Insulin Receptor Substrate-2 Proteasomal Degradation Mediated by a Mammalian Target of Rapamycin (mTOR)-induced Negative Feedback Down-regulates Protein Kinase B-mediated Signaling Pathway in β-Cells*

Regulation of insulin receptor substrate (IRS)-2 expression is critical to β-cell survival, but the mechanisms that control this are complex and undefined. Here in pancreatic β-cells (INS-1), chronic exposure (>8 h) to 15 mm glucose and/or 5 nm IGF-1, increased Ser/Thr phosphorylation of IRS-2, which correlated with decreased IRS-2 levels. This glucose/IGF-1-induced decrease in IRS-2 levels was prevented by the proteasomal inhibitor, lactacystin. In addition, the glucose/IGF-1-induced increase in Ser/Thr phosphorylation of IRS-2 and the subsequent decrease in INS-1 cell IRS-2 protein levels was thwarted by the mammalian target of rapamycin(mTOR) inhibitor, rapamycin. Moreover, adenoviral-mediated expression of constitutively active mTOR (mTORΔ) further increased glucose/IGF-1-induced Ser/Thr phosphorylation of IRS-2 and decreased IRS-2 protein levels, whereas adenoviral-mediated expression of “kinase-dead” mTOR (mTOR-KD) conversely reduced Ser/Thr phosphorylation of IRS-2 and maintained IRS-2 protein levels. In adenoviral-infected β-cells expressing mTORΔ, the decrease in IRS-2 protein levels was also prevented by rapamycin or lactacystin, further indicating a proteasomal mediated degradation of IRS-2 mediated via mTOR-induced Ser/Thr phosphorylation of IRS-2. Finally, we found that chronic activation of mTOR leading to decreased levels of IRS-2 in INS-1 cells led to a significant decrease in PKB activation and consequently increased β-cell apoptosis. Thus, chronic activation of mTOR by glucose (and/or IGF-1) in β-cells leads to increased Ser/Thr phosphorylation of IRS-2 that targets it for proteasomal degradation, resulting in decreased IRS-2 expression and increased β-cell apoptosis. This may be a contributing mechanism as to how β-cell mass is decreased by chronic hyperglycemia in the pathogenesis of type-2 diabetes.

Regulation of ␤-cell mass has important implications for the pathogenesis of obesity-linked type 2 diabetes (1). Obesity is linked to an increase in peripheral insulin resistance, but this can be compensated by an increase in ␤-cell function and mass so that the onset of type 2 diabetes is avoided (10). However, with time and/or increased insulin resistance, ␤-cell mass eventually becomes insufficient to cope with the increased insulin demand and type 2 diabetes ensues (1,10). Indeed, there is a decreased ␤-cell mass associated with the onset of type 2 diabetes in human and rodent models of type 2 diabetes, caused by increased ␤-cell apoptosis (10 -13). The balance between ␤-cell mass and insulin resistance relative to the pathogenesis of type 2 diabetes is nicely illustrated by comparing the IRS-1 and IRS-2 knock-out mice phenotypes (4,5,14). The IRS-1 Ϫ/Ϫ mice are insulin resistant but this is adequately compensated for by an increased ␤-cell mass and as such diabetes is avoided in these animals. In contrast, IRS-2 Ϫ/Ϫ mice have a marked decrease in ␤-cell mass so that the inherent insulin resistance is not compensated for in these animals and they become severely diabetic (4,5,14).
Although the pathophysiological relevant factors that instigate a decrease in ␤-cell mass associated with the onset of type 2 diabetes have not been elucidated, chronic hyperglycemia and/or hyperlipemia are thought to be the prime suspects (10,15,16). Both chronic exposure to elevated glucose levels and/or FFA levels cause ␤-cell apoptosis (16 -22). FFA are thought to decrease ␤-cell survival, at least in part, via inhibition of IGF-1-induced PKB activation (10,22,23). In skeletal muscle, FFAinduced inhibition of insulin-mediated PI3K/PKB activation has been suggested to be via FFA-induced activation of novel PKC isoforms that increase Ser/Thr phosphorylation of IRS-1 that, in turn, has a negative effect of dampening downstream IRS-1 signal transduction (24,25). This is thought to make a significant contribution to the mechanism of insulin resistance. In ␤-cells, FFA also activate novel PKC isoforms that could increase the IRS-2 Ser/Thr phosphorylation status (22) and, in turn, contribute to FFA-induced inhibition of downstream PKB activation (22,23). As such, it is possible to envisage commonalities between mechanisms of FFA-induced insulin resistance in muscle and FFA-induced ␤-cell apoptosis (10). However, it has been shown that other protein kinases, not just certain PKC isoforms, are capable of Ser/Thr phosphorylating IRS proteins, including mTOR (26 -28), but this has yet to be examined in pancreatic ␤-cells.
Under normal circumstances, relatively short term fluctuations in glucose concentrations in the 3-15 mM range appear protective for ␤-cell (9,22). Glucose can induce ERK-1/2 phosphorylation activation in ␤-cells, independently of IRS-2 (7, 29 -31). However, activation of the ERK-1/2 signaling pathway in ␤-cells has yet to be associated with control of ␤-cell survival (10). Glucose can also activate mTOR independently of IRS-2 in ␤-cells (29,30), which contributes to up-regulation of general protein synthesis in ␤-cell via phosphorylation of the eIF-4Ebinding protein (4E-BP1) and p70 S6K (32)(33)(34)(35). However, it is possible that prolonged activation of mTOR by chronically elevated glucose concentrations could be detrimental to IRS-mediated signaling in ␤-cells. It has been described in insulin target tissues that chronically activated mTOR increased the Ser/Thr phosphorylation state of IRS-1 and IRS-2 (27,28). This increased Ser/Thr phosphorylation of IRS-1 or IRS-2 via mTOR subsequently leads to increased degradation of IRS-1/2 proteins and a consequential down-regulation of insulin signaling transduction pathways, which also contributes to insulin resistance in peripheral tissues (27,28,36,37). If chronic activation of mTOR leads to Ser/Thr phosphorylation of IRS-2 in ␤-cells, then this would decrease IRS-2 protein levels and be detrimental for ␤-cell survival. Indeed, in this present study we present experimental evidence to indicate that this might well be the case, and in turn may be a contributing mechanism as to how chronic hyperglycemia eventually leads to a decrease in ␤-cell mass in the pathogenesis of type 2 diabetes.
Cell Culture and Treatment-The pancreatic ␤-cell line, INS-1, was maintained in the complete medium as previously described (38). In general, INS-1 cells were subcultured on 6-well plates to ϳ80% confluence and then incubated 16 h at 37°C in serum-free modified RPMI 1640 medium containing 3 mM glucose and 1% (w/v) bovine serum albumin as described (7,29). After the quiescent period the cells were incubated at 37°C with fresh serum-free RPMI 1640 medium with 3 or 15 mM glucose Ϯ 5 nM IGF-1 Ϯ inhibitors for 8 or 24 h as indicated.
Adenovirus Infection-All mTOR constructs have an NH 2 -terminal AU1 epitope tag. The "kinase-dead" mTOR KD variant contains an Asp 2338 to Ala mutation that renders it catalytically inactive (39,40). The mTOR⌬ variant is a mutant form of mTOR that was generated by deleting amino acids 2430 -2450 (41). This region contains sites phosphorylated by Akt (41), as well as the epitope for the activating antibody, mTAb1 (42). The intrinsic kinase activity of mTOR⌬ is ϳ5-fold higher than that of wild type mTOR when assayed in vitro (43) and overexpression of mTOR⌬ in cells results in constitutive phosphorylation of the mTOR targets, 4E-BP1 and p70 S6K (41). Adenoviruses expressing wild type (AdV-mTOR-WT), constitutively activated (AdV-mTOR⌬), and kinase-dead mTOR (AdV-mTOR-KD) were generated, amplified, and purified as previously described (7,29), using the bacterial recombination method in BJ5183 Escherichia coli cells (44). Adenovirus expressing GFP (AdV-GFP) and luciferase (AdV-Luc) were used as control viruses. The determination of the appropriate titer of adenovirus and the adenovirus infection were as previously described (7,29).
Immunoblot Analysis-Cell lysates were prepared as previously described first and assayed for total protein levels as determined using the BCA protein assay kit (Pierce). Equivalent protein sample aliquots were then subjected to immunoblot analysis as previously described (7,29). For alkaline phosphatase treatment, INS-1 cells were lysed in phosphatase inhibitor-free lysis buffer. These lysates were then treated with 1.725 DEA units/100 l of alkaline phosphatase attached to agarose beads with shaking at 37°C for 1 h. The alkaline phosphatase was removed by microcentrifugation prior to the immunoblot analysis.
Immunofluorescence-AdV-infected INS-1 cells were subcultured on glass coverslips within 6-well plates to 60 -70% confluence, and after incubation were washed with ice-cold PBS, fixed with 4% paraformaldehyde/PBS, washed with 0.1% Triton/PBS, and then re-washed with PBS. Fixed cells were first incubated for 1 h with 5% donkey serum in 5% bovine serum albumin/PBS prior to incubation with primary antibody against activated caspase-9 overnight at a 1:100 dilution in 5% bovine serum albumin/PBS at 4°C. The cells were then washed 3 times with PBS and incubated with a secondary antibody conjugated to fluorescein isothiocyanate (1:500 in 5% bovine serum albumin/PBS) (Amersham Biosciences) for 1 h at 25°C. The cells were then washed 3 times in PBS and mounted onto glass slides with mounting media (Molecular Probes). Activated caspase-9 was visualized using an Olympus FV500 confocal microscope.
Expression of Data and Statistics-Data are presented as mean Ϯ S.E. Statistically significant differences between groups were analyzed using Student's t test, where p Ͻ 0.05 was considered statistically significant.

Chronic Exposure to Glucose and/or IGF-1 Induces Ser/Thr
Phosphorylation of IRS-2 in Pancreatic ␤-Cells-Quiescent INS-1 cells were incubated for 8 h in the presence of basal 3 mM glucose or stimulatory 15 mM glucose Ϯ 5 nM IGF-1. Immunoblot analysis of IRS-2 indicated different stages of a decreased electrophoretic mobility of IRS-2 protein on a 7% polyacrylamide gel analysis (Fig. 1A). A stimulatory 15 mM glucose induced an upward mobility shift in IRS-2 protein migration compared with that at basal 3 mM glucose (Fig. 1A). IGF-1 added at 3 mM basal glucose also induced a slight upward shift in IRS-2 migration, which was further increased when IGF-1 was added together with a stimulatory 15 mM glucose concentration (Fig. 1A). IRS-2 immunoblot analysis by the 10% polyacrylamide gel was used to more tightly resolve IRS-2 electrophoretic migration and better evaluate total IRS-2 protein levels. Intriguingly, IRS-2 protein levels were increased 2-3fold at 15 mM glucose compared with basal 3 mM glucose (Fig.  1A). The addition of IGF-1 at basal 3 mM glucose had no effect on IRS-2 protein levels, and addition of IGF-1 at a stimulatory 15 mM glucose tended to reduce IRS-2 protein levels compared with that at 15 mM glucose alone (Fig. 1A). Immunoblot analysis of the p85 subunit of PI3K was used as a loading control and its protein expression levels were unchanged under all mTOR-mediated IRS-2 Proteasomal Degradation in ␤-Cells conditions emphasizing the specific effect on IRS-2 (Fig. 1A).
The time course of this glucose-induced IRS-2 gel mobility shift was examined. Quiescent INS-1 cells were incubated in the presence of basal 3 mM or stimulatory 15 mM glucose for 1, 4, 8, and 24 h. There was little change in IRS-2 electrophoretic mobility at basal 3 mM glucose during early time points, although IRS-2 protein levels were decreased by ϳ50% at 24 h compared with that at the preceding 8-h time point (Fig. 1B). At 15 mM glucose there was an increased IRS-2 gel mobility shift apparent by 4 h and remained up to 24 h (Fig. 1B). Stimulatory 15 mM glucose increased IRS-2 protein levels, which reached a maximum 2-3-fold increase at 8 h (Fig. 1B). IRS-2 expression levels at 15 mM glucose remained elevated by 24 h compared with that at basal 3 mM glucose, although this was actually decreased by about 50% compared with that at the preceding 8-h time point at a stimulatory 15 mM glucose concentration (Fig. 1B).
Glucose/IGF-1-induced IRS-2 Electrophoretic Mobility Shift Is Partly Because of the Increased IRS-2 Ser/Thr Phosphorylation State-To determine whether IRS-2 gel shift was because of an increase in the Ser/Thr phosphorylation state, INS-1 cell lysates were first treated with alkaline phosphatase to specifically dephosphorylate phosphoserine and phosphothreonine but not the phosphotyrosine residues on IRS-2. As a positive control it was found that both IGF-1 and the glucose-induced Ser/Thr phosphorylation state of ERK-1/2 and PKB was markedly decreased by alkaline phosphatase treatment ( Fig. 2A). The IGF-1 or 15 mM glucose-induced IRS-2 electrophoretic upward mobility shift was decreased after alkaline phosphatase treatment, indicating that, at least in part, this shift was because of the increased Ser/Thr phosphorylation state of IRS-2. However, the shift was not completely resolved to that at basal 3 mM glucose by alkaline phosphatase treatment, suggesting that something else might have contributed to the glucose/IGF-1-induced IRS-2 electrophoretic mobility shift.
Glucose/IGF-1-induced IRS-2 Electrophoretic Mobility Shift Is Also Contributed by Targeting IRS-2 for Proteasomal Degradation-Previous studies in adipocytes have shown a controlled proteasome-mediated degradation of IRS-1 and IRS-2 proteins that correlated with the IRS-1 or IRS-2 Ser/Thr phosphorylation states (27,36,45,46). Thus, it was investigated whether the mobility shift of IRS-2 might also be contributed by a ubiquitin-dependent targeting of IRS-2 for proteasomal degradation. Lactacystin, a specific proteasome inhibitor, was used to prevent ubiquitin-dependent proteasomal protein degradation in pancreatic ␤-cells. Quiescent INS-1 cells were incubated for 8 h in the presence of 3 mM glucose Ϯ 5 nM IGF-1, or 15 mM glucose, in the presence or absence of 10 M lactacystin. The addition of lactacystin rendered an increased mobility shift of IRS-2 in itself at basal 3 mM glucose, and further increased the IGF-1 and 15 mM glucose-induced shift in IRS-2 electrophoretic mobility ( Fig 3A). The IRS-2 gel mobility shift induced by lactacystin was then examined to see how much of it was contributed to by IRS-2 Ser/Thr phosphorylation. Alkaline phosphatase-mediated removal of the phosphates from IRS-2 Ser/Thr sites did not completely resolve the upward gel mobility shift of IRS-2 induced by IGF-1 or glucose in the presence of lactacystin (Fig. 3B). These results suggest that an alternative post-translational modification of IRS-2, other than Ser/Thr phosphorylation, induced by glucose/IGF-1 stimulation contributes to the IRS-2 upward electrophoretic mobility shift on SDS-PAGE. Although lactacystin inhibits ubiquitin-dependent proteasomal degradation it does not affect the ubiquitinylation of proteins. As lactacystin further increased the glucose/
mTOR Mediates IGF-1/Glucose-induced Ser/Thr Phosphorylation of IRS-2-It was examined whether mTOR mediated the glucose/IGF-1-induced increase of IRS-2 Ser/Thr phosphorylation in ␤-cells, initially by using the relatively specific mTOR inhibitor, rapamycin. Quiescent INS-1 cells were cultured at basal 3 mM or stimulatory 15 mM glucose Ϯ IGF-1, in the presence or absence of rapamycin (50 nM), for 8 h. The IRS-2 upward gel mobility shift induced by IGF-1 or glucose was reduced in the presence of rapamycin (Fig. 4). Interestingly, IRS-2 protein expression levels appeared to be enhanced by rapamycin, especially at stimulatory 15 mM glucose concentrations (Fig. 4). The PI3K p85 subunit protein levels did not change indicating specific effects on IRS-2 protein expression levels (Fig. 4). The inhibitory effect of rapamycin on mTOR activation was confirmed by observing a prevention of glucose/ IGF-1-induced phosphorylation activation of p70 S6K (data not shown). To further examine mTOR-mediated phosphorylation of IRS-2 we investigated the effect of adenoviral-mediated expression of a wild type (AdV-mTOR-WT), constitutively active (AdV-mTOR⌬), or a kinase-dead form (AdV-mTOR-KD) of mTOR in pancreatic ␤-cells on IRS-2 Ser/Thr phosphorylation. A GFP-expressing adenovirus (AdV-GFP) was used as a control. The mTOR adenoviral constructs all had a NH 2 -terminal AU1 tag that was used for specific immunoblot analysis of adenoviral-mediated expression of mTOR. Twenty-four hours after infection of INS-1 cells with increasing titer of AdV-mTOR-WT, AdV-mTOR⌬, or AdV-mTOR-KD adenovirus, a corresponding increase in AU1-tagged mTOR expression was observed (Fig. 5A). The activity of adenoviral mediated expression of the mTOR variants was assessed by evaluating the phosphorylation state of the substrates of mTOR, p70 S6K , and 4E-BP1, in INS-1 cells infected with equivalent titer (750 plaque forming units/cell) of AdV-mTOR-WT, AdV-mTOR⌬, and AdV-mTOR-KD or the control AdV-GFP. Infected cells were incubated for 16 h at basal 3 mM glucose and then for either 1 h (for p70 S6K analysis (Fig. 5B)) or 30 min (for 4E-BP1 analysis (Fig. 5C)) at 3 or 15 mM glucose Ϯ IGF-1, and Ϯrapamycin, as indicated (Fig. 5). Glucose-induced p70 S6K phosphorylation was further increased by the addition of IGF-1 in AdV-GFP-and AdV-mTOR-WT-infected cells, as previously shown (29) (Fig. 5B). In AdV-mTOR⌬-infected cells, p70 S6K phosphorylation activation was significantly increased at basal 3 mM glucose as indicated by an increased upward mobility shift of total p70 S6K because of its increased phosphorylation state (Fig. 5B), consistent with the degree of constitutive activation of the mTOR⌬ variant (41,43). Likewise, p70 S6K activation by IGF-1 and/or 15 mM glucose was enhanced in AdV-mTOR⌬-infected cells (Fig. 5B). In contrast, p70 S6K phosphorylation induced by glucose and/or IGF-1 was decreased in AdV-mTOR-KD-infected cells, consistent with the lack of mTOR activity in the mTOR-KD variant (Fig. 5B). As a control to indicate these effects of p70 S6K phosphorylation were mediated by mTOR. It was found that rapamycin inhibited glucose-and glucose/IGF-1-induced p70 S6K phosphorylation in AdV-GFP and AdV-mTOR⌬-infected cells (Fig. 5B). As previously described (33), a stimulatory 15 mM glucose concentration induced 4E-BP1 phosphorylation in ␤-cells, which was further increased by IGF-1, as indicated by an increase in the upper migrating band corresponding to the highly phosphorylated ␥-form of 4E-BP-1, in AdV-GFP and AdV-mTOR-WT-infected cells (Fig. 5C). In AdV-mTOR⌬-infected cells, 4E-BP1 phosphorylation was slightly increased at basal 3 mM and enhanced further at 15 mM glucose, compared with that in AdV-GFP-and AdV-mTOR-WT-infected cells (Fig. 5C). This was also consistent with the constitutive activation of the mTOR⌬ variant (41,43). In contrast, 4E-BP1 phosphorylation induced by glucose and/or IGF-1 was decreased in AdV-mTOR-KD-infected cells mTOR-mediated IRS-2 Proteasomal Degradation in ␤-Cells (Fig. 5C), again consistent with the lack of mTOR activity in the mTOR-KD variant. Rapamycin also inhibited glucose and IGF-1-induced 4E-BP1 phosphorylation in AdV-GFP and AdV-mTOR⌬-infected cells (Fig. 5C), reaffirming that the 4E-BP1 phosphorylation was mediated by mTOR.
IRS-2 gel mobility shift was assessed in INS-1 cells infected with AdV-mTOR-WT, AdV-mTOR⌬, and AdV-mTOR-KD or the control AdV-GFP recombinant adenoviruses. Infected cells were incubated for 16 h at basal 3 mM glucose and then for 8 h in the presence of basal 3 mM or stimulatory 15 mM glucose Ϯ 5 nM IGF-1, as indicated (Fig. 6). Glucose and IGF-1 caused an upward electrophoretic mobility shift of IRS-2 on the 7% PAGE immunoblot analysis, in control AdV-GFP-infected cells (Fig. 6) and as shown previously in uninfected INS-1 cells (Fig. 1). This IGF-1/glucose-induced upward shift of IRS-2 electrophoretic mobility was further increased in both AdV-mTOR-WT-and AdV-mTOR⌬-infected cells (Fig. 6). In contrast, in AdV-mTOR-KD-infected cells the glucose/IGF-1-induced upward mobility shift of IRS-2 was prevented (Fig. 6), similarly to that observed in rapamycin-treated cells (Fig. 4). These results further indi-

mTOR-mediated IRS-2 Proteasomal Degradation in ␤-Cells
infected cells, IRS-2 protein levels were generally better preserved and significantly increased at 15 mM glucose and 15 mM glucose/IGF-1 compared with that in AdV-GFP control cells (p Ͻ 0.05; Fig. 7A). Inhibition of endogenous mTOR by rapamycin had no additional effect on IRS-2 expression levels in AdV-mTOR-KD-infected INS-1 cells (Fig. 7A). These observations on IRS-2 protein expression levels in AdV-GFP, AdV-mTOR⌬, or AdV-mTOR-KD-infected ␤-cells were specific because protein levels of p85(PI3K) were similar between samples (Fig. 7A). Thus, chronic activation of mTOR in mTOR⌬ expressing ␤-cells decreased IRS-2 levels, while inhibiting mTOR Ser/Thr kinase activity, by either rapamycin or expression of mTOR-KD, preserved IRS-2 protein expression levels.
It was also investigated whether lactacystin prevented the mTOR-mediated decrease of IRS-2 protein levels. The AdV-mTOR⌬ and AdV-mTOR-KD or the control AdV-GFP-infected INS-1 cells were incubated at basal 3 mM glucose for 16 h and then for a further 24 h at basal 3 mM or stimulatory 15 mM glucose Ϯ 5 nM IGF-1, in the presence or absence of lactacystin for the last 6 h. In AdV-GFP-infected cells, the addition of lactacystin preserved the increased IRS-2 protein levels at 15 mM glucose and 15 mM glucose/IGF-1 (Fig. 7B). In AdV-mTOR⌬-infected INS-1 cells, IRS-2 protein expression levels were generally lower compared with that in control AdV-GFP-infected INS-1 cells (Fig. 7B). Glucose-induced increase in IRS-2 protein levels in AdV-mTOR⌬-infected ␤-cells was significantly reduced in the absence of IGF-1, and further decreased in the presence of IGF-1, compared with AdV-GFPinfected control INS-1 cells (p Ͻ 0.05; Fig. 7B). However, this decrease was partially rescued by addition of lactacystin so that IRS-2 protein levels were not significantly different from that found in AdV-GFP-infected control cells, particularly at 15 mM glucose (Fig. 7B). In contrast, in AdV-mTOR-KD-infected cells, IRS-2 protein levels were higher, especially at 15 mM glucose and 15 mM glucose ϩ IGF-1, compared with that in AdV-GFP control cells, and remained unchanged after lactacystin treatment (Fig. 7B). The protein levels of p85(PI3K) were similar between samples (Fig. 7B), underlining the specific effect on IRS-2 protein levels. Taken together, the data indicate that mTOR-decreased IRS-2 protein levels are mediated by proteasomal degradation of IRS-2 and is likely via prior ubiquitination of the protein as previously shown in insulin target cells (45,46). mTOR-mediated IRS-2 Proteasomal Degradation in ␤-Cells phosphorylation activation of PKB and ERK-1/2 in AdV-mTOR⌬, AdV-mTOR-KD, or the control AdV-GFP-infected INS-1 cells in the presence or absence of rapamycin or lactacystin. In AdV-GFP-infected cells, 24 h incubation at a stimulatory 15 mM glucose significantly increased Ser 473 PKB phosphorylation above that at basal 3 mM glucose, which was further enhanced by the addition of IGF-1 (Fig. 8, A and B). Neither rapamycin (Fig. 8A) nor lactacystin (Fig. 8B) significantly changed glucose/IGF-1-induced phosphorylation activation of PKB in AdV-GFP-infected control cells. However, in AdV-mTOR⌬-infected cells, both 15 mM glucose-and 15 mM glucose/IGF-1-induced phosphorylation activation of PKB were significantly inhibited by Ն50% (p Ͻ 0.05, n ϭ 6) compared with that in AdV-GFP-infected control cells (Fig. 8, A  and B). However, the addition of rapamycin (Fig. 8A), or lactacystin (Fig. 8B), significantly restored glucose and glucose/IGF-1-induced Ser 473 phosphorylation of PKB in AdV-mTOR⌬-infected INS-1 cells to that in AdV-GFP-infected control cells (Fig. 8). This correlated with the decreased IRS-2 protein levels in these AdV-mTOR⌬-infected cells that was prevented by rapamycin or lactacystin (Fig. 7, A and B). In AdV-mTOR-KD-infected cells, PKB Ser 473 phosphorylation activation induced by 15 mM glucose or 15 mM glucose/IGF-1 was preserved and comparable with that in AdV-GFP control cells (Fig. 8). The addition of rapamycin (Fig. 8A) or lactacystin (Fig. 8B) to AdV-mTOR-KD-infected INS-1 cells did not significantly affect PKB phosphorylation activation (Fig.  8). PKB total protein levels were similar between samples emphasizing the specific effect on PKB phosphorylation activation (Fig. 8, A and B). The glucose/IGF-1-induced phosphorylation activation of ERK-1/2 was examined in the same INS-1 cell lysates. Expression of mTOR⌬ or mTOR-KD did not significantly affect ERK-1/2 phosphorylation induced by 15 mM glucose and 15 mM glucose/IGF-1 compared with that in AdV-GFP-infected control INS-1 cells (Fig. 8, A and B). Glucose/IGF-1-induced ERK-1/2 phosphorylation activation was unaffected by rapamycin (Fig. 8A). In contrast, ERK-1/2 phosphorylation at basal 3 mM glucose was enhanced by lactacystin as well as 15 mM glucose/IGF-1-induced stimulation of ERK-1/2 phosphorylation activation in AdV-GFP control cells. Similarly, the enhanced glucose/IGF-1-induced ERK-1/2 phosphorylation by lactacystin was observed in AdV-mTOR⌬ and AdV-mTOR-KD-infected INS-1 cells in parallel to that in AdV-GFP control cells (Fig. 8B). As such, lactacystin-increased ERK-1/2 phosphorylation was likely independent of mTOR (Fig. 8B). It should be noted that ERK-1/2 total protein levels were unaffected in these experiments underlining the specific effects of lactacystin on ERK-1/2 phosphorylation activation (Fig. 8B).

mTOR-mediated Decrease in IRS-2 Protein Levels Correlates with Down-regulation of IGF-1/Glucose-induced PKB
Chronic Activation of mTOR Decreases ␤-Cell Survival-Decreasing IRS-2 expression levels promotes ␤-cell apoptosis (4,5,9). Thus, it was investigated whether the mTOR-mediated decrease in IRS-2 protein levels was associated with an increase in ␤-cell apoptosis. The degree of apoptosis was indicated by immunoblot analysis of cleaved/activated caspase-9. The final stages of apoptosis are executed by the family of

mTOR-mediated IRS-2 Proteasomal Degradation in ␤-Cells
caspases, including the initiator caspase-9 that is proteolytically cleaved from inert procaspase-9 (47). The amount of cleaved caspase-9 has been correlated to the degree of ␤-cell apoptosis (9,22). The AdV-mTOR⌬ and AdV-mTOR-KD or the control AdV-GFP-infected INS-1 cells were incubated at basal 3 mM glucose for 16 h and then for a further 24 h at basal 3 mM or stimulatory 15 mM glucose Ϯ 5 nM IGF-1. A low degree of basal caspase-9 activation was seen in AdV-GFP-infected con-trol INS-1 cells that was not significantly changed between 3 or 15 mM glucose, or 15 mM glucose ϩ IGF-1 (Fig. 9, A and B). However, in AdV-mTOR⌬-infected cells, activated caspase-9 was significantly increased compared with that in control AdV-GFPinfected cells (p Յ 0.05; Fig. 9B). In contrast, in AdV-mTOR-KDinfected cells minimal activated caspase-9 was detected, even below the degree of apoptosis found in AdV-GFP-infected control INS-1 cells, whether at basal 3 or 15 mM glucose, or 15 mM FIG. 9. Chronic activation of mTOR decreases ␤-cell survival. A, INS-1 cells infected with AdV-mTOR⌬, AdV-mTOR-KD, or AdV-GFP as a control were made quiescent and then incubated for 24 h with 3 mM glucose and 15 mM glucose Ϯ 5 nM IGF-1. The cell lysates were resolved by 10 and 15% SDS-PAGE and immunoblotted (as described under "Experimental Procedures"), respectively, with antibodies against phospho-Ser 473 -PKB and activated caspase-9. The 10% blots were reprobed with PKB as indicated. IB, immunoblot. B, quantification of the activated caspase-9 was performed by densitometric scanning with Optiquant software analysis. The data are expressed as a mean Ϯ S. E. (n ϭ 3). C, AdV-mTOR⌬, AdV-mTOR-KD, or AdV-Luc control-infected cells subcultured on glass coverslips were incubated for 36 h with 3 mM glucose. After fixing, cells were analyzed by immunofluorescence for activated caspase-9 as described under "Experimental Procedures." Rounded cells, positive for activated caspase-9 and undergoing apoptosis, are indicated by white arrowheads in these representative images. mTOR-mediated IRS-2 Proteasomal Degradation in ␤-Cells glucose ϩ IGF-1 (Fig. 9, A and B). Phosphorylation activation of PKB was assessed in the same INS-1 cell lysates. Consistent with the observations in Fig. 8, PKB Ser 473 phosphorylation activation induced by glucose/IGF-1 was reduced in AdV-mTOR⌬-infected cells and remained unchanged in AdV-mTOR-KD -infected INS-1 cells compared with that in AdV-GFP-infected control cells (Fig. 9A). Total PKB protein levels were similar between samples (Fig. 9A), emphasizing the specific effect on PKB and caspase-9 activation. Immunocytochemical analysis of activated caspase-9 in AdV-mTOR⌬, AdV-mTOR-KD, and AdV-Luc control-infected INS-1 cells incubated for 36 h at basal 3 mM glucose are shown in Fig. 9C. Immunofluorescent analysis of activated caspase-9 was markedly increased in AdV-mTOR⌬-infected cells compared with that in AdV-Luc control cells. In contrast, the immunofluorescence analysis of activated caspase-9 revealed was almost undetectable in AdV-mTOR-KD-infected cells (Fig. 9C), in accordance with the low incidence of activated caspase-9 by immunoblot analysis even below that of AdV-Luc-infected control INS-1 cells (Fig. 9, A and B). DISCUSSION Here, we have demonstrated that chronic activation of mTOR leads to an increased Ser/Thr phosphorylation state of IRS-2 in pancreatic ␤-cells. Activation of IRS-mediated signaling pathways requires tyrosine phosphorylation of IRS molecules by certain tyrosine kinase activities (e.g. ligand-activated insulin or IGF-1 receptors (2,48)). The phosphorylated tyrosine motifs on IRS proteins act as docking sites for Src homology domain 2 containing proteins (such as the p85 regulatory subunit of PI3K or Grb2) to associate with IRS and consequently become activated (2). There are actually more potential Ser/Thr phosphorylation sites on IRS proteins than there are tyrosine phosphorylation sites, but the functional consequence of these has not yet been well characterized. It appears that a few Ser/Thr phosphorylation sites on IRS might have a positive role in potentiating IRS-mediated signal transduction, but most appear to have a negative role opposing the effect of tyrosine phosphorylation of IRS proteins (26). Indeed, a common mechanism for down-regulating IRS-mediated signaling pathways is via Ser/Thr phosphorylation of IRS-1 or IRS-2 (26). In insulin responsive cell types, the Ser/Thr phosphorylation of IRS-1 can disrupt the association between the insulin receptor and IRS and inhibit tyrosine phosphorylation (49 -51), and/or can facilitate a subcellular redistribution of IRS linked to its ubiquitinmediated proteasomal degradation (27,28,52).
Several Ser/Thr kinases (e.g. certain PKC isoforms, mTOR, JNK, IKK␤, and mitogen-activated protein kinase (26)) have been described to promote Ser/Thr phosphorylation of the IRS proteins in insulin target tissues such as myocytes, hepatocytes, and adipocytes (24,25,27,28,53,54). Here, we find in pancreatic ␤-cells that chronic activation of the mTOR Ser/Thr protein kinase by glucose/IGF-1, or via adenoviral-mediated expression of a constitutive activated variant of mTOR, specifically increased the Ser/Thr phosphorylation state of IRS-2 that was correlated with a decrease in IRS-2 protein expression levels. Conversely, inhibition of mTOR activity, by either rapamycin or the expression of the kinase-dead mTOR variant, inhibited Ser/Thr phosphorylation of IRS-2 and correlated with increased IRS-2 protein levels in ␤-cells. Although we were unable to directly test whether IRS-2 is ubiquitinated in ␤-cells, the use of the proteasomal degradation inhibitor, lactacystin, indicated that the increased Ser/Thr phosphorylation state of IRS-2 mediated by mTOR was associated with IRS-2 ubiquitination. Lactacystin upwardly shifted the electrophoretic mobility of IRS-2 caused by glucose or IGF-1, and this could never be completely resolved by alkaline phosphatase treatment. Thus, another post-translational modification was contributing to the apparent increased molecular mass of IRS-2 in addition to the Ser/Thr phosphorylation, and was most likely an ubiquitination. Moreover, lactacystin prevented the decrease in IRS-2 protein levels in AdV-mTOR⌬-infected cells. This, in turn, indicated that the decrease in ␤-cell IRS-2 protein levels associated with mTOR-mediated Ser/Thr phosphorylation of IRS-2 was most likely via ubiquitin-dependent proteasomal degradation, rather than any effect at the level of IRS-2 protein synthesis. Intriguingly, a similar pathway has been described in adipocytes (45), where Ser/Thr-phosphorylated IRS-2 become ubiquitinated and are subsequently degraded by the 26 S proteasome in response to a chronic insulin stimulation or cellular stress (46).
Although it was observed that mTOR is involved in Ser/Thr phosphorylation of IRS-2, it is not yet clear whether mTOR directly phosphorylates IRS-2 in pancreatic ␤-cells. It has recently been revealed that other protein factors, such as raptor and mLST8, are required to modify mTOR activity and/or present substrates such as 4E-BP1 and p70 S6K to mTOR for Ser/ Thr phosphorylation (55)(56)(57). The increased expression of a constitutively active mTOR variant markedly increased IRS-2 Ser/Thr phosphorylation in ␤-cells, apparently independently of a need for protein cofactors of mTOR. Thus, either these mTOR cofactors are expressed in excess in ␤-cells that cater for the chronic activation of mTOR, or that another protein kinase(s) lies between mTOR and IRS-2 Ser/Thr phosphorylation in an as of yet undefined signaling pathway. Nonetheless, the observation that rapamycin inhibits Ser/Thr phosphorylation of IRS-2 in ␤-cells, as well as preserves IRS-2 protein expression levels, indicates that mTOR plays a critical part in instigating Ser/Thr phosphorylation of IRS-2 and its subsequent proteasomal degradation. It should be noted that the mTOR⌬ variant was able to increase phosphorylation of p70 S6K and 4E-BP1, particularly under basal conditions, consistent with its degree of constitutive activation. This suggested that there were sufficient protein cofactors of mTOR expressed in ␤-cells for suitable p70 S6K and 4E-BP phosphorylation activation. Another note should be made that the constitutively activated mTOR⌬ variant was inhibited by rapamycin that is consistent with the so-called rapamycin-binding FRB domain being retained in the mTOR⌬ variant (41,43,57).
Maintenance of optimal IRS-2 expression levels in ␤-cells and downstream activation of the PI3K/PKB signaling is essential for promoting ␤-cell survival (10). The mTOR-mediated decrease in IRS-2 protein levels in ␤-cells was also associated with decreased glucose/IGF-1-induced PKB phosphorylation activation. Both rapamycin and lactacystin prevented the IRS-2 protein degradation and rescued PKB phosphorylation activation in response to glucose/IGF-1, indicating that the extent of PKB activation is linked to the level of IRS-2 expression. This is corroborated by previous findings that increased expression of IRS-2 enhanced PKB phosphorylation activation and prevented ␤-cells from FFA-increased apoptosis (9), while decreasing expression of IRS-2 using adenoviral-mediated expression of IRS-2 antisense-inhibited PKB phosphorylation activation and induced ␤-cell apoptosis (9). Adenoviral-mediated expression of constitutively active PKB in ␤-cells is also protective against FFA-induced apoptosis (22).
In this study, it was also found that decreased IRS-2 protein expression in ␤-cells mediated via chronic activation of mTOR correlated with decreased PKB phosphorylation activation and that this, in turn was associated with increased caspase-9 activation, a marker of ␤-cell apoptosis. Thus, there is a general interrelationship between IRS-2 expression and downstream activation of PKB in ␤-cells, which is key to maintaining ␤-cell survival. However, a small proviso should be made in regard to the combination of stimulatory glucose and IGF-1 that occasionally caused a modest decrease in IRS-2 protein levels ( Fig.  1), yet normal PKB phosphorylation activation was maintained (Fig. 8). In this instance it should be noted that glucose can activate mTOR independently of IRS-2 in ␤-cells (30), and is a relatively poor activating stimulus for PKB requiring Ͼ4-h exposure to glucose (29). In contrast, IGF-1 induced activation of PKB and mTOR in ␤-cells is IRS-2 dependent, and IGF-1 renders a rapid and sustained activation of PKB in ␤-cells within minutes (7,29,30). As such, although there may be reduced IRS-2 protein expression levels, mediated by chronic glucose/IGF-1 induced activation of mTOR, IGF-1 (but not necessarily glucose) is still able to give an effective activation of PKB via the residual IRS-2 present in ␤-cells. Notwithstanding, in contrast to the mTOR-mediated decrease in IRS-2 protein levels generally correlating with decreased PKB phosphorylation, there was no effect on ERK-1/2 phosphorylation activation in the same ␤-cells. This underlines the importance of the specific glucose-induced activation of ERK-1/2 in ␤-cells that is independent of IRS-2 (30,31). Moreover, it reaffirms that the ERK-1/2 signaling pathway is not necessarily involved in maintaining ␤-cell survival (9,10).
The reduction of ␤-cell mass via increased ␤-cell apoptosis is key to the pathogenesis of type 2 diabetes (10,11,13). Chronic hyperglycemia has been proposed to be at least one of the major factors that instigates ␤-cell apoptosis. Indeed, in vitro studies have shown that chronically elevated concentrations of glucose induced ␤-cell apoptosis (17)(18)(19)(20). In contrast, over a shorter time frame, glucose can induce ␤-cell proliferation (7,8,29,30,58) as well as protect ␤-cells from apoptosis (9,22). Thus, the effect of glucose on ␤-cell growth and survival depends on the duration of exposure. It has been previously shown that glucose can independently activate mTOR in ␤-cells and this is further augmented by IGF-1-induced signaling via IRS-2 (29,30). It is therefore possible to envisage that chronic exposure to elevated glucose concentrations might induce a decrease in IRS-2 protein levels, at least in part, via chronic activation of mTOR to mediate increased IRS-2 Ser/Thr phosphorylation, leading to proteasomal degradation of IRS-2, decreased ␤-cell IRS-2 levels, and increased ␤-cell apoptosis. This potential mechanism has yet to be confirmed in primary ␤-cells and/or in models of type 2 diabetes. However, it raises the prospect that inhibition of mTOR using rapamycin might have a role in preserving ␤-cell mass as a potential therapy for type 2 diabetes. In this regard, it should be noted that as well as having immunosuppressive activity, rapamycin should also promote ␤-cell survival of transplanted islets used as a therapy for type 1 diabetes (59). Finally, considering that hyperglycemia and/or hyperinsulinemia also promotes mTOR-mediated Ser/Thr phosphorylation of IRS-1/2 and subsequent IRS-1/2 degradation in insulin target tissues that in turn contributes to the mechanism of insulin resistance (26 -28, 36, 37), a therapeutic strategy to protect ␤-cells by inhibiting mTOR should also alleviate a degree of insulin resistance and perhaps even delay the onset of type 2 diabetes. Notwithstanding, our findings in this study further emphasize the importance that control of IRS-2 levels in pancreatic ␤-cells is critical for maintaining their survival and plays a pivotal point in the pathogenesis of type 2 diabetes.