Intact Regulation of the AMPK Signaling Network in Response to Exercise and Insulin in Skeletal Muscle of Male Patients With Type 2 Diabetes: Illumination of AMPK Activation in Recovery From Exercise

AMPK has emerged in recent years as an attractive drug target for the prevention and treatment of diseases associated with insulin resistance (e.g., type 2 diabetes [T2D]) given its key role in metabolically relevant tissues such as adipose tissue, liver, and skeletal muscle (1). It has been suggested that AMPK plays a central role in the regulation of exercise metabolism in skeletal muscle (2) and as a regulator of muscle gene transcription and insulin sensitivity in exercise recovery (3–5). Consequently, elucidating the AMPK signaling network in muscle of patients with T2D is of great interest. On the basis of current evidence, AMPK expression and activity in the rested nonstimulated state have generally been reported to be similar in skeletal muscle of subjects with T2D and matched control subjects (6–9). In contrast, studies evaluating exercise-mediated AMPK signaling in skeletal muscle from patients with T2D are inconclusive (8,9). Hence, during moderate-intensity cycling exercise, Musi et al. (9) found a similar increase in total AMPK-α2 activity in muscle of lean subjects with or without T2D, whereas Sriwijitkamol et al. (8) reported impaired total AMPK-α2 activation in muscle of obese subjects with or without T2D compared with lean control subjects. In accordance, dysregulation of exercise-mediated AMPK signaling has been proposed as a possible mediator of impaired adaptations to exercise, including decreased induction of PGC-1α expression, potentially leading to muscle insulin resistance in T2D (8,10).

Human skeletal muscle apparently expresses only three AMPK heterotrimeric complexes: α2β2γ1 (∼65%), α2β2γ3 (∼20%), and α1β2γ1 (∼15%) (7). In healthy lean subjects, these complexes are regulated by exercise in a differential manner, depending on exercise duration and intensity (11). Evidently, measures of total AMPK activity or phosphorylation (as routinely reported) that reflect the activity of more than one heterotrimer are potentially biased because such measurements may conceal differential regulation among the three complexes (11,12). In the current study, we took advantage of this knowledge to investigate whether a possible dysregulation of AMPK signaling during exercise in skeletal muscle of patients with T2D is founded in differential trimer complex activation.

TBC1D4 and TBC1D1 are points of convergence for insulin- and exercise-dependent signaling (13,14). Akt2 and AMPK are putative upstream kinases of both TBC1D1 and TBC1D4 (15,16). Compelling evidence indicates that TBC1D4 regulates insulin-stimulated glucose transport in skeletal muscle (17). Interestingly, impaired insulin-regulated phosphorylation of TBC1D4 is observed in skeletal muscle of patients with T2D, potentially contributing to muscle insulin resistance (18–20). We previously reported that exercise training improves insulin signaling defects at the level of TBC1D4 in muscle of patients with T2D concomitantly with improved insulin action on glucose metabolism (20). Also, insulin-stimulated TBC1D4 signaling is upregulated in recovery from a single bout of exercise in rodents and healthy subjects (21–23). We thus investigated the hypothesis that TBC1D4 signaling defects in muscle of patients with T2D would be rescued by a single bout of exercise, potentially leading to enhanced insulin action in the period after exercise.

Exercise increases TBC1D1 phosphorylation on multiple sites in skeletal muscle from healthy humans (24–26), which has been proposed to regulate glucose transport during muscle contraction (27). Previous studies have shown that the increment in muscle glucose uptake and whole-body glucose disposal during exercise is normal in patients with T2D (28–30). Hence, we hypothesized that exercise-mediated regulation of TBC1D1 phosphorylation is also intact in patients with T2D.

To investigate these hypotheses, we examined a cohort of overweight/obese male subjects because this phenotype characterizes most male patients with T2D. We studied skeletal muscle signaling in patients with T2D and weight-matched control subjects at rest, immediately after, and in recovery from cycling exercise. Furthermore, insulin-stimulated muscle signaling was studied on two separate occasions: one trial was conducted 7 h into recovery from cycling exercise and another trial in the rested condition.

This study is part of a larger investigation designed to assess the effect of insulin, exercise, and postexercise insulin stimulation on multiple metabolic proteins in skeletal muscle of patients with T2D. The experimental design has been described thoroughly in a previous publication on this study in which detailed information can be found (31).

The study participants included overweight or obese male subjects: 13 with T2D and 14 without T2D. The two groups were matched according to age (T2D: 55 ± 2, control [Con]: 55 ± 2 years), BMI (T2D: 29.7 ± 1.0, Con: 29.0 ± 0.9 kg · m⁻²), and self-reported physical activity level (T2D: 5,558 ± 943, Con: 5,248 ± 952 MET min/week [International Physical Activity Questionnaire score]) (31). Patients with T2D were treated by diet alone (n = 4), by diet combined with metformin (n = 5), or by metformin and sulfonylurea (n = 4). All medications were withdrawn 1 week before both trials, and subjects were instructed to avoid strenuous exercise 48 h before each study day. Informed consent was obtained from all subjects before participation. The study was approved by The Regional Scientific Committees for Southern Denmark (Project-ID S-20100093) and performed in accordance with the Helsinki Declaration.

One week before study participation, maximal aerobic (VO_2max) (T2D: 3.22 ± 0.23, Con: 3.50 ± 0.17 L · min⁻¹) and maximal workload (T2D: 196 ± 20, Con: 236 ± 12 Watt) capacity were determined in all subjects. No significant differences were observed between the two groups (VO_2max: P = 0.34; maximal workload: P = 0.09). All subjects underwent two separate clamp trials (baseline day and exercise day) separated by 4–8 weeks (Fig. 1). On the baseline day, overnight-fasted subjects underwent a 4 h euglycemic-hyperinsulinemic clamp (40 mU · m⁻² · min⁻¹ of insulin) with tracer 3-[³H]glucose. These clamp data have previously been reported (31). Muscle biopsy specimens were obtained from two insertions (4–5 cm apart) in the vastus lateralis of one leg before and 4 h after insulin stimulation. When participants returned to the clinic on the exercise day after an overnight fast, they rested supine for 30 min. Next, a muscle biopsy specimen was obtained from the vastus lateralis, after which the subjects exercised on a cycle ergometer for 60 min at an intensity of 70% of VO_2max. Immediately after completion of exercise, a second muscle biopsy specimen was obtained from the vastus lateralis of the same leg (new incision) as before exercise. Whole-body VO₂ and VCO₂ were measured at 15 min and 45 min of exercise (Jaegers MasterScreen-CPX-system; Intramedic, Gentofte, Denmark). Subjects then rested for 3 h before undergoing a postexercise euglycemic-hyperinsulinemic clamp identical to the clamp on the baseline day. Muscle biopsy specimens were obtained from two incisions (4–5 cm apart) in the other leg before and after insulin stimulation.

Plasma glucose was measured on an ABL 800 Flex (Radiometer, Copenhagen, Denmark). Plasma lactate and serum insulin were measured as previously described (31).

Muscle lysates, SDS-PAGE, and Western blot analyses were prepared and performed as previously described (5,31).

Phosphorylation of Akt, ACC, TBC1D1, and TBC1D4 was measured using site-specific antibodies against TBC1D4 (Ser³¹⁸, Ser³⁴¹, Ser⁵⁸⁸, Thr⁶⁴², and Ser⁷⁰⁴), TBC1D1 (Ser²³⁷ and Thr⁵⁹⁶), Akt (Thr³⁰⁸ and Ser⁴⁷³), and ACC (Ser²²¹), as previously described (5,25). Besides antibodies against pTBC1D1 Ser²³⁷ (Millipore, Temecula, CA), pTBC1D4 Ser³⁴¹ (provided by Prof. Carol MacKintosh, University of Dundee, Scotland, U.K.), and Ser⁷⁰⁴ (provided by Dr. L.J. Goodyear, Joslin Diabetes Center, Boston, MA), all phospho-specific antibodies were purchased from Cell Signaling Technology (Danvers, MA). Total Akt2, ACC, TBC1D1, and TBC1D4 protein were determined using specific antibodies, as previously specified (25). Antibodies used for the AMPK activity assay were raised against AMPK-α2, AMPK-α1, and AMPK-γ3. All three AMPK antibodies were provided by Prof. D.G. Hardie (University of Dundee, U.K.).

TBC1D4 14-3-3 overlays were determined and performed as previously described (15,25).

Heterotrimer specific AMPK activity was measured on 300 μg of muscle lysate as previously described (11). In short, AMPK α2β2γ3 activity was measured on AMPK-γ3 immunoprecipitations (IPs) using AMPK-γ3 antibody. Activity of AMPK α2β2γ1 was measured on supernatants from the AMPK-γ3 IPs using AMPK-α2 antibody for a second IP, and AMPK α1β2γ1 activity was measured on supernatants from the AMPK-α2 IPs using AMPK-α1 antibody for a third IP. AMPK activity was analyzed on P81 filter paper using a Storm 850 PhosphorImager (Molecular Dynamics) (11). Total AMPK activity was calculated as the summed activity of the three heterotrimeric complexes.

Total RNA extraction and quantitative real-time RT-PCR analyses of skeletal muscle tissue biopsy specimens were performed as previously described (32). In short, quantitative real-time PCR was performed on a 7900HT Fast Real-Time PCR System (Applied Biosystems, Life Technologies) using TaqMan Custom Arrays. The data were analyzed using qBase+ Biogazelle software (Zwijnaarde, Belgium), with normalization to the geometric mean of two reference genes, PPIA and B2M.

Statistical analyses were performed using SigmaStat 3.5 software (Systat Software, San Jose, CA). Two-way repeated-measures ANOVA was used to assess the effect of insulin (baseline day and exercise day) and exercise/recovery (preexercise, exercise, and 3 h recovery data on the exercise day). Significant main effects or interactions were further analyzed by the Student-Newman-Keul post hoc test. Correlation analyses were assessed by the Spearman rank correlation coefficient. Differences were considered statistically significant at P < 0.05. Data are presented as means ± SEM.

During the 1-h cycling exercise, the subjects performed a similar relative workload reaching 71 ± 1% and 69 ± 3% of VO_2max (P = 0.47) corresponding to 141 ± 7 (Con) and 114 ± 12 (T2D) Watt (P = 0.052). By the end of exercise, the rate of perceived exhaustion (20 grade Borg scale; 15 ± 1 [Con] and 15 ± 1 [T2D]) and the pulmonary respiratory exchange ratio (0.89 ± 0.01 [Con] and 0.89 ± 0.01 [T2D]) were similar in the two groups. As previously reported from this study (31), muscle glycogen levels and utilization (∼50%) during exercise were similar between the two groups. In response to exercise, serum insulin concentrations decreased from 63 ± 12 to 34 ± 7 pmol/L (T2D) and from 46 ± 7 to 21 ± 3 pmol/L (Con), with no significant differences between the groups (Table 1). Serum insulin concentrations 3 h after exercise had returned to preexercise levels. The plasma glucose concentration at rest was higher in the patients with T2D than in control subjects and remained higher during exercise despite a modest decrease in the subjects with T2D only (Table 1). Plasma lactate concentrations were similar at rest and increased with exercise in both groups (Table 1). Plasma lactate concentrations were significantly higher in subjects with T2D (∼30%) than in control subjects during exercise. However, plasma lactate concentrations returned to resting levels in both groups 3 h after exercise cessation. These latter observations may relate to a higher glycolytic flux during exercise in the hyperglycemic patients with T2D (28).

Basal and insulin-stimulated whole-body glucose disposal rates (GDRs) during both clamp trials have previously been reported (31). In brief, the basal GDR was similar between groups on the baseline day (90 ± 4 [T2D] and 80 ± 2 [Con] mg · m⁻² · min⁻¹) and exercise day (93 ± 3 [T2D] and 87 ± 4 [Con] mg · m⁻² · min⁻¹). Insulin-stimulated GDR was ∼30% lower in patients with T2D compared with weight-matched control subjects (242 ± 35 [T2D] and 349 ± 35 [Con] mg · m⁻² · min⁻¹, baseline day) and was surprisingly not altered by prior exercise in either group (245 ± 28 [T2D] and 333 ± 28 [Con] mg · m⁻² · min⁻¹, exercise day). Serum insulin levels during the clamp were similar between groups and were similar (Con) or modestly lower (T2D) on the exercise day compared with the baseline day (Table 1).

We and others have previously reported no differences in basal AMPK activity in skeletal muscle of lean and obese subjects with or without T2D (6,8,9). This is consistent with the current study, in which no significant differences in total AMPK activity were found between male subjects with T2D and control subjects at any time on both study days (Fig. 2A). On the exercise day, total AMPK activity increased by ∼1.3-fold in response to exercise (P < 0.05) and, remarkably, remained elevated 3 h into recovery compared with preexercise (P < 0.05). From 3 to 7 h into recovery (postexercise insulin stimulation), total AMPK activity declined (P < 0.05) (Fig. 2A). Because insulin did not affect total AMPK activity on the baseline day, we interpret this decline primarily to be related to the time elapsed after cessation of exercise rather than to insulin per se. Changes in total AMPK activity were comparable to changes in AMPK Thr¹⁷² phosphorylation, although no significant increase in AMPK Thr¹⁷² phosphorylation was detected 3 h into exercise recovery compared with preexercise (31).

To clarify possible differences in AMPK heterotrimer complex activity, we analyzed the three specific trimers separately. On the exercise day, exercise increased α2β2γ3 activity similarly by approximately ninefold in both groups (P < 0.001) (Fig. 2B). Activity of α2β2γ3 at 3 h into recovery had declined (P < 0.001) but remained elevated in both groups (∼1.7-fold; P < 0.01) compared with preexercise levels. AMPK α2β2γ1 activity was unaffected by exercise (Fig. 2C), whereas α1β2γ1 activity decreased significantly in both groups (∼27%) (P < 0.05) (Fig. 2D). Compared with preexercise, at 3 h into exercise recovery, the activity of α2β2γ1 and α1β2γ1 both increased (P < 0.05 and P < 0.01, respectively). No significant differences in AMPK α1β2γ1 and α2β2γ1 activities were found between the groups on the exercise day. These findings demonstrate that activation of AMPK in skeletal muscle during exercise is not impaired by T2D per se. Furthermore, our results reveal a previously unrecognized activation of the two AMPK heterotrimeric complexes containing the γ1 regulatory subunit (α2β2γ1 and α1β2γ1) during exercise recovery. Interestingly, higher α2β2γ3 activity was observed among subjects with T2D 3 and 7 h into exercise recovery (P < 0.05) (Fig. 2B), reflecting somewhat differential regulation of the α2β2γ3 complex between the two groups after exercise.

To further clarify AMPK signaling, we evaluated the phosphoregulation of AMPK downstream targets ACC Ser²²¹ and TBC1D1 Ser²³⁷. On the exercise day, phosphorylation of these targets increased with exercise similarly in the two groups (ACC Ser²²¹, ∼6-fold [P < 0.001] [Fig. 3A] and TBC1D1 Ser²³⁷, ∼1.8 fold [P < 0.01] [Fig. 3B]). At 3 h into recovery, phosphorylation of both targets had declined (P < 0.01) but remained elevated compared with preexercise (P < 0.001). In accordance with the α2β2γ3 activity, phosphorylation of TBC1D1 Ser²³⁷ was higher in the patients with T2D than in control subjects during exercise recovery (P < 0.01). Thus, regulation of these two targets highly reflects regulation of AMPK α2β2γ3 activity as also indicated by the significant associations between these parameters; that is, the individual increments (exercise values − preexercise values) for AMPK α2β2γ3 activity significantly correlated with the increments in pACC Ser²²¹ (r = 0.56, P < 0.01, n = 27) as well as pTBC1D1 Ser²³⁷ (r = 0.47, P < 0.05, n = 27). As expected, neither of the two targets was regulated by insulin on the baseline day or exercise day.

Previous studies have suggested that AMPK regulates PGC-1α gene transcription (33). Measuring PGC-1α gene expression as a potential functional consequence of AMPK activation, we found that PGC-1α mRNA increased ∼6.5-fold 3 h into exercise recovery, with no significant differences between the groups (P < 0.001) (Fig. 4). Taken together, the regulation of downstream targets TBC1D1, ACC, and PGC-1α indicates intact AMPK signaling during exercise in muscle of male patients with T2D.

Being a critical signaling node for insulin action on glucose metabolism previously found to be dysregulated in T2D (18–20,35), we measured the ability of insulin to increase phosphorylation of Akt (Ser⁴⁷³ [∼2.3-fold, P < 0.001] [Fig. 5A] and Thr³⁰⁸ [∼5.7-fold, P < 0.001] [Fig. 5B]) and various downstream target sites on both TBC1D4 (Ser³¹⁸ [∼2.4-fold, P < 0.001] [Fig. 6A], Ser³⁴¹ [∼1.9-fold, P < 0.001] [Fig. 6B], Ser⁵⁸⁸ [∼1.7-fold, P < 0.001] [Fig. 6)], Thr⁶⁴² [∼1.9-fold, P < 0.001] [Fig. 6D], and Ser⁷⁰⁴ [∼1.5-fold, P < 0.001] [Fig. 6E]) and TBC1D1 (Ser⁵⁹⁶ [∼1.3-fold, P < 0.001] [Fig. 6F]) on the baseline day. A functional consequence of the changes in phosphorylation of TBC1D4 during insulin stimulation was an increase in the ability of TBC1D4 to bind 14-3-3 protein (∼1.6-fold, P < 0.001) (Fig. 6G). The effects of insulin on all aforementioned readouts were similar in the two groups on both experimental days, except for a reduced pAkt Thr³⁰⁸ in T2D on the exercise day. Thus, we did not find any impairment in muscle insulin signaling in this cohort of male patients with T2D compared with the weight-matched control subjects despite impaired GDR. Furthermore, neither group displayed upregulated insulin signaling after exercise alongside no improvement in whole-body GDR.

Some of the signaling events were decreased acutely during exercise (pAkt Ser⁴⁷³, pTBC1D4 Ser³¹⁸, Ser⁵⁸⁸, and Thr⁶⁴²), returning to (pAkt Ser⁴⁷³, pTBC1D4 Ser⁵⁸⁸, and Thr⁶⁴²) or increasing above (pTBC1D4 Ser³¹⁸) preexercise levels 3 h into recovery and further increasing during the insulin clamp. This pattern was similar in the two groups and coincided with changes in plasma insulin concentrations (Table 1).

Other signaling was upregulated during exercise (pTBC1D4 Ser⁷⁰⁴ [∼1.9-fold, P < 0.001] [Fig. 6E] and pTBC1D1 Thr⁵⁹⁶ [∼1.3-fold, P < 0.001] [Fig. 6F]) or 3 h into exercise recovery (pTBC1D4 Ser³⁴¹ [∼1.5-fold, P < 0.001] [Fig. 6B] and 14-3-3 protein binding [∼1.3-fold, P < 0.01] [Fig. 6G]). Besides pTBC1D1 Thr⁵⁹⁶, these signaling events were not further regulated by subsequent insulin stimulation during the clamp. In fact, pTBC1D4 Ser⁷⁰⁴ decreased from 3 to 7 h of recovery (Fig. 6E), likely as a consequence of the time elapsed from the cessation of exercise.

Here, we report that the ability of exercise to stimulate the AMPK signaling network in skeletal muscle from male subjects is not compromised by T2D per se. These observations question previous hypotheses indicating exercise intolerance at the level of AMPK and downstream target PGC-1α in patients with T2D (8,36). Normal metabolic signaling during exercise is also implied by other observations in patients with T2D; for example, normal glucose uptake across the working limb (28,30), similar changes in metabolic flexibility and transcriptional profile (29,37), and similar changes in skeletal muscle protein adaptations to exercise (20). In the study by Sriwijitkamol et al. (8) no significant activation of AMPK was observed in skeletal muscle in response to whole-body exercise at ∼50 and ∼75 Watt in obese and patients with T2D. The low number of subjects, the heterogeneity in age, sex, and fitness level, as well as the low whole-body exercise intensity might well have limited the ability to obtain a consistent and measurable activation of AMPK in that study (8). By AMPK heterotrimer complex separation, we found that only α2β2γ3 activity increased after 60 min of cycle exercise. This is similar to previous findings in lean healthy subjects (12). Also, in lean subjects with T2D, cycling exercise (at the same relative intensity as in the current study) increased total AMPK-α2 activity by ∼2.7-fold, a response similar to that observed in lean healthy subjects (9). This increment in AMPK-α2 activity is similar to the current study (∼2.1-fold increase in the summed activity of α2β2γ1 and α2β2γ3). Thus, intact AMPK signaling at rest and in response to exercise in skeletal muscle of male patients with T2D argues against a functional defect of AMPK in the pathogenesis of T2D. Bearing this in mind, activation of an intact AMPK signaling network by exercise or pharmacological interventions is promising for the treatment/prevention of muscle insulin resistance in patients with T2D.

In the current study, we extend to overweight/obese and patients with T2D our previous observations in lean healthy male subjects (11,12) that measurements of total AMPK activity may conceal differential regulation among the three heterotrimers. As a novel finding, we report that AMPK α1β2γ1 and α2β2γ1 activity increased during exercise recovery. Other studies in lean, obese, and T2D subjects have not been able to detect an increase in AMPK-α1–associated activity in recovery from exercise (8,9,12,38). This discrepancy may be related to differences in experimental design, including absolute exercise intensity, sex, and time. Elevated total AMPK-α2 activity postexercise, previously observed in lean healthy and T2D subjects (9), seems to be based on both elevated α2β2γ3 and α2β2γ1 activity. Because α2β2γ3 activity decreases and α2β2γ1 activity increases after exercise cessation, the relative contribution of the two complexes to total AMPK-α2–associated activity shifts during exercise recovery. Such differential regulation suggests different roles for the two heterotrimeric proteins. In skeletal muscle of humans, pigs, and mice, expression of an activating γ3 mutation induces glycogen storage (39–41). A similar phenotype is reported in muscle expressing an activating γ1 mutation (42). However, it may be speculated that this similar muscle-glycogen phenotype is brought about by different mechanisms. Thus, in rodent skeletal muscle, AMPK-γ3 is reported to be involved in glucose uptake, insulin sensitivity, and mitochondrial biogenesis (5,41,43), whereas overexpression of an active AMPK-γ1 subunit increases GLUT4 protein content and decreases PDK4 mRNA, potentially enhancing glucose uptake and glycolytic flux, respectively (42). Whether such roles of the heterotrimeric proteins also occur in human skeletal muscle awaits further studies. Furthermore, the mechanism(s) for the differential regulation of the three AMPK heterotrimeric complexes is at present unknown but highly relevant.

Evidence supports that the AMPK-regulated gene transcription factor PGC-1α is a master regulator of mitochondrial biogenesis in skeletal muscle (44) and may therefore play an important role in exercise adaptations. We found an identical response in PGC-1α mRNA expression after acute exercise in skeletal muscle of control subjects and subjects with T2D. This is in agreement with some (8,37) but not all studies (10). This further questions the concept of impaired exercise-mediated AMPK signaling in male patients with T2D. In support of this, the metabolic and transcriptional profile in response to exercise revealed no impairments in subjects with T2D compared with healthy control subjects (29).

We are the first to demonstrate intact exercise-induced phosphorylation of TBC1D1 Ser²³⁷ and Thr⁵⁹⁶ in skeletal muscle from overweight/obese male patients with T2D compared with weight-matched control subjects. These results are in line with recent studies in healthy lean (24,25) and obese (26) subjects. Mutational studies in mice have shown that lack of phosphorylation on multiple sites on TBC1D1, including Ser²³⁷ and Thr⁵⁹⁶, reduce contraction-stimulated glucose uptake in skeletal muscle (27,45). Furthermore, the TBC1D1 conventional knockout mouse model shows impaired muscle glucose uptake in response to exercise (46). Thus, our findings and those from mice are in accordance with the concept of normal exercise-induced metabolic regulation in skeletal muscle of patients with T2D (28,29,47). We previously found that contraction-stimulated phosphorylation of TBC1D1 Ser²³⁷ and Thr⁵⁹⁶ is abolished in AMPK-deficient murine skeletal muscle (15,25,48). Here, we report that the exercise-induced increase in Ser²³⁷ phosphorylation correlates with the increase in AMPK α2β2γ3 activity, in line with our previous observations in skeletal muscle from healthy lean subjects (25). Collectively, this indicates that the AMPK α2β2γ3 complex may regulate phosphorylation of TBC1D1 in skeletal muscle of healthy individuals and subjects with T2D during exercise.

Our study groups of male overweight/obese and subjects with T2D displayed characteristics indicating a healthier metabolic phenotype compared with cohorts of similar age and BMI previously investigated by us—this includes insulin sensitivity, physical fitness level, and habitual physical activity levels (20,35). Since we previously reported that exercise training normalizes molecular insulin resistance in patients with T2D at the level of TBC1D4 signaling (20), we speculate that the healthier phenotype of the present cohorts may explain the lack of impaired muscle insulin signaling in our patients with T2D compared with control subjects. Alternatively, defects in TBC1D4 phosphorylation may only be detectable at higher levels of insulin stimulation used by the latter study (∼900 vs. ∼400 pmol/L) (20). For whatever reason, this fact excluded us from testing the hypothesis that a single bout of exercise (rather than exercise training) is sufficient to normalize insulin signaling in patients with T2D.

We cannot rule out that the finding of intact AMPK signaling in our relatively insulin-sensitive cohort with T2D may not apply to more severe cases of T2D. Also, it is possible that exercise-mediated AMPK signaling is compromised in both of our study groups compared with healthy lean subjects. However, if such a defect was present, impairments in AMPK signaling would be a consequence of overweight/obesity per se rather than T2D. AMPK activity and signaling are dependent on exercise intensity (11), and thus, our data interpretation depends on the assumption that the exercise elicited equal stress to the muscle of the subjects in each group. In support of this, similar relative whole-body exercise intensity (∼70% VO_2max) was performed by the subjects eliciting similar muscle glycogen utilization. Although not measured in the current study, equal changes in muscle metabolites in control subjects and subjects with T2D during exercise (60% VO_2max) have been reported by others (28). Moreover, in muscle of healthy subjects, the exercise-induced gene response is indeed related to the relative and not the absolute workload (49), and thus, the similar induction of PGC-1α mRNA observed in the current study supports an equal muscular stress response in the two groups. Importantly, the higher plasma lactate accumulation during exercise in subjects with T2D compared with control subjects does not imply different metabolic stress at the level of muscle. This is so because glucose uptake and glycolytic flux are expected to be higher during exercise in the hyperglycemic (T2D) compared with the euglycemic condition (control), as also reported by others (28). Because the combustion of carbohydrate (based on the respiratory exchange ratio values) and muscle glycogen utilization (31) are similar in the two groups during exercise, higher glucose uptake and glycolytic flux would be expected to elicit a higher lactate release from the working muscle of the subjects with T2D.

In several previous reports on rodents and healthy lean humans, we and others have found that a single bout of exercise increases the ability of insulin to stimulate glucose uptake in the previously exercised muscle (22,50–52). Several studies suggest that a cross talk between exercise and insulin signaling may take place at TBC1D4, leading to this enhanced effect on glucose metabolism (5,23). As previously reported from this study (31), we did not detect improvements in whole-body GDRs after exercise. This was unexpected, but in a sense, the observation does not argue against the abovementioned relationship between TBC1D4 and insulin sensitivity after exercise. This is so because insulin-stimulated whole-body GDRs and phosphorylation of TBC1D4 were not improved after exercise. Although puzzling, one hypothesis for future evaluation is whether such potentiation by prior exercise is primarily seen in lean but not in obese subjects, although one report argues against this view (53).

Defects in glycogen synthase (GS) activity have consistently been observed in muscle of patients with T2D (35,54). This likely contributes to muscle insulin resistance and decreased ability to partition glucose toward storage in skeletal muscle (55). From the current study, we have previously reported lower GS activity and increased phosphorylation at sites 2+2a in patients with T2D during exercise recovery, indicating a diminished exercise response at the level of GS (31). A mutational study in mice has identified AMPK-α2 as a regulator of GS site 2 phosphorylation in skeletal muscle (56). Thus, the higher AMPK α2β2γ3 activity observed in patients with T2D during exercise recovery may account for the enhanced phosphorylation of GS site 2+2a compared with weight-matched control subjects. These changes, however, did not seem to affect glucose metabolism. Thus, whole-body glucose partitioning toward storage was not attenuated in subjects with T2D versus control subjects during a euglycemic-hyperinsulinemic clamp (31). Insulin-stimulated phosphorylation of Akt Thr³⁰⁸ was diminished in patients with T2D after exercise. Yet, this potential reduction in Akt activation was not reflected in downstream targets such as GSK-3 (31), TBC1D1, and TBC1D4 (present study), and thus, at present, we interpret this to be of minor importance for glucose metabolism.

The current study demonstrated that the AMPK signaling network is normally regulated in skeletal muscle of overweight/obese male patients with T2D in response to a single bout of exercise, including differential regulation of the AMPK heterotrimeric complexes as well as gene induction. We speculate that the activation of certain AMPK complexes long into exercise recovery is important for adaptations to exercise. We conclude that the adaptability of insulin resistant skeletal muscle to activate AMPK and downstream signaling after exercise is not compromised in male patients with T2D, leaving exercise interventions for treatment/prevention of diabetes applicable and attractive.

Acknowledgments. The authors thank L. Hansen, C.B. Olsen (Department of Endocrinology, Odense University Hospital, Denmark) and B. Bolmgren, I.B. Nielsen, and N.R. Andersen (Department of Nutrition, Exercise and Sports, University of Copenhagen, Denmark) for their skilled technical assistance.

Funding. This study was supported by grants from the Danish Council for Independent Research Medical Sciences (including Sapere Aude DFF Starting grant), The Novo Nordisk Foundation (including Excellence Grant 2009), The European Foundation for the Study of Diabetes (EFSD), The Danish Diabetes Academy supported by the Novo Nordisk Foundation, the research program (2016) “Physical Activity and Nutrition for Improvement of Health” funded by the University of Copenhagen Excellence Programme for Interdisciplinary Research, and Odense University Hospital.

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. R.K., J.R.H., R.S., and J.B.B. conducted the laboratory experiments. R.K., A.J.T.P., J.R.H., R.S., J.B.B., J.M.K., and K.H. contributed to analysis of data. R.K., A.J.T.P., and J.F.P.W. wrote the first version of the manuscript. A.J.T.P. conducted the in vivo experiments. A.J.T.P., K.H., and J.F.P.W. were responsible for the conception and design of study. All authors contributed to the interpretation of the results, revised the manuscript, and approved the final version. K.H. and J.F.P.W. are the guarantors of this work and, as such, had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Prior Presentation. Parts of this study were presented as an oral presentation at the 75th Scientific Sessions of the American Diabetes Association, Boston, MA, 5–9 June 2015.