Role of the Insulin Receptor Substrate 1 and Phosphatidylinositol 3-Kinase Signaling Pathway in Insulin-Induced Expression of Sterol Regulatory Element Binding Protein 1c and Glucokinase Genes in Rat Hepatocytes

Adenovirus vectors encoding a dominant-negative mutant of PI 3-kinase (AxCAΔp85) (20), a constitutively active form of PI 3-kinase (AxCAMyr-p110) (21), and an Akt mutant in which Thr³⁰⁸ and Ser⁴⁷³ are replaced by alanine (AxCAAkt-AA) (22) were described previously. A control adenovirus encoding β-galactosidase was provided by I. Saito. Complementary DNAs encoding NH₂-terminal portions of human IRS-1 (amino acids 2–401) (IRS-1N) and mouse IRS-2 (amino acids 2–430) (IRS-2N) tagged at their NH₂-termini with the hemagglutinin epiptope (HA) were synthesized with the use of PCR. Adenovirus vectors encoding HA-tagged IRS-1N or IRS-2N (AxCAIRS-1N and AxCAIRS-2N, respectively) were generated with the use of an adenovirus expression kit (Takara, Tokyo) as described previously (22).

Hepatocytes were isolated from the liver of male Wistar rats (body mass 150–250 g) with the use of collagenase (23). The isolated hepatocytes were seeded at a density of 2.6 × 10⁶ cells per 60-mm dish (coated with rat collagen-I) in medium A (medium 199 containing Earle’s salt, penicillin [100 units/ml], and streptomycin [100 μg/ml]) supplemented with 5% (vol/vol) fetal bovine serum (FBS), 100 nmol/l dexamethasone, 1 nmol/l insulin, and 100 nmol/l triiodothyronine. At 8–12 h after seeding, hepatocytes were infected with adenovirus vectors by incubation for 2 h at 37°C in medium A supplemented with 5% FBS and the indicated amounts of virus. The virus-containing medium was then aspirated, and the cells were incubated for an additional 6–8 h in medium A containing 5% FBS. For Northern blot analysis of SREBP-1c mRNA, cells infected (or not) with adenovirus vectors were incubated in medium A (serum free) for 16 h and then treated with 100 nmol/l insulin in the same medium for 6 h; for analysis of GK mRNA, the cells (infected or not) were incubated with 100 nmol/l insulin in medium A for 18 h. After insulin treatment, total RNA was extracted from the cells with the use of an RNeasy kit (Qiagen), and portions (20 μg) of the isolated RNA were subjected to Northern analysis with ³²P-labeled probes prepared with a BcaBEST labeling kit (Takara). Probes for SREBP-1c (24) and GK (25) mRNAs were provided by H. Shimano and T. Noguchi, respectively. Autoradiograms of Northern blots were visualized with a BAS2000 image analyzer (Fuji, Tokyo).

For immunoblot analysis and assays of PI 3-kinase and Akt activity, hepatocytes infected (or not) with adenovirus vectors were incubated in medium A (serum free) for 16 h, treated (or not) with 100 nmol/l insulin for the indicated times, and lysed. The cell lysates were then subjected to immunoprecipitation or immunoblot analysis. Antibodies to glycogen synthase kinase (GSK)-3β were obtained from Transduction Laboratories. Antibodies to IRS-1, IRS-2, Akt2, or phosphotyrosine (4G10) were obtained from Upstate Biotechnology, and those specific for phospho-Ser⁴⁷³ forms of Akt or phosphorylated forms of GSK3 were from Cell Signaling. PI 3-kinase activity in immunoprecipitates prepared with antibodies to IRS-1 or IRS-2 was assayed as described (20), and Akt activity in immunoprecipitates with antibodies to Akt2 was assayed with the use of Crosstide peptide as a substrate as described (26).

In primary cultured hepatocytes, insulin induced a marked increase in the amount of PI 3-kinase activity associated with IRS-1 or IRS-2 immunoprecipitates (Fig. 1A). Infection of the cells with an adenovirus vector encoding a dominant-negative mutant of PI 3-kinase (AxCAΔp85) resulted in almost complete inhibition of the insulin-induced increase in the amount of PI 3-kinase activity that coprecipitated with IRS-1 or IRS-2. Cells that had been infected (or not) with AxCAΔp85 were then incubated in the absence or presence of insulin, after which total RNA was isolated and subjected to Northern blot analysis with a fragment of SREBP-1c cDNA as a probe. Although this probe reacts with both SREBP-1a and SREBP-1c mRNAs, insulin increases the amount of SREBP-1c mRNA, but not that of SREBP-1a mRNA, in primary cultured hepatocytes (17); the insulin-induced increase in the hybridization signal intensity apparent with this probe for these cells therefore reflects the increase in the abundance of SREBP-1c mRNA. Whereas insulin induced a marked increase in the amount of SREBP-1c mRNA in noninfected hepatocytes (Fig. 1B), this effect was inhibited in a multiplicity of infection (MOI)-dependent manner in the cells infected with AxCAΔp85 (Fig. 1B). Given that SREBP-1c is thought to regulate expression of the GK gene (15), we also examined the effect of Δp85 on the insulin-induced expression of this gene. The increase in the amount of GK mRNA triggered by insulin that was apparent in noninfected hepatocytes was inhibited in an MOI-dependent manner in the cells infected with AxCAΔp85 (Fig. 1C). Infection of hepatocytes with AxCAMyr-p110, an adenovirus vector that encodes a constitutively active form of PI 3-kinase, resulted in an increase in the abundance of both SREBP-1c and GK mRNAs in the absence of insulin (Fig. 1B and C). These results indicate that PI 3-kinase is required for the insulin-induced expression of SREBP-1c and GK genes.

The NH₂-terminal regions of IRS proteins contain pleckstrin homology (PH) and phosphotyrosine binding (PTB) domains, both of which have been suggested to contribute to the interaction between IRS and the insulin receptor (27,28). We investigated the effects of an NH₂-terminal portion of IRS-1 (IRS-1N) that contains the PH and PTB domains on insulin-induced tyrosine phosphorylation of IRS-1 and IRS-2. Infection of hepatocytes with an adenovirus vector encoding IRS-1N (AxCAIRS-1N) resulted in an MOI-dependent increase in the abundance of the mutant protein (Fig. 2A). The expression of IRS-1N did not affect the abundance of IRS-1 and IRS-2 in the cells (Fig. 2B and C). Expression of this mutant protein did, however, result in an MOI-dependent inhibition of insulin-induced tyrosine phosphorylation of IRS-1. In contrast, tyrosine phosphorylation of IRS-2 was slightly increased by IRS-1N. IRS-1N did not affect insulin-induced tyrosine phosphorylation of the insulin receptor, and a control virus encoding β-galactosidase had no effect on the insulin-induced tyrosine phosphorylation of IRS-1 or IRS-2 at an MOI of up to 30 plaque-forming units (PFUs)/cell (data not shown). The insulin-induced increases in the amount of PI 3-kinase activity associated with IRS-1 and IRS-2 were inhibited by >90% and increased by ∼30%, respectively, as a result of infection of hepatocytes with AxCAIRS-1N at an MOI of 10 PFUs/cell (Fig. 2D). These results thus indicate that IRS-1N inhibits the association of PI 3-kinase with IRS-1 by attenuating the tyrosine phosphorylation of endogenous IRS-1.

On the basis of our observation that IRS-1N selectively inhibited the IRS-1–PI 3-kinase signaling pathway, we next investigated the contribution of this pathway to the insulin-induced expression of the SREBP-1c gene with the use of IRS-1N. Infection of hepatocytes with AxCAIRS-1N resulted in an MOI-dependent inhibition of the insulin-induced increase in the abundance of SREBP-1c mRNA (Fig. 3A); this effect of insulin was almost completely blocked at an MOI of 10 PFUs/cell, a virus dose sufficient to inhibit the insulin-induced increase in IRS-1–associated PI 3-kinase activity by >90% (Fig. 2D). The insulin-induced increase in the amount of GK mRNA was also similarly inhibited by IRS-1N in a dose-dependent manner (Fig. 3B). Infection of the cells with a control adenovirus encoding β-galactosidase did not affect the abundance of SREBP-1c or GK mRNAs at an MOI of up to 30 PFUs/cell (data not shown). In contrast, the insulin-induced phosphorylation of Akt, which is thought to be essential for the activation of this kinase, was not affected by expression of IRS-1N, even at an MOI of 10 PFUs/cell, as assessed by immunoblot analysis with antibodies specific for phosphorylated forms of the enzyme (Fig. 3C). These results suggest that the insulin-induced expression of SREBP-1c and GK genes depends on the IRS-1–PI 3-kinase pathway, whereas inhibition of this signaling pathway does not prevent the insulin-induced phosphorylation of Akt.

We next investigated the effects of a dominant-negative mutant of Akt (Akt-AA), which has been shown to inhibit the insulin-induced activation of Akt as well as various metabolic actions of insulin (21,22,29), on insulin-induced expression of SREBP-1c and GK genes. Hepatocytes were infected (or not) with an adenovirus vector encoding Akt-AA (AxCAAkt-AA), incubated in the absence or presence of insulin, and then lysed. Cell lysates were subjected to immunoprecipitation with antibodies to Akt2, and the resulting precipitates were subjected to immunoblot analysis with antibodies to Akt2 or phosphorylated Akt. Because the Akt-AA cDNA was constructed from rat Akt1 cDNA, the Akt2-specific antibodies did not recognize the Akt-AA protein. Expression of Akt-AA resulted in inhibition of the insulin-induced phosphorylation of Akt2, as revealed both by immunoblot analysis with the antibodies specific for the phosphorylated protein and by the mobility shift detected with antibodies to Akt2 (Fig. 4A), the latter of which also reflects phosphorylation status. Moreover, infection of the cells with AxCAAkt-AA at an MOI of 20 inhibited insulin-induced activity of endogenous Akt2 by ∼90% (Fig. 4B). The insulin-induced phosphorylation of GSK3β, which is a direct substrate of Akt (30), was also inhibited by Akt-AA (Fig. 4C). These results suggest that Akt-AA exerts a dominant-negative effect on the insulin-induced activation of Akt as well as on signaling downstream of this protein in primary cultured hepatocytes. However, expression of Akt-AA did not affect the insulin-induced increases in the abundance of SREBP-1c and GK mRNAs; rather, Akt-AA increased the amounts of both mRNAs in a dose-dependent manner (Fig. 5).

The abundance of IRS-1 and IRS-2 was not affected by Akt-AA (Fig. 6A and B). Although insulin-induced tyrosine phosphorylation of IRS-1 was enhanced in cells expressing Akt-AA, that of IRS-2 was slightly decreased. Furthermore, the amount of PI 3-kinase activity associated with IRS-1 was increased, whereas that associated with IRS-2 was slightly reduced in cells expressing Akt-AA (Fig. 6C). These results thus suggest that inhibition of Akt activity promoted the association of PI 3-kinase with IRS-1 by increasing the extent of tyrosine phosphorylation of IRS-1, consistent with the recent observation that Akt contributes to a negative feedback pathway directed at IRS-1 (31).

Finally, we investigated the effects of an NH₂-terminal portion of IRS-2 that contains the PH and PTB domains (IRS-2N) on insulin action. Infection of hepatocytes with an adenovirus vector encoding IRS-2N (AxCAIRS-2N) resulted in an MOI-dependent increase in the abundance of the mutant protein (Fig. 7A). Expression of IRS-2N resulted in a dose-dependent inhibition of the insulin-induced tyrosine phosphorylation of both IRS-1 and IRS-2 without an effect on the abundance of these proteins (Fig. 7B and C). However, the tyrosine phosphorylation of IRS-1 was inhibited more efficiently by IRS-2N than was that of IRS-2; whereas tyrosine phosphorylation of IRS-2 was not affected in cells infected with AxCAIRS-2N at an MOI of 1 PFU/cell, that of IRS-1 was markedly inhibited.

Infection of hepatocytes with AxCAIRS-2N at an MOI of 1 PFU/cell, a virus dose sufficient to inhibit insulin-induced tyrosine phosphorylation of IRS-1, did not inhibit insulin-induced phosphorylation of Akt (Fig. 8A). However, in cells infected with AxCAIRS-2N at an MOI of 10 or 30 PFUs/cell, the insulin-induced phosphorylation of Akt was inhibited in a dose-dependent manner. Insulin-induced phosphorylation of GSK3β was inhibited by the expression of IRS-2N (Fig. 8B), and the dose-response relation of inhibition was similar to that apparent for insulin-induced phosphorylation of Akt and of IRS-2. The insulin-induced increases in the abundance of SREBP-1c and GK mRNAs were almost completely inhibited by AxCAIRS-2N at an MOI of 1 PFU/cell (Fig. 8C and D), a virus dose sufficient to inhibit tyrosine phosphorylation of IRS-1 but insufficient to inhibit that of IRS-2.

The NH₂-terminal portion of IRS is thought to participate in the interaction of IRS with the insulin receptor (27,28). We have now shown that an NH₂-terminal fragment of IRS-1 that contains the PH and PTB domains (IRS-1N) prevented the insulin-induced tyrosine phosphorylation of IRS-1, but not that of IRS-2, when expressed in primary cultured hepatocytes. An equivalent fragment of IRS-2 (IRS-2N) prevented the insulin-induced tyrosine phosphorylation of both IRS-1 and IRS-2, although that of IRS-1 was inhibited more efficiently than that of IRS-2. These results are consistent with the notion that the affinity of the NH₂-terminal portion of IRS-2 for the insulin receptor is higher than that of the equivalent portion of IRS-1. Sawka-Verhelle et al. (32) have identified a region of IRS-2 that has no counterpart in IRS-1. This region, termed the kinase regulatory loop binding (KRLB) domain, also binds directly to the insulin receptor (32). It is thus possible that, whereas IRS-1 binds to the insulin receptor solely through its NH₂-terminal portion, IRS-2 interacts with the receptor through both the NH₂-terminal portion and the KRLB domain. Although it remains unclear why IRS-1N slightly enhances the tyrosine phosphorylation of IRS-2 induced by insulin, overexpression of the KRLB domain inhibited the binding of IRS-2 to the insulin receptor and, concomitantly, enhanced that of IRS-1 (33). IRS-1 and IRS-2 might therefore compete with each other for binding to the insulin receptor; therefore, a decrease in the binding of one of these proteins to the receptor may result in a reciprocal increase in the binding of the other.

We have also shown that the insulin-induced increases in the abundance of SREBP-1c and GK mRNAs, both of which are mediated by PI 3-kinase, were blocked in cells in which the insulin-induced tyrosine phosphorylation of IRS-1 (but not that of IRS-2) was inhibited by expression of either IRS-1N or a small amount of IRS-2N. These results suggest that the IRS-1–PI 3-kinase pathway is essential for insulin-induced expression of the SREBP-1c and GK genes, whereas the IRS-2–PI 3-kinase pathway alone is not sufficient to mediate these effects of insulin. In contrast, the insulin-induced phosphorylation of Akt was not affected by IRS-1N, indicating that IRS-1 is responsible for some, but not all, of the PI 3-kinase–dependent actions of insulin in hepatocytes. Recent studies have revealed that defects in IRS-1– or IRS-2–dependent signaling result in distinct effects on cellular functions. For example, ablation of the IRS-1 gene in immortalized brown preadipocytes or embryonic fibroblasts markedly impaired the ability of these cells to acquire adipocyte characteristics, whereas a deficiency of IRS-2 in these cells had little effect on this ability (12,13). In contrast, the attenuation of insulin-stimulated glucose uptake apparent in IRS-2–deficient brown adipocytes was partly reversed by expression of IRS-2 but not by that of IRS-1 (11). These data, together with our present results, suggest that, despite their structural and biochemical similarities, IRS-1 and IRS-2 are not completely redundant, but rather are selectively linked to specific actions of insulin. The mechanisms by which each IRS protein contributes differentially to various actions of insulin in the same cells remain to be determined. The relative abundance and subcellular localizations of IRS-1 and IRS-2 as well as their interactions with specific molecules and the time courses of their upregulation or downregulation may be important determinants of their distinct contributions to insulin actions.

Shimomura et al. (34) showed that the expression of IRS-2 and the phosphorylation of Akt are markedly reduced, whereas the expression of SREBP-1c is increased in the liver of lipodystrophic or leptin-deficient (ob/ob) mice, both of which exhibit whole-body insulin resistance and chronic hyperinsulinemia. These investigators concluded that insulin resistance (impaired activation of Akt) and hypersensitivity to insulin (increased expression of SREBP-1c) exist concomitantly in the liver of these mice. This apparently paradoxical phenomenon may be explained if the IRS-1–PI 3-kinase and IRS-2–PI 3-kinase pathways diverge in hepatocytes, with the former pathway being more closely linked to regulation of the SREBP-1c gene, as demonstrated in the present study.

Expression of a dominant-negative mutant of Akt (Akt-AA) enhanced the insulin-induced tyrosine phosphorylation of IRS-1 in hepatocytes. Evidence suggests that PI 3-kinase–Akt signaling contributes to serine phosphorylation of IRS-1 and, in turn, attenuates tyrosine phosphorylation of IRS-1 (31). Expression of Akt-AA thus likely disrupted an Akt-mediated negative feedback pathway leading to IRS-1, resulting in upregulation of tyrosine phosphorylation of IRS-1. Given that Akt-AA did not augment IRS-2–PI 3-kinase signaling, Akt-mediated negative feedback does not appear to be directed at IRS-2.

Insulin induced the expression of SREBP-1c and GK genes in cells expressing Akt-AA. Moreover, the effects of insulin on the expression of these genes were actually enhanced in these cells. These results are also consistent with the observation that the expression of SREBP-1c was increased, whereas the insulin-induced activation of Akt was inhibited in the liver of insulin-resistant mice (34). Although expression of an active form of Akt has been shown to increase the abundance of SREBP-1c mRNA in hepatocytes (18), it is likely that Akt signaling, at least that mediated through Akt2, is not required for the insulin-induced expression of SREBP-1c and GK genes. The enhanced tyrosine phosphorylation of IRS-1, but not of IRS-2, as well as the increased expression of SREBP-1c and GK genes induced by insulin in Akt-AA–expressing cells, also support our hypothesis that the IRS-1–PI 3-kinase pathway plays a major role in mediating the insulin-induced expression of these genes.

Although IRS-1N did not prevent insulin-induced phosphorylation of Akt, expression of IRS-2N at a level sufficient to inhibit insulin-induced tyrosine phosphorylation of both IRS-1 and IRS-2 also blocked insulin-induced phosphorylation of Akt. One possible explanation for these results is that the IRS-2–PI 3-kinase pathway, but not the IRS-1–PI 3-kinase pathway, is responsible for insulin-induced phosphorylation of Akt. This notion is consistent with the observation that long-term exposure of hepatocytes to insulin results in marked decreases in both the amount of IRS-2 and the extent of insulin-induced phosphorylation of Akt, with no effect on the amount of IRS-1 (34). Another possibility is that the insulin-induced phosphorylation of Akt is mediated through both IRS-2– and IRS-1–dependent pathways and that the inhibition of only one pathway is not sufficient to prevent this effect. Regardless, the IRS-2–PI 3-kinase pathway appears to participate in signaling leading to Akt in hepatocytes. Given that the insulin-induced phosphorylation of Akt was shown to be reduced in brown preadipocytes derived from IRS-1–deficient mice, whereas IRS-2–associated PI 3-kinase activity was increased in these cells (35), the relative contributions of IRS-1 and IRS-2 to the activation of Akt may thus differ in different cell types.

This study was supported by a grant-in-aid for the Research for the Future Program from the Japan Society for the Promotion of Science (to M.K.); grants from the Ministry of Education, Culture, Sports, Science, and Technology (to M.K. and W.O.); and a grant from Naito Foundation (to W.O.).

We thank I. Saito for AxCALacZ; H. Shimano and T. Noguchi for probes for SREBP-1c and GK mRNAs, respectively; and T. Kadowaki and K. Tobe for helpful discussions.