Increased p85/55/50 Expression and Decreased Phosphotidylinositol 3-Kinase Activity in Insulin-Resistant Human Skeletal Muscle

Eight nondiabetic lean subjects, seven obese subjects with type 2 diabetes, and eight obese subjects without diabetes participated in the study. The clinical characteristics of the subjects are given in Table 1. Studies in subjects with type 2 diabetes were performed after discontinuation of their antihyperglycemic medications (five were taking sulfonylureas and two were taking metformin plus sulfonylurea) for at least 2 weeks. All participants were in good general health and had normal kidney and liver function. None of the control subjects were taking any medications to alter glucose tolerance. During the clamp study, a biopsy from the right leg was performed before the insulin infusion was started (Humulin; Eli Lilly, Indianapolis, IN) at a rate of 80 mU · m⁻² · min⁻¹. After 300 min of infusion, a second biopsy was performed from the left vastus lateralis. Tissues were blotted free of blood and immediately frozen in liquid nitrogen. To reach a steady state of insulin-stimulated glucose metabolism in the insulin-resistant subjects, 4–5 h of insulin infusion were required. Consequently, the muscle biopsies were taken at time 0 and 5 h. However, previous results have indicated that post–receptor signaling defects are observed as early as 2 h, 3 h, and even 30 min after insulin infusion (34–36). In a previous study, we showed that defects in insulin-stimulated tyrosine phosphorylation were easily detected after 30 min of insulin stimulation and fully sustained throughout the 5-h time course (37).

The study protocol was approved by the Internal Review Board for Human Subjects of the University of California at San Diego. Written informed consent was obtained from each subject. The procedures for hyperinsulinemic-euglycemic clamps and analysis of clinical parameters were described previously in detail (38,39).

All buffer components, protease and phosphatase inhibitors, and lipids were purchased from Sigma (St. Louis, MO). Phosphospecific antibodies were from Cell Signaling Technology (Beverly, MA). Rabbit anti-pan p85 (no. 06-496), anti-IRS-1, and p110 were from Upstate Biotechnology (Lake Placid, NY).

Muscle samples (50–100 mg) were powdered under liquid nitrogen and homogenized by Polytron (3 × 15 s) in 2 ml ice-cold buffer containing 20 mmol/l Tris/HCl, pH 7.5, 150 mmol/l NaCl, 0.25 mol/l sucrose, 1.2 mmol/l EGTA, 1 mmol/l EDTA, and a cocktail of protease and phosphatase inhibitors as described previously (36). After centrifugation at 1,000g for 10 min at 4°C, the supernatants were analyzed as homogenates and fractionated into total membrane and cytosol by centrifugation at 100,000g for 1 h at 4°C. For comparison of signal intensity between samples, equal amounts of protein from each sample were loaded per lane during gel electrophoresis. Immunoblotting of the membranes after SDS-PAGE, enhanced chemiluminescence detection, and quantitation of the signals by densitometry were carried out following standard protocols.

For myosin heavy chain (MHC) analysis by SDS-PAGE, muscle homogenates were prepared by following the protocol described by Klitgaard et al. (40) and electrophoresis was carried out at 100 V for 16 h using 6% separating gel and standard running buffer containing 35% glycerol. Proteins were visualized by silver staining with a kit from Pierce (Rockford, IL).

To measure PKC activity, 0.5–1.0 mg proteins were immunoprecipitated with rabbit polyclonal anti–COOH-terminal PKC-ζ/λ or anti–PKC-θ antisera or preimmune serum in the presence of protein-A/G agarose (all from Santa Cruz Biotechnology, Santa Cruz, CA). The washed immunobeads were assayed for PKC-ζ/λ and PKC-θ activities following previously published methods (12,31,33,36).

IRS-1–bound PI 3-kinase activity was assayed in the immunobeads after anti–IRS-1 immunoprecipitation. The agarose beads were washed following a procedure described by Caldwell et al. (41), and the PI 3-kinase activity was assayed as described by Kruszynska et al. (7) and Walsh et al. (42).

Results are presented as means ± SE. Significance of a difference between two groups was determined by ANOVA and paired two-tailed t tests. The linear regression and multivariate analyses (MINOVA and discrimination function tests) were carried out by Minitab statistical software. P values ≤0.05 were considered significant.

Western blotting with a pan-reactive anti-p85 antibody detected all three regulatory subunits (p85, p55, and p50) in the muscle lysates. When normalized against actin, the content of the three regulatory subunits together (p85/p55/p50) was significantly increased (∼40–50%) in muscle lysates from obese subjects and even further increased (80–90%) in type 2 diabetic patients compared with lean control subjects (Fig. 1). In terms of individual subunits, the greatest increase was in the level of p55 (P < 0.01), followed by p50 (P < 0.01) and p85 (P < 0.05). The protein expression level of p85/p55/p50 was the same both in the basal state and after the 5-h insulin infusion. Because there was a small but significant (P < 0.05) decrease (10–20% compared with lean) in IRS-1 content in the insulin-resistant tissues (Fig. 2A), the ratio of regulatory subunits to IRS-1 is even higher than the ratio calculated with respect to actin (results not shown). A similar decrease in IRS-1 level was reported by Brozinick et al. (43) in obese muscle.

The phosphorylation of IRS-1 at Ser³⁰⁷ was measured in the muscle lysates by Western blotting with a phosphospecific anti-pSer³⁰⁷ IRS-1 antibody. As seen in Fig. 2B, basal Ser³⁰⁷ phosphorylation is low in the lean subjects with a two- to threefold increase after 5 h of in vivo insulin stimulation. In the obese and type 2 diabetic subjects, there is a marked fourfold increase in basal Ser³⁰⁷ phosphorylation and insulin stimulation was without further effect. This increase in Ser³⁰⁷ phosphorylation would predict impaired insulin-stimulated IRS-1 tyrosine phosphorylation, which is precisely what is shown in Fig. 2C. Thus, in lean subjects insulin treatment leads to a fivefold increase in IRS-1 tyrosine phosphorylation; however, there is a marked decrease in the obese group and an even greater degree of impairment in the type 2 diabetic subjects.

The increased concentration of p85-derived regulatory subunits, coupled with the decrease in IRS-1 tyrosine phosphorylation, should lead to a substantial decrease in insulin-stimulated IRS-1–associated PI 3-kinase activity. As summarized in Fig. 3, we have measured the amount of p85 and p110 protein recovered in IRS-1 immunoprecipitates before and after insulin stimulation in the three study groups. A very large decrease in p110 association with IRS-1 is observed in the insulin-resistant subjects (Fig. 3B), fully consistent with the combination of increased regulatory subunit concentration and decreased IRS-1 tyrosine phosphorylation. Interestingly, as demonstrated in Fig. 3C, although a significant decrease in IRS-1–associated p85 protein was quite evident in the insulin-resistant groups, the magnitude of the decrease was much less than the decrease in p110 association. Most likely, this reflects the increased adaptor subunit concentration in the muscle lysates, which leads to competition between monomeric regulatory subunits (p85, p55, and p50) with the heterodimers (p110/p85, p110/p55, and p110/p50) for binding to tyrosine phosphorylated IRS-1. This leads to a greater decrease in formation of the trimeric complex (IRS-1–p110 adaptor protein) compared with the dimeric IRS-1–adaptor protein complex.

We also measured the PI 3-kinase enzymatic activity in the IRS-1 precipitates, and these data show a remarkable decrease in insulin-stimulated IRS-1–associated PI 3-kinase activity (Fig. 4), fully consistent with the decrease in formation of IRS-1/p110 complexes.

Akt and PKC-ζ/λ are two enzymes downstream of PI 3-kinase that mediate PI 3-kinase–directed signals to facilitate GLUT-4 translocation in skeletal muscle and adipocytes. In lean subjects, in vivo insulin stimulation led to a twofold increase in PKC-ζ/λ activity, and this effect was markedly blunted in the insulin-resistant groups (Fig. 5). Insulin also caused a marked stimulation of Akt phosphorylation in the lean subjects, and this effect was only modestly blunted in the insulin-resistant groups (Fig. 6). These results are consistent with previous studies (36,44,45), including an earlier report that showed a dissociation between decreased PI 3-kinase and Akt activity (4,7). It should be noted that Akt2 is more directly related to stimulation of glucose transport (46,47), and the available antibodies do not distinguish between Akt1 and Akt2. Therefore, it is possible that if we were able to measure Akt2 activity separately, a greater defect would be observed.

As shown earlier, IRS-1 Ser³⁰⁷ phosphorylation is elevated in the insulin-resistant groups. It is widely held that activation of the proinflammatory pathway contributes to insulin resistance (27,29–33) and that specific kinases in this pathway can phosphorylate IRS-1 at positions 307 and 1101 (27,28,29,30). Three of these candidate kinases were assessed. As seen in Fig. 7A, basal PKC-θ activity was elevated in the insulin-resistant groups compared with lean control subjects. More strikingly, when subcellular fractionation was performed (Fig. 7B), the appearance of PKC-θ activity in the skeletal muscle membrane fraction is enhanced in the obese subjects and is even further increased in the type 2 diabetic group. These results on PKC-θ activity are consistent with a previously published report (33).

We also assessed JNK-1 activity by measuring the JNK phosphorylation state using a phosphospecific antibody (Figs. 7C and D). We found a two- to threefold elevation in the level of activated JNK-1 in the insulin-resistant groups. In contrast, levels of phosphorylated IKK-β, which is also a proinflammatory serine/threonine kinase implicated in IRS-1 serine phosphorylation, were only minimally increased in the insulin-resistant subjects (Figs. 7E and F).

Electrophoretic analysis of MHC composition showed a significant increase in the percent composition of MHC IIx fibers (P < 0.005) in the insulin-resistant muscle (from 5.0 ± 1.2% in lean to 15.9 ± 2.7% and 16.1 ± 1.9% in obese and type 2 muscles, respectively). The increase in MHC IIx was accompanied by a decrease in both MHC I and IIa fibers. MHC IIa decreased from 56.4 ± 1.0% in lean to 46.4 ± 4.9% and 52.5 ± 3.7% in obese and type 2 diabetic muscles, respectively. In comparison to lean subjects, the proportion of MHC I did not change in obese muscle, whereas it fell from 38.5 ± 0.8% in lean to 31.4 ± 3.1% in type 2 diabetic muscle. These observations are consistent with other reports (48–50) showing that insulin resistance is associated with a decrease in oxidative and an increase in glycolytic fiber types in muscle. Although the decreased oxidative and increased glycolytic types of fibers in type 2 diabetic subjects has been previously observed, it is unclear whether this represents a primary or secondary event and further studies into this question seem warranted.

Linear regression analyses were carried out using glucose infusion rate as the dependent variable and the ratio of p85/55/50 to actin, basal PKC-θ activity, phosphorylation of JNK and Ser³⁰⁷ IRS-1, and the percent composition of MHC IIx fiber type as the independent variables. Pairwise analysis (glucose infusion rate versus a variable) showed significant inverse correlations between glucose infusion rate values and the levels of p85/55/50 (r = −0.88, P < 0.0001), PKC-θ activity (r = −0.89, P < 0.0001), pJNK-1 (r = −0.66, P < 0.001), pSer³⁰⁷ IRS-1 (r = −0.65, P < 0.001), and the proportion of type IIx fibers (r = −0.6, P < 0.004). Multilinear regression and multivariate analyses (MANOVA and discrimination function tests) showed that a combination of all five independent variables accounted for 82% of the variation (r² = 0.82, F = 14.7, P < 0.001) in the glucose infusion rate value. The residual errors, t, and F values produced by the regression model suggest that the predictive ability of the variables would be PKC-θ activity > p85/55/50 expression > pJNK > pSer³⁰⁷ IRS-1 > MHC IIx fiber content.

Because insulin-resistant patients become hyperinsulinemic, the increased circulating insulin levels could lead to overstimulation of the MAP kinase pathways. We measured extracellular signal–related kinase (ERK)-1 and -2 phosphorylations in our experimental groups, and the results are seen in Figs. 8A and B. The results show that ERK-2 content and phosphorylation state are relatively constant in the basal and insulin-stimulated states and are the same across the various study groups. In contrast, ERK-1 phosphorylation was markedly stimulated (fivefold) by insulin in the lean subjects, whereas in the hyperinsulinemic insulin-resistant subjects, the basal level of ERK-1 phosphorylation was increased three- to fourfold with little further effect of insulin infusion. Qualitatively similar results were seen when phospho-p38 was measured, with modestly elevated levels seen in the insulin-resistant groups compared with the lean control subjects (Figs. 8C and D). Interestingly, after insulin stimulation there was a small effect to increase phospho-p38 levels in the lean subjects, whereas insulin led to a paradoxical decrease in phospho-p38 in the insulin-resistant subjects, similar to a previous report (51).

Decreased insulin sensitivity is a widely described metabolic characteristic of patients with obesity and type 2 diabetes, and in the current studies we have explored the cellular mechanisms underlying this phenomenon. Because a major manifestation of insulin resistance is decreased insulin-stimulated glucose disposal by skeletal muscle, we have focused our attention on the mechanisms of skeletal muscle insulin resistance. To do this, we analyzed skeletal muscle biopsies taken from normal lean, nondiabetic obese, and obese type 2 diabetic subjects in the basal state and after a 5-h insulin infusion in the course of a euglycemic glucose clamp experiment.

One major finding in this study is increased expression of p85 protein and its splice variants p55 and p50 in the insulin-resistant groups, in the order of p55 > p50 > p85. This finding of increased p85/55/50 expression and the excellent correlation with decreased in vivo insulin sensitivity (glucose infusion rate values) suggest that increased expression of these PI 3-kinase adaptor subunits represents an important cellular mechanism that contributes to skeletal muscle insulin resistance in obesity and type 2 diabetes. A previous report in animals is consistent with the human data, as the combined levels of p85, p55, and p50 in liver of ob/ob mice were elevated compared with lean controls (25). Because the published data on the expression of regulatory subunit mRNA in human muscle showed discordant results (52,53) and changes in protein expression are often not reflective of changes in mRNA levels, we have restricted our analysis to protein expression. Furthermore, free p85 subunits are degraded more rapidly than p110-bound p85 (54), creating an additional complicating issue in attempting to correlate p85/55/50 proteins to mRNA levels. In a previous study with control and diabetic groups matched for obesity (BMI values of 26–27 kg/m²), Tsuchida et al. (52) did not find statistically significant changes in protein expression of the regulatory subunits. Because the obese and diabetic subjects in our study had significantly higher BMI values than the nonobese lean subjects (Table 1), obesity might be a key factor magnifying the differences in subunit protein expression between lean and insulin-resistant subjects. Alternatively, it is possible that other unknown clinical or physiological differences between our patients and those in the earlier report might be responsible.

If competition between the excessive adaptor protein expression and the p85/55/50-p110 heterodimer occurs, then one would predict a greater decrease in IRS-1–associated p110 content compared with IRS-1–associated p85. As seen in Fig. 3, this is precisely what we observed, consistent with the view that excessive adaptor subunits compete with the PI 3-kinase heterodimer for binding to protein targets such as IRS-1, contributing to decreased insulin-stimulated PI 3-kinase activity.

The ability of IRS-1 to interact with the PI 3-kinase heterodimer is even further impaired due to abnormalities in insulin-stimulated IRS-1 tyrosine phosphorylation. Thus, the infused insulin had a four- to fivefold effect to stimulate IRS-1 tyrosine phosphorylation in the lean control subjects, and this effect was blunted by ∼50% in the obese subjects and even further impaired in the type 2 diabetic patients. It is well known that phosphorylation of IRS-1 at Ser³⁰⁷ can impair subsequent insulin-stimulated tyrosine phosphorylation (26,27). In lean subjects, we found low levels of Ser³⁰⁷ phosphorylation in the basal state, which was stimulated approximately threefold by insulin. In the insulin-resistant obese and type 2 diabetic subjects, basal Ser³⁰⁷ phosphorylation was markedly increased (fivefold) with no further effect of insulin. Based on the known effects of Ser³⁰⁷ phosphorylation to impair tyrosine phosphorylation (26,27), it is likely that this increased basal level of Ser³⁰⁷ phosphorylation is responsible for the defect in insulin-stimulated tyrosine phosphorylation of IRS-1.

The results also provide evidence on the mechanism of the increased Ser³⁰⁷ phosphorylation. Thus, PKC-θ, JNK-1, and IKK-β are three serine/threonine kinases in the proinflammatory pathway (27,29–33), and all three were assessed in the muscle samples before and after insulin stimulation. There was a threefold increase in basal JNK-1 phosphorylation and three- and fivefold increases in basal activity in the muscle samples from the obese and type 2 diabetic subjects, respectively, compared with the lean control subjects. Interestingly, in the lean control subjects, activation of PKC-θ was stimulated by insulin, whereas basal PKC-θ activity was three- to fivefold higher in the insulin-resistant subjects with no further stimulation by insulin. These results demonstrate that these components of the proinflammatory pathway are hyperactivated in skeletal muscle from insulin-resistant subjects, providing a plausible explanation for the increased phosphorylation of IRS-1 at Ser³⁰⁷. Thus, a combination of two deleterious effects, decreased IRS-1 tyrosine phosphorylation and increased p85/55/50 expression, could coordinately impair insulin stimulation of PI 3-kinase activity.

It is possible that competition with the PI 3-kinase heterodimer is not the only way in which an excessive adaptor protein subunit concentration can lead to impaired insulin action (54). In mice, it has been observed that inactivation of the p85 gene leads to impaired activation of JNK-1 (55) and that restoration of the p85-α level stimulates JNK-1 activity (55), suggesting that p85 can directly regulate JNK-1 activity without requiring PI 3-kinase activity. This may be relevant to our finding that JNK1 was hyperactivated in the insulin-resistant subjects concomitant with the increased expression of the adaptor subunits, despite a decrease in PI 3-kinase activity.

Linear regression analysis indicated that the increased expression of p85/55/50, PKC-θ activity, phosphorylation levels of Ser³⁰⁷ IRS-1 and JNK-1, and the proportion of MHC IIx fiber can predict 82% of the decrease in insulin sensitivity (glucose infusion rate values). When the relationship between the five independent variables was examined, a remarkably high correlation (r = 0.92, P < 0.0001) was noted between p85/55/50 and PKC-θ activity. This may suggest a currently unknown biological correlation between these two signaling proteins.

PI 3-kinase mediates insulin-stimulated glucose transport, which results in enhanced glucose disposal into skeletal muscle in vivo (1,4–8). Although the precise mechanisms whereby activation of PI 3-kinase leads to glucose transport stimulation are incompletely understood, stimulation of PI 3-kinase leads to downstream activation of Akt and PKC-λ, and numerous reports exist implicating these kinases in stimulation of GLUT-4 translocation and glucose transport (9,10,11,12,13,14). The results presented here demonstrate a modest decrease in Akt phosphorylation in insulin-resistant subjects, with a much more dramatic decrease in PKC-λ activation. Although it is clear that inhibition of either Akt2 or PKC-λ will lead to reduced insulin-stimulated glucose transport (9,10,11,12,13), the relative role and importance of these two kinases are unknown. Although our results would suggest a more dominant role for PKC-λ compared with Akt in the insulin-resistant state, it should be noted that the phosphospecific antibody used does not distinguish between Akt1 and Akt2. It has been shown that Akt2 is more important for glucose transport stimulation than Akt1 (46,47). If Akt1 is more abundant in muscle tissue and if no defect in Akt1 exists, then despite the mild decrease in total phospho-Akt, a large decrease in Akt2 phosphorylation and activation might be present. Along these lines, Brozinick et al. (43) demonstrated a decrease in insulin-stimulated Akt2 and Akt3 (but not Akt1) activities in muscle from obese subjects.

Another important result from these studies is the marked increase in basal levels of activated ERK-1 as well as p38 kinase in the obese and type 2 diabetic subjects. Because obese and type 2 diabetic subjects were hyperinsulinemic, it is possible that this contributed to the higher basal phosphorylation of JNK-1, ERK-2, and p38 by hyperstimulation of these pathways. It is also possible that elevated levels of p85/55/50 may inhibit the interaction of p85/p110 complexes with their normal SH2-binding partners, directing a greater proportion of the p85/p110 complexes to p21 Ras, where it will stimulate downstream activation of the ERK pathway.

In summary, we have observed a strong inverse correlation between in vivo insulin sensitivity and increase in the expression of PI 3-kinase adaptor subunits, basal phosphorylation in Ser³⁰⁷ IRS-1 and JNK-1, basal PKC-θ activity, and MHC IIx fiber type in skeletal muscle of obese and type 2 diabetic subjects. These increases correspond with a decrease in IRS-1 tyrosine phosphorylation, leading to marked impairment of the ability of insulin to activate PI 3-kinase in skeletal muscle from insulin-resistant subjects, and probably represent an important cellular mechanism for decreased insulin sensitivity in obesity and type 2 diabetes.

This work was supported by grants from the National Institutes of Health (DK 33651 and DK 16964).