A Novel Domain Mediates Insulin-Induced Proteasomal Degradation of Insulin Receptor Substrate 1 (IRS-1)

Abstract

Insulin receptor substrate-1 (IRS-1) plays a pivotal role in insulin signaling, therefore its degradation is exquisitely regulated. Here, we show that insulin-stimulated degradation of IRS-1 requires the presence of a highly conserved Ser/Thr-rich domain that we named domain involved in degradation of IRS-1 (DIDI). DIDI (amino acids 386–430 of IRS-1) was identified by comparing the intracellular degradation rate of several truncated forms of IRS-1 transfected into CHO cells. The isolated DIDI domain underwent insulin-stimulated Ser/Thr phosphorylation, suggesting that it serves as a target for IRS-1 kinases. The effects of deletion of DIDI were studied in Fao rat hepatoma and in CHO cells expressing Myc-IRS-1^WT or Myc-IRS-1^Δ386–430. Deletion of DIDI maintained the ability of IRS-1^Δ386–434 to undergo ubiquitination while rendering it insensitive to insulin-induced proteasomal degradation, which affected IRS-1^WT (80% at 8 h). Consequently, IRS-1^Δ386–434 mediated insulin signaling (activation of Akt and glycogen synthesis) better than IRS-1^WT. IRS-1^Δ386–434 exhibited a significant greater preference for nuclear localization, compared with IRS-1^WT. Higher nuclear localization was also observed when cells expressing IRS-1^WT were incubated with the proteasome inhibitor MG-132. The sequence of DIDI is conserved more than 93% across species, from fish to mammals, as opposed to approximately 40% homology of the entire IRS-1. These findings implicate DIDI as a novel, highly conserved domain of IRS-1, which mediates its cellular localization, rate of degradation, and biological activity, with a direct impact on insulin signal transduction.

A novel, conserved domain of IRS-1 (aa 386-430), mediates its cellular localization, rate of degradation and biological activity, with a direct impact on insulin signaling.

Insulin receptor substrate (IRS) proteins are key players in insulin signal transduction and are the best studied targets of the insulin receptor (reviewed in Refs. 1, 2). IRS proteins contain a conserved pleckstrin homology (PH) domain, located at their amino terminus, that serves to anchor the IRS proteins to membrane phosphoinositides in close proximity to the insulin receptor (3). The PH domain is flanked by a P-Tyr binding (PTB) domain that functions as a binding site to the NPXY motif at the juxtamembrane domain of the insulin receptor (4). The C-terminal region of IRS proteins contains multiple Tyr phosphorylation motifs that act as a signaling scaffold, providing a docking interface for SH2 domain-containing proteins like the p85α regulatory subunit of phosphatidylinositol 3 kinase (PI3K), Grb2, Nck, Crk, Fyn, and SHP-2, which further propagate the metabolic and growth-promoting effects of insulin (5, 6).

The cellular content of IRS proteins is regulated at several levels. These include modulation of IRS proteins gene transcription and regulation of IRS proteins degradation. Early studies proposed that the degradation of IRS proteins is mediated by calpain, a calcium-dependent protease (7). However, more recent studies have shown that insulin-induced degradation of IRS proteins is regulated by the 26S proteasome complex (8, 9, 10, 11). Proteins targeted for degradation by the proteasome are subjected to ubiquitination by a complex containing a ubiquitin-activating enzyme, a ubiquitin-conjugating enzyme, and a ubiquitin-protein ligase (E3) (12). Additional postubiquitination processes, which include binding of chaperons and accessory factors, finalize the process (13).

Ser/Thr phosphorylation promotes protein ubiquitination and degradation (14, 15). Indeed, insulin-simulated Ser/Thr kinases have been implicated in the induction of degradation of IRS proteins (9, 10, 16, 17, 18). Phosphorylation is mediated by mammalian target of rapamycin (mTOR), a downstream effector of PI3K, and by S6K1 (ribosomal protein S6 kinase) (19) independent of the Ras/MAPK pathway (8, 9). Other stimuli, such as hyperosmotic (20) or oxidative stress (21), can also induce degradation of IRS proteins. These effects are insensitive to mTOR and proteasome inhibitors, suggesting the involvement of lysosomal degradation (20, 21). IRS-1 can also be degraded by caspases, such as caspase-10 (22, 23), activated as a result of an apoptotic stimuli.

The structural elements of IRS-1 regulating its degradation start being characterized. The PH and PTB domains are presumably involved (17), as well as the region spanning residues 522–574 (24). However, a comprehensive picture of these elements is missing. In this study, we provide evidence that a short domain between amino acids (aa) 386 and 430 of IRS-1, which we named domain involved in degradation of IRS-1 (DIDI), mediates a novel ubiquitination-independent step in the process of insulin-induced proteasomal degradation of IRS-1. Our studies further implicate DIDI as being involved in regulating nuclear translocation of IRS-1, where it can modulate protein transcription (25). Hence, DIDI seems to regulate both the intracellular localization and the rate of degradation of IRS-1.

Results

Insulin-induced degradation of IRS-1 involves cell-specific signaling pathways

Ser/Thr phosphorylation of IRS proteins triggers their degradation by the proteasome (8, 9, 10, 11). Consistent with these findings, insulin treatment of Fao cells for 3 to 9 h reduced the cellular content of IRS-1 by 47 to 60%, respectively (Fig. 1A). This was accompanied by decreased electrophoretic mobility of IRS-1 due to its Ser/Thr phosphorylation, which precedes its degradation (26). Insulin did not affect the cellular content of the p85α regulatory subunit of PI3K (Fig. 1A) and of annexin II (data not shown), which served as control proteins. Inhibition of the PI3K or the mTOR cascades (by wortmannin and rapamycin, respectively), blocked both the mobility shift and the reduction in cellular content of IRS-1 (Fig. 1A) and restored IRS-1 content to 90–100% of its initial level. Using phosphorylation of Akt and p70S6K as biological readouts of the PI3K and mTOR pathway, respectively, we verified that the activation of these kinases, induced by insulin, was indeed blocked by these inhibitors (Fig. 1A).

Fig. 1.

Insulin-induced degradation of IRS-1 in Fao and CHO-T cells. A, Fao cells were preincubated with wortmannin and rapamycin for 60 min. Cells were then further incubated with 100 nm insulin for the indicated times. Extracts (150 μg) were immunoblotted with the indicated antibodies. A representative experiment out of three is shown. B, CHO-T cells were preincubated with the indicated inhibitors for 60 min, followed by insulin treatment for the indicated times. Extracts were immunoblotted with anti-IRS-1 or anti-p85 antibodies. The intensity of selected bands was determined by densitometry (A and B). C, CHO-T cells overexpressing GFP-Akt or GFP alone (control) were serum starved and subjected to insulin treatment for the indicated times. Extracts were immunoblotted with anti-IRS-1, anti-Akt, or anti-actin antibodies. Data shown are of two independent experiments. D, Serum-starved Fao or CHO-T cells overexpressing IRS-1^WT were treated with or without insulin for 16 h. Extracts (100 μg) were immunoblotted with anti-IRS-1, anti-Shc, anti-annexin II, or anti-Myc antibodies. Data shown are representative of two independent experiments.

In CHO cells that overexpress the insulin receptor (CHO-T cells), 41% degradation of IRS-1 was observed after a 4-h insulin treatment. Interestingly, insulin-induced degradation of IRS-1 in these cells was blocked only by wortmannin but not by rapamycin (Fig. 1B), suggesting that IRS-1 degradation, which involves the mTOR pathway, is not functioning in CHO cells. To further establish the involvement of Akt in mediating insulin-induced Ser phosphorylation (27) and degradation of IRS-1, CHO-T cells were transfected with green fluorescent protein (GFP)-Akt fusion protein or with GFP alone (control), and the extent of IRS-1 degradation in the two cell lines was compared. As shown in Fig. 1C, insulin-induced degradation of IRS-1 occurred much faster in cells overexpressing GFP-Akt, indicating that Akt, a downstream effector of the insulin receptor, indeed mediates IRS-1 degradation.

Unlike IRS-1, Shc, another substrate of the insulin receptor, failed to undergo insulin-stimulated degradation (Fig. 1D). These results suggest that insulin-induced Ser/Thr phosphorylation of IRS-1, mediated by the PI3K/mTOR cascades, regulates the cellular content of IRS proteins. This regulation depends on the cellular context as shown here and in previous studies (9, 28). Furthermore, insulin-induced degradation seems to be specific, because not all substrates of the insulin receptor are subjected to this proteolytic pathway.

A truncated form of IRS-1 (IRS-1^1–430) undergoes insulin-induced degradation comparable with the endogenous IRS-1

The N-terminal segment of IRS-1, including its PH and PTB domains, is essential for targeting IRS-1 to proteasomal degradation (17). We have previously shown that a truncated form of IRS-1 (IRS-1^1–430) that includes these domains undergoes insulin-induced Ser/Thr phosphorylation (26). Hence, IRS-1^1–430 contains enough structural information that enables it to serve as a target for different kinases along the insulin signaling pathway. IRS-1^1–430 was therefore used in search for structural elements involved in IRS-1 degradation. Stable CHO-T cell lines overexpressing Myc-IRS-1^1–430 were generated, and the degradations of the endogenous IRS-1 and of Myc-IRS-1^1–430 were compared (Fig. 2). Insulin induced a decrease in mobility of the endogenous IRS-1 already by 3.5 h, which was followed by degradation of approximately 60% of the protein by 7.5 h. Inclusion of cycloheximide (an inhibitor of protein synthesis) potentiated the effects of insulin and caused degradation of approximately 80% of the endogenous IRS-1 by 7.5 h, whereas addition of the proteasome inhibitor MG-132 blocked IRS-1 degradation. Similarly, a mobility shift of the Myc-IRS-1^1–430 was observed after 3.5 h of insulin treatment (Fig. 2), and this was followed by degradation of the fragment by approximately 40 and approximately 70% after 7.5 and 24 h, respectively, of insulin treatment. Addition of cycloheximide, together with insulin, enhanced the degradation of Myc-IRS-1^1–430, but this process was completely inhibited in the presence of MG-132. These results indicate that Myc-IRS-1^1–430 exhibits a similar pattern of insulin-induced proteasomal degradation as the endogenous IRS-1 protein.

Fig. 2.

Insulin-induced degradation of IRS-1^1–430 in CHO-T cells. Serum-starved CHO-T cells overexpressing Myc-IRS-1^1–430 were preincubated with cycloheximide or MG-132 for 1 h as indicated, followed by insulin treatment for the indicated times. Thereafter, extracts were immunoblotted with anti-IRS-1 (for endogenous IRS-1) or anti-Myc antibodies (for IRS-1^1–430). Densitometry analysis of at least three experiments (mean ± sd) is presented.

To determine whether the degradation of the truncated IRS-1 is also regulated by phosphorylation, CHO-T cells overexpressing Myc-IRS-1^1–430 were treated with different kinase inhibitors, and the degradation of Myc-IRS-1^1–430 was examined. In CHO-T cells, insulin-induced degradation of IRS-1^1–430, like that of IRS-1^WT, was blocked by wortmannin but not by rapamycin, whereas in Fao cells, inhibitors of mTOR and PI3K blocked the degradation of Myc-IRS-1^1–430 (data not shown). These results support the idea that degradation of IRS-1^1–430, like that of IRS-1^WT, depends upon insulin-stimulated activation of cell-specific IRS kinases.

Insulin fails to induce Ser phosphorylation and degradation of IRS-1^1–365

We have previously shown (26) that insulin fails to induce Ser phosphorylation (mobility shift) (Fig. 3A) of a shorter truncated form of IRS-1, IRS-1^1–365. To determine whether IRS-1^1–365 is subjected to degradation, CHO-T and Fao cells overexpressing this domain were treated with insulin for 12 h. Prolonged insulin treatment failed to reduce the mobility (induce phosphorylation) of Myc-IRS-1^1–365 (Fig. 3, B and C). Furthermore, a 12-h incubation of CHO-T cells with insulin caused only a 12% reduction in the cellular content of Myc-IRS-1^1–365, as compared with a 40% reduction in the cellular content of Myc-IRS-1^1–430 (Fig. 3B). Similarly, a 10-h treatment of Fao cells with a combination of insulin and cycloheximide reduced by 65% the cellular content of Myc-IRS-1^1–430 while having no effect on the cellular content of Myc-IRS-1^1–365 (Fig. 3C). These results indicate that aa 365–430 contain a domain that renders IRS-1 sensitive to insulin-induced degradation.

Fig. 3.

Insulin-induced phosphorylation and degradation of IRS-1^1–430 and IRS-1^1–365. A, CHO-T cells were transiently transfected with plasmids encoding IRS-1^1–430 or IRS-1^1–365; 24 h after transfection, cells were serum-starved for 16 h followed by insulin treatment as indicated. Cell extracts were immunoblotted with anti-Myc antibodies. B, Stable lines of CHO-T cells overexpressing IRS-1^1–430 or IRS-1^1–365 were serum starved for 16 h. The cells were then incubated with cycloheximide and/or insulin for 12 h. Cell extracts were immunoblotted with anti-Myc antibody. Densitometry analysis of three experiments (mean ± sd) is presented. Western blottings of one representative experiment are displayed. C, Fao cells were transfected with plasmids encoding IRS-1^1–430 or IRS-1^1–365; 48 h after transfection, cells were starved for 16 h, and the indicated stimuli were applied. Cell extracts were immunoblotted with anti-Myc or anti-actin antibodies, and the intensity of selected bands was determined by densitometry.

Identification of DIDI

To narrow down the domain that regulates IRS-1 degradation, additional truncated forms of IRS-1, Myc-IRS-1^1–386, and Myc-IRS-1^1–407 were generated. Stable lines of CHO-T cells overexpressing these IRS-1 fragments were produced, and their insulin-induced degradation was analyzed. A correlation existed between the length of the IRS-1 fragments: their mobility shift (after a 12-h insulin treatment) and the extent to which they were degraded in response to insulin. The mobility shifts of Myc-IRS-1^1–386 and Myc-IRS-1^1–407 were much smaller than that of Myc-IRS-1^1–430 (data not shown). Similarly, the cellular content of Myc-IRS-1^1–407 and Myc-IRS-1^1–386 were reduced by 30 and 18%, respectively, whereas that of IRS-1^1–430 was reduced by 45% (Table 1). These results indicate that residues between aa 386 and 430 modulate the degradation of IRS-1. This region was therefore named DIDI. Because 43% of the aa of this domain are Ser/Thr (Table 2), we checked whether DIDI undergoes insulin-induced Ser phosphorylation. To this end, the 386- to 430-aa fragment of IRS-1 was expressed in CHO-T cells as a triple-repeat peptide fused to GFP. GFP-IRS-1^{(386–430)x3} underwent Ser phosphorylation under basal conditions, which was elevated after insulin stimulation, as was evident by its reduced mobility (Fig. 4A). Addition of alkaline phosphatase abolished the mobility shift, indicating that this shift could be accounted for by Ser/Thr phosphorylation. Phosphorylation of GFP-IRS-1^{(386–430)x3} was further established when we could show that insulin increased the binding of antibodies directed toward phospho-S408 located to this domain (26, 29). Phosphorylation of S408 involved the PI3K pathway as wortmannin partially inhibited phosphorylation of this site, whereas inhibitors of c-Jun NH2-terminal kinase (SP600125), p38 MAPK (SB202190), and MAPK kinase (PD98059) had no effects (Fig. 4B). These results suggest that DIDI undergoes insulin-stimulated Ser/Thr phosphorylation and could potentially regulate IRS-1 degradation.

Table 1.

Effects of insulin on the extent of degradation of different truncated forms of IRS-1

Construct	Amino acids	% Degradation (12 h)
Full-length IRS-1	1–1233	75 ± 7
Truncated IRS-1	1–430	45 ± 5
		1–407	30 ± 5
		1–386	18 ± 5
		1–365	14 ± 3

Stable clones of CHO-T cells overexpressing Myc-IRS-1^WT, Myc-IRS-1^1–430, Myc-IRS-1^1–407, Myc-IRS-1^1–386, and Myc-IRS-1^1–365 were generated. The extent of degradation of the different IRS-1 forms, after a 12-h insulin treatment, were measured. Shown are results of at least two independent experiments.

Table 2.

Sequence homology of amino acids 386–430 of IRS-1 across species

Accession no.			Protein length (aa)	Species	Sequence homology vs. Mus musculus
								Total sequence	DIDI sequence
1	NP_005535.1	1242	Homo sapiens	88	100
2	NP_034700.2	1231	M. musculus	100	100
3	XP_426682.2	836	Gallus gallus (Chicken)	54	100
4	AAA73572.1	885	Xenopus laevis (Frog)	36	93
5	XP_687702.3	1136	Danio rerio (Zebrafish)	40	98

Multiple sequence alignment of aa 386–430 of human (1), mouse IRS-1 (2), chicken (3), frog (4), and zebrafish (5) were analyzed by CLUSTALW. Asterisks represent identical residues, a colon represents conservative replacement, and aa heterogeneities are underlined:

1) 391 SLSSSSTSGHGSTSDCLFPRRSSASVSGSPSDGGFISSDEYGSSPCDFRSSFRS 435.

2) 386 SLSSSSTSGHGSTSDCLFPRRSSASVSGSPSDGGFISSDEYGSSPCDFRSSFRS 430.

3) 287 SLSSSSTSGHGSTSDCLFPRRSSASVSGSPSDGGFISSDEYGSSPCDFRSSFRS 331.

4) 184 SLSSSSTSGHGSTSDCMCPRRSSASISGSPSDGGFISSDEYGSSPCDFRSSFRS 228.

5) 372 SLSSSSTSGHGSTSDCLFPRRSSASISGSPSDGGFISSDEYGSSPCDFRSSFRS 416.

****************: *******:***********************************

Fig. 4.

Effect of insulin on Ser/Thr phosphorylation of GFP-IRS-1^{(386–430)x3}. A, CHO-T cells were transiently transfected with plasmids encoding GFP or GFP-IRS-1^{(386–430)x3}; 24 h after transfection, cells were serum starved for 16 h followed by a 1-h insulin treatment. Cell extracts were incubated with or without alkaline-phosphatase for 1 h at 37 C and were immunoblotted with anti-GFP antibodies. Results of one out of three similar experiments are shown. B, CHO-T cells overexpressing GFP-IRS-1^{(386–430)x3} were serum starved for 16 h. Cells were treated for 1 h with wortmannin (Wort; 0.1 μm), PD98059 (PD; 25 μm), SB202190 (SB; 10 μm), and SP600125 (SP; 10 μm) followed by a 1-h insulin stimulation. Cell extracts were immunoblotted with anti-GFP or anti-P-S408 antibodies. Densitometry values (mean ± sd) of three independent experiments are presented.

Deletion of DIDI inhibits the degradation of IRS-1

To further establish the role of DIDI in regulating IRS-1 degradation, we constructed an IRS-1 mutant in which DIDI was deleted and stable lines of CHO-T cells overexpressing Myc-IRS-1^Δ386–434 were generated. As shown in Fig. 5A, Myc-IRS-1^Δ386–434 underwent insulin-stimulated Tyr phosphorylation, indicating that it was properly folded to the extent that it serves as a comparable substrate to the insulin receptor as IRS-1^WT. In contrast, the rate of insulin-induced degradation of Myc-IRS-1^Δ386–434 was significantly reduced when compared with the wild-type (WT) protein. This was evident after a 7-h insulin treatment (Fig. 5A) and even more so after a 16-h of insulin treatment (Fig. 5B).

Fig. 5.

Effect of insulin on the phosphorylation and degradation of Myc-IRS-1^Δ386–434. A, CHO-T cells stably overexpressing Myc-IRS-1^WT or Myc-IRS-1^Δ386–434 were serum starved for 16 h before being incubated with 50 μg/ml cycloheximide and/or 100 nm insulin for the indicated times. Thereafter, cell extracts were immunoblotted with the indicated antibodies. Blots are representatives of three independent experiments with similar results. B, CHO-T cells overexpressing IRS-1^WT or IRS-1^Δ386–434 were incubated with MG-132, cycloheximide, or insulin for 16 h in serum-free medium. Cell extracts were immunoblotted with anti-Myc or anti-actin antibodies. Shown are results of one out of three independent experiments. C, CHO-T cells overexpressing Myc-IRS-1^WT, Myc-IRS-1^Δ386–434, or Myc-IRS-1^S408A were serum starved for 16 h followed by treatment ±100 nm insulin and 50 μg/ml cycloheximide for the indicated times. Cell extracts were immunoblotted with anti-Myc antibodies. Densitometry analysis (mean ± sd) of three independent experiments performed in duplicates is shown. Each point represents the content-ratio of each IRS protein in cells treated with cycloheximide ± insulin.

Of note, the rate of insulin-induced degradation of the overexpressed Myc-IRS-1^WT (Fig. 5) was reduced when compared with the endogenous IRS-1 (Fig. 1), due to the continuous synthesis of Myc-IRS-1^WT, driven by the constitutively active cytomegalovirus promoter. To partially overcome this problem, an inhibitor of protein synthesis, cycloheximide, was added to the cells together with insulin. As shown in Fig. 5B, a 16-h treatment with insulin and cycloheximide resulted in a 60% reduction in the cellular content of Myc-IRS-1^WT, whereas the addition of MG-132 completely inhibited this reduction. In contrast, no degradation of Myc-IRS-1^Δ386–434 was observed under these conditions. Interestingly, insulin-induced phosphorylation of S307, implicated in regulating the degradation of IRS-1 (18), was readily detected in Myc-IRS-1^Δ386–434 (Fig. 5A), suggesting that phosphorylation of this site might be necessary, but it is insufficient to promote IRS-1 degradation. Similarly, mutation of S408, a negative regulator of IRS-1 function (26, 29), failed to affect IRS-1 degradation. As shown in Fig. 5C, degradation of IRS-1^S408A was similar to that of IRS-1^WT. Thus, IRS-1 degradation could not be attributed to single phosphorylation of S307 or S408.

In accordance with its resistance to degradation, Myc-IRS-1^Δ386–434 could better maintain its interactions with downstream effectors. For example, activation of protein kinase B (PKB) (Akt) after 7 h of insulin treatment was better maintained in cells expressing Myc-IRS-1^Δ386–434 than in cells expressing Myc-IRS-1^WT (Fig. 5A). Similarly, prolonged (16 h) insulin treatment did not affect the ability of IRS-1^Δ386–434 to bind the p85 regulatory subunit of PI3K (data not shown). To determine whether IRS-1^Δ386–434 provides a physiological advantage to the cells, the induction of glycogen synthesis after insulin treatment was determined. As shown in Figure 6, glycogen synthesis increased by 30% in cells overexpressing IRS-1^Δ386–434 compared with those overexpressing IRS-1^WT.

Fig. 6.

Effects of IRS-1^WT and IRS-1^Δ386–430 on insulin action. Serum-starved CHO-T cells overexpressing IRS-1^WT or IRS-1^Δ386–430 were incubated with or without insulin for 2 h as indicated. Glycogen synthesis was then assessed as described in Materials and Methods. Results (mean ± sd) of three independent experiments performed in duplicates are shown as glycogen synthase activity (I) and as fold induction by insulin (II). *, P < 0.05; **, P < 0.01.

Deletion of DIDI does not impair the ability of IRS-1 to bind murine double minute (Mdm2) or to undergo ubiquitination

Ser phosphorylation of IRS-1 is expected to promote binding of E3 ligases that induce its ubiquitination and degradation (30). To this end, we compared the ability of IRS-1^WT and IRS-1^Δ386–434 to bind Mdm2, an E3 ligase of IRS-1 (30). We could demonstrate that Myc-IRS-1^WT and Myc-IRS-1^Δ386–434 were equipotent in their ability to bind Mdm2 either in the absence or in the presence of insulin (Fig. 7A) and that insulin enhanced Mdm2 binding to the IRS proteins. To compare the extent of their ubiquitination, CHO-T cells expressing Myc-IRS-1^WT or Myc-IRS-1^Δ386–434 were transfected with hemagglutinin (HA)-ubiquitin. The cells were then treated with or without insulin, and the IRS proteins were immunoprecipitated from cell extracts and treated with 1% sodium dodecyl sulfate to avoid nonspecific binding to the immunocomplexes. As shown in Fig. 7B, ubiquitination was evident only when proteasomal protein degradation was inhibited by MG-132. Similar results were obtained when the experiments were carried out without expressing HA-ubiquitin (Fig. 7C). Both Myc-IRS-1^WT and Myc-IRS-1^Δ386–434 underwent ubiquitination to a similar extent, and this process was insulin independent. These results suggest that insulin presumably enhances the degradation of IRS-1 by promoting DIDI phosphorylation at a postubiquitination stage.

Fig. 7.

Effects of insulin on ubiquitination of IRS-1. A, CHO-T cells overexpressing IRS-1^WT or IRS-1^Δ386–434 were transiently transfected with a 5 μg plasmid encoding Mdm2; 24 h after transfection, cells were serum starved for 16 h followed by a 1-h insulin (100 nm) treatment. Cell lysates (0.5–1 mg) were immunoprecipitated with anti-Myc antibody and immunoblotted with anti-Myc or anti-Mdm2 antibodies. Blots are a representative of two independent experiments with similar results. B and C, CHO-T cells overexpressing IRS-1^WT or IRS-1^Δ386–434 were transiently transfected with (B) or without (C) plasmids encoding HA-ubiquitin; 24 h after transfection, cells were treated, as indicated, with 10 μm MG-132 and/or 100 nm insulin for 16 h in serum-free-medium. Cell extracts were immunoprecipitated with anti-IRS-1 antibody and were immunoblotted with anti-HA, anti-Myc, or anti-actin antibodies. Blots of a representative of three independent experiments are shown.

Elimination of DIDI does not protect IRS-1 from apoptosis-induced degradation

To determine whether IRS-1^Δ386–434 is specifically resistant to insulin-induced degradation, we tested its sensitivity to degradation induced by apoptotic process. As shown in Fig. 8A, both IRS-1^WT and IRS-1^Δ386–434 were equally degraded in response to the proapoptotic agent 3,3′-methylene-bis (4-hydroxycoumarin) (dicumarol) (31). Degradation could not be inhibited, and was even somewhat potentiated, in the presence of the proteasome inhibitors MG-132. Conversely, Z-VAD-fmk (benzyloxycarbonyl-valinyl-alaninyl-aspartyl fluoromethyl ketone), a pan-caspase inhibitor (32), partially reversed the effects of dicumarol on IRS-1 degradation (Fig. 8B). Staurosporine, as well as proteasome and caspase inhibitors (33), also promoted the degradation of IRS-1^Δ386–434 (Fig. 8C). Hence, although deletion of DIDI confers resistance from insulin-induced degradation, it fails to protect IRS-1 from degradation triggered by apoptotic stimuli.

Fig. 8.

Effects of dicumarol and staurosporine, as well as proteasome and caspase inhibitors, on the cellular content of IRS-1^WT and IRS-1^Δ386–434. A, CHO-T cells overexpressing IRS-1^WT or IRS-1^Δ386–434 were serum starved for 16 h. Cells were then treated with MG-132 (50 μm) for 1 h, followed by treatment with dicumarol (100 μm) for 4 h. Cell extracts were immunoblotted with anti-Myc or anti-actin antibodies. Densitometry analysis (mean ± sd) of three independent experiments is shown. Western blottings of one representative experiment are displayed. B, CHO-T cells overexpressing IRS-1^WT were serum starved for 16 h. Cells were then treated with Z-VAD-fmk (benzyloxycarbonyl-valinyl-alaninyl-aspartyl fluoromethyl ketone) for 1 h, followed by a 4-h treatment with dicumarol. Cell extracts were immunoblotted with anti-Myc or anti-actin antibodies. Representative blots and densitometry values (mean ± sd) of two independent experiments are shown. C, CHO-T cells overexpressing IRS-1^Δ386–434 were serum starved for 16 h. Cells were then treated with staurosporine for the indicated times. Cell extracts were immunoblotted with anti-Myc or anti-actin antibodies. Shown are results of one out of two independent experiments.

DIDI affects the cellular distribution of IRS-1

It has been previously shown that IRS-1 can localize to the nucleus (34), where it acts to promote gene transcription (35). In accordance with these findings, we could show (Fig. 9) that 38% of CHO-T cells that express IRS-1^WT exhibit nuclear (and cytoplasmic) staining, as opposed to 60% of the cells that exhibited exclusive cytoplasmic staining. Expression of Myc-IRS-1^Δ386–434 increased the incidence of cells expressing both nuclear and cytoplasmic staining almost 2-fold (to 66%). Both constructs were expressed to comparable levels (Fig. 9A). The increase in nuclear localization of Myc-IRS-1^Δ386–434 could be attributed, at least in part, to its resistance to proteosomal degradation, because treatment of cells expressing IRS-1^WT with the proteasome inhibitor MG-132 also increased the incidence of cells expressing nuclear IRS-1 to 57%. Insulin treatment reduced the fraction of cells expressing nuclear IRS-1, whether these were cells expressing IRS-1^WT, IRS-1^Δ386–434, or cells treated with MG-132 (Fig. 9B). These results suggest that the effects of insulin on nuclear localization of IRS-1 are independent of its effects on the degradation of the protein.

Fig. 9.

Effects of insulin or proteasome inhibitor on the cellular content and localization of IRS-1^WT and IRS-1^Δ386–434. A, CHO-T cells stably overexpressing Myc-IRS-1^WT or Myc-IRS-1^Δ386–434 were serum starved for 16 h before being incubated for 7 h with 100 nm insulin and/or 10 μm MG-132 as indicated. Cell extracts were immunoblotted with the indicated antibodies. B, Cells grown in 12-well plates on cover slips were serum starved for 16 h. Cells were then treated as indicated in A and were washed with PBS. After fixation, cells were immunostained for Myc (IRS-1) and 4′,6-diamidino-2-phenylindole as indicated in Materials and Methods. IRS-1 localization was examined under a fluorescence microscope. Samples were analyzed in an unbiased blinded manner. The results are mean ± sem of at least three independent experiments.

Discussion

IRS-1 plays a pivotal role in several signaling cascades (1). Therefore, its degradation is tightly controlled. In the present study, we provide evidence that IRS-1 is degraded by at least two independent processes. One is triggered by chronic insulin treatment, whereas the other is initiated once an apoptotic process sets off. Insulin-stimulated proteasomal degradation of IRS-1 involves a ubiquitination-independent step that requires the presence of a defined Ser/Thr-rich domain, which we named DIDI. Deletion of DIDI (aa 386–434) maintains the ability of IRS-1^Δ386–434 to bind E3 ligases and undergo ubiquitination while rendering it insensitive to insulin-induced degradation. These results suggest that insulin triggers ubiquitination-independent phosphorylation of IRS-1, which depends upon the presence of DIDI and enables IRS-1 to undergo proteasomal degradation. DIDI is selectively involved in insulin-induced degradation, because this domain is not required for caspase-mediated degradation of the IRS-1 protein.

Several lines of evidence support this hypothesis. First, we could show that a truncated form of IRS-1 that contains only aa 1–430 (IRS-1^1–430) undergoes insulin-stimulated degradation almost as effectively as the naive IRS-1, suggesting that IRS-1^1–430 contains enough structural information to enable it to interact with the insulin-induced degradative machinery. IRS-1^1–430 also undergoes insulin-stimulated Ser/Thr phosphorylation (26), which suggests that it serves as a substrate for insulin-induced IRS-1 kinases. This capacity is lost upon further truncation of IRS-1 down to 365 aa. IRS-1^1–365 not only fails to undergo insulin-stimulated Ser/Thr phosphorylation, but it is also resistant to insulin-induced degradation. Hence, the region including aa 365–430 appears to be a critical domain both for the interaction of IRS-1 with IRS kinases and for the induction of IRS-1 degradation.

By generating additional truncated forms of IRS-1, we could pinpoint DIDI to aa 386–430 (Fig. 10). Deletion of DIDI from the naive IRS-1 rendered it insensitive to insulin-induced degradation, most likely due to the loss of critical Ser/Thr residues, the phosphorylation of which is necessary to promote IRS-1 degradation. Indeed, the isolated DIDI domain, when expressed in cells, undergoes insulin-stimulated Ser/Thr phosphorylation. Furthermore, deletion of DIDI reduces the extent of insulin- stimulated Ser/Thr phosphorylation of IRS-1^Δ386–434, assessed by its mobility shift. One of DIDI’s phosphorylation sites is S408, which negatively regulates IRS-1 function (26). However, phosphorylation of S408 might be necessary, but it is insufficient to promote insulin-induced IRS-1 degradation, because mutation of this site alone does not affect the rate of degradation of IRS-1^S408A, thus supporting the notion that phosphorylation of more than one Ser residue is required to promote degradation of IRS-1.

Fig. 10.

A scheme illustrating the different truncated/mutated forms of IRS-1 being used.

The attenuated degradation of IRS-1^Δ386–434 could not be attributed to a general nonspecific impaired activity of the ubiquitin-proteasome system induced by IRS-1^Δ386–434, because overexpression of IRS-1^Δ386–434 did not inhibit degradation of other nonrelevant proteins, such as p53, which undergoes degradation to a similar extent in cells expressing either IRS-1^Δ386–434 or IRS-1^WT (data not shown). Finally, IRS-1^Δ386–434 can interact with IR; undergo insulin-induced Tyr phosphorylation and form complexes with downstream effectors such as PI3K, as effectively as WT IRS-1, again suggesting that its overall conformation does not largely differ from that of IRS-1^WT. As a result, IRS-1^Δ386–434 better propagates insulin action in selected cells, exemplified by the enhanced insulin-induced glycogen synthesis in CHO cells expressing IRS-1^Δ386–434.

Ser/Thr phosphorylation has been implicated as a prerequisite for degradation of a number of proteins (36, 37, 38). This process enables binding of E3 ligases that ubiquitinate the phosphorylated target proteins (39). Indeed, DIDI is highly enriched in Ser/Thr residues, which encompass 43% of this domain. However, contrary to our expectations, deletion of DIDI did not impede the ability of IRS-1^Δ386–434 to bind Mdm2, an E3 ligase of IRS-1 (30), nor did it impair the ability of IRS-1^Δ386–434 to undergo ubiquitination, suggesting that Ser sites that facilitate binding of Mdm2 or other E3 ligases, e.g. cullin 7 (CUL 7) (24) and elongin B and elongin C complex of ubiquitin-ligase (40), are preserved in IRS-1^Δ386–434. This conclusion is supported by the fact that a region spanning residues 522–574 of IRS-1, outside the DIDI domain, has been implicated in promoting degradation of IRS-1, mediated by the CUL 7 E3 ligase (24). Furthermore, IGF-I still promotes degradation of IRS-1 even in CUL7^−/− cells (24). Hence, our findings suggest that DIDI is required for an insulin-induced phosphorylation at a postubiquitination step that facilitates the degradation of IRS-1. This conclusion is supported by the fact that ubiquitination of IRS-1 (either the WT or the Δ386–434) occurs in an insulin-independent manner, as was previously shown (11), in spite of higher levels of Mdm2 binding to the IRS proteins in insulin-treated cells.

Several postubiquitination steps are known to promote protein degradation. These include binding of chaperons, deubiquitinating enzymes or other accessory proteins that facilitate binding of target proteins to the 19S “lid” complex and their introduction into the “core” 20S proteasomal complex (13, 41). Hence, insulin could facilitate binding of chaperons or other accessory factors either directly to DIDI or to other IRS-1 domains after phosphorylation of DIDI by insulin-stimulated IRS kinases.

Although the ability of insulin to promote protein degradation is well established, the number of proteins subjected to this regulatory process is rather limited. In fact, other protein substrates of the insulin receptor, such as the Shc family members, fail to undergo significant insulin-induced degradation. IRS-2 is also a much poorer target for insulin-induced degradation, when compared with IRS-1 (data not shown). In the case of IRS-2, the reduced ability of insulin to promote its degradation could be attributed, at least in part, to the lack of an intact DIDI domain within the IRS-2 structure. Conversely, the DIDI domain is highly conserved across species. We found more than 93% homology of the DIDI region when comparing fish or frog to mouse/human IRS-1, whereas the overall homology of IRS-1 is approximately 40% between these species (Table 2). These findings attest as to the importance of this domain for IRS-1 function.

We can only speculate why insulin preferentially controls the degradation of a selected subset of its substrates (e.g. IRS-1 but not Shc). One possibility is that IRS-1, more than IRS-2, is associated with the growth-promoting functions of insulin in several tissues (42), risking a possible undesired growth when cells are subjected to prolonged insulin stimulation. The ability of insulin to inhibit its growth-promoting functions through the induction of IRS-1 degradation could serve as an additional negative feedback control mechanism used by insulin to turn off its own signals. This mechanism joins other negative feedback control mechanisms used by insulin, which involve Ser/Thr phosphorylation of the IRS proteins that uncouples them from the insulin receptor itself and from their downstream effectors (43).

Although DIDI is essential for insulin-dependent degradation, it does not regulate the caspase-mediated degradation of IRS-1. Although the sites of IRS-1 cleavage by caspases are presently unknown, our results suggest that at least some of these sites are located N-terminally to DIDI (between aa 1 and 385), because a truncated form of IRS-1, IRS-1^1–430, is also subjected to apoptosis-induced proteolysis (data not shown).

Phospho-Ser/Thr motifs have been recently implicated as mediators of nuclear translocation of proteins lacking nuclear localization signal (44). Furthermore, several studies suggest that IRS-1 can translocate to the nucleus, where it interacts with nuclear proteins, including the upstream binding factor 1 (45) and the estrogen receptor (46) to support selected gene transcription (25). Although phosphorylation of DIDI does not seem to be involved in promoting nuclear translocation of IRS-1 (because IRS-1^Δ386–434 exhibits higher prevalence of nuclear localization than IRS-1^WT), DIDI might mediate the subsequent degradation of IRS-1 by nuclear proteasome. This hypothesis is supported by the fact that inclusion of proteasomal inhibitors doubles the prevalence of cells having nuclear WT IRS-1. However, addition of insulin reduces the cellular localization of both IRS-1^WT and IRS-1^Δ386–434, indicating that DIDI is not involved in nuclear exclusion of IRS-1 induced by insulin.

In summary, our studies suggest that IRS-1 undergoes ubiquitination in an insulin-independent manner. Insulin promotes the degradation of IRS-1 by inducing the phosphorylation of a highly conserved Ser/Thr-rich domain, unique to IRS-1, which we named DIDI. The phosphorylation of DIDI negatively regulates IRS-1 function: limits its nuclear accumulation and facilitates its degradation by the proteasome. Our studies further suggest that DIDI is not required to promote degradation of IRS-1 as a result of an apoptotic process. Hence, IRS-1 degradation is controlled by at least two signaling pathways, one involves the insulin signaling cascade with DIDI as a target that is proteasome dependent, whereas the other involves caspases cascade. Further studies are required to identify the binding partners of DIDI and the way they regulate the degradation of IRS-1.

Materials and Methods

Materials

Polyclonal p70S6K, PKB, and ERK1/2 antibodies and monoclonal phospho-specific ERK1/2 (Thr 183, Tyr 185) antibodies were kindly provided by R. Seger (The Weizmann Institute of Science). Mdm2 antibody (4B2, 2A10) was kindly provided by M. Oren (The Weizmann Institute of Science). The following materials were from commercial sources: protease inhibitor cocktail, wortmannin, cycloheximide, dicumarol, SB203580, Collagenase Type II, insulin-free BSA, insulin (30 USP/mg), and puromycin (Sigma Chemicals Co., St. Louis, MO); MG-132, PD98059, SB202190, and rapamycin (Calbiochem, La Jolla, CA); Lipofectamine and OptiMEM (GIBCO-BRL, Grand Island, NY); pGEM-T vector system (Promega, Madison, WI); Nucleofector kit T (Amaxa Biosystems, Walkersville, MD); Bio-X-ACT DNA polymerase (Bioline, Taunton, MA); monoclonal Myc (9E10) antibody and protein G-PLUS Agarose beads (Santa Cruz Biotechnology, Inc., Santa Cruz, CA); polyclonal anti-IRS-1 pSer312 and SP600125 (BioSource, Carlsbad, CA); monoclonal P-Tyr (PY-20) and monoclonal anti-annexin II antibodies (Transduction Labs, Lexington, KY); alkaline phosphatase (New England Biolabs, Inc., Beverly, MA); polyclonal IRS-1 and p85α-PI3K antibodies (Upstate, Lake Placid, NY); polyclonal phospho-specific PKB (Ser 473) (Cell Signaling, Inc., Beverly, MA); monoclonal ubiquitin antibody (Covance, Berkeley, CA); and D-[U-¹⁴C] glucose (Amersham, Buckinghamshire, UK).

Cell cultures

Rat hepatoma (Fao) cells and CHO cells overexpressing the insulin receptor (CHO-T cells) (47) were grown in RPMI1640 or F-12 medium, respectively, supplemented with 10% fetal calf serum. Twenty-four hours after cell culturing, at 70–80% confluence, cells were deprived of serum for 16 h and were then subjected to the indicated treatments. After treatment, cells were washed three times with ice-cold PBS and were harvested in buffer A [25 mm Tris-HCl, 2 mm sodium orthovanadate, 0.5 mm EGTA, 10 mm NaF, 10 mm sodium pyrophosphate, 80 mm β glycerophosphate, 25 mm NaCl, 1% Triton X-100, and protease inhibitor cocktail, 1:1000 (pH 7.4)]. Cell lysates were centrifuged at 12,000 × g for 15 min at 4 C. Supernatants were collected, mixed with 5× Laemmli sample buffer, resolved by 8% SDS-PAGE, and then subjected to Western blotting with the indicated antibodies. Quantification of the immunoblots was done using the NIH Image 1.62b7.

Generation of truncated/mutated Myc-tagged IRS-1 isoforms

Truncated Myc-tagged IRS-1 isoforms were generated by PCR using pcDNA3-Myc-IRS-1 as a template and the following sets of primers (start and stop codons are in bold and restriction sites are underlined). The sense (S) primer: 5′-CAG GAT CCG CAT ATG GAA CAA AAG CTC-3′ was common for all isoforms. Antisense (AS) primers were as follows: 5′-GAG AAT TCA TCA GGG ACT AGA ACC ATA-3′ for MycIRS-1^1–430, 5′-GAG AAT TCA TCA GCG CCT CGG GAA GAG-3′ for MycIRS-1^1–407, and 5′-GAG AAT TCA TCA ACT CAC TGG GCT GG-3′ for MycIRS-1^1–386. The PCR products were ligated into pGEM-T construct (pGEM-T kit). The inserts were excised using BamHI and EcoRI and were religated into pcDNA3 at the same restriction sites. MycIRS-1^1–365 was subcloned from pcDNA3-Myc-IRS-1 by PCR with the following primers: S, 5′-AA GCT TAA GAT ATC GAT CAT ATG-3′ and AS, 5′-TTA GTT GAG TGG GGG GTG CAG CCT-3′. The PCR product, encoding MycIRS-1^1–365, was first ligated into a pGEM-T construct, which was then digested with EcoRV and NotI at sites located within the pGEM-T plasmid. The insert was excised from pGEM-T and was religated into pcDNA3 at the same restriction sites.

A Myc-tagged mouse IRS-1^Δ386–434 was generated by addition of two NheI sites at nucleotides 386 and 434 of mIRS-1 using two PCRs with the following primers (start and stop codons are in bold and restriction sites are underlined): N-terminal MycIRS-1 (1–1158 bp): S, 5′-CAG GAT CCG CAT ATG GAA CAA AAG CTC-3′ and AS, 5′-CT ACT AGA TGA CAG GCT AGC TGG GCT GGT GGC TGA AGG-3′; C-terminal MycIRS-1 (1302–3702 bp): S, 5′-GGT TCT AGT CCC TGC GCT AGC CGA AGT TCC TTC CGC-3′ and AS, 5′-CTA TTG ACG ATC CTC TGG CTG C-3′. The PCR product encoding the N-terminal MycIRS-1 was first ligated into pGEM-T plasmid. It was digested with BamHI and NotI (located in the pGEM-T plasmid) and was ligated into pcDNA3 at the same restriction sites. The PCR product encoding the C-terminal part was ligated into pGEM-T plasmid. It was then ligated into pcDNA3-N-terminal MycIRS-1 with Nhe-I and NotI (located in the pGEM-T plasmid). The predicted sequence, generated as a result of the deletion, was confirmed by sequencing.

Generation of GFP-myc-IRS-1^{(386–430)x3}

GFP-myc-IRS-1^{(386–430)x3} was generated by three PCRs, amplifying the coding sequence of aa 386–430 of mIRS-1, with different restriction sites at the edges of the primers, using pcDNA3-Myc-IRS-1 as a template. The following sets of primers were used sequentially (start and stop codons are in bold and restriction sites are underlined): 1) S, 5′-TCC GGA ATG GAA CAA AAG CTC ATC TCA GAA G-3′ and AS, 5′-GGA AGA TCT GCA GGG ACT AGA ACC ATA-3′; 2) S, 5′-GGA AGA TCT AGC CTG TCA TCT AGT AGT ACC-3′ and AS, 5′-CGC GGC CGC GCA GGG ACT AGA ACC ATA-3′; and 3) S, 5′-GCG GCC GCG AGC CTG TCA TCT AGT AGT ACC-3′ and AS, 5′-CCG GAA TTC CTA GCA GGG ACT AGA ACC ATA-3′. The first PCR product, encoding Myc-IRS-1^386–430, was ligated into pGEM-T. Next, BglII site within the Myc sequence was knocked-out by a silent mutation using a QuickChange kit (Stratagene, La Jolla, CA) according to the manufacturer’s instructions with the following primers (mutated nucleotides are in bold): S, 5′-G CTC ATC TCA GAA GAA GAC TTG AAT GCT AGC CTG TCA TC-3′ and AS, 5′-GA TGA CAG GCT AGC ATT CAA GTC TTC TTC TGA GAT GAG C-3′. The second PCR product was ligated into pGEM-T, digested with BglII and NotI, and religated into a pGEM-T that contained the first PCR product ending up with pGEM-T-mycIRS-1^{(386–430)x2}. The third PCR product was ligated into a pGEM-T, digested with NotI and NdeI, and religated into a pGEM-T containing mycIRS-1^{(386–430)x2} to generate pGEM-T-mycIRS-1^{(386–430)x3}. Finally, mycIRS-1^{(386–430)x3} was digested from the pGEM-T-mycIRS-1^{(386–430)x3} with BspEI and EcoRI and ligated into a pEGFP-C1 vector (Clontech, Mountain View, CA) to generate pEGFP-C1mycIRS-1^{(386–430)x3} that encodes the GFPmycIRS-1^{(386–430)x3} peptide.

Transfection of Fao and CHO cells

Fao cells were transiently transfected using the Nucleofector system (Amaxa Biosystems). Cells were then transferred into RPMI1640 medium in a six-well plate; 24 h after transfection, the cells were split according to the experiment requirement. After additional 24 h, the cells were starved for 16 h, and the indicated stimuli were applied. CHO-T cells, at 70–80% confluence, were transiently transfected with Lipofectamine (Life Technologies, Carlsbad, CA). After a 3- to 5-h incubation at 37 C, the medium was replaced by F-12 medium supplemented with 10% fetal calf serum; 24 h after transfection, the cells were starved for 16 h, and the indicated stimuli were applied.

Establishment of stable cell lines

CHO-T cells, grown in 60-mm plates at 60–70% confluence, were cotransfected with 2 μg of the target plasmid (e.g. pcDNA3-IRS-1) and helper plasmid (0.20 μg, pBabe-puro) using Lipofectamine as described above; 24 h after transfection, the cells were treated with10 μg/ml puromycin. The puromycin-containing medium was replaced every 3–4 d. After 2 wk of selection, stable cell lines were either frozen or were further cultured for the indicated experiments in puromycin-containing medium.

In vivo ubiquitination

CHO-T cells overexpressing IRS-1^WT or IRS-1^Δ386–434 were treated with 10 μm MG-132 and/or insulin in serum-free medium, as indicated. Cells were then washed three times with ice-cold PBS and were lysed in 200 μl of 1% sodium dodecyl sulfate in TBS [150 mm NaCl and 10 mm Tris-HCl (pH 7.5)] by boiling twice for 5 min after vigorous vortexing. Next, 400 μl of TBS containing 1.5% Triton X-100 was added to the lysates. After centrifugation for 15 min at 12,000 × g at 24 C, the supernatants were collected for immunoprecipitation with anti-IRS-1 antibodies. Samples were resolved by SDS-PAGE and were Western blotted with the indicated antibodies.

Alkaline phosphatase treatment

Cell lysates were prepared in 100 mm NaCl, 50 mm Tris-HCl, 10 mm MgCl₂, and 1 mm dithiothreitol (pH 7.9). After three freeze-thaw cycles, the cells were centrifuged at 12,000 × g for 15 min at 4 C, and the supernatants were collected. Aliquots (300 μg) were incubated with 30 units of alkaline phosphatase for 1 h at 37 C. After incubation, samples were resolved by SDS-PAGE and were Western blotted with anti-GFP or P-Tyr antibodies.

Immunoprecipitation

Protein G-PLUS Agarose beads in 0.1 m Tris-HCl (pH 8.5) were incubated with monoclonal Myc antibodies for 2 h at 4 C. Extracts derived from frozen cell lysates (0.5–1.0 mg protein) were added and further incubated for 16 h at 4 C. Immunocomplexes were then washed three times with buffer A, twice with ice-cold PBS, and were then resuspended in 70 μl 1× Laemmli sample buffer. Immunocomplexes were resolved by SDS-PAGE and were Western blotted with the indicated antibodies. The same protocol was used for coimmunoprecipitation experiments, save for the fact that the lysates were prepared from freshly isolated cells in buffer A containing 0.5% Nonidet P-40 instead of 1% Triton X-100.

Glycogen synthesis

CHO cells (10⁵ per well) were seeded in a 24-well plate. After 16 h of serum starvation, cells were washed twice with buffer J (1 mm CaCl₂, 20 mm HEPES, 140 mm NaCl, 5 mm KCl, and 2.5 mm MgCl₂) and were incubated for 2 h at 37 C in 500 μl buffer J containing 2.5 μm glucose, 0.2 μCi D-[U-¹⁴C] glucose, and 100 nm insulin as indicated. Cells were then washed twice with cold PBS and were extracted with preheated (55 C) 2 n NaOH, followed by a 10-min incubation at 37 C. Glycogen (1% final concentration) was then added to the extract. Cell extracts were spotted onto Whatman paper discs and were washed intensively for with 70% ethanol followed by a wash in acetone. Papers were then transferred into vials containing scintillation liquid, and the ¹⁴C radioactivity was counted. Nonspecific counts were subtracted from sample measurements.

Immunofluorescence microscopy

Cellular localization of IRS-1 was analyzed in CHO-T cells overexpressing either Myc-IRS-1^WT or Myc-IRS-1^Δ386–434. Cells were grown in 12-well plates on cover slips, and 24 h later, they were serum starved for 16 h. Treatments, as indicated, were carried out for 7 h, and the cells were then rinsed three times with PBS supplemented with 0.1 mm CaCl₂ and 1 mm MgCl₂. Cells were fixed in 4% paraformaldehyde for 15 min, rinsed three times with PBS supplemented with 100 mm glycine, and permeabilized with 0.1% Triton X-100 in PBS. Cells were incubated in blocking solution (0.1% Triton X-100, 0.1% glycine, and 1% normal goat mouse in PBS) for 1 h followed by incubation for 1 h with primary antibodies (anti-Myc D9) in blocking solution. Cells were extensively washed and incubated with specific secondary antibodies coupled to Cy3 together with 4′,6-diamidino-2-phenylindole (100 ng/ml) in blocking solution. Cells were then extensively washed with PBS and mounted for over night. Cells were examined under fluorescence microscope. Samples were analyzed in an unbiased blinded manner.