A Single Bout of One-Legged Exercise to Local Exhaustion Decreases Insulin Action in Nonexercised Muscle Leading to Decreased Whole-Body Insulin Action

A single bout of exercise enhances insulin sensitivity for glucose uptake in the exercised muscle (1–4). However, while the evidence for improved insulin sensitivity in the prior exercised muscle is compelling, the evidence for enhanced whole-body insulin sensitivity after a single bout of exercise is not. This could indicate that insulin action of nonexercised muscle and/or other organs is decreased by a single bout of exercise. Some studies report enhanced whole-body insulin sensitivity (∼18–29%) after a single bout of exercise (1–48 h post exercise) (5–7), but a substantial number of studies do not (8–17). Although study design and subject characteristics vary among the studies, none of these factors solely seem to explain why many studies do not find enhanced whole-body insulin sensitivity after a single bout of exercise. To our knowledge, two studies have measured insulin sensitivity of nonexercised muscle in a body in which other muscles have been active. Annuzzi et al. (11) observed similar insulin-stimulated glucose uptake in the nonexercised forearm 1 day after leg exercise compared with the day before exercise. Despite enhanced insulin-stimulated glucose uptake in the prior exercised legs (cycling exercise), whole-body glucose disposal rate (R_d) was similar on the 2 days. In contrast, Devlin et al. (9) reported reduced insulin-stimulated glucose uptake in the nonexercised forearm muscles after a cycling exercise bout but no difference in whole-body R_d. They speculated that whole-body R_d might be the summation of the effects of various muscle groups (i.e., exercised and nonexercised muscle). However, insulin-stimulated glucose uptake was not measured in the exercised muscle, and its contribution to whole-body R_d was unknown. Thus, the aim of the current study was to investigate whether a single bout of exercise with one muscle group induces systemic changes that affect insulin action in nonexercised muscle. To explore this, we compared insulin-stimulated glucose uptake in the nonexercised and exercised leg on a day with prior one-legged knee-extensor exercise to insulin-stimulated glucose uptake in a rested leg on a control day with no exercise.

The study was approved by the Regional Ethics Committee for Copenhagen (H-6–2014–038; Copenhagen, Denmark) and complied with the guidelines of the 2013 Declaration of Helsinki. Written informed consent was obtained from all participants prior to entering the study.

Eight young (26 ± 1 years), lean (BMI 23.5 ± 0.4 kg/m²), moderately trained (48.5 ± 2.4 mL · min⁻¹ · kg⁻¹), and healthy men were included in the study and underwent a euglycemic-hyperinsulinemic clamp (EHC) on 2 separate days: one day with prior one-legged exercise to local exhaustion (exercise day) and another day without prior exercise (rest day). The 2 experimental days were performed in randomized order and separated by at least 11 days (Fig. 1).

A minimum of 1 week prior to the first experimental day, VO_2peak was determined by an incremental cycle test to exhaustion (Monark cycle ergometer, Ergomedic 839E) using breath-by-breath measurements of VO₂ (MasterScreen CPX; Intramedic A/S, Gentofte, Denmark). Body composition was measured by DXA (DPX-IQ Lunar; Lunar Corporation, Madison, WI). After familiarization to the one-legged knee-extensor ergometer (18), peak workload (PWL) of the knee extensors was determined in both legs by an incremental test. Prior to the first experimental day, subjects recorded food intake for 3 days, which they repeated prior to the second experimental day. Subjects abstained from alcohol, caffeine, and strenuous physical activity for 48 h prior to both experimental days.

On the morning of the experimental days, subjects arrived at the laboratory 1 h after having ingested a small breakfast (oatmeal, skimmed milk, and sugar; 5% of daily energy intake) (19). Upon arrival, they either rested for 2.5 h or performed one-legged exercise until local exhaustion with a randomized leg. The exercise protocol consisted of one-legged knee-extensor exercise for 1 h at 80% PWL with 5-min intervals at 90% PWL every 10 min. This was followed by 4-min intervals until exhaustion with 1 min at 50% PWL between intervals. The intensity for the 4-min intervals started at 100% PWL. When subjects were unable to maintain a kicking frequency of 60 rpm, intensity was lowered by 10 percentage points until subjects were unable to finish 4 min at 60% PWL. The average duration of the one-legged exercise was 2 h and 21 ± 6 min.

Subjects then rested in the supine position, and catheters (Pediatric Jugular Catheterization set; Teleflex Inc., Wayne, PA) were inserted into the femoral vein of both legs and a dorsal hand vein (Venflon Pro Safety; Mediq Danmark A/S, Brøndbyvester, Denmark) for sampling of arterialized venous blood (heated hand vein). In seven of the eight subjects, a primed (priming dose 2.6 mg · kg⁻¹), constant [6,6-²H₂]-glucose tracer infusion (0.044 mg · kg⁻¹ · min⁻¹) was initiated after 2 h of rest. Four hours after the exercise bout, the EHC was initiated with a bolus of insulin (9 mU · kg⁻¹) (Actrapid; Novo Nordisk, Bagsværd, Denmark) followed by 120 min of constant insulin infusion (1.42 mU · min⁻¹ · kg⁻¹). Glucose was infused during the EHC from a 20% glucose solution enriched with 1.9% [6,6-²H₂]-glucose and adjusted throughout the clamp to maintain euglycemia. Blood samples were drawn from all three catheters before (−60, −30, and 0 min) and during (15, 30, 45, 60, 80, 100, and 120 min) the EHC. Prior to blood sampling, femoral arterial blood flow was measured using the ultrasound Doppler technique (Philips iU22; ViCare Medical A/S, Birkerød, Denmark). Blood samples for the measurement of plasma enrichment of the glucose isotope were drawn from the heated hand vein prior to (−30, −15, and 0 min), in the beginning (25, 30, and 35 min) and in the end of the EHC (100, 110, and 120 min). Muscle biopsies of m. vastus lateralis were obtained in both legs immediately before and after the clamp under local anesthesia (∼3 mL Xylocaine 2%; AstraZeneca, Copenhagen, Denmark) using the Bergström needle technique with suction (20). We were unable to obtain a full set of biopsies from one subject, and thus, analyses performed on the muscle biopsies are n = 7.

Plasma glucose concentration was measured by a blood-gas analyzer (ABL800 FLEX; Radiometer Danmark, Brønshøj, Denmark). Plasma insulin concentration was measured using an Insulin ELISA kit (ALPCO). Plasma fatty acids (FAs) (NEFA C kit; Wako Chemicals Europe GmbH, Neuss, Germany) were measured using an enzymatic colorimetric method (Hitachi 912 automatic analyzer). Plasma enrichment of the labeled glucose isotope was measured using liquid chromatography mass spectrometry as previously described (21).

Muscle biopsies were freeze-dried for 48 h and dissected free of visible blood, fat, and connective tissue. Muscle biopsies were homogenized as previously described (22), and lysates were recovered by centrifuging the homogenates (20 min, 18,320g, at 4°C). Homogenate and lysate protein content were determined by the bicinchoninic acid method (Pierce Biotechnology Inc., Waltham, MA).

To measure protein expression and phosphorylation, samples were separated on self-cast gels using SDS-PAGE followed by semidry transfer of proteins on polyvinylidene difluoride membranes. Membranes were blocked for 5 min in 2% skimmed milk in TBS containing 0.05% Tween-20 followed by overnight incubation at 4°C in primary antibodies: anti–phospho-AktSer473 (Research Resource Identifier: AB_329825), anti–phospho-Akt^Thr308 (AB_329828), anti-Akt2 (AB_2225186), anti–phospho-TBC1D4^Thr642 (AB_2651042), anti–phospho-TBC1D4^Ser588 (AB_10860251), anti–phospho-TBC1D1^Thr596 (AB_10828720), and anti–phospho-GSK3α^Ser21/GSKβ^Ser9 (AB_329830) (Cell Signaling Technology, Danvers, MA); anti-GLUT4 (AB_2191454) (Thermo Fisher Scientific, Waltham, MA); anti–phospho-TBC1D4^Ser704 (custom made; Capra Science Antibodies AB, Ängelholm, Sweden); anti-TBC1D4 (AB_2818964) and anti-TBC1D1 (AB_2814949) (Abcam, Cambridge, U.K.); anti-GSK3β (AB_397601) (BD Transduction Laboratories, San Diego, CA); anti–glycogen synthase (GS) (custom made; Oluf Pedersen, University of Copenhagen); and anti–phospho-TBC1D4^Ser341, anti–pyruvate dehydrogenase (PDH), anti–phospho-PDH^site1, anti–phospho-PDH^site2, anti–phospho-GS^Ser7+10 (GS^2+2a), and anti–phospho-GS^Ser640+644 (GS^3a+3b) (custom made; Dr. David Grahame Hardie, University of Dundee, Dundee, U.K.). Membranes were incubated with horseradish peroxidase–conjugated secondary antibodies (Jackson ImmunoResearch Europe Ltd, Ely, U.K.) for 45 min at room temperature before visualizing protein bands with chemiluminescence (Millipore, Burlington, MA) and a ChemiDoc MP imaging system (Bio-Rad Laboratories, Hercules, CA). Some membranes were stripped and reprobed with a new primary antibody against another phosphorylation site or corresponding total protein after removal of the first antibody by incubation in stripping buffer (62.3 mmol/L Tris-HCl, 69.4 mmol/L SDS, double-distilled H₂O, and 0.08% β-mercaptoethanol, pH 6.7). The membranes were checked for successful removal of the initial primary antibody before reprobing.

Muscle glycogen was measured in homogenates (150 µg protein) as glycosyl units after acid hydrolysis determined by a fluorometric method (23). Muscle glucose, glucose-6-phosphate (G6P), and lactate were extracted with perchloric acid and measured fluorometrically (23).

GS activity was measured in homogenates in the presence of 0.02, 0.17, and 8 mmol/L G6P as previously described (24).

Insulin receptor substrate 1 (IRS1)–associated phosphoinositide 3-kinase (PI3K) activity was measured on immunoprecipitates of 200 μg of lysate protein using Protein G agarose and an IRS1 C-terminal antibody (Dr. K. Siddle, Cambridge University, Cambridge, U.K.) as previously described (2). Briefly, the PI3K activity reaction ran for 15 min at 30°C with L-α-phosphatidylinositol (PtdIns) as substrate (Merck, Branchburg, NJ) and ³³P γ-labeled ATP (Hartmann Analytic, Braunschweig, Germany). The reaction was stopped by addition of 5 N HCl, and PtdIns was extracted with chloroform:methanol (1:1) and spotted on a thin-layer chromatography silica gel plate (Merck). After thin-layer chromatography migration, the amount of PtdIns-incorporated ³³P was analyzed with exposure to phosphor imager screens followed by scanning in a Typhoon FLA 7000 IP+ (GE Healthcare, Brøndbyvester, Denmark).

Hepatic glucose production was calculated using Steele’s steady-state equations (25). Rates of glucose oxidation were calculated using the stoichiometric equations of Frayn (26). Nonoxidative R_d (NOGD) was calculated by subtracting the glucose oxidation rate from R_d. Exchange of substrates (glucose and lactate) was calculated by multiplying arteriovenous (A-V) differences with femoral arterial blood flow expressed per kilogram of lean leg mass (LLM). During one-legged knee-extensor exercise, only the quadriceps muscle is active (18). Assuming that the prior exercised quadriceps muscle caused the difference in insulin-stimulated glucose uptake between the two legs on the exercise day, we calculated an estimate for the insulin-stimulated glucose uptake exclusively for the quadriceps muscle:

where LGU is leg glucose uptake.

Based on the link between body compartments for a reference man (27), we calculated total leg muscle mass, which on average constituted ∼83% of the LLM measured by DXA. Thus, we calculated the muscle mass for the active quadriceps muscle based on the following assumptions: 1) 83% of lean thigh mass measured by DXA is muscle mass; and 2) quadriceps constitutes 40% of all thigh muscles (28). For the rested muscle, muscle mass was calculated as 83% of LLM with the assumption that all muscles in the rested leg behaved similarly.

To evaluate whether the reduced insulin action in nonexercised muscle was responsible for the reduced whole-body insulin action, we calculated an estimate of the “expected” difference in whole-body glucose disposal between the 2 days based on differences in glucose uptake of the muscles. For the simplicity of the calculation, we assumed that all nonexercised muscles had the same reduction in insulin-stimulated glucose uptake as the nonexercised leg (calculated as the difference between the nonexercised and rested leg), although indications of differential insulin-stimulated glucose uptake between arm and leg muscles exist (29). The reduction in insulin-stimulated glucose uptake of the nonexercised leg was multiplied by the estimated total inactive muscle mass. Total muscle mass was estimated based on assumptions that skeletal muscle constitutes ∼43% of total body weight (30). The amount of inactive muscle mass was calculated as total muscle mass minus the active muscle mass, which was estimated as described above. The total reduction of insulin-stimulated glucose uptake in nonexercised muscle minus the increased uptake in the exercised muscle will be termed the “expected” difference of whole-body glucose disposal:

where AUC is the area under the leg glucose uptake curve for the entire 120-min clamp, TI is total inactive, and m.m. is muscle mass.

This estimate was compared with the actual difference in whole-body glucose disposal between the 2 days (AUC for glucose infusion rate plus hepatic glucose production for the entire 120-min clamp).

Data are presented as means ± SEM. Glucose infusion rate was evaluated by paired t test. Insulin-stimulated glucose uptake, glucose A-V difference, and femoral arterial blood flow were evaluated by one-way repeated-measures ANOVA. Insulin-stimulated glucose uptake in the exercised muscle versus rested muscle was evaluated by paired t test. Two-way repeated-measures ANOVA was used to evaluate hepatic glucose production, whole-body R_d, arterial plasma glucose, insulin, FAs, respiratory exchange ratio, lactate release, PI3K, glycogen content, G6P, glucose, GS activity, and all data obtained by Western blotting. Significant main effects and interactions were evaluated by Tukey post hoc test. Differences were considered significant when P < 0.05. All statistical analyses were performed using Sigma Plot (version 13; Systat Software); N = 8, unless otherwise stated. Because there were no significant differences between the two rested legs on the rest day on all parameters, we have chosen to present these data as mean of the two legs on the rest day as “rest leg.”

The data generated and analyzed during the current study are available from the corresponding author upon reasonable request.

Glucose infusion rate was significantly lower (∼20%) on the exercise day compared with the rest day (Table 1). This was not due to differences in arterial plasma insulin and glucose concentrations during the clamp between the 2 days (Table 1). Hepatic glucose production was suppressed in response to insulin to a similar extent on the rest day and exercise day (∼71% and ∼76%, respectively) (Table 1). Thus, whole-body R_d was lower (∼18%) on the exercise day compared with the rest day (Table 1). The reduced R_d was not due to differences in glucose oxidation but likely due to a lower whole-body NOGD in response to insulin on the exercise day (∼28%; P = 0.12) (Table 1).

On the exercise day, 4 h after one-legged exercise to local exhaustion, insulin-stimulated glucose uptake was higher in the prior exercised leg compared with the nonexercised leg (Fig. 2A). During one-legged knee-extensor exercise, only the quadriceps muscle is active (18), but the leg glucose uptake measured represents the entire exercise leg, which also includes glucose uptake of other nonexercised muscle groups. Thus, we calculated an estimate of the insulin-stimulated glucose uptake specifically for the prior exercised quadriceps muscle. This revealed a significantly higher (∼139%) insulin-stimulated glucose uptake in the quadriceps muscle of the exercised leg compared with the rested quadriceps muscle on the rest day (Fig. 2B). In contrast, insulin-stimulated glucose uptake was reduced in the nonexercised leg compared with the rest leg (∼37%) (Fig. 2A). This was due to reduced glucose extraction (Fig. 2C) and not reduced femoral arterial blood flow (Fig. 2D). To evaluate if the reduced insulin action in the nonexercised muscle could account for the reduced whole-body insulin action, we used the reduction in insulin-stimulated glucose uptake in nonexercised muscle to estimate the expected difference in whole-body glucose disposal. This was done with the assumption that all nonexercised muscles of the body had similar insulin-stimulated glucose uptake. The expected difference in whole-body glucose disposal could account for ∼79% of the actual difference in whole-body glucose disposal, suggesting that the reduced insulin-stimulated glucose uptake in nonexercised muscle was mainly responsible for the reduction in whole-body R_d.

Immediately after cessation of the one-legged exercise, the levels of plasma adrenaline and noradrenaline were higher (101% and 225%, respectively) compared with samples obtained after 2.5 h of rest on the rest day (Fig. 3A and B). This was measured in five of the eight subjects and occurred in all five. Prior to initiation of the EHC, the levels had returned to baseline and were similar between the 2 days. Likely as a consequence of the catecholamine response to the one-legged exercise bout, the FA concentration in plasma was higher (41%) on the exercise day compared with the rest day before but not during EHC (Fig. 3C). This was evident in all of the eight subjects.

Phosphorylation of PDH in the muscle (indirect measure of PDH activity, the gateway to glucose oxidation) (data not shown) was not different between the rested and nonexercised leg. In addition, neither muscle lactate content at the end of the clamp (71.5 ± 3.4 vs. 69.1 ± 3.8 mmol · kg dw⁻¹, rested and nonexercised muscle, respectively) nor lactate release from the muscle in the last 40 min of the clamp (3.0 ± 0.8 vs. 3.5 ± 0.7 µmol · min⁻¹ · kg⁻¹, rested and nonexercised leg, respectively) were different between the 2 days. Collectively, these results indicate that glucose oxidation was not impaired in the nonexercised muscle. This is in accordance with the calculated whole-body glucose oxidation rates, which increased in response to insulin to the same extent on both days (Table 1).

As the majority of insulin-stimulated glucose uptake is directed toward glucose storage (31), we speculated whether dysregulation of GS could be involved in the reduced insulin-stimulated glucose uptake of the nonexercised leg. However, the phosphoregulation of key regulatory sites on GS (Ser^2+2a and Ser^3a+3b) (Fig. 4A and B), activation by insulin (Fig. 4C), or the total activity of GS (data not shown) were not different between the rested and nonexercised leg. In addition, glycogen, G6P, and glucose levels in muscle were similar between the rested and nonexercised leg (Fig. 4D–F). Thus, the intracellular machinery to handle glucose seems not to be impaired in the nonexercised muscle, leading to the question whether the transport step of glucose was affected in the nonexercised muscle on the exercise day.

Muscle GLUT4 expression was similar between the legs (Fig. 5A). To investigate whether insulin signaling to GLUT4 translocation was impaired, we measured proximal insulin signaling at the level of PI3K and Akt2. In response to insulin, IRS1-associated PI3K activity increased in the muscles of both the rested, nonexercised and exercised leg. Although the absolute activity of PI3K seemed to be reduced on the exercise day, this was not significant (P = 0.37), perhaps due to low power (0.06) (Fig. 5B). Downstream of PI3K, Akt2 was phosphorylated in response to insulin on both Akt2^Thr308 and Akt2^Ser473 (Fig. 5C and D). Phosphorylation of Akt2^Thr308, which has previously been shown to reflect Akt2 activity (32), was slightly impaired on the exercise day in both the nonexercised and exercised leg compared with the rest leg on the rest day (Fig. 5C). Akt2 protein expression was not altered by prior exercise (Fig. 5E).

The phosphorylation of Akt downstream targets TBC1D4^Thr642, TBC1D4^Ser588, TBC1D4^Ser704, TBC1D4^Ser341, and TBC1D1^Thr596 all increased similarly in response to insulin on both days (Fig. 6A–E), except for TBC1D4^Ser704 in the prior exercised leg, as it was already highly phosphorylated prior to insulin stimulation (Fig. 6C). Phosphoregulation of other Akt substrates, GSK3β^Ser9 and GSK3α^Ser21, was also not impaired in the nonexercised muscle (Fig. 7A and B), questioning the cellular consequence of the impaired Akt signaling at least toward GLUT4 translocation (TBC1D1/TBC1D4) and glycogen synthase activation (GSK3).

In the prior exercised leg, phosphorylation of TBC1D4, but not TBC1D1 and GSK3, was significantly higher compared with the rested and nonexercised leg with insulin stimulation (Fig. 6B–D). This is line with our previous findings (33) and is hypothesized to be a key event regulating the insulin-sensitizing effect of prior exercise (34,35). TBC1D4, TBC1D1, and GSK3 protein expression was not altered by prior exercise (Fig. 6F).

In this study, we report that a single bout of one-legged knee-extensor exercise to local exhaustion (∼2.5 h) reduces whole-body glucose infusion rate compared with a day without prior exercise. This was mainly due to reduced insulin-stimulated glucose uptake in the nonexercised muscle and not due to impaired insulin suppression of hepatic glucose production. Indications of this phenomenon were revealed by us recently (17). However, due to design of that study (17), we were unable to determine whether the lower insulin-stimulated glucose uptake at the whole-body level and in nonexercised muscle on the exercise day was due to the exercise bout, consumption of a carbohydrate-rich diet, or lack of randomization between the experimental days. Based on the findings in the current study, it seems evident that one-legged knee-extensor exercise to local exhaustion reduces insulin action in the nonexercised muscle.

The findings from the current study cannot answer whether this relates to all exercise modalities, exercise intensity, and/or duration. In previous one-legged exercise studies of ours, using a less severe exercise protocol (1 h at ∼80 PWL) with a lower glycogen degradation during the exercise bout compared with the current study, the average insulin-stimulated glucose uptake of the nonexercised leg was ∼40 µmol · min⁻¹ · kg LLM⁻¹ (4,36,37). Compared with the insulin-stimulated glucose uptake of the nonexercised leg in the current study (23.5 ± 4.3 µmol · min⁻¹ · kg LLM⁻¹), our data suggest that the reduced insulin action in nonexercised muscle may associate to the duration and/or intensity of the exercise protocol.

In two studies, insulin action in the nonexercised forearm muscle was measured in response to a single bout of cycling exercise (9,11). In the early recovery (2 h) from an exhaustive cycling exercise bout (∼84 min, 70% VO_2max for 15-min intervals, 5-min recovery), the increase in insulin-stimulated glucose uptake was completely abolished in the nonexercised forearm muscle (9). This supports our findings of decreased insulin action in the nonexercised leg. In contrast, the day after a long-duration, low-intensity cycling exercise bout (3 h at 50% VO_2max), insulin-stimulated glucose uptake was not impaired in the nonexercised muscle (11). Collectively, these studies support that exercise intensity and/or duration likely influence the response in nonexercised muscle. In line with this, Bergström and Hultman (38) reported a lower rise in muscle glycogen concentration in the nonexercised leg after an exhaustive one-legged cycling bout upon glucose infusion when compared with a group who had not exercised. The findings were, however, not significant, perhaps due to an unpaired study design (38). Collectively, these results suggest that exercise may, but not always, lead to reduced insulin action in nonexercised muscle, and this may lead to decreased whole-body R_d. Interestingly, both of the previous studies investigating insulin action in nonexercised muscle found that whole-body R_d was unaffected by the prior exercise bout (9,11). Because the subjects in both studies performed exercise with both legs (i.e., more muscle mass), we speculate that the whole-body response to a single bout of exercise might be dependent on the amount of active muscle mass, counteracting a potential reduction in insulin action of the inactive muscles. We propose that this complex relationship is part of the, at times, unpredictable effects of insulin in daily life of physically active patients treated with insulin—in particular, patients with type 1 diabetes in whom peripheral insulin action is not compromised. Fear of losing glycemic control often keeps insulin-dependent patients from engaging in regular physical activity (39,40). Thus, from an applied point of view, our observations may improve the understanding of the complexity of controlling glucose homeostasis in physically active insulin-dependent patients, and our data support that including “exercise factors” such as exercise intensity/duration and the amount of muscle mass activated could be important for the prediction of the optimal insulin-dose algorithm (41). Further investigation is, however, needed to evaluate if the response at a whole-body level and in nonexercised muscles is similar in other groups of subjects (e.g., females, obese, subjects with type 1 diabetes, etc.) and in more applied settings.

The molecular mechanisms for the reduced insulin-stimulated glucose uptake in the nonexercised muscle have not previously been investigated. Noradrenaline has been shown to decrease insulin sensitivity (42), and although we found higher levels of noradrenaline and adrenaline immediately after exercise, the levels were not different between the 2 days prior to initiation of the clamp. This suggests that catecholamines per se were not responsible for the reduced insulin-stimulated glucose uptake in the nonexercised muscle. However, the catecholamine response likely increased lipolysis in the adipose tissue, leading to the increased FA levels in plasma on the exercise day. Similarly, Devlin et al. (9) also found higher FA levels in plasma on the exercise day concomittantly with reduced insulin action in the nonexercised forearm. Markedly increased levels of plasma FAs can induce insulin resistance at the whole-body level (36,43–49) and in skeletal muscle (36,44,45,50). In our study, plasma FA levels were moderately elevated on the exercise day compared with the rest day. Lipid-induced insulin resistance may involve impaired intracellular metabolism (45,46,51) and/or insulin-stimulated glucose transport (43,47–49). We did not find any evidence of impaired intracellular glucose oxidation or impaired activation of GS (rate limiting for glycogen synthesis). Yet, glucose storage may still be compromised if the transport of glucose is inhibited. Such interpretation is supported by our findings of the trend toward reduced IRS-associated PI3K activation and the significantly reduced phosphorylation of Akt2^Thr308, which likely reflects Akt2 activity (32). Yet, our analysis of downstream targets of Akt did not reveal impaired distal signaling. Thus, our results do not support that TBC1D4-regulated GLUT4 translocation was compromised in the nonexercised leg. Akt has numerous downstream targets, some of which are involved in other parts of the GLUT4 machinery (52). Thus, it is possible that other aspects of GLUT4 trafficking are affected.

In conclusion, a single bout of one-legged knee-extensor exercise to local exhaustion enhances insulin action in the exercised muscle while at same time reducing insulin action in nonexercised muscle. Thus, the effect of a single bout of exercise on whole-body insulin action depends on the balance between local effects increasing and systemic effects decreasing insulin action. The molecular explanations remain to be elucidated but could involve impairments in the PI3K–Akt signaling axis. Under the condition applied, these changes may secure that glucose is spared (decreased action in nonexercise leg) and directed (increased action in exercise leg) to the muscle in need. As we in this study present another factor in the equation for the whole-body insulin response following physical activity, we may have provided insights to improve the insulin-dose algorithm to the future benefit of insulin-treated, physically active patients.

Acknowledgments. The authors thank Nicoline R. Andersen, Betina Bolmgren, and Irene B. Nielsen (University of Copenhagen, Copenhagen, Denmark) for the skilled technical assistance and Dr. O.B. Pedersen (University of Copenhagen, Copenhagen, Denmark) and Dr. K. Siddle (Cambridge University, Cambridge, U.K.) for the kind donation of material essential for this work.

Funding. D.E.S. was supported by a research grant from the Danish Diabetes Academy funded by the Novo Nordisk Foundation. K.A.S. was supported by a postdoctoral research grant from the Council for Independent Research/Medicine (grant 4092-00309). This study was supported by grants from the Danish Council for Independent Research Medical Sciences (to J.F.P.W.), the Novo Nordisk Foundation (to J.F.P.W.), the Lundbeck Foundation (to J.F.P.W.), the Ministry of Culture Denmark (FPK 2016-0027 to J.F.P.W.), and the research program “Physical Activity and Nutrition for Improvement of Health” (2016), funded by the University of Copenhagen Excellence Programme for Interdisciplinary Research.

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

Author Contributions. D.E.S. and J.F.P.W. were responsible for conception and design of research. D.E.S., J.R.H., A.T., J.M.K., K.A.S., B.K., E.A.R., and J.F.P.W. performed experiments. D.E.S. and J.B.B. performed analyses. All authors interpreted results. D.E.S. and J.F.P.W. drafted the manuscript. All authors edited and revised the manuscript and read and approved the final version. J.F.P.W. is the guarantor of this work and, as such, had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.