. 2 A). In contrast, insulin was unable to prevent caspase-3 activation in IRS-2–deficient hepatocytes. Insulin treatment for 24 hours suppressed DNA fragmentation in wild-type hepatocytes partly at 10 nM and wholly at 100 nM (Fig. 2B). However, insulin failed to rescue IRS-2−/− hepatocytes from serum withdrawal–induced DNA fragmentation. As a result, insulin increased the percentage of viability after serum withdrawal only in wild-type cells. Furthermore, insulin was unable to rescue IRS-2−/− neonatal hepatocytes from apoptosis induced with the anti-Fas antibody Jo2, which mimics the effect of Fas ligand (FasL) (Fig. 2C).
Studies of the signaling pathways involved in the antiapoptotic effect induced by insulin indicate that both PI 3-kinase/Akt and Ras/MAPK cascades participate in the control of this process. Accordingly, we analyzed these signaling pathways in our cell lines. First, we measured whether IRS-1–associated Grb-2, which leads to MAPK activation, was able to compensate the lack of IRS-2 in deficient cells. As shown in Fig. 3 A, insulin induced a substantial increase in IRS-1/Grb-2 association in both cell lines. As a result, no differences in the phosphorylation of MAPK were observed, indicating that IRS-1 signaling compensates for the lack of IRS-2 in activating MAPK in neonatal hepatocytes.
Activation of PI 3-kinase leads to the recruitment of Akt to the membrane and promotes its phosphorylation by other membrane-associated phospholipid-dependent kinases.30 We have recently demonstrated that IRS-2 deficiency in neonatal hepatocytes results in a strong inhibition of Akt phosphorylation on insulin stimulation.11 Downstream from Akt, we analyzed the phosphorylation of Bad and nuclear/cytosolic translocation of Foxo1. Survival factors acting through Akt signaling inhibit the apoptotic activity of Bad by phosphorylation at serine 136.31 To test whether this phosphorylation depends on IRS-2 signaling in neonatal hepatocytes, we performed anti-phospho-BadSer136 Western blot analysis in cytosolic extracts of wild-type and IRS-2−/− cells stimulated with insulin for 10 minutes. As shown in Fig. 3B, insulin induced the phosphorylation of Bad in wild-type cells, but this response was lost in the absence of IRS-2. Conversely, transcription factors of the Foxo family have been recently implicated in the regulation of several proapoptotic genes.32–34 Foxo transcription factors translocate from the nucleus to the cytosol after Akt-mediated phosphorylation by growth factors, including insulin.35 We have recently shown that the lack of IRS-2/Akt signaling results in the loss of Foxo1Ser256 phosphorylation in response to insulin in neonatal hepatocytes.11 To further assess whether IRS-2 deficiency modifies Foxo1 subcellular distribution, we performed Western blot analysis in nuclear and cytosolic extracts of insulin-stimulated cells. Insulin induced a substantial increase in the amount of cytosolic Foxo1 in parallel to a decrease of Foxo1 in the nuclear compartment in wild-type cells (Fig. 3C, left panel). However, insulin failed to increase cytosolic and to decrease nuclear Foxo1 in IRS-2−/− hepatocytes. These results were confirmed by immunofluorescence analysis (Fig. 3C, right panel).
Differential Expression of Pro- (Foxo1, Bim) and Anti- (Bcl-xL) Apoptotic Genes in Wild-Type and IRS-2−/− Hepatocytes.
Our next aim was to elucidate the differential gene expression regarding apoptosis in wild-type and IRS-2–deficient hepatocytes. Thus, we studied the expression of pro- and antiapoptotic genes of the Bcl-2 family, as well as nuclear Foxo1 protein content, in the presence or in the absence of 10% serum in a time-dependent manner (2-24 hours). In addition, insulin was added for 24 hours as indicated. The expression of the antiapoptotic protein Bcl-xL was high in wild-type cells under growing conditions, and low levels of this protein were detected already after 24 hours of serum deprivation (Fig. 4). By contrast, IRS-2 deficiency accelerated the loss of Bcl-xL expression, low levels of this protein being detected after 8 hours of serum deprivation. Interestingly, insulin treatment for 24 hours counteracted starvation-induced Bcl-xL downregulation in wild-type cells, but not in IRS-2−/− cells. Regarding proapoptotic genes, Bim expression was almost undetectable in both cell types under growing conditions. However, on serum deprivation, Bim was detected in control cells after 8 hours and in IRS-2−/− hepatocytes after only 2 hours, reaching the maximal expression at 15 hours in both cell types. Insulin treatment downregulated Bim expression in wild-type cells but not in cells lacking IRS-2. Similarly, nuclear Foxo1 was visible after 8 hours in wild-type cells or after 4 hours of serum withdrawal in IRS-2−/−cells, demonstrating that insulin was unable to facilitate export of nuclear Foxo1 in IRS-2−/− cells.
Constitutively Active Foxo1 Induces Apoptosis in Wild-type Hepatocytes.
To assess whether accumulation of nuclear Foxo1 might induce apoptosis in neonatal hepatocytes, wild-type cells were infected with an adenovirus encoding a constitutively active Foxo1 mutant (Foxo1 ADA), in which the Akt phosphorylation sites (Thr24Ala, Ser256Asp, Ser316Ala) have been mutated.36 As a control, parallel dishes were infected with an adenovirus encoding truncated Foxo1 mutant (Δ256) that lacks the transactivation domain and competes with endogenous Foxo1. We monitored adenoviral infection by measuring FasL overexpression, a direct target of Foxo1. As shown in Fig. 5 A, constitutively active Foxo1 increased FasL expression by 2-fold. Next, we analyzed the effect of accumulation of nuclear Foxo1 in the expression of several members of the Bcl-2 family. Constitutively active Foxo1 increased Bim expression and induced Bid cleavage with the concomitant presence of the 15-kd Bid truncated fragment (tBid) (Fig. 5B). ADA overexpression also resulted in a marked decrease in the antiapoptotic protein Bcl-xL compared with the controls. Moreover, we found that constitutively active Foxo1 upregulates MnSOD (antioxidant function) and p27kip (cell-cycle arrest). Regarding caspase activation, overexpression of constitutively active Foxo1 increased both caspase-8 and caspase-3 activities (Fig. 5C). Finally, DNA fragmentation was visualized in Foxo1 ADA–transduced cells. All of the effects described above were not observed in the cells expressing dominant negative Foxo1 (Fig. 5D).
IRS-2–Deficient Neonatal Hepatocytes Reconstituted With IRS-2 Restore the Antiapoptotic Effect of Insulin.
To confirm the role of IRS-2 in the antiapoptotic effect of insulin in neonatal hepatocytes, IRS-2−/− cells were infected with adenoviruses expressing wild-type IRS-2 (adeno IRS-2) or β-gal (mock), as previously described.11 Expression of IRS-2 in deficient cells downregulated nuclear Foxo1 and Bim, as well as upregulated Bcl-xL expression in response to insulin (Supplementary Fig. 2A). Alternatively, IRS-2−/− hepatocytes were infected and caspase-3 activity was determined after 8 hours of serum deprivation either in the absence or in the presence of insulin. Whereas insulin reduced caspase-3 activity in cells infected with adenoIRS-2, this response was prevented in β-gal–infected cells (Supplementary Fig. 2B). Finally, we studied whether the reconstitution with IRS-2 in deficient hepatocytes might alter the pattern of DNA fragmentation on insulin treatment. Supplementary Fig. 2C shows a representative gel of extranuclear DNA obtained from IRS-2−/− hepatocytes infected with IRS-2 or mock adenoviruses and further deprived of serum for 24 hours, in the absence or presence of insulin. IRS-2–reconstituted cells completely restored the inhibition of DNA laddering by insulin. Together, these results indicate that IRS-2 mediates the antiapoptotic effect of insulin in neonatal hepatocytes.
EGF Rescues From Apoptosis Through PI 3-Kinase/PIP3/Akt/Foxo1 Signaling in IRS-2–Deficient Neonatal Hepatocytes.
In addition to insulin, PI 3-kinase/Akt signaling is activated by a variety of growth factors, including EGF. Therefore, we investigated whether EGF signals through PI 3-kinase/Akt in immortalized neonatal hepatocytes, particularly in IRS-2–deficient cells. First, we analyzed the tyrosine phosphorylation of the EGF receptor (EGF-R) in neonatal hepatocytes lacking IRS-2. As shown in Fig. 6 A, EGF, but not insulin, induced a marked increase in the tyrosine phosphorylation of the EGF-R in IRS-2−/− cells. As observed previously, insulin, but not EGF, induced the tyrosine phosphorylation of the IR in both cell types. These data clearly indicate the absence of cross-talk between both receptors in immortalized neonatal hepatocytes. In addition, both EGF and insulin induced PI 3-kinase activity, enhanced PIP3 content, and phosphorylated Akt and Foxo1 in wild-type cells (Fig. 6B-C). However, insulin induced PI 3-kinase activity by 50% but failed to increase PIP3 content in IRS-2–deficient hepatocytes, as previously shown.11 Furthermore, in these cells, the phosphorylation of Akt and Foxo1 in response to insulin was almost impaired. However, EGF induced PI 3-kinase activity, enhanced PIP3 content, and phosphorylated Akt and Foxo1 in IRS-2–deficient hepatocytes (Fig. 6B-C).
Next, we investigated whether EGF could protect IRS-2–deficient cells from serum withdrawal-induced apoptosis. IRS-2−/− hepatocytes were cultured for 24 hours in serum-free medium in the presence of EGF or insulin (as a control), and the expression of pro- and antiapoptotic genes was analyzed. EGF downregulated nuclear Foxo1 and Bim and upregulated Bcl-xL expression (Fig. 7 A). These effects were impaired in response to insulin. Moreover, EGF, but not insulin, prevented the activation of caspase-3 (Fig. 7B) and completely suppressed DNA laddering evoked by serum-free conditions (Fig. 7C). Thus, these results show that EGF rescues neonatal hepatocytes from serum withdrawal–induced apoptosis in an IRS-2–independent manner.
Hepatoprotection remains one of the major challenges of clinical therapies aimed at limiting the damage of liver injuries such as chronic hepatitis and cholestasis. It is now well established that most cells in higher animals may require continuous tropic stimulation to survive37; otherwise, apoptosis occurs after serum/growth factor deprivation.38, 39 Insulin and IGF-I elicit survival effect in a number of cell types.15–20 Multiple signaling pathways have been implicated in the antiapoptotic functions of insulin and IGF-I; however, it is still unclear exactly which signaling molecules mediate this response.
The IRSs are docking proteins that connect the insulin receptor with the SH2 domain-containing proteins that initiate the signaling pathways responsible for the biological effects of the hormone. Among them, IRS-2 plays an essential role in mediating the metabolic actions of insulin in the liver.7–10 Thus, the generation of immortalized hepatocyte cell lines derived from individual livers of wild-type and IRS-2 knockout mice has provided an excellent tool for the in vitro study of insulin signaling pathways, which led to the expression of genes involved in the regulation of glucose metabolism in the liver.11 In the current study, we used these cell lines to evaluate the role of IRS-2 and its signaling in mediating the survival effects of insulin in neonatal hepatocytes. Our results demonstrate that IRS-2–deficient hepatocytes have increased susceptibility to undergo apoptosis under conditions of serum deprivation, as compared with wild-type cells. Interestingly, the lack of IRS-2 triggers substantial changes in pro- and antiapoptotic gene expression in response to serum deprivation. Regarding proapoptotic genes, activation of Foxo transcription factors is associated with increased cell death. In the unphosphorylated state, these proteins predominantly localize in the nucleus, where they bind to promoters of various proapoptotic target genes. These include FasL,34 Bim,32, 40 TRAIL,41 TRADD,42 and BCL-6, a transcriptional repressor of Bcl-xL.43 Our results indicate that as early as 4 hours after serum deprivation, a substantial amount of nuclear Foxo1 appears in IRS-2−/− cells. However, nuclear localization of Foxo1 is detected after 8 hours of serum withdrawal in wild-type cells. Bim is a well-known target of Foxo transcription factors that exerts its proapoptotic activity through heterodimerization with antiapoptotic Bcl-2 members. Ectopic expression of Bim is sufficient to induce apoptosis in Ba/F3 cells.32 Moreover, Bim−/− lymphocytes and neurones have an increased resistance to cell death induced by cytokine withdrawal.44 The results here presented demonstrate that neonatal hepatocytes lacking IRS-2 show upregulation of Bim expression at earlier times of serum deprivation than wild-type cells, in agreement with nuclear Foxo1 content. In addition, a more rapid downregulation of antiapoptotic gene expression (Bcl-xL) occurs in serum-deprived IRS-2−/− cells compared with the wild-type. As a result, IRS-2−/− cells show massive DNA fragmentation after 12 hours of serum withdrawal (Fig. 1). Therefore, these results together with those previously reported10, 11 indicate that the deletion of IRS-2 has deleterious effects on hepatic carbohydrate and lipid metabolism, as well as on hepatocyte survival.
Insulin elicits a survival effect in wild-type neonatal hepatocytes by preventing caspase-3 activation, proapoptotic gene expression and DNA laddering. Although chronic insulin treatment has been found to induce apoptosis,45, 46 insulin is able to protect immortalized hepatocytes from cell death up to 72 hours (results not shown). However, insulin is unable to protect neonatal hepatocytes from serum withdrawal-induced apoptosis in the absence of IRS-2. Whereas phosphorylation of MAPK is unchanged in IRS-2−/− hepatocytes, phosphorylation of Akt11 and, subsequently, Bad and Foxo1 is severely impaired. Phosphorylation is an essential requirement to eliminate the proapoptotic effects of both molecules. Once phosphorylated, Bad is sequestrated in the cytosol by 14-3-3 proteins and canno longer heterodimerize with the antiapoptotic member Bcl-xL.31 The fact that IRS-2–deficient hepatocytes did not respond to insulin in regulating the phosphorylation of Bad indicates that IRS-2/PI 3-kinase/Akt is essential for this antiapoptotic effect. Indeed, IRS-1/Grb-2/MAPK signaling is unable to compensate the lack of this response. Foxo1 is phosphorylated in an insulin-responsive manner by Akt and then excluded from the nucleus,22, 35 protecting cells from apoptosis. The results presented here demonstrate that in the absence of IRS-2, insulin stimulation failed to exclude Foxo1 from the nucleus toward the cytosol. The critical role of Foxo transcription factors in cell death has been shown by overexpression of Foxo proteins in various cell types.32, 33 In neonatal hepatocytes, we have shown a direct role for Foxo1 in controlling survival by performing adenoviral infections with a mutant of Foxo1 that cannot be phosphorylated (ADA) and, therefore, it is localized exclusively into the nucleus. Increased expression of proapoptotic proteins such as Bim and FasL was observed in ADA-infected cells. Likewise, the detection of the truncated Bid (tBid) indicated caspase-8 activation in these cells. Conversely, the antiapoptotic protein Bcl-xL was significantly decreased. As a result, caspase-3 activation and DNA fragmentation occurred exclusively in ADA-infected cells. Thus, the ability of insulin to regulate Foxo1 activity in neonatal hepatocytes constitutes one of the mechanisms by which the hormone elicits the survival effect on the liver. On the other hand, two recent papers47, 48 show that in mammalian cells Foxo proteins confer resistance to oxidative stress. In fact, MnSOD is a Foxo1 target gene also induced by the active form of Foxo1 (ADA) in wild-type hepatocytes. However, the overall effect of ADA overexpression in hepatocytes is proapoptotic. These results together with those that indicate the inability of insulin to exclude Foxo1 from the nucleus in IRS-2–deficient hepatocytes outline the essential role played by the IRS-2/Akt/Foxo1 signaling in the antiapoptotic effect of insulin in neonatal hepatocytes. These results are also reinforced by the reconstitution experiments with adenoviral IRS-2 shown in Supplementary Fig. 2. Thus, in addition to its role in hepatic carbohydrate metabolism, IRS-2 is an essential mediator for protecting hepatocytes from death.
Signals from other growth factors such as EGF have been proposed to protect against apoptosis in different liver cell systems. Chen et al.49 found that the suppression of the apoptotic effect of transforming growth factor-beta in human hepatoma cells was dependent on the PI 3-kinase/Akt pathway. Recently, Roberts et al.50 showed that EGF-mediated survival effect requires both PI 3-kinase and MAPK pathways in adult rat hepatocytes. In fetal hepatocytes, EGF prevents TGF-β–induced caspase-3 activity, cytochrome c release, and downregulation of Bcl-xL in a PI 3-kinase–dependent manner.28 Interestingly, we noted that despite the absence of IRS-2 signaling in immortalized neonatal hepatocytes, EGF stimulated a substantial activation of PI 3-kinase. More importantly, EGF, but not insulin, increased PIP3 content in IRS-2–deficient hepatocytes. This increase in PIP3 leads to the phosphorylation of Akt/Foxo1 on EGF treatment in an IRS-2–independent manner. These results prompted us to consider that EGF might be an alternative survival factor to counteract the apoptotic effect of IRS-2 deficiency in neonatal hepatocytes. Several lines of evidence from the current study support this idea. First, expression of nuclear Foxo1 and Bim (proapoptotic) were markedly decreased after treatment of IRS-2−/− hepatocytes with EGF, but not with insulin. Second, the antiapoptotic protein Bcl-xL is upregulated in IRS-2−/− cells in the presence of EGF. Finally, caspase-3 activation and DNA fragmentation are attenuated by EGF. Thus, EGF promotes cell survival through the activation of PI-3 kinase and the enhancement of PIP3 in an IRS-2–independent manner.
In conclusion, lack of IRS-2 in neonatal hepatocytes accelerates cell death on serum withdrawal. This effect is reflected on the earlier induction of proapoptotic and repression of antiapoptotic genes, leading to a rapid activation of caspase-3. However, constitutively active Foxo1 induces apoptosis in the presence of serum in wild-type cells. The survival effect of insulin is prevented by the absence of IRS-2/PIP3/Akt signaling but recovered by reconstitution with IRS-2. Alternatively, EGF rescues from apoptosis in an IRS-2–independent manner. Thus, our results demonstrate the unique role of IRS-2 in maintaining the balance of cell death and survival in the liver.
The authors thank Domenico Accili (Berrie Research Pavilion, NY) for supplying the adenoviruses encoding Foxo1 ADA and Foxo1 Δ256. We acknowledge Alberto Álvarez (Centro Nacional de Investigaciones Cardiovasculares, Spain) for his expert assistance with the confocal microscopy. We also recognize the technical skill of M. López.