Akt2 influences glycogen synthase activity in human skeletal muscle through regulation of NH₂-terminal (sites 2 + 2a) phosphorylation

Abstract

Type 2 diabetes is characterized by reduced muscle glycogen synthesis. The key enzyme in this process, glycogen synthase (GS), is activated via proximal insulin signaling, but the exact molecular events remain unknown. Previously, we demonstrated that phosphorylation of Thr³⁰⁸ on Akt (p-Akt-Thr³⁰⁸), Akt2 activity, and GS activity in muscle were positively associated with insulin sensitivity. Here, in the same study population, we determined the influence of several upstream elements in the canonical PI3K signaling on muscle GS activation. One-hundred eighty-one nondiabetic twins were examined with the euglycemic hyperinsulinemic clamp combined with excision of muscle biopsies. Insulin signaling was evaluated at the levels of the insulin receptor, IRS-1-associated PI3K (IRS-1-PI3K), Akt, and GS employing activity assays and phosphospecific Western blotting. The insulin-stimulated GS activity was positively associated with p-Akt-Thr³⁰⁸ (P = 0.01) and Akt2 activity (P = 0.04) but not p-Akt-Ser⁴⁷³ or IRS-1-PI3K activity. Furthermore, p-Akt-Thr³⁰⁸ and Akt2 activity were negatively associated with NH₂-terminal GS phosphorylation (P = 0.001 for both), which in turn was negatively associated with insulin-stimulated GS activity (P < 0.001). We found no association between COOH-terminal GS phosphorylation and Akt or GS activity. Employing whole body Akt2-knockout mice, we validated the necessity for Akt2 in insulin-mediated GS activation. However, since insulin did not affect NH₂-terminal phosphorylation in mice, we could not use this model to validate the observed association between GS NH₂-terminal phosphorylation and Akt activity in humans. In conclusion, our study suggests that although COOH-terminal dephosphorylation is likely necessary for GS activation, Akt2-dependent NH₂-terminal dephosphorylation may be the site for “fine-tuning” insulin-mediated GS activation in humans.

type 2 diabetes (T2D) is characterized by reduced insulin-stimulated glucose uptake, which can be attributed largely to reduced muscle glycogen synthesis (37). Although glucose uptake/transport is an important rate-limiting process regulating glycogen synthesis (29, 34), several studies have demonstrated that the activity of the key enzyme in glycogen synthesis, glycogen synthase (GS), is rate limiting under various conditions in both animals (1, 8, 26) and humans (29, 39, 42) as well. Accordingly, it has been estimated that the “control coefficient” for GS activity in the regulation of whole body glucose disposal is 30%, underlining the importance of GS activity in the control of glucose homeostasis (23). Thus, a thorough understanding of the regulation of this important enzyme is crucial to finally elucidate the mechanisms responsible for insulin-mediated glucose uptake.

The regulation of GS activity is complex and involves allosteric regulation by glucose 6-phosphate (38) and glycogen (21) as well as inhibitory phosphorylations of GS. Although several important phosphorylation sites have been identified, sites 2 and 2a (NH₂ terminus) and 3a and 3b (COOH terminus) are generally considered the most important for GS activity (6, 26, 38). GS site 2 is phosphorylated by several kinases, including AMP-activated protein kinase (AMPK) and protein kinases A (PKA) and C (PKC) (6, 17), allowing for sequential phosphorylation of site 2a by casein kinase-1 (17). Phosphorylation of sites 3a and 3b on GS (p-GS 3a + 3b) is regulated mainly by GS kinase-3 (GSK-3) (6). Dephosphorylation of GS may involve protein phosphatase 1 (26, 35) or occur spontaneously due to high phosphate turnover in GS (17).

Overall, activation of GS by insulin is believed to involve the insulin receptor, insulin receptor substrate-1 (IRS-1), phosphatidylinositol 3-kinase (PI3K), Akt-dependent inhibition of GSK-3, and dephosphorylation of sites 3a and 3b (6, 7, 17, 19, 26). Previous studies have demonstrated significant dysregulation of muscle GS activity in insulin-resistant and type 2 diabetic individuals (13, 15, 17–19, 33, 42). However, the nature and importance of this abnormality is still debated. A genetic origin of the reduced GS activity in insulin-resistant individuals has been hypothesized, and although polymorphisms in the muscle GS gene have been linked to T2D (14), we demonstrated previously that GS activity could be attributed largely to nongenetic regulation (33).

Two studies on the present data have been published previously. First, we investigated the regulation of proximal insulin signaling (from insulin receptor to Akt) and its association with whole body insulin sensitivity (9). Second, we investigated the heritability of GS activity/phosphorylation and its association with insulin sensitivity (33). The novel aspect of this study is to focus on the regulation of GS activity by proximal insulin signaling. This was not addressed in our previous studies, and we are unaware of other data sets allowing such in-depth association studies of the (currently known) most important molecular mediators of muscle insulin action. This is the first major study, including almost 200 participants, to investigate numerous insulin signaling molecules ranging from the most proximal to the resulting activation of muscle GS, uniquely allowing us to differentiate the significance of each step in proximal insulin signaling on GS activation in humans. Specifically, we investigated the association between GS activity and insulin receptor tyrosine kinase (IRTK), IRS-1-associated PI3K (IRS-1-PI3K), and Akt1 and Akt2 activity as well as phosphorylation of Thr³⁰⁸ on Akt (p-Akt-Thr³⁰⁸) and Ser⁴⁷³ on Akt (p-Akt-Ser⁴⁷³).

METHODS

Young (22–31 yr) and old (57–66 yr) monozygotic and same-sex dizygotic twin pairs (n = 98 twin pairs) born at term (<3 wk from expected term) were identified through the Danish Twin Register (31–33) (Table 1). Zygosity was determined by polymorphic genetic markers (30). The twin study design facilitated evaluation of the genetic component in insulin signaling. These data have been reported previously (9, 33) and thus will not be discussed here. Glucose tolerance status was defined according to the 1999 World Health Organization criteria (12). The study population was generally healthy but included three participants with screen-detected T2D that were excluded from the analyses to avoid issues with hyperglycemia-induced alterations of insulin signaling. The study was approved by the regional ethics committee for the counties of Funen and Vejle, Denmark, and performed according to the Declaration of Helsinki. Informed consent was obtained from the study participants.

Table 1.

Clinical characteristics and insulin signaling parameters

Clinical Characteristic and Insulin Signaling Parameter	Means (SD)
Zygosity (MZ/DZ)	101/80
Age, yr	43 (17)
Birth weight, g	2,640 (480)
BMI, kg/m2	25 (3.8)
Total fat, %	24.5 (8.6)
V̇o2max, ml·min−1·kg−1	34.1 (9.8)
Rd clamp, mg·kg fat-free mass−1·min−1	11.0 (3.4)
Glycogen content, basal (mmol/kg wet wt muscle	89.4 (24.3)
Glycogen content, insulin stimulated, mmol/kg wet wt muscle	89.0 (24.4)
IRTK (AU)	4.5 (1.9)
IRS-1-PI3K (AU)	54.3 (33.6)
p-Akt-Thr308 (AU)	35.6 (21.9)
p-Akt-Ser473 (AU)	40.7 (22.7)
Akt1 activity, fmol·min−1·mg protein−1	52.9 (32.4)
Akt2 activity, fmol·min−1·mg protein−1	71.8 (38.9)
p-GS 2 + 2a, basal (AU)	3.8 (2.7)
p-GS 2 + 2a, insulin stimulated (AU)	3.1 (2.0)*
p-GS 3a + 3b, basal (AU)	0.26 (0.19)
p-GS 3a + 3b, insulin stimulated (AU)	0.16 (0.10)*
GSK-3α activity, basal, nmol·min−1·mg protein−1	5.1 (1.1)
GSK-3α activity, insulin stimulated, nmol·min−1·mg protein−1	3.6 (0.9)*
GS activity, basal, 0.17 mmol/l G6P, nmol·min−1·mg wet wt muscle−1	0.8 (0.3)
GS activity, basal, 8.00 mmol/l G6P, nmol·min−1·mg wet wt muscle−1	4.6 (1.1)
GS activity, basal (FV%)	16.6 (6.0)
GS activity, insulin, 0.17 mmol/l G6P, nmol·min−1·mg wet wt muscle−1	1.5 (0.6)*
GS activity, insulin, 8.00 mmol/l G6P, nmol·min−1·mg wet wt muscle−1	4.6 (1.1)
GS activity, insulin stimulated (FV%)	32.6 (9.5)*
δ-GS activity (FV%)	15.9 (6.7)

Data are means (SD); n = 181 (93 males and 88 females). MZ, monozygotic; DZ, dizygotic; R_d clamp, rate of glucose infusion during the euglycemic hyperinsulinemic clamp; IRTK, insulin receptor tyrosine kinase; IRS-1-PI3K, insulin receptor substrate-1-associated phosphatidylinositol 3-kinase; p-Akt-Ser⁴⁷³, phosphorylation of Ser⁴⁷³ on Akt; p-Akt-Thr³⁰⁸, phosphorylation of Thr³⁰⁸ on Akt; GS, glycogen synthase; AU, arbitrary units; p-GS 2 + 2a, phosphorylation of sites 2 + 2a on GS; p-GS 3a + 3b, phosphorylation of sites 3a + 3b on GS; %FV, %fractional velocity; δ-GS activity, insulin-mediated increase in GS fractional activity; G6P, glucose 6-phosphate. Only insulin-stimulated protein activities/phosphorylations are stated, unless otherwise indicated. GS phosphorylation was corrected for total GS protein content. Since the antibodies for total Akt protein and phosphorylation status were not isoform specific, this could not be done for Akt phosphorylation. For basal levels of protein activities/phosphorylations of IRTK and IRS-1-PI3K, please see previous studies of this study population (9).

^*P < 0.0001 vs. basal values.

Clinical examination.

Participants underwent a 2-day clinical examination, including a standard 75-g oral glucose tolerance test, anthropometric measurements that included weight, height, and waist and hip circumferences, and determination of body composition by dual energy X-ray absorptiometry scanning and maximal aerobic capacity (V̇o_{2 max}) by bicycle ergometer testing (31–33). Insulin sensitivity was examined by a 2-h (40 mU·m⁻²·min⁻¹) euglycemic hyperinsulinemic clamp. Insulin levels were ∼400 pmol/l during insulin infusion (32). A primed, constant, continuous infusion of [3-³H]tritiated glucose (bolus 22 μCi, 0.22 μCi/min) was initiated at 0 min and continued throughout the clinical investigation [basal period (120 min) and clamp period (120 min)]. Steady state (defined as the last 30 min of the basal and insulin-stimulated periods) was achieved (32). The glucose disappearance rate during insulin stimulation [glucose disposal (R_d) clamp] was calculated using the non-steady-state equations by Steele (31). Muscle biopsies (m. vastus lateralis) were taken during these periods [n = 181 individuals (101 young and 80 old)] using a Bergström needle with suction applied. Biopsies were frozen in liquid nitrogen and stored at −80°C. For a more elaborate description of the clinical examination, we refer to previous studies of this study population (31–33).

Transgenic mice analyses.

All animal experiments were conducted in accordance with the guidelines of the Institutional Animal Care and Use Committee of the Joslin Diabetes Center and the National Institutes of Health. Four male and two female Akt2 knockout (KO) and four male and two female wild-type mice on a C57BL/6N background (10–12 wk of age) (4) were studied. Mice were kept on a 12:12-h light-dark cycle with ad libitum access to standard chow and water. After 12–14 h without access to food, extensor digitorum longus (EDL) muscles were quickly removed from the euthanized mouse (cervical dislocation) and incubated in Krebs-Henseleit solution (117 mM NaCl, 4.7 mM KCl, 2.5 mM CaCl₂, 1.2 mM KH₂PO₄, 1.2 mM MgSO₄, and 24.6 mM NaHCO₃ with addition of 2 mM pyruvate, pH 7.4) at 37°C and oxygenated with gas containing 95% O₂ and 5% CO₂. Incubations were carried out in the presence or absence of insulin (150 nM, Humulin R, human insulin; Eli Lilly, Indianapolis, IN). All muscles were preincubated for 10 min following dissection and subsequently incubated for 30 min in the presence or absence of insulin. After incubation, muscles were quickly frozen in liquid nitrogen and stored at −80°C.

Protein analyses.

The methodology for the preparation of lysate for protein analyses has been described previously (10). In brief, 50 mg of muscle was homogenized (Polytron PT 3100; Kinematica) in a buffer [1:20 (wt/vol)] containing 50 mM HEPES (pH 7.5), 150 mM NaCl, 20 mM Na-pyrophosphate, 20 mM β-glycerophosphate, 10 mM NaF, 2 mM Na-ortovanadate, 2 mM EDTA, 1% Nonidet P-40, 10% glycerol, 2 mM PMSF, 1 mM MgCl₂, 1 mM CaCl₂, 10 μg/ml leupeptin, 10 μg/ml aprotinin, and 3 mM benzamindine. Homogenates were rotated end over end for 1 h (4°C) and cleared by centrifugation at 17,500 g for 1 h (4°C). The lysate was stored at −80°C for later analysis. All chemicals were obtained from Sigma-Aldrich (St. Louis, MO). Protein content was determined by the bicinchoninic acid assay (Pierce, Rockford, IL).

IRTK activity was measured in muscle lysate after immunopurification of the insulin receptor (anti-IR, AB3; Oncogen Science), using a microtiter assay as described previously (44). IRS-1-associated PI3K activity was measured in muscle lysate after immunopurification of IRS-1 (anti-IRS-1, no. 06-248; Upstate Biotechnology), as described previously (19). Sequential immunopurification of Akt2 (anti-Akt2, no. 06-606; Upstate Biotechnology) and then Akt1 (anti-Akt1, no. 06-558; Upstate Biotechnology) was performed on 400 μg of muscle lysate before Akt activity measurements. Prior depletion of Akt2 protein was important to avoid coimmunopurification of Akt1 as well as Akt2 using the anti-Akt1 antibody. The immunopurification was performed employing G-Sepharose beads. The pellet was washed three times in a buffer containing 20 mmol/l Tris (pH 7.4), 5 mmol/l EDTA, 10 mmol/l Na₄P₂O₇, 100 mmol/l NaF, 1% NP-40, and 3 mmol/l Na₃VO₄ and twice in a buffer containing 20 mmol/l Tris (pH 7.4), 10 mmol/l MgCl₂, and 1 mmol/l dithiothreitol. The kinase activity assay was performed in a buffer containing 50 mmol/l Tris (pH 7.4), 10 mmol/l MgCl₂, 1 mmol/l dithiothreitol, and 1 μmol/l protein kinase inhibitor (no. P-0300; Sigma-Aldrich) employing 30 μmol/l Akt/SGK substrate peptide (no. 12-340; Upstate Biotechnology) and [γ-³²P]ATP (111 kBq/sample; Perkin-Elmer, Boston, MA) as substrates. The kinase assay ran for 30 min at 30°C before the reaction was stopped by the addition of a buffer containing 0.6% HCl, 1 mmol/l ATP, and 1% BSA. Akt and GS phosphorylation status was evaluated by Western blotting [see Højlund et al. (20) for a detailed description] with phospho- and site-specific antibodies raised against p-Akt-Thr³⁰⁸ (no. 06-678; Upstate Biotechnology), p-Akt-Ser⁴⁷³ (9271; Cell Signaling Technology, Beverly, MA), p-GS 2 + 2a (phosphorylation of Ser⁷ and Ser¹⁰), and p-GS 3a + 3b (phosphorylation of Ser⁶⁴⁰ and Ser⁶⁴⁴). The antibodies raised against p-GS 2 + 2a and p-GS 3a + 3b were a gift from Prof. Grahame D. Hardie (Division of Molecular Physiology, University of Dundee, Dundee, UK). GS phosphorylation was corrected for total GS protein content. This could not be done for Akt phosphorylation since the antibodies for total Akt protein and phosphorylation status were not isoform specific. Furthermore, only insulin-stimulated measurements of proximal insulin signaling were used, since basal Akt1 and Akt2 activity as well as p-Akt-Thr³⁰⁸ could not be reliably determined; e.g., measurement of basal Akt2 activities in a subpopulation of 43 individuals provided paradoxical negative activity values for more than one-third of the basal samples. To obtain a rough estimate of the effect of insulin on Akt2 activity, we employed these measurements for basal Akt2 activity and found that insulin stimulation resulted in approximately fourfold increased Akt2 activity. However, due to the uncertainties associated with these measurements, these data will not be discussed further in this article. For GSK-3α activity measurements, GSK-3α was immunoprecipitated from lysate using an anti-GSK-3α antibody (no. 06-391; Upstate Biotechnology) bound to protein G-Sepharose. Kinase activity was measured as described previously (20). GS activity in the presence of 0.17 or 8 mmol/l glucose 6-phosphate was measured in duplicate, employing a 96-well plate assay (Unifilter 350 plates; Whatman, Cambridge, UK) (19). GS activity is given as the fractional activity (100 × activity at 0.17 mmol/l glucose 6-phosphate ÷ activity at 8 mmol/l glucose 6-phosphate). Both the absolute insulin-stimulated values and the insulin-mediated increase of (insulin-stimulated GS activity − basal GS activity, δ) GS fractional activity were used as measurements of GS activity. Glycogen content was determined as glycosyl units after acid hydrolysis, as described previously (43). Protein activity/phosphorylation status was related to a standard loaded on each gel/assay, allowing for comparison between different gels/assays.

Statistics.

All statistical tests were performed employing SAS statistical software (SAS Institute, Cary, NC), using the proc mixed procedure (multiple linear regression) to adjust for twin pair and zygosity status and other contributing variables, including age, sex, V̇o_{2 max}, total fat percentage, and basal glycogen content. A statistical model to test the association between GS activity and Akt2 activity looks like this: ln(insulin-stimulated GS activity) = age + sex + total fat percentage + V̇o_{2 max} + basal glycogen content + Akt2 activity. The model calculates an estimate for the β-coefficients (with corresponding P values) of each explanatory variable. These estimates for the β-coefficients represent the increase/decrease in the response variable associated with an increase of the explanatory variable of one unit (e.g., 1 unit of fat percentage from 24 to 25%). In the statistical model employed throughout this study, we had to log transform the response variables (e.g., GS activity) to avoid skewness of the residuals (a normal distribution of residuals is a prerequisite for the statistical model). This log transformation resulted in estimates expressing percentage-wise and not absolute changes of the response variable. Since activity/phosphorylation of proteins in the insulin signaling cascade was measured in arbitrary units with no direct physiological relevance, the association between the explanatory variables and the response variables in Table 2 was calculated per doubling of the explanatory variable from the average (e.g., what percentage of GS activity increased/decreased when Akt2 activity was doubled?). A doubling of protein activity/phosphorylation was generally within 2.5 SD (Table 1) and was thus biologically/statistically possible. In the mouse study, the data were analyzed employing a repeated two-way ANOVA (the data were paired with both basal and insulin-stimulated samples coming from the same mouse; n = 6 wild-type and 6 Akt2 KO mice). If the interaction or a main effect was significant, we examined the possible differences, employing paired and unpaired t-tests where appropriate.

Table 2.

Association between proximal insulin signaling and GS activity

	Insulin-Stimulated GS Activity (FV%)	δ-GS Activity (FV%)
IRTK activity (AU)	1%↑	No change
IRS-1-PI3K activity (AU)	3%↑	4%↑
p-Akt-Thr308 (AU)	9%↑*	18%↑**
p-Akt-Ser473 (AU)	3%↓	8%↑
Akt1 activity (AU)	No change	1%↓
Akt2 activity (AU)	9%↑*	19%↑**

The statistical model can be written as ln(insulin-stimulated GS activity) = age + sex + total fat percentage + V̇o_2max + basal glycogen content + measurement of proximal insulin signaling activity. The regression coefficient expresses the change in the response variable (e.g., insulin-stimulated GS activity) associated with a doubling of the explanatory variable (e.g., IRTK activity) from the average. All measurements of insulin signaling were measured in insulin-stimulated biopsies.

^*P < 0.05;

^**P < 0.01; ↑increased GS activity; ↓decreased GS activity.

RESULTS

The clinical characteristics of our human population have been described previously in detail (9, 31–33) and are summarized in Table 1. We have demonstrated previously that GS activity was positively correlated with whole body insulin sensitivity in this population (33). To illustrate GS's potential effect on glucose metabolism, we calculated the effect sizes using the statistical model described in methods. We found that a doubling of insulin-stimulated GS activity was associated with a 47% increased insulin-stimulated whole body glucose disposal rate (R_d clamp; P < 0.0001) and a 90% increased nonoxidative glucose metabolism (P < 0.0001). Furthermore, we found a significant positive correlation between δ-GS activity (insulin-stimulated GS activity − basal GS activity) and δ-whole body glucose disposal [R_d clamp (glucose disposal during insulin stimulation) − R_d basal (basal glucose disposal)] (r = 0.44, P < 0.0001). The effect of age on GS activity and phosphorylation has been published previously (33). In this cohort, insulin-stimulated GS activity was not different in males compared with females [males: 32.6% (SD: 9.3%); females: 32.5% (SD: 9.8%); P = 0.82]. In accord with previous studies, 2 h of insulin stimulation did not result in elevated intramuscular glycogen levels (Table 1). This likely reflects that the intraindividual variation is too large to detect the expected increase in intramuscular glycogen levels of 2–4% during the current study setup (43, 44).

Proximal insulin signaling and GS activity in humans.

Insulin-stimulated GS activity was significantly associated with insulin-stimulated levels of p-Akt-Thr³⁰⁸ (P = 0.01; Table 2 and Fig. 1A) and Akt2 activity (P = 0.04) but not with IRTK, IRS-1-PI3K, Akt1 activity, or p-Akt-Ser⁴⁷³ (P ≥ 0.48 for all analyses). A doubling of p-Akt-Thr³⁰⁸ or Akt2 activity from the average was associated with an increase in insulin-stimulated GS activity of 9% (Table 2). Similar results were obtained for δ-GS activity [insulin-stimulated GS activity − basal GS activity; 18 (P = 0.003) and 19% (P = 0.008) increases for doublings of p-Akt-Thr³⁰⁸ and Akt2 activity, respectively].

Fig. 1.

Insulin-stimulated levels of glycogen synthase (GS) fractional activity (A), phosphorylation of sites 2 + 2a on GS (p-GS 2+ 2a; B), phosphorylation of sites 3a + 3b on GS (p-GS 3a + 3b; C), and GSK-3 activity (D) as a function of phosphorylation of Thr³⁰⁸ on Akt (p-Akt-Thr³⁰⁸). Additionally, insulin-stimulated levels of GS fractional activity were plotted against insulin-stimulated levels of p-GS 2 + 2a (E) and p-GS 3a + 3b (F). GS phosphorylation was corrected for total GS protein content. Since the antibodies for total Akt protein and phosphorylation status were not isoform specific, this could not be done for Akt phosphorylation. Correlation coefficients (Spearman's rank correlation) and P values for each correlation have been included; n = 181 individuals. Each black dot represents data from 1 twin subject. AU, arbitrary units.

Using simple univariate analyses, Akt2 activity was negatively correlated with p-GS 2 + 2a (r = −0.24, P = 0.001), similarly to p-Akt-Thr³⁰⁸, and was not significantly correlated with p-GS 3a + 3b (r = −0.11, P = 0.16) or GSK3 activity (r = 0.09, P = 0.22). Although Akt2 activity was not significantly correlated with insulin-stimulated GS activity (r = 0.11, P = 0.12), it was significantly correlated with δ-GS activity (r = 0.17, P = 0.02), supporting the findings in Table 2, where multiple regression analyses were employed. Since the study population was very diverse, including males and females, young and old, lean and obese, physically active and inactive, etc., the correlation coefficients of these unadjusted analyses are not very high.

GS phosphorylation in humans.

Since only p-Akt-Thr³⁰⁸ and Akt2 activity were associated with GS activity, only these were included in further analyses of GS regulation. p-Akt-Thr³⁰⁸ was significantly associated with p-GS 2 + 2a (P = 0.001; Fig. 1B) but not p-GS 3a + 3b (P = 0.93; Fig. 1C) or GSK-3α activity (P = 0.41; Fig. 1D). A doubling of p-Akt-Thr³⁰⁸ was associated with 26% decreased p-GS2 + 2a. Similar results were obtained for Akt2 activity, where a significant association was found between Akt2 activity and p-GS 2 + 2a (29% decreased p-GS2 + 2a, P = 0.001) but not p-GS 3a + 3b (P = 0.07) and GSK-3α activity (P = 0.11).

Insulin decreased p-GS 2 + 2a [basal: 3.8 (SD: 2.7) arbitrary units (AU) vs. insulin: 3.1 (SD: 2.0) AU, P < 0.001], p-GS 3a + 3b [0.26 (SD: 0.19) AU vs. 0.16 (SD: 0.10) AU, P < 0.001], and GSK-3α activity significantly [5.1 (SD: 1.1) vs. 3.6 (SD: 0.9) AU, P < 0.001].

Regulation of GS activity by phosphorylation in humans.

Insulin-stimulated GS activity was negatively associated with p-GS 2 + 2a (a doubling of p-GS 2 + 2a was associated with 12% decreased GS activity, P < 0.001; Fig. 1E) but not p-GS 3a + 3b (P = 0.32; Fig. 1F) or GSK-3α activity (P = 0.28). Similar results for regulation of δ-GS activity by p-GS 2 + 2a were demonstrated. Additionally, δ-GS activity was negatively associated with δ-p-GS 2 + 2a (P = 0.03) but not δ-p-GS 3a + 3b (P = 0.47).

Regulation of GS activity in mice.

EDL muscles from wild-type and Akt2-KO mice were sampled to validate our results from the human study. Western blot analyses showed complete loss of expression of Akt2 in skeletal muscle (P < 0.001; Fig. 2A). Furthermore, both insulin-mediated p-Akt-Thr³⁰⁸ (P = 0.003; Fig. 2B) and p-Akt-Ser⁴⁷³ (P = 0.003; Fig. 2C) were reduced markedly. Employing our transgenic animal model, we confirmed Akt2's involvement in insulin-mediated GS activation by demonstrating that insulin-stimulated GS activity was reduced markedly in Akt2-KO mice compared with wild-type mice (P = 0.05; Fig. 2D and Table 3). Furthermore, the effect of insulin was abolished completely in Akt2-KO mice (P = 0.66; Fig. 2D and Table 3). Next, we investigated whether Akt2 was necessary for insulin-mediated GS dephosphorylation in rodent skeletal muscle. Insulin stimulation resulted in dephosphorylation of sites 3a + 3b in wild-type mice (P = 0.01; Fig. 2F). This effect of insulin was not evident in the Akt2-KO mice (P = 0.66; Fig. 2F). However, we observed no effect of insulin on p-GS 2 + 2a in either the wild-type (P = 0.98; Fig. 2E) or Akt2-KO mice (P = 0.53; Fig. 2E).

Fig. 2.

GS activity/phosphorylation in Akt2-knockout (KO) mice. Basal and insulin-stimulated levels of Akt2 protein (A), p-Akt-Thr³⁰⁸ (B), phosphorylation of Ser⁴⁷³ on Akt (p-Akt-Ser⁴⁷³; C), GS fractional activity (D), p-GS 2 + 2a (E), and p-GS 3a + 3b (F) in wild-type (WT) and whole body Akt2-KO mice. The data were analyzed by employing a repeated 2-way ANOVA. If the interaction or a main effect was significant, we examined the possible differences, employing paired and unpaired t-tests where appropriate. Findings are shown in representative immunoblots. GS phosphorylation was corrected for total GS protein content. Since the antibodies for Akt phosphorylation status were not isoform specific, this could not be done for Akt phosphorylation. Data are means ± SE; n = 6 mice in each group. *P < 0.05 vs. basal; †P < 0.05 vs. WT.

Table 3.

GS activity in wild-type and whole body Akt2-KO mice

	Wild Type		Akt2-KO
	Basal	Insulin	Basal	Insulin
GS activity, 0.17 mmol/l G6P	7.9 (0.8)	9.4 (2.2)	8.3 (0.7)	8.7 (1.4)
GS activity, 8.00 mmol/l G6P	16.1 (3.2)	15.1 (3.8)	16.1 (1.9)	16.7 (3.8)
GS activity (FV%)	50 (7)	63 (8)*	51 (3)	53 (8)

Data are means (SD). KO, knockout. GS activity in the presence of 0.17 and 8 mmol/l G6P is given as nmol·min⁻¹·mg protein⁻¹. Data are paired (basal and insulin-stimulated samples were from the same animal); n = 6 mice in each group.

^*P < 0.05 vs. basal. P values for this table were calculated using paired t-tests.

DISCUSSION

In this study, we found that in humans GS activity was positively associated with p-Akt-Thr³⁰⁸ and Akt2 activity but not IRS-1-PI3K activity. Furthermore, p-Akt-Thr³⁰⁸ and Akt2 activity were negatively associated with p-GS 2 + 2a, which in turn was negatively associated with GS activity. In contrast, p-Akt-Thr³⁰⁸ and Akt2 activity were not associated with p-GS 3a + 3b or GSK-3 activity in our human population. Thus, although dephosphorylation of sites 3a + 3b is necessary for GS activation, phosphorylation status of sites 2 + 2a may be critical for “fine-tuning” GS activity in humans. To validate these results, we employed a transgenic animal model with whole body Akt2-KO. We validated that Akt2 indeed is necessary for insulin-mediated GS activation. However, since p-GS 2 + 2a was not regulated by insulin in rodent muscle [in contrast to human muscle (13, 18, 42)], we could not use this model to validate that Akt2 regulates p-GS 2 + 2a, as indicated in our cross-sectional human study.

p-Akt-Thr³⁰⁸ and Akt2 activity were the only measures of proximal insulin signaling significantly associated with GS activity. Because p-Akt-Thr³⁰⁸ was strongly associated with Akt2 activity in this study population (9), we hypothesize that its effect was mediated mainly through Akt2 activation. Previously, we demonstrated that muscle GS activity was positively associated with p-Akt-Thr³⁰⁸ but not with IRS-1-PI3K activity or p-Akt-Ser⁴⁷³ (15, 18). The putative importance of p-Akt-Thr³⁰⁸ but not p-Akt-Ser⁴⁷³ in the control of GS activity may seem strange, since both phosphorylations are needed for full Akt activation and thus probably also for full Akt-dependent GS activation. Nonetheless, a study has suggested that phosphorylation of residue Ser⁴⁷³ precedes phosphorylation of residue Thr³⁰⁸ (36). Thus, it is possible that extensive regulation of the insulin signal occurs at the level of p-Akt-Thr³⁰⁸ after phosphorylation of residue Ser⁴⁷³. A weak association between PI3K and GS activity is supported by studies demonstrating dissociation between IRS-1-PI3K activity and GS activity in human participants (5, 19, 42). Although several studies have provided indirect evidence suggesting that reduced GS activity in insulin-resistant participants was associated with reduced PI3K but normal Akt activity, direct statistical analyses of the association between GS and PI3K as well as Akt activity were not performed in those studies (17, 19). It is well established that PI3K activity is necessary for GS activation (45); however, our study suggests that the signal is “fine-tuned” at the level of Akt, since only Akt was significantly associated with GS activity in this study and insulin-mediated glucose disposal in our previous study (9). This indicates strong regulation/modification of the insulin signal at this level and suggests a role for Akt as a “master switch” in insulin-dependent GS activation in humans. Finally, we demonstrated the importance of Akt2 over Akt1 in the regulation of GS, thus supporting studies indicating that Akt2 is the major Akt isoform involved in glucose metabolism in humans (3, 42). Uniquely, we were able to confirm the importance of Akt2 in the regulation of muscle GS activity in our transgenic rodent study, where loss of Akt2 in skeletal muscle abolished insulin-mediated GS activation completely. These results elaborate on the conclusions from an earlier study demonstrating that inhibition of all known Akt isoforms, Akt1, Akt2, and Akt3, in rodent muscle cells (L6 cells) resulted in decreased insulin-mediated GS activation (40). Previously, we demonstrated that reduced p-Akt-Thr³⁰⁸ and Akt2 activity was associated with insulin resistance in this human study population (9). The present findings suggest that part of this effect may be mediated through decreased GS activity (33).

Investigating potential mechanisms by which Akt could regulate GS activity, we found that insulin-stimulated p-Akt-Thr³⁰⁸ and Akt2 activity were negatively associated with p-GS 2 + 2a in our human population. Supporting our results, previous studies have demonstrated that reduced insulin-stimulated p-Akt-Thr³⁰⁸ coincided with reduced dephosphorylation of sites 2 + 2a in studies of insulin-resistant human muscle (15, 18). The mechanistic link between Akt activity and reduced p-GS 2 + 2a is unknown but may involve activation of phosphatases, including protein phosphatase 1, which is indeed regulated by insulin (35), or inhibition of kinases responsible for the phosphorylation of sites 2 + 2a, including PKA, which is inhibited by insulin in primates (28). We were not able to validate the importance of Akt2 for insulin-mediated GS dephosphorylation of sites 2 + 2a in our mouse model, since p-GS 2 + 2a was not affected by insulin in this rodent model. Similarly, insulin appears to not be able to induce dephosphorylation of sites 2 + 2a in rat skeletal muscle (25). This is in contrast to insulin's effect on p-GS 2 + 2a in humans (13, 18, 42). Low antibody specificity cannot explain our “missing” effect of insulin on p-GS 2 + 2a in mice since we have demonstrated previously that stimulation with AICAR, an activator of AMPK, results in two- to threefold increased p-GS 2 + 2a, using the same antibody (22). Thus, it will require another study model, possibly transgenic human muscle stem cells, to validate the putative association between Akt2 activity and dephosphorylation of sites 2 + 2a in humans. We found no association between Akt and p-GS 3a + 3b in our human study population, which is in agreement with the lack of association between Akt and GSK3 activity. Most evidence supporting a direct role of Akt in the regulation of muscle GSK-3 activity derives from studies in rodents (2, 27) and rodent cell lines (7), whereas previous human studies have provided mostly indirect evidence (15). In line with previous studies (2, 7, 27), using our unique animal model, we could provide direct evidence for the involvement of Akt2 in insulin-mediated dephosphorylation of sites 3a + 3b in rodents. In this study we were able to measure only GSK-3α activity in our human population. Nonetheless, a study has demonstrated that GSK-3β is the most important isoform in the regulation of murine GS activity (27). However, since GSK-3β would mediate its effect through p-GS 3a + 3b, which was not associated with Akt or GS activity in this human study population, it seems unlikely that GSK-3β would be strongly associated with Akt or GS activity. Our study is the first major study to investigate the association between Akt and GSK-3 activity in humans, and although we cannot rule out an effect of Akt on GSK-3 activity, our results indeed question the common perception of Akt primarily acting through GSK-3 in humans.

In accord with Akt activity being negatively associated with p-GS 2 + 2a and positively associated with GS activity in our human population, we demonstrated that p-GS 2 + 2a was strongly negatively associated with GS activity. This is in accord with previous studies demonstrating that insulin-resistant individuals, including patients with HIV-associated lipodystrophy (15), obesity (18), T2D (18), and polycystic ovary syndrome (13), displayed both reduced insulin-mediated dephosphorylation of sites 2 + 2a on GS as well as reduced GS activity. p-GS 3a + 3b was not associated with GS activity, which opposes several studies demonstrating that p-GS 3a + 3b inhibits GS activity in rodents (2, 27, 38). However, in studies investigating whether p-GS 3a + 3b can be linked to GS activity in human skeletal muscle, this was seen only in patients with HIV (15) or polycystic ovary syndrome (13) but not in healthy controls (13, 18). Thus, our results suggest that insulin-mediated dephosphorylation of sites 3a + 3b on GS may be permissive rather than directly controlling GS activity in human skeletal muscle from healthy individuals. Although a permissive role of insulin-mediated dephosphorylation of sites 3a + 3b is in contrast to previous rodent and cell model studies (2, 27, 38), it is important to keep in mind that our study is the first human study large enough to investigate this in a human physiologically relevant in vivo setting. The transition from transgenic animals and rodent cell models to humans might explain partly the different results regarding the role of p-GS 3a + 3b on GS activity, highlighting the need for further studies of this important aspect of GS activation in human skeletal muscle. Nonetheless, we acknowledge that our cross-sectional study design does not permit us to exclude a direct role for p-GS 3a + 3b in the regulation of human GS activity.

The regression and correlation analyses of the association between GS activity and R_d clamp as well as nonoxidative glucose metabolism indicate that changes in muscle GS activity may have very large effects on whole body glucose metabolism. Our study suggests that Akt2 is not the only major regulator of GS activity, since a doubling of Akt2 phosphorylation/activity was associated with “only” a 10–20% increased GS activity. In this regard, it is important to keep in mind that additional signaling mediators, including the kinases AMPK, PKA, and PKC, are involved in regulating GS activity (6, 17). Nonetheless, this study implicates Akt2 as a critical node in the control of GS in human muscle and presents data on the possible mechanism for this regulation, Akt2-dependent dephosphorylation of sites 2 + 2a on GS.

It could be hypothesized that the timing of biopsy excision (after 2 h of insulin stimulation) could affect the results of this study, since the insulin signal might have been attenuated by this “long-term” insulin stimulation. However, this is unlikely, since both we and others have demonstrated that IRTK, IRS-1-PI3K, Akt, and GS activity were not attenuated after 2 h of physiological insulin stimulation in humans (11, 16, 24, 41, 43, 44).

In conclusion, our results suggest that the association between proximal insulin signaling and GS activity in healthy individuals may be mediated primarily through Akt-dependent regulation of NH₂-terminal GS phosphorylation. From this study we hypothesize that although dephosphorylation of sites 3a and 3b seems to be required for GS activation, it is the Akt2-dependent dephosphorylation of sites 2 + 2a that “fine-tunes” the insulin-mediated activation of GS in humans. Nonetheless, additional pathways apart from the canonical insulin signaling pathway seem to be involved in GS activation.

GRANTS

M. Friedrichsen was supported by a grant from the Danish Agency for Science Technology and Innovation. This study was supported by grants from the Danish Medical Research Council, the Danish Strategic Research Council, the Novo Nordisk Foundation, the Danish Diabetes Association, the Lundbeck Foundation, The UNIK Project: Food, Fitness, & Pharma for Health and Disease, supported by the Danish Ministry of Science, Technology, and Innovation, and an integrated 6th Frame Work EU project (EXGENESIS) (contract no. LSHM-CT-2004-005272) from the European Union. This work was supported in part by the National Institutes of Health (R01-AR-42238 to L. J. Goodyear).

DISCLOSURES

B. F. Hansen is employed at Novo Nordisk. A. Vaag has received a grant from Novo Nordisk for this research. M. Friedrichsen and A. Vaag are shareholders at Novo Nordisk.

AUTHOR CONTRIBUTIONS

M.F., H.B.-N., A.V., P.P., and J.F.W. contributed to the conception and design of the research; M.F., J.B.B., E.A.R, C.P., B.F.H., M.F.H., L.J.G., P.P., and J.F.W. performed the experiments; M.F., P.P., and J.F.W. analyzed the data; M.F., R.R.-M., and J.F.W. interpreted the results of the experiments; M.F. and J.B.B. prepared the figures; M.F. drafted manuscript; M.F., J.B.B., E.A.R., R.R.-M., C.P., B.F.H., H.B.-N., M.F.H., L.J.G., A.V., P.P., and J.F.W. edited and revised the manuscript; M.F., J.B.B., E.A.R., R.R.-M., C.P., B.F.H., H.B.-N., M.F.H., L.J.G., A.V., P.P., and J.F.W. approved the final version of the manuscript.