Insulin Antiapoptotic Signaling Involves Insulin Activation of the Nuclear Factor κB-dependent Survival Genes Encoding Tumor Necrosis Factor Receptor-associated Factor 2 and Manganese-superoxide Dismutase*

We recently showed that the antiapoptotic function of insulin requires nuclear factor κB (NF-κB) activation (Bertrand, F., Atfi, A., Cadoret, A., L'Allemain, G., Robin, H., Lascols, O., Capeau, J., and Cherqui, G. (1998) J. Biol. Chem.273, 2931–2938). Here we sought to identify the NF-κB-dependent survival genes that are activated by insulin to mediate this function. Insulin increased the expression of tumor necrosis factor receptor-associated factor 2 (TRAF2) mRNA and protein in Chinese hamster ovary cells overexpressing insulin receptors (IRs). This effect required (i) IR activation since it was abrogated by IR mutation at tyrosines 1162 and 1163 and (ii) NF-κB activation since it was abolished by overexpression of dominant-negative IκB-α(A32/36) and mimicked by overexpression of the NF-κB c-Rel subunit. TRAF2 contributed to insulin protection against serum withdrawal-induced apoptosis since TRAF2 overexpression mimicked insulin protection, whereas overexpression of dominant-negative TRAF2-(87–501) reduced this process. Along with its protective effect, overexpressed TRAF2 increased basal and insulin-stimulated NF-κB activities. All effects were inhibited by IκB-α(A32/36), suggesting that an amplification loop involving TRAF2 activation of NF-κB is implicated in insulin antiapoptotic signaling. We also show that insulin increased manganese-superoxide dismutase (Mn-SOD) mRNA expression through NF-κB activation and that Mn-SOD contributed to insulin antiapoptotic signaling since expression of antisense Mn-SOD RNA decreased this process. This study provides the first evidence that insulin activates the NF-κB-dependent survival genes encoding TRAF2 and Mn-SOD and thereby clarifies the role of NF-κB in the antiapoptotic function of insulin.

Programmed cell death or apoptosis is a fundamental event in the developmental and homeostatic processes of all multicellular organisms. Apoptosis is inhibited by various growth factors, including insulin. The antiapoptotic function of insulin requires the integrity of the insulin receptor (IR) 1 tyrosine kinase domain (1,2) and involves several intracellular signaling molecules, including insulin receptor substrate 1 (3), Ras (2), Raf-1 (1), and phosphatidylinositol 3-kinase (1,4). In addition, recent evidence from our laboratory (1) indicates that insulin antiapoptotic signaling involves the activation of nuclear factor B (NF-B), a transcription factor playing a critical role in apoptosis inhibition (5).
NF-B is a member of the NF-B/Rel family. This family is composed of NF-B1 (p50), NF-B2 (p52), RelA (p65), c-Rel, and RelB. Prototypical NF-B is a p50/p65 heterodimer that is usually retained in the cytoplasm of unstimulated cells in an inactive form by IB-␣, one of the most important members of the IB inhibitory protein family. In response to various stimuli, IB-␣ is phosphorylated at Ser 32 and Ser 36 by IB kinases ␣ and ␤ (6). Once phosphorylated, IB-␣ is rapidly ubiquitinated and subsequently degraded by the 26 S proteasome complex (7). The degradation of IB-␣ unmasks the nuclear localization signal of the NF-B heterodimer, which then translocates into the nucleus, where it directly binds to its cognate DNA sequence to regulate gene transcription.
The antiapoptotic role played by NF-B involves the ability of this transcription factor to induce the expression of genes that promote cell survival such as the genes coding for tumor necrosis factor (TNF) receptor-associated factors 1 and 2 (TRAF1 and TRAF2, respectively) (8), the cellular inhibitors of apoptosis 1 and 2 (c-IAP1 and c-IAP2, respectively) (8,9), the protein A20 (10), and manganese-superoxide dismutase (Mn-SOD) (11,12). TRAF1 and TRAF2 are cytoplasmic adapter proteins that interact directly or indirectly with the intracellular domains of members of the TNF receptor superfamily (13). Both TRAF1 and TRAF2 are able to recruit other adapter proteins such as c-IAP1 and c-IAP2 (14,15) and the protein A20 (13). Of interest, TRAF2, the most extensively studied TRAF family member to date, was recently shown to be also localized in the nucleus (16). TRAF2 is known to activate NF-B through its interaction with NIK, a serine/threonine-specific NF-Binducible kinase that activates the IB kinase complex, thus leading to IB phosphorylation and degradation and subsequent NF-B activation (17). TRAF2 has a clear antiapoptotic function, as supported by the findings that thymocytes from TRAF2 null mice (18) or from transgenic mice expressing a dominant-negative form of TRAF2 (19) are highly sensitive to TNF-␣-induced cell death. In addition, the overexpression of the dominant-negative TRAF2-(87-501) mutant lacking the RING finger domain increases TNF-␣-induced apoptosis in MCF-7 and HeLa cells (20). The antiapoptotic function of TRAF2 is mediated by NF-B-dependent (13,20) and NF-Bindependent (18,19,21) pathways and may involve the ability of TRAF2 to interact with c-IAP1 and c-IAP2 (13). As regards Mn-SOD, it is known as an enzyme of the mitochondrial matrix functioning as a scavenger of superoxide radicals. Like TRAF2, Mn-SOD is involved in apoptosis inhibition since its overexpression suppresses TNF-␣-induced apoptosis in MCF-7 cells (22).
This study was designed to investigate whether the NF-Bdependent survival genes coding for TRAF2, c-IAP1, c-IAP2, and Mn-SOD are activated by insulin and, if so, to analyze their role in the antiapoptotic function of insulin. We used CHO cells overexpressing wild-type IRs (CHO-R and CHO-R15) or IRs made kinase-defective by mutation at Tyr 1162 and Tyr 1163 autophosphorylation sites (CHO-Y2). We show that insulin increased the expression of TRAF2 mRNA and protein at the cytoplasmic and nuclear levels in CHO-R15 and CHO-R cells, but not in CHO-Y2 cells. Insulin induced TRAF2 overexpression through NF-B activation, and overexpressed TRAF2 contributed to insulin antiapoptotic signaling by exerting positive feedback control on NF-B. We also show that insulin increased Mn-SOD (but not c-IAP1 or c-IAP2) mRNA expression in CHO-R15 cells and that Mn-SOD, like TRAF2, contributed to insulin antiapoptotic signaling. These results clarify the role of NF-B in the antiapoptotic function of insulin and provide new insight into the nuclear mechanisms involved in this function.
Northern Blot Analysis-Total RNA and poly(A) ϩ RNA were isolated from CHO-R and CHO-R15 cells using kits from QIAGEN Inc. according to the manufacturer's instructions. Total RNA (20 g) or poly(A) ϩ RNA (2.5 g) was electrophoresed and transferred to Hybond-N ϩ nylon membranes (Amersham Pharmacia Biotech) as described (24). Membranes were hybridized (3 h at 65°C) in Rapid-Hyb buffer (Amersham Pharmacia Biotech) to a random-primed labeled cDNA probe specific for TRAF2, c-IAP1, c-IAP2, or Mn-SOD and subsequently to a labeled human elongation factor 1 probe as a control for RNA loading. Membranes were then washed to high stringency as described (24). Signals were quantified and normalized by differential densitometric scanning of the specific bands versus the human elongation factor 1 band.
Apoptosis Assays-The analysis of DNA laddering in stable transfectants was performed as described (1). For evaluating apoptosis in transient transfectants, CHO-R15 cells were cotransfected with the pCMV5/ lacZ reporter plasmid and the expression vector coding for TRAF2, TRAF2-(87-501), IB-␣(A32/36), or antisense Mn-SOD RNA. The subsequent steps were performed as described in the legend to Fig. 3.
Luciferase Assays-The lysates from transfected cells were prepared and assayed for luciferase activity as described (1,25). Results were normalized as described in the legend to Table I.
Statistical Analysis-Results are given as the means Ϯ S.E. for the indicated numbers of independently performed experiments. Differences between the mean values were evaluated by Student's t test for unpaired data.

Insulin Stimulates TRAF2 mRNA and Protein Expression in CHO Cells
Overexpressing IRs-To examine the effect of insulin on the expression of TRAF2 mRNA, total RNA, or poly(A) ϩ RNA prepared from CHO-R and CHO-R15 cells, two different clones of CHO cells overexpressing wild-type human IRs (1,23) were analyzed by Northern blotting. Insulin stimulated the expression of the 2.2-kilobase TRAF2 transcript at 6 and 24 h in total RNA from CHO-R and CHO-R15 cells (Fig. 1A). Quantification of the results by scanning densitometry of the autoradiograms after normalization to the human EF1 signal indicated that insulin increased TRAF2 mRNA expression by ϳ1.5and 2.2-fold at 6 and 24 h, respectively (Fig. 1A). Higher increases (3-and 6-fold at 6 and 24 h, respectively) were measured on Northern blots performed with poly(A) ϩ RNA from CHO-R15 cells (Fig. 1B). Accordingly, insulin increased the expression of the 50-kDa TRAF2 immunoreactive protein in cytosolic and nuclear fractions from CHO-R15 cells (Fig. 1C). These findings provide the first evidence that insulin stimulates the expression of TRAF2 mRNA and protein at the cytoplasmic and nuclear levels in CHO cells overexpressing IRs.
Insulin Stimulation of TRAF2 Expression Is Mediated by IRs and Involves NF-B Activation-We then evaluated the specificity of the stimulation of TRAF2 expression by insulin and investigated the role of NF-B in this effect. As shown in Fig.  2A, the insulin-induced increase in TRAF2 protein expression observed in total cell lysates from CHO-R15 cells was lost in CHO-Y2 cells overexpressing kinase-defective IRs mutated at tyrosines 1162 and 1163. As shown in Fig. 2B, the stable expression of the dominant-negative IB-␣(A32/36) mutant in CHO-R15 cells (IB-␣(A32/36) cells), which almost completely abolished basal and insulin-stimulated NF-B-mediated luciferase activities (Table I), dramatically decreased insulin stimulation of TRAF2 mRNA expression as compared with CHO-R15 cells stably transfected with an empty vector and a pcDNA3.1/zeo selection vector (pCMV-zeo cells). Reciprocally, the overexpression of the c-Rel subunit of NF-B in CHO-R cells (Rel-3 cells) (Fig. 2B), which stimulated NF-B-mediated activity by 2.2-fold (Table I), increased TRAF2 mRNA expression by 2.3-fold (Fig. 2B). Together, these results indicate that insulin stimulates TRAF-2 expression in CHO-R and CHO-R15 cells through IR activation and through an NF-B-dependent pathway.
TRAF2 Is Involved in Insulin Antiapoptotic Signaling-To examine the role of TRAF2 in insulin antiapoptotic signaling, CHO-R15 cells were transiently transfected with the expression vector coding for TRAF2 or its dominant-negative TRAF2-(87-501) mutant lacking the RING finger domain or with an empty vector, together with a ␤-galactosidase reporter plasmid (pCMV5/lacZ). Transfected cells were then incubated for 24 h in serum-free medium (SFM) with or without 10 Ϫ7 M insulin. After fixation and staining, the percentage of apoptotic cells was evaluated by scoring blue ␤-galactosidase transfectants as healthy or apoptotic, as judged by blebbing of the membranes and shrinkage of the cell bodies. As shown in Fig. 3A, the percentage of apoptotic cells that amounted to 48 Ϯ 2% in control cells maintained in SFM in the absence of insulin fell to 15 Ϯ 1% in the presence of insulin. In TRAF2-transfected cells maintained for 24 h in SFM in the absence and presence of 10 Ϫ7 M insulin, the percentages of apoptotic cells were 24 Ϯ 2 and 16 Ϯ 1%, respectively (Fig. 3A). In contrast, the percentages of apoptotic cells determined in TRAF2-(87-501)-transfected cells rose to 48 Ϯ 3 and 32 Ϯ 3% in the absence and presence of insulin, respectively. To strengthen the above findings supporting a role for TRAF2 in insulin antiapoptotic signaling, CHO-R15 cells were stably transfected with the TRAF2 or TRAF2-(87-501) expression vector. Two clones designated as TRAF2 and TRAF2-(87-501), which exhibited TRAF2 and TRAF2-(87-501) overexpression as compared with control pCMV-zeo cells (Fig. 3B), respectively, were maintained in SFM for 24 h with or without 10 Ϫ7 M insulin and then analyzed for apoptosis by the DNA fragmentation assay (1). In controls, insulin almost completely abolished the DNA degradation induced by a 24-h serum deprivation (Fig. 3C). In TRAF2 cells maintained in SFM in the absence of insulin, the extent of DNA degradation was reduced as compared with controls, indicating that TRAF2 overexpression partially protected CHO-R15 cells from apoptosis. Insulin further protected TRAF2 cells from this process (Fig. 3C). In contrast, TRAF2-(87-501) cells maintained in SFM in the absence of insulin exhibited a ladder of DNA fragmentation similar to that observed in controls. Most important, Fig. 3C shows that insulin protection against apoptosis was markedly decreased in these cells. Altogether, these findings indicate that the overexpression of TRAF2 partially mimicked the antiapoptotic effect of insulin, whereas the overexpression of TRAF2-(87-501) markedly reduced this effect. They strongly argue for a role of TRAF2 in insulin antiapoptotic signaling in CHO-R15 cells.  The Role of TRAF2 in Insulin Antiapoptotic Signaling Involves TRAF2 Activation of NF-B-Since the role played by TRAF2 in insulin antiapoptotic signaling in CHO-R15 cells could involve the well known ability of TRAF2 to activate NF-B (13,20), we investigated the effect of the stable overexpression of TRAF2 or TRAF2-(87-501) on basal and insulinstimulated NF-B activities. As reported in Table I, insulin stimulated basal NF-B-mediated luciferase activity by ϳ3fold in pCMV-zeo cells. As compared with controls, TRAF2 cells exhibited 3.4-and 2.3-fold increases in basal and insulin-stimulated NF-B activities, respectively. In contrast, TRAF2-(87-501) cells displayed 3.6-and 3.8-fold decreases in basal and insulin-stimulated NF-B activities, respectively, as compared with controls. Noteworthy is the finding that the stimulation of NF-B activity induced by insulin in control cells persisted in TRAF2-(87-501)-transfected cells despite the marked decrease in basal NF-B activity exhibited by these cells. This strongly suggests that TRAF2-(87-501) did not affect NF-B activation by insulin.
In view of the above results, we investigated the role of NF-B in the ability of TRAF2 to mimic the antiapoptotic effect of insulin. To this end, pCMV-zeo and TRAF2 cells were trans-fected with the IB-␣(A32/36) expression plasmid or an empty vector in the presence of either the (Ig)3-conaluc or pCMV5/ lacZ reporter plasmid. Transfected cells were treated with or without insulin (24 h, 10 Ϫ7 M) and then examined for NF-Bmediated activity and the extent of apoptosis. In pCMV-zeo cells, the transient expression of IB-␣(A32/36) inhibited both insulin-stimulated NF-B activity (Table I) and insulin antiapoptotic signaling (Fig. 4). Similarly, we found that IB-␣(A32/36) expression in TRAF2 cells reduced both TRAF2 stimulation of NF-B-mediated luciferase activity (Table I) and TRAF2 antiapoptotic function (Fig. 4). Altogether, the above results are consistent with the notion that TRAF2 is not involved in insulin activation of NF-B, but contributes to insulin antiapoptotic signaling, at least in part, through its ability to initiate an amplification loop involving NF-B activation.
Insulin Stimulates the Expression of Mn-SOD mRNA, but Not of c-IAP1 and c-IAP2 mRNAs-To determine whether insulin activates other antiapoptotic genes known as targets of NF-B in mammalian cells, we examined the effect of insulin on the expression of c-IAP1, c-IAP2, and Mn-SOD mRNAs in CHO-R15 cells. To this end, total RNA or poly(A) ϩ RNA prepared from cells treated with or without 10 Ϫ7 M insulin for 6 or 24 h was analyzed by Northern blotting. As shown in Fig. 5A, the 6-kilobase c-IAP2 transcript was undetectable in poly(A) ϩ RNA from CHO-R15 cells, whereas it was well detected in IAP2 cells, a clone of CHO-R15 cells stably transfected with a pCMV4-c-IAP2 expression plasmid. Insulin had no effect on c-IAP2 mRNA expression in CHO-R15 cells at either time tested (Fig. 5A). Similar results were obtained when using a c-IAP1-specific probe (data not shown). In contrast, insulin increased the expression of the 1-and 4-kilobase transcripts of Mn-SOD at 6 and 24 h, as determined by Northern blot analysis of total RNA from CHO-R15 cells (Fig. 5B). Quantification of the results showed that insulin stimulated the expression of the major 1-kilobase transcript by 1.4-and 1.7-fold at 6 and 24 h, respectively. In addition, Fig. 5B shows that the stimulation of Mn-SOD mRNA expression induced by insulin was lost in IB-␣(A32/36) cells, indicating that insulin increased Mn-SOD expression through NF-B activation.
Role of Mn-SOD in Insulin Antiapoptotic Signaling-To in- vestigate whether Mn-SOD participates in insulin antiapoptotic signaling, CHO-R15 cells were transiently transfected with an empty vector or an antisense Mn-SOD plasmid, together with pCMV5/lacZ. After a 24-h incubation in SFM with or without 10 Ϫ7 M insulin, cells were examined for the extent of apoptosis. In the absence of insulin, the percentage of apoptotic cells determined in CHO-R15 cells transfected with the antisense Mn-SOD plasmid was slightly but not significantly higher than that determined in control cells (Fig. 5C). In contrast, in the presence of insulin, the percentage of apoptotic cells in cells transfected with the antisense Mn-SOD plasmid was significantly increased as compared with controls (Fig.  5C). These results are consistent with the notion that Mn-SOD, like TRAF2, contributes to the antiapoptotic function of insulin in CHO-R15 cells. DISCUSSION We (25) and others (26) previously reported that insulin activates NF-B in mammalian cells. More recently, we provided evidence for a role of NF-B in the antiapoptotic function of insulin (1). To clarify the role of NF-B in this function, we now investigated whether the NF-B-dependent genes coding for the antiapoptotic proteins TRAF2, c-IAP1, c-IAP2, and Mn-SOD were activated by insulin and, if so, whether these proteins contributed to insulin antiapoptotic signaling.
Insulin caused a time-dependent increase in TRAF2 mRNA and/or protein expression in CHO-R and CHO-R15 cells. In these cells, the TRAF2 immunoreactive protein was localized in the cytoplasm and the nucleus. Accordingly, Min et al. (16) observed a cytoplasmic and nuclear localization of endogenous and transfected TRAF2 in human endothelial cells and proposed a role for nuclear TRAF2 in the regulation of gene transcription. In CHO-R15 cells, insulin increased the amount of TRAF2 protein in both the cytosolic and nuclear fractions. These findings provide the first evidence that insulin stimulates the expression of an adapter protein recruited by activated TNF receptors and thus reveal a novel aspect of the interplay between insulin and TNF-␣ signaling (27).
Insulin failed to increase TRAF2 expression in CHO-Y2 cells overexpressing tyrosine kinase-deficient IRs mutated at Tyr 1162 and Tyr 1163 autophosphorylation sites. This finding has three important implications. First, it excludes the possibility that the stimulation of TRAF2 expression elicited by 10 Ϫ7 M insulin could be mediated by insulin-like growth factor 1 receptors in CHO-R and CHO-R15 cells. Second, it indicates that insulin induction of TRAF2 requires the presence of the IR Tyr 1162 and Tyr 1163 residues, similar to what was found for insulin activation of NF-B (25) and insulin antiapoptotic signaling (1). Third, the inability of insulin to induce TRAF2 expression in CHO-Y2 cells that had lost their response to insulin for the activation of NF-B (25) suggests a role for NF-B in insulin induction of TRAF2. This hypothesis is further supported by the two following findings. (i) Insulin failed to increase TRAF2 mRNA in IB-␣(A32/36) cells, a clone of CHO-R15 cells expressing a dominant-negative form of IB-␣ and exhibiting a dramatic decrease in basal and insulin-stimulated NF-B activities; and (ii) reciprocally, the insulin-induced increase in TRAF2 mRNA found in CHO-R and CHO-R15 cells was mimicked in Rel-3 cells, a clone of CHO-R cells exhibiting a constitutive activation of NF-B activity due to c-Rel overexpression (Ref. 1 and this study). Altogether, these results indicate that insulin increases TRAF2 expression through a pathway dependent on the IR tyrosine kinase activity and through a transcriptional mechanism involving NF-B activation. However, the possibility that insulin may also increase TRAF2 expression through a post-transcriptional mechanism as has been shown for other insulin-responsive genes (28) cannot be excluded. Of interest, our results extend the recent findings of Wang et al. (8), who were the first to report a role for NF-B in TNF-␣ induction of TRAF2 in the HT1080V fibrosarcoma cell line. In this cell line, TNF-␣ also increased the expression of other NF-B-regulated genes such as the genes coding for c-IAP1 and c-IAP2 (8). In the present study, the c-IAP1 and c-IAP2 mRNAs could not be detected in CHO-R15 cells in the absence or presence of insulin.
Several lines of evidence indicate that TRAF2 has an antiapoptotic function. Thymocytes from TRAF2 null mice (18) or from transgenic mice expressing a dominant-negative form of TRAF2 (19) were highly sensitive to TNF-␣-induced cell death. In addition, the overexpression of dominant-negative TRAF2-(87-501) in MCF-7 and HeLa cells potentiated TNF-␣-induced apoptosis (20). In the present study, TRAF2-(87-501) overexpression in CHO-R15 cells did not modify the extent of serum withdrawal-induced apoptosis, as supported by the results of the ␤-galactosidase and DNA fragmentation assays. A possible explanation for this difference is that endogenous TRAF2 is thought to modulate TNF-␣-induced apoptosis (20), whereas it is unlikely to play a role in the control of serum withdrawalinduced apoptosis (29). Of interest, the transient and stable overexpression of TRAF2-(87-501) markedly but not totally decreased the antiapoptotic effect of insulin in CHO-R15 cells. Consistent with this result, the transient and stable overexpression of wild-type TRAF2 partially mimicked the antiapoptotic effect of insulin. This is supported by the finding that overexpressed TRAF2 significantly reduced the extent of serum withdrawal-induced apoptosis in the ␤-galactosidase and DNA fragmentation assays, but to a level that remained higher than that obtained with insulin and that could be further reduced by the hormone. Together, these results strongly argue that the increase in TRAF2 expression that insulin induces in CHO-R and CHO-R15 cells serves its antiapoptotic function.
The overexpression of TRAF2 markedly stimulated NF-Bmediated activity in CHO-R15 cells, in agreement with what was found in other cell types (20). TRAF2 activation of NF-B involves the ability of cytoplasmic TRAF2 to bind to NIK (17), an NF-B-inducible kinase that activates the IB kinase complex, thus leading to IB phosphorylation and degradation and subsequent NF-B activation (6). In addition, nuclear TRAF2 could participate in the activation of NF-B-regulated genes (16). Here we show that, together with increasing basal and insulin-stimulated NF-B activities, overexpressed TRAF2 inhibited serum withdrawal-induced apoptosis in CHO-R15 cells. Moreover, NF-B inhibition by IB-␣(A32/36) reduced both TRAF2 and insulin antiapoptotic signaling in these cells, indicating that the inhibition of apoptosis induced by TRAF2 and insulin relies, at least in part, on the capacity of these molecules to activate NF-B. The pathways whereby insulin and TRAF2 activate NF-B appear to be distinct. Indeed, TRAF2 was not involved in insulin stimulation of NF-B, as supported by the failure of overexpressed TRAF2-(87-501) to affect insulin-stimulated NF-B activity despite the fact that TRAF2-(87-501) markedly reduced basal NF-B activity in CHO-R15 cells. Considered altogether, our results suggest that TRAF2 is not required for insulin activation of NF-B, but that an amplification loop involving NF-B activation by overexpressed TRAF2 is required for insulin antiapoptotic signaling. This does not exclude the possibility that overexpressed TRAF2 may also contribute to this process through an NF-B-independent pathway (18,19,21). On the other hand, the finding that insulin induced TRAF2 overexpression in cytosolic and nuclear fractions in CHO-R15 cells raises the question of whether it is cytoplasmic or nuclear TRAF2 or both that play a role in insulin antiapoptotic signaling.
Besides TRAF2, insulin increased Mn-SOD mRNA expression in CHO-R15 cells. Consistent with this, insulin-like growth factor 1, a growth factor activating several signaling molecules also activated by insulin, including NF-B (25), enhanced Mn-SOD expression in MCLM colon cancer cells (30). The finding that insulin no longer increased Mn-SOD mRNA expression in IB-␣(A32/36) cells indicates that insulin induction of Mn-SOD involves NF-B, similar to what was reported for TNF-␣ (31). The overexpression of Mn-SOD, an enzyme known to function as a scavenger of superoxide radicals, conferred cell resistance to TNF-␣-induced apoptosis (22). Reciprocally, the reduction in endogenous Mn-SOD levels by expression of antisense Mn-SOD RNA increased cell susceptibility to TNF-␣, presumably because Mn-SOD is a protective protein induced by this factor (11). In our study, the transient expression of antisense Mn-SOD RNA did not significantly modify the sensitivity of CHO-R15 cells to serum withdrawal-induced apoptosis, which may be explained by the fact that this process involves a signaling apoptotic pathway distinct from that initiated by TNF-␣ (29). In contrast, the antisense Mn-SOD RNA significantly reduced insulin protection against apoptosis, indicating that Mn-SOD is one of the protective proteins whose synthesis is induced by insulin to mediate cell survival. In this regard, it is worth noting that the protection exerted by insulin against apoptosis induced by growth factor deprivation in primary cultures of mouse proximal tubular cells involved its ability to inhibit the production of superoxide radicals in the culture medium (32).
In conclusion, our study provides the first evidence that the antiapoptotic function of insulin involves the activation by insulin of the NF-B-dependent survival genes coding for TRAF2 and Mn-SOD. Since a recent study reported that the bcl-2 and bcl-x survival genes were induced by TNF-␣ through NF-B activation (33), it would be interesting to investigate whether these genes could also be activated by insulin and mediate this function in concert with the genes encoding TRAF2 and Mn-SOD. In addition to the NF-B pathway, insulin-mediated cell survival involves the activation of the phosphatidylinositol 3-kinase pathway (1,4). Recent evidence indicates that Akt, the effector of phosphatidylinositol 3-kinase, promotes cell survival through several mechanisms, including the inhibition of the nuclear translocation of FKHRL1, a transcription factor of the Forkhead family that is likely to regulate the expression of apoptotic genes (34 -36). Further studies will be required to examine whether the antiapoptotic function of insulin involves the inhibition of this transcription factor in addition to the activation of the NF-B-dependent survival genes coding for TRAF2 and Mn-SOD.