Exercise training reduces the insulin‐sensitizing effect of a single bout of exercise in human skeletal muscle

Abstract

Key points

• A single bout of exercise is capable of increasing insulin sensitivity in human skeletal muscle. Whether this ability is affected by training status is not clear.

• Studies in mice suggest that the AMPK‐TBC1D4 signalling axis is important for the increased insulin‐stimulated glucose uptake after a single bout of exercise.

• The present study is the first longitudinal intervention study to show that, although exercise training increases insulin‐stimulated glucose uptake in skeletal muscle at rest, it diminishes the ability of a single bout of exercise to enhance muscle insulin‐stimulated glucose uptake.

• The present study provides novel data indicating that AMPK in human skeletal muscle is important for the insulin‐sensitizing effect of a single bout of exercise.

Abstract

Not only chronic exercise training, but also a single bout of exercise, increases insulin‐stimulated glucose uptake in skeletal muscle. However, it is not well described how adaptations to exercise training affect the ability of a single bout of exercise to increase insulin sensitivity. Rodent studies suggest that the insulin‐sensitizing effect of a single bout of exercise is AMPK‐dependent (presumably via the α₂β₂γ₃ AMPK complex). Whether this is also the case in humans is unknown. Previous studies have shown that exercise training decreases the expression of the α₂β₂γ₃ AMPK complex and diminishes the activation of this complex during exercise. Thus, we hypothesized that exercise training diminishes the ability of a single bout of exercise to enhance muscle insulin sensitivity. We investigated nine healthy male subjects who performed one‐legged knee‐extensor exercise at the same relative intensity before and after 12 weeks of exercise training. Training increased ${\dot{V}}_{{\mathrm{O}}_2\mathrm{peak}}$ and expression of mitochondrial proteins in muscle, whereas the expression of AMPKγ3 was decreased. Training also increased whole body and muscle insulin sensitivity. Interestingly, insulin‐stimulated glucose uptake in the acutely exercised leg was not enhanced further by training. Thus, the increase in insulin‐stimulated glucose uptake following a single bout of one‐legged exercise was lower in the trained vs. untrained state. This was associated with reduced signalling via confirmed α₂β₂γ₃ AMPK downstream targets (ACC and TBC1D4). These results suggest that the insulin‐sensitizing effect of a single bout of exercise is also AMPK‐dependent in human skeletal muscle.

Key points

• A single bout of exercise is capable of increasing insulin sensitivity in human skeletal muscle. Whether this ability is affected by training status is not clear.

• Studies in mice suggest that the AMPK‐TBC1D4 signalling axis is important for the increased insulin‐stimulated glucose uptake after a single bout of exercise.

• The present study is the first longitudinal intervention study to show that, although exercise training increases insulin‐stimulated glucose uptake in skeletal muscle at rest, it diminishes the ability of a single bout of exercise to enhance muscle insulin‐stimulated glucose uptake.

• The present study provides novel data indicating that AMPK in human skeletal muscle is important for the insulin‐sensitizing effect of a single bout of exercise.

Introduction

Muscle insulin sensitivity increases as an adaptive response to chronic exercise training (Dela et al. 1992; Holten et al. 2004; Frøsig et al. 2007a) involving changes in muscle size, morphology, capillarization and protein composition. In addition, in response to a single exercise bout, the prior exercised muscle responds with an acute increase in insulin sensitivity to stimulate glucose uptake lasting for up to 48 h (Mikines et al. 1988). Together, these adaptations secure enhanced insulin sensitivity during a period of exercise training and may partly explain the health‐promoting effects of physical activity. Improved insulin sensitivity following a single bout of exercise is described in skeletal muscle of various species (Richter et al. 1982, 1989; Bonen et al. 1984; Garetto et al. 1984; McConell et al. 2015). Observations from isolated rodent muscle preparations (Richter et al. 1982; Garetto et al. 1984) and one‐legged exercise models in humans (Richter et al. 1989; Wojtaszewski et al. 1997; Frøsig et al. 2007b) indicate a central role for local contraction‐induced mechanisms within the skeletal muscle in the improved insulin sensitivity after a single bout of exercise (Richter et al. 1984; Cartee, 2015). In rodents, the improved insulin sensitivity associates with an increased abundance of glucose transporter 4 (GLUT4) at the muscle cell surface membrane (Hansen et al. 1998). Accordingly, the point of convergence in cellular signalling events leading to GLUT4 translocation elicited by exercise and insulin has been a matter of active research for years. Because prior exercise does not alter the proximal insulin signalling (Bonen et al. 1984; Wojtaszewski et al. 1997, 2000; Hamada et al. 2006; Frøsig et al. 2007b), current hypotheses suggest that more distal signalling molecules might be involved. In this context, TBC1D4, which is involved in insulin‐stimulated glucose transport (Sano et al. 2003; Kramer et al. 2006), has been proposed as a signalling point of convergence between exercise and insulin (Cartee & Wojtaszewski, 2007; Cartee, 2015). In support, both human and rodent studies find increased phosphor‐regulation of TBC1D4 by insulin in the recovery period from acute exercise concomitantly with increased insulin sensitivity (Funai et al. 2009; Treebak et al. 2009; Castorena et al. 2014).

We recently demonstrated, in the skeletal muscle of mice, that both AICAR, a potent AMPK activator, and contraction/exercise increased insulin‐stimulated glucose uptake in an AMPK‐dependent manner (Kjøbsted et al. 2015, 2017). This was associated with site‐specific phosphorylation of TBC1D4, which specifically depended on the AMPK heterotrimeric complex, α₂β₂γ₃ (Kjøbsted et al. 2017). Whether this also applies to humans is unclear. In human skeletal muscle, three heterotrimeric complexes are detectable (α₁β₂γ₁, α₂β₂γ₁ and α₂β₂γ₃). Of these, the α₂β₂γ₃ complex is activated potently and rapidly in response to acute exercise (Birk & Wojtaszewski, 2006). Intriguingly, the expression of the γ₃ subunit is highly responsive to muscle use (expression decreases) (Frøsig et al. 2004; Wojtaszewski et al. 2005; Mortensen et al. 2013) and disuse (expression increases) (Kostovski et al. 2013). Accordingly, the AMPK α₂β₂γ₃ complex is much less activated during exercise in the trained compared to the untrained muscle, even when exercise is performed at the same relative intensity (Mortensen et al. 2013). If the α₂β₂γ₃ complex is central for the insulin‐sensitizing effect of acute exercise also in human skeletal muscle, a consequence of the above observations would be that the ability of the trained muscle to improve insulin sensitivity in response to acute exercise would be diminished.

Only two human studies bring some but inconclusive insights to this scenario. Mikines et al. (1989) did not find an increased insulin sensitivity 1 h after acute cycling exercise (75% ${\dot{V}}_{{\mathrm{O}}_2\mathrm{peak}}$) in a group of well‐trained subjects. However, the insulin sensitivity measured after acute exercise was compared with a ‘control/resting’ condition, which was only 15 h after an exercise training session. It is thus conceivable that insulin sensitivity under both conditions were improved by the prior exercise bout (Mikines et al. 1988). Furthermore, in the ‘control’ condition, muscle glycogen stores were not fully replenished (Mikines et al. 1989). As glycogen levels have been reported to influence the ability of insulin to increase glucose uptake (Jensen et al. 1997; Derave et al. 2000; Richter et al. 2001; Wojtaszewski et al. 2003a), this could also contribute to the absence of improved insulin sensitivity after acute exercise reported in their study (Mikines et al. 1989). In another study, whole‐body insulin sensitivity index was increased after acute exercise in a group of sedentary obese subjects but not in a group of regularly physically active obese subjects (Nelson & Horowitz, 2014). The physical fitness was, however, low in both groups and did not differ significantly between the groups. Together, these two studies suggest that training status might affect the ability of acute exercise to enhance insulin sensitivity. Yet the results are inconclusive and these observations at the whole‐body level cannot be related specifically to skeletal muscle.

Thus, in the present study, we tested the hypothesis that exercise training diminishes the ability of acute exercise to enhance insulin‐stimulated glucose uptake. We predicted that this was associated with decreased AMPKα₂β₂γ₃ expression/activation and thus lesser exercise‐ and insulin‐induced TBC1D4 signalling after training. Such observations will strengthen the case in favour for AMPKα₂β₂γ₃ being a key regulator of insulin sensitivity in human skeletal muscle. We explored this hypothesis by comparing muscle insulin‐stimulated glucose uptake 4 h after one‐legged knee‐extensor exercise before and after 12 weeks of whole‐body cycling exercise training.

Methods

Ethical approval

Nine young (aged 25 ± 1 years), lean (body mass index 23.5 ± 0.5 kg m^–2) and healthy men gave their written, informed consent to participate in the study approved by the Regional Ethics Committee for Copenhagen (H‐6‐2014‐038) and complied with the ethical guidelines of the Declaration of Helsinki II, except for registration in a database.

Experimental protocol

The experimental protocol consisted of two experimental days separated by 12 weeks of endurance cycling training (Fig. 1). Minimum one week prior to the PRE training experimental day, peak oxygen uptake (${\dot{V}}_{{\mathrm{O}}_2\mathrm{peak}}$) was determined by an incremental test to exhaustion on a Monark ergometer cycle (Ergomedic 839E; Monark Exercise AB, Vansbro, Sweden) using breath by breath measurements of ${\dot{V}}_{{\mathrm{O}}_2}$ (Masterscreen CPX; IntraMedic, Gentofte, Denmark). Body composition was measured by dual x‐ray absorptiometry (DPX‐IQ Lunar; Lunar Corporation, Madison, WI, USA). After familiarization to the one‐legged knee‐extensor ergometer (Andersen et al. 1985), peak workload (PWL) of the knee‐extensors was determined in both legs by an incremental test. Subjects were instructed to record food intake for 3 days and to abstain from alcohol, caffeine and strenuous physical activity for 48 h prior to the PRE training experimental day.

Figure 1. Experimental study design.

Subjects underwent two experimental days separated by 12 weeks of endurance cycling training (PRE and POST training experimental day). On each experimental day, subjects performed an acute bout of one‐legged exercise (80% peak workload (PWL) interspersed with 3 × 5 min intervals at 100% PWL). After 4 h of rest, a 2 h euglycemic hyperinsulinaemic clamp (EHC) was initiated. Biopsies (B) in the prior exercised and rested legs were taken immediately before and after EHC. Prior to the PRE experimental day and in the 12th training week, body composition, ${\dot{V}}_{{\mathrm{O}}_2\mathrm{peak}}$ and PWL were determined and fasting plasma glucose and plasma insulin levels were measured.

On the morning of the experimental day, subjects arrived at the laboratory 1 h after having ingested a small breakfast (oatmeal, skimmed milk, sugar; 5% of daily energy intake) (Henry CJ, 2005). Upon arrival, they performed 1 h of dynamic knee‐extensor exercise, with a randomized leg, at 80% of PWL interspersed with 3 × 5 min at 100% of PWL. After exercise, subjects rested in the supine position and catheters (Pediatric Jugular Catherization set; Arrow International, Reading, PA, USA) were inserted into the femoral vein of both legs and in a dorsal hand vein (Venflon Pro Safety; Mediq, Brøndby, Denmark) for sampling of arterialized venous blood (heated hand vein). After 4 h of rest, a euglycaemic hyperinsulinaemic clamp (EHC) was initiated with a bolus of insulin (9 mU kg⁻¹; Atrapid; Novo Nordisk, Bagsværd, Denmark) followed by 120 min of constant insulin infusion (1.4 mU min⁻¹ kg⁻¹). Blood samples were drawn simultaneously from all three catheters before (−60, −30 and 0 min) and during the EHC (15, 30, 45, 60, 80, 100 and 120 min). Prior to each blood sampling, femoral arterial blood flow was measured using the ultrasound Doppler technique (Philips iU22; ViCare Medical A/S, Birkerød, Denmark). This allowed for calculation of the glucose uptake using Fick's principle across the previously exercised and rested leg, respectively. Muscle biopsies of musculus vastus lateralis were obtained in both legs immediately before and after the clamp under local anaesthesia (∼3 mL of xylocaine 2%; AstraZeneca, Copenhagen, Denmark) using the Bergström needle technique with suction (Bergström, 1962).

The experimental day was repeated in the same way after completing 12 weeks of training (POST training experimental day). For the one‐legged knee‐extensor exercise to be performed at the same relative intensity, the absolute workload was increased compared to the PRE training experimental day. Subjects repeated their 3 day diet regime and abstained from alcohol, caffeine and strenuous physical activity for 48 h prior to the POST training experimental day. Both PRE and POST training, a venous blood sample was taken on a separate day after an overnight fast to measure fasting plasma glucose and insulin concentrations (antecubital vein). By using the one‐legged knee extensor model, the resting leg serves as a within subject control leg. This design also gives the advantages that the subjects only had to go through the invasive procedure twice (one before and one after training).

Training regime

To ensure feasibility and high compliance to the training regime, we chose to use indoor cycling for the 12 weeks of training consisting of 4 × 1 h of indoor cycling exercise per week (both legs) (Body bike supreme classic; Pedan, Køge, Denmark). The intensity of the training sessions ranged from 75% to 90% of maximal heart rate measured by Polar heart rate monitors (Polar CS400; Polar, Kempele, Finland). Three of the four weekly training sessions were performed at the subjects' home residence, whereas the fourth training session was performed at our laboratory. Throughout the 12 weeks of training, subjects were instructed to continue their habitual diet, remain weight stable and measure resting heart rate in the morning (3 days a week). In the 12th training week, two of the training sessions were substituted with a ${\dot{V}}_{{\mathrm{O}}_2\mathrm{peak}}$ and PWL test, respectively. The last training session was performed 48–72 h prior to the POST training experimental day.

Analysis of plasma samples

Plasma glucose concentration was measured by a blood‐gas analyser (ABL800 FLEX; Radiometer, Copenhagen, Denmark). Plasma insulin concentration was measured using an insulin enzyme‐linked immunosorbent assay kit (ALPCO, Salem, NH, USA). The concentration of plasma fatty acids (NEFA C kit; Wako Chemicals GmbH, Neuss, Germany) and triacylglycerol (GPO‐PAP kit; Roche Diagnostics, Mannheim, Germany) were measured using enzymatic colorimetric methods (Hitachi 912 automatic analyser; Hitachi, Mannheim, Germany).

Muscle homogenate and lysate preparation

Muscle biopsies were freeze dried for 48 h and dissected free of visible blood, fat and connective tissue. Muscle homogenates were generated as described previously (Kristensen et al. 2015). Lysates were recovered by centrifuging the homogenates (18,320 g for 20 min at 4°C). Homogenate and lysate protein content were determined by the bicinchoninic acid method (Pierce Biotechnology, Rockford, IL, USA).

SDS‐PAGE and Western blotting

To measure protein expression and phosphorylation, samples were separated on self‐cast gels using SDS‐PAGE followed by semi‐dry transfer of proteins on polyvinylidene fluoride membranes. Membranes were blocked for 15 min in 2% skimmed milk in TBS containing 0.05% Tween‐20 followed by overnight incubation at 4°C in primary antibodies against: anti‐ACC (streptavidin) Dako, Glostrup, Denmark); anti‐phospho‐ACC^Ser221, anti‐phospho‐Akt^Ser473, anti‐phospho‐Akt^Thr308, anti‐Akt2, anti‐phospho‐AMPK^Thr172, anti‐HKII, anti‐phospho‐TBC1D4^Thr642 and anti‐phospho‐TBC1D4^Ser588 (Cell Signaling Technology, Beverly, MA, USA). Anti‐GLUT4 (Thermo Fisher Scientific, Waltham, MA, USA); anti‐α₁ AMPK, anti‐γ₁ AMPK, anti‐CS and anti‐OXPHOS total cocktail (human) (Abcam, Cambridge, UK); anti‐α₂ AMPK, anti‐β₁ AMPK (Santa Cruz Biotechnology, Dallas, TX, USA); anti‐β₂ AMPK (kindly provided by Dr D. G. Hardie, University of Dundee, Dundee, UK); anti‐GS (custom made, Oluf Pedersen, University of Copenhagen, Copenhagen, Denmark); anti‐γ₃ AMPK (Zymed Laboratories Inc., San Francisco, CA, USA); anti phospho‐TBC1D4^Ser704 (custom made, Professor Laurie Goodyear, Joslin Diabetes Centre and Harvard Medical School, Boston, MA, USA); anti‐TBC1D4 (Upstate; Millipore, Billerica, MA, USA). The next day, membranes were incubated with horseradish peroxidase‐conjugated secondary antibodies (Jackson ImmunoResearch, West Grove, PA, USA) for 1 h at room temperature before visualizing protein bands with chemiluminescence (Millipore) and a ChemiDoc MP imaging system (Bio‐Rad, Hercules, CA, USA). Some membranes were stripped and re‐probed 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 mm Tris‐HCl, 69.4 mm SDS, ddH₂O and 0.08% β‐mercaptoethanol, pH 6.7). The membranes were checked for successful removal of the initial primary antibody before re‐probing.

Muscle glycogen

Muscle glycogen content was measured in homogenates (150 μg of protein) as glycosyl units after acid hydrolysis determined by a fluorometric method (Lowry & Passoneau, 1972).

AMPK and glycogen synthase (GS) activity

Isoform‐specific AMPK activity was measured in muscle lysate (250 μg of protein) by sequential immunoprecipitation of the γ₃, α₂ and α₁ subunit. AMPKγ3 antibody used for immunoprecipitation was custom made at Yenzym Antibodies (San Francisco, CA, USA); AMPK α₂ was obtained from Santa Cruz Biotechnology and α₁ was custom made at Genscript USA Inc. (Piscataway, NJ, USA). AMPK activity was measured in the presence of 200 μm AMP and 100 μm AMARA‐peptide (Schafer‐N, Copenhagen, Denmark) as substrate, as described previously (Birk & Wojtaszewski, 2018). GS activity was measured in homogenates (duplicates) in the presence of 0.02, 0.17 and 8 mm glucose‐6‐phosphate (G6P), as described previously (Højlund et al. 2009).

Statistical analyses

Data are presented as the mean ± SEM. Subject characteristics, fold changes in muscle protein expression, Δ glycogen content between legs and Δ glucose uptake between legs were evaluated by paired t tests. To evaluate changes during the last 40 min of the EHC on leg glucose uptake, arterial blood flow and plasma glucose concentration arterial–venous (A–V) difference, two‐way repeated‐measures (RM) ANOVAs were performed. A two‐way RM ANOVA was also applied to evaluate changes in clamp parameters. For all protein phosphorylation and activity measurements, four two‐way RM ANOVAs were applied; one two‐way RM ANOVA was used to test the factors ‘acute exercise’ (rested leg vs. acutely exercised leg) and ‘insulin’ (basal vs. insulin) for the PRE training experimental day alone. Similarly, a second two‐way RM ANOVA was used to test the factors ‘acute exercise’ and ‘insulin’ for the POST training experimental day alone. A third and fourth two‐way RM ANOVA tested the factors ‘acute exercise’ and ‘training’ (PRE training vs. POST training) for the basal samples and the insulin‐stimulated samples, separately. Significant interactions were evaluated by Tukey's post hoc test. P < 0.05 was considered statistically significant. N = 9, except otherwise stated. All statistical analyses were performed using Sigma Plot, version 13 (Systat Software Inc., Chicago, IL, USA).

Results

Adaptations to 12 weeks of endurance training

All subjects completed 94 ± 1% of all training sessions and remained weight stable throughout the 12 weeks of training (Table 1). Adaptations in several parameters signify an effective training regime i.e. increased ${\dot{V}}_{{\mathrm{O}}_2\mathrm{peak}}$ (17%) and peak knee‐extensor workload (PWL, 18%), as well as decreased fat mass (−6%) and resting heart rate (−5 beats min^–1) (Table 1). In skeletal muscle, protein expression of glucose handling enzymes (hexokinase II, GS) and mitochondrial markers (citrate synthase, complex II, III and V) was increased, whereas GLUT4 remained unchanged (Fig. 2). Protein expression of some regulatory signalling components (AMPKα₁, AMPKβ₁, AMPKβ₂, AMPKγ₁ and TBC1D4) increased in response to training, whereas others (AMPKα₂, acetyl‐CoA carboxylase) remained unchanged (Fig. 2). As expected, AMPKγ₃ protein expression decreased with training (Fig. 2).

Table 1.

Subject characteristics PRE and POST training

	PRE training	POST training
Age (years)	25 ± 1
Weight (kg)	78.3 ± 1.7	78.0 ± 1.8
Body mass index	23.5 ± 0.5	23.4 ± 0.6
Lean mass (kg)	56.6 ± 1.9	57.5 ± 1.9
Fat mass (kg)	19.0 ± 1.2	17.8 ± 0.9**
Fat mass (%)	24.2 ± 1.5	22.8 ± 1.2**
Visceral adipose tissue (g)	464 ± 75	351 ± 74*
${\dot{V}}_{{\mathrm{O}}_2\mathrm{peak}}$ (mL min−1 kg−1)	43.6 ± 1.6	50.9 ± 1.2***
HRmax	197 ± 2	194 ± 1
HRrest	60 ± 2	55 ± 2*
PWL (W)	39 ± 4	46 ± 3**
Fasting glucose (mmol L−1)	5.2 ± 0.1	5.4 ± 0.1*
Fasting insulin (μIU mL−1)	5.1 ± 0.5	5.6 ± 0.5

Values are the mean ± SEM. ^* P < 0.05, ^** P < 0.01 and ^*** P < 0.001 vs. PRE training. HR, heart rate; PWL, peak workload for the knee‐extensors of one leg. Lean mass, fat mass and visceral adipose tissue were determined by dual X‐ray absorptiometry.

Figure 2. Change in protein expression after 12 weeks of exercise training.

Protein expression of ACC, AMPK α₁, α₂, β_1, β₂, γ₁ and γ₃, CS, GLUT4, GS, HKII, TBC1D4 and complex II, III and V of the electron transport chain was evaluated by Western blotting PRE training (white bars) and POST training (black bars) (A). Data are calculated as the mean value of protein expression measured in the prior exercised and rested leg before and after insulin stimulation. Representative blots (B). Data are expressed as the mean ± SEM. (*) = 0.058, ^* P < 0.05, ^** P < 0.01 and ^*** P < 0.001 PRE vs. POST training.

Glucose infusion rate during the euglycaemic hyperinsulinaemic conditions was increased by training from 4.2 ± 0.3 mg min⁻¹ kg⁻¹ to 5.0 ± 0.3 mg min⁻¹ kg⁻¹ (Table 2) reflecting increased whole body insulin‐stimulated glucose uptake. Plasma glucose (∼5 mmol L⁻¹) and insulin (∼120 μIU mL⁻¹) concentrations were similar during the insulin clamp performed before and after the training period (Table 2). Plasma concentrations of fatty acids and triacylglycerol decreased in response to insulin but were unaffected by training (Table 2). Improved insulin‐stimulated glucose uptake was also evident in the skeletal muscle by an increased insulin‐stimulated glucose uptake in the rested leg after training compared to before training (Fig. 3 A). This was largely a result of enhanced glucose extraction rather than changes in muscle blood perfusion (Fig. 3 C and D). The increased glucose extraction was likely associated with the increased capacity for glucose handling within the muscle cell by increased HKII and GS protein expression (Fig. 2)

Table 2.

Clamp parameters PRE and POST training

	PRE training		POST training
Glucose infusion rate (mg min−1 kg−1)	4.2 ± 0.3		5.0 ± 0.4*
	Pre clamp	End of clamp	Pre clamp	End of clamp
Arterial plasma glucose (mmol L−1)	5.1 ± 0.1	5.1 ± 0.1	5.1 ± 0.1	5.0 ± 0.1
Arterial plasma Insulin (μIU mL−1)	5.7 ± 0.7	117.2 ± 7.7‡‡‡	5.7 ± 0.6	121.8 ± 5.6‡‡‡
Arterial plasma fatty acids (μmol L−1)	641 ± 42	37 ± 5‡‡‡	557 ± 66	37 ± 1‡‡‡
Arterial plasma triacylglycerol (mmol L−1)	1.1 ± 0.3	0.9 ± 0.2‡‡‡	1.1 ± 0.1	0.9 ± 0.2‡‡‡
RER	0.75 ± 0.02	0.83 ± 0.02‡‡‡	0.76 ± 0.01	0.84 ± 0.01‡‡‡

Values are the mean ± SEM. ^* P < 0.05 vs. PRE training. ^‡‡‡ Main effect of insulin (P < 0.001). RER, respiratory exchange ratio. Glucose infusion rate is expressed as weighted mean of 120 min of euglycaemic hyperinsulinaemic clamp (EHC). Pre clamp: mean of blood samples taken 30 min and 0 min before EHC. End of clamp: mean last 40 min of EHC. RER pre clamp: 60 min before EHC, RER end of clamp: 110 min into the EHC.

Figure 3. Acute one‐legged exercise on leg glucose uptake, glycogen utilization, A–V difference and arterial blood flow.

Leg glucose uptake (A) was calculated using Fick's principle by multiplying glucose A–V difference (C) with leg arterial blood flow (D) and divided by lean leg mass (LLM). The difference in leg glucose uptake between the rested and acutely exercised leg is expressed as Δ leg glucose uptake (B). Leg glucose uptake, glucose A–V difference and arterial blood flow are expressed as mean values of the last 40 min of the 2 h EHC in the rested (white bars) and prior exercised leg (black bars), PRE and POST training. Muscle glycogen measured in the rested (white bars) and prior exercised leg (black bars) 4 h after acute one‐legged exercise (basal) and after a 2 h EHC (insulin) PRE and POST training (E). The difference in glycogen content between the rested and acutely exercised leg 4 h after acute exercise is depicted as glycogen degradation (F). Data are expressed as the mean ± SEM. ME, main effect. ^** P < 0.01 PRE vs. POST training; †P < 0.05, ††P < 0.01 and †††P < 0.001 rested leg vs. acutely exercised leg.

Acute one‐legged exercise was performed at the same relative intensity

Both PRE and POST training muscle glycogen was reduced in response to acute exercise in the prior exercised leg compared to the rested leg (Fig. 3 E). Although training increased the glycogen content within the muscles (Fig. 3 E), the absolute glycogen utilization during acute one‐legged exercise was similar PRE and POST training (Fig. 3 F). In a subset of subjects (n = 6), the percentage maximum heart rate (53 ± 3% vs. 50 ± 2%) and rate of perceived exertion (16 ± 0 vs. 16 ± 0 on the Borg (6–20) scale) were measured during acute one‐legged exercise with no difference between the PRE and POST training experimental day. Together, these results support that the subjects (as per the design) performed the acute bout of one‐legged exercise at the same relative intensity PRE and POST training (absolute workload was 33 ± 3 W PRE training and 39 ± 2 W POST training).

Diminished ability of acute exercise to enhance insulin‐stimulated leg glucose uptake after training

Both PRE and POST training, insulin increased leg glucose uptake to a greater extent in the prior exercised leg compared to the rested leg (Fig. 3 A). However, the difference in insulin‐stimulated glucose uptake between the prior exercised and rested leg was markedly smaller (−50 ± 12%) POST training compared to PRE training (Fig. 3 B). This reflects a diminished ability of acute exercise to enhance muscle insulin‐stimulated glucose uptake in the trained state.

During the insulin clamp, the plasma glucose concentration A–V difference was higher in the prior exercised leg compared to the rested leg both PRE and POST training. In response to training, the plasma glucose concentration A–V difference was increased in the rested leg, whereas the A–V difference in the prior exercised leg was unaffected by training (Fig. 3 C). Femoral arterial blood flow was slightly higher in the prior exercised leg compared to the rested leg and was unaffected by training (Fig. 3 D). Thus, in the prior exercised leg, lower glucose extraction rather than glucose delivery was responsible for the diminished ability to increase insulin‐stimulated glucose uptake after training, reflecting important intramuscular changes.

Muscle signalling

To investigate the effects of prior exercise on insulin signalling to GLUT4 translocation, phosphorylation of Akt and TBC1D4 was measured. Insulin increased, as expected, the phosphorylation of Akt^Ser473 and Akt^Thr308, whereas prior exercise did not, nor did exercise improve the effect of insulin on Akt phosphorylation (Fig. 4 A and B). Akt2 protein expression increased in response to training (Fig. 4 C).

Figure 4. Phosphorylation and expression of Akt2.

Site‐specific phosphorylation of Akt^Ser473 (A) and Akt^Thr308 (B) and Akt2 protein expression (C) measured by Western blotting in the rested (white bars) and prior exercised leg (black bars) 4 h after acute one‐legged exercise (basal) and after a 2 h EHC (insulin) PRE and POST training. Representative blots (D). Data are expressed as the mean ± SEM. AU, arbitrary units; ME, main effect. ^** P < 0.01 PRE vs. POST training; ^† P < 0.05, rested leg vs. acutely exercised leg.

Phosphorylation of TBC1D4^Ser704, a confirmed AMPKα₂β₂γ₃ target site (Treebak et al. 2010), was significantly higher in the prior exercised vs. rested muscle both PRE and POST training before the clamp was initiated (Fig. 5 A). However, the TBC1D4^Ser704phosphorylation in the prior exercised muscle was significantly lower POST training compared to PRE training. In response to insulin, the phosphorylation increased in the rested muscle but did not increase further in prior exercised muscle PRE training. Phosphorylation of TBC1D4^Ser704 increased in response to insulin in the muscles of both legs POST training (Fig. 5 A).

Figure 5. Phosphorylation of TBC1D4.

Site‐specific phosphorylation of TBC1D4^Ser704 (A), TBC1D4^Thr642 (B) and TBC1D4^Ser588 (C) measured by Western blotting in the rested (white bars) and prior exercised leg (black bars) 4 h after acute one‐legged exercise (basal) and after a 2 h EHC (insulin) PRE and POST training. Representative blots (D). Data are expressed as the mean ± SEM. AU, arbitrary units; ME, main effect. ^* P < 0.05, ^** P < 0.01 PRE vs. POST training; ^† P < 0.05 and ^††† P < 0.001 rested leg vs. acutely exercised leg; ^‡ P < 0.05 and ^‡‡‡ P < 0.001 basal vs. insulin.

TBC1D4^Thr642 is a confirmed Akt target site and, although this site is not a direct AMPK target site, some data suggest that phosphorylation of TBC1D4^Thr642 is dependent on TBC1D4^ser704 phosphorylation and thus becomes indirectly dependent of AMPK (Kjøbsted et al. 2015). Accordingly, phosphorylation of TBC1D4^Thr642 was higher in the prior exercised muscle compared to rested muscle 4 h after acute exercise PRE training but not POST training. In response to insulin, phosphorylation of TBC1D4^Thr642 was increased in muscles of both legs both PRE and POST training (Fig. 5 B).

TBC1D4^Ser588 is also a confirmed Akt target site reported to be independent of TBC1D4^Ser704 phosphorylation. TBC1D4^Ser588 phosphorylation was modestly higher in the prior exercised vs. rested muscle 4 h after acute one‐legged exercise both PRE and POST training prior to insulin stimulation (Fig. 5 C). In response to insulin, phosphorylation of TBC1D4^Ser588 increased in muscles of both legs both at the PRE and POST training experimental day (Fig. 5 C).

Thus, a major difference between the PRE and POST experimental day is the markedly elevated TBC1D4^Ser704 phosphorylation in the prior exercised muscle observed PRE training but not POST training. To gain further support for a potential role of AMPK, we measured phosphorylation of ACC^Ser221 a confirmed AMPK target site (Abu‐Elheiga et al. 1997; Munday, 2002). Accordingly, phosphorylation of ACC^Ser221 was only increased by exercise PRE training and not POST training (Fig. 6 A). Together, the phosphorylation patterns of both TBC1D4^Ser704 and ACC^Ser221 suggest a lower activation of AMPK in response to acute exercise in the trained state compared to the untrained state. As a result of the study design, biopsies were not obtained during exercise in the present study. Thus, we cannot directly confirm the reduced AMPK activation during exercise POST training compared to PRE training. Four hours after exercise, phosphorylation of AMPK^Thr172 and activity of the AMPKα₂β₂γ₃ complex showed a rather fast reversal toward basal levels (Fig. 6 B and C), confirming previous observations (Mortensen et al. 2013).

Figure 6. ACC and AMPK activation.

Site‐specific phosphorylation of ACC^Ser221 (A) and AMPK^Thr172 (B) measured by Western blotting in the rested (white bars) and prior exercised leg (black bars) 4 h after acute one‐legged exercise (basal) and after a 2 h EHC (insulin) PRE and POST training. Representative blots (C). Activity of AMPKα₂β₂γ₃ in the rested (white bars) and prior exercised leg (black bars) 4 h after acute one‐legged exercise (basal) and after a 2 h EHC (insulin) PRE and POST training (D). Data are expressed as the mean ± SEM. AU, arbitrary units; ME, main effect. ^* P < 0.05 PRE vs. POST training.

Activation of GS in response to acute exercise and insulin PRE and POST training

As the major part of muscle glucose metabolism is non‐oxidative during insulin stimulation, the activity of GS (rate‐limiting enzyme in glycogenesis) was measured. In accordance with the higher GS protein expression (Fig. 2), total GS activity was also increased by training (Fig. 7 A). The activity of GS measured in the presence of 0.02 and 0.17 mm G6P (Fig. 7 B and C) was increased by prior exercise and insulin both PRE and POST training. Yet, the ability of insulin to activate GS was not dependent of prior exercise either at the PRE or POST training experimental day. Thus, the activity pattern of GS does not explain the reduced effect of acute exercise to enhance insulin‐stimulated glucose uptake in the trained state.

Figure 7. GS activity.

GS activity measured in the presence of 8 mm G6P (A) 0.02 mm G6P (B) and 0.17 mm G6P (C) in the rested (white bars) and prior exercised leg (black bars) 4 h after acute one‐legged exercise (basal) and after a 2 h EHC (insulin) PRE and POST training. Data are expressed as the mean ± SEM. ME, main effect.

Discussion

In the present study, we found that 12 weeks of exercise training increased insulin‐stimulated glucose uptake on a whole body level, as well as in skeletal muscle. Interestingly, insulin‐stimulated glucose uptake in the acutely exercised leg was not further enhanced by training. This led to our primary finding that the ability of acute exercise to enhance insulin‐stimulated glucose uptake is reduced in the trained state. Importantly, this was seen despite that the acute bout of exercise was performed at the same relative intensity eliciting similar absolute glycogen degradation during exercise before and after training.

In contrast to the findings reported by Mikines et al (1989) and Nelson and Horowitz (2014), we found increased insulin‐stimulated glucose uptake following acute exercise even in the trained muscle. This could be a result of differences in study design (cross‐sectional vs. intervention study, study population, etc.), although it indeed reflects a lack of knowledge about the dynamics and magnitude of the response to acute exercise and to the mechanisms regulating these factors, including regulation of microvascular perfusion (Sjøberg et al. 2017). It may be speculated that an upper limit of muscle insulin sensitivity is obtainable in the highly trained muscle leaving only very little or no additional effects of a single exercise bout in these subjects. An important premise for such an interpretation is that leg glucose uptake has not reached the maximal obtainable level. We did not evaluate this in the present study. However previous studies, on comparable groups of subjects, have reported higher (∼60–80%) levels of leg glucose uptake in response to a maximal dose of insulin (Richter et al. 1989; Dela et al. 1992). Furthermore, in studies in which leg glucose uptake has been evaluated under hyperinsulinaemic–hyperglycaemic conditions (Hansen et al. 1999) or when insulin and exercise have been superimposed (Dela et al. 1994), rates of leg glucose uptake exceeded by far (∼4‐fold) the values obtained in the present study. Thus, we assume that, under the conditions applied, leg glucose uptake has not reached saturation.

From an applied view, the observation that the same exercise bout elicits markedly different responses in insulin‐stimulated glucose uptake depending on the person's fitness level may add additional explanatory insight to the complexity of controlling glucose homeostasis in patients dependent on insulin treatment. Indeed, it may add insights that are important for prediction of the optimal insulin dose algorithm (Riddle, 2008).

The present study provided the opportunity to identify possible molecular mechanisms in skeletal muscle explaining the reduced ability of acute exercise to enhance insulin‐stimulated glucose uptake in the trained state. A series of studies in rodents and transgenic animal models have signified the importance of the signalling node at TBC1D4 integrating inputs from AMPK (‘exercise’) and Akt (‘insulin’), indicating both the necessity and sufficiency of AMPK in muscle insulin sensitization (Fisher et al. 2002; Kjøbsted et al. 2015, 2017; Wang et al. 2018). More specifically, these studies point to interactions between AMPKα₂β₂γ₃ and Akt2 and the phosphor‐regulation of TBC1D4^{Ser711/Thr642}. Indications of similar interactions in human muscle have also been reported. Thus, the level of phosphorylation of the equivalent TBC1D4^Ser704 (in humans) is associated with AMPKα₂β₂γ₃ activity during exercise in samples of whole muscle as well as in samples representing either type I or type II muscle fibres (Kristensen et al. 2015). Also, we recently reported enhanced phosphor‐regulation of TBC1D4 by insulin concomitantly with enhanced insulin‐stimulated glucose uptake following exercise (Treebak et al. 2009; Pehmøller et al. 2012; Sjøberg et al. 2017; Hingst et al. 2018). In the present study, we found reduced phosphorylation of TBC1D4^{Ser704/Thr642} 4 h after exercise in the acutely exercised leg after training compared to before training, whereas phosphorylation of TBC1D4^{Ser704/Thr642} during insulin stimulation reached similar levels before and after training. Exactly how this may relay to the regulation of insulin action following exercise is still unclear. Rodent studies suggest that the localization of TBC1D4 and protection against dephosphorylation mediated by prior exercise might be of importance (Zheng & Cartee, 2016; Arias et al. 2018). In the present study, we showed decreased AMPKγ₃ expression in skeletal muscle in response to 12 weeks of cycling exercise, confirming previous observations (Frøsig et al. 2004; Wojtaszewski et al. 2005; Mortensen et al. 2013). We have previously shown that the AMPKα₂β₂γ₃ is the primary complex activated during acute exercise (Birk & Wojtaszewski, 2006) and that the protein expression of AMPKγ₃ correlates with the AMPKα₂β₂γ₃ activity (Mortensen et al. 2009), suggesting that alterations in AMPKγ₃ expression probably affect AMPKα₂β₂γ₃ activation during acute exercise. As a result of the study design, we did not obtain muscle biopsies immediately after exercise. Therefore, we were unable to directly confirm such differences in AMPKα₂β₂γ₃ activation during the acute one‐legged exercise before and after training. However, two arguments suggest that, also in the present study, AMPKα₂β₂γ₃ activation has been higher during exercise before training vs. after training. First, in the present study, the phosphor‐regulation of the confirmed AMPK targets ACC^Ser221 and TBC1D4^Ser704 was increased in the previously exercised muscle before but not after training. Second, previous cross‐sectional and training intervention studies have reported lesser activation of the AMPK‐ACC signalling axis in response to acute exercise in the trained vs. the untrained state. (Nielsen et al. 2003; McConell et al. 2005; Lee‐Young et al. 2009; Mortensen et al. 2013). The mechanism responsible for the lesser AMPK activation in the trained muscle, even by exercise loads eliciting the same relative intensity, is unclear. Glycogen might be a regulator of AMPK activity (Wojtaszewski et al. 2003b; McConell et al. 2005) via the ability of AMPK to bind to the glycogen particles (McBride et al. 2009). Rodent studies also suggest that the absolute level of glycogen influences the ability of insulin to increase glucose uptake (Jensen et al. 1997; Derave et al. 2000) and the insulin‐sensitizing effect of exercise has previously been proposed to be dependent on the glycogen utilization during exercise (Richter et al. 2001). Thus, the higher absolute levels of glycogen in muscle after training might reduce the ability of acute exercise to enhance insulin‐stimulated glucose uptake by reducing activation of AMPK.

In the present study, phosphorylation of TBC1D4^Thr642 responded similarly to acute exercise as TBC1D4^Ser704. This is in line with previous studies suggesting a mutual dependency of these two phosphor‐sites (Kjøbsted et al. 2015, 2017), such that, even though TBC1D4^Thr642 is not an AMPK target site, its regulation is dependent on the degree of TBC1D4^Ser704 phosphorylation (Kjøbsted et al. 2015). By contrast, such an interaction is not seen in the phosphor‐regulation of TBC1D4^Ser588 (Kjøbsted et al. 2015), probably explaining why the phosphorylation of TBC1D4^Ser588 increased to the same extent in response to acute exercise before and after training in the present study. As TBC1D4^Ser588 is not an AMPK target site (Kjøbsted et al. 2015), it is plausible that the increased phosphorylation of TBC1D4^Ser704 and ^Thr642 in response to acute exercise before but not after training was mediated via AMPK and not another upstream kinase. In line with this, the activation (by phosphorylation) of the insulin‐regulated TBC1D4 kinase Akt2 was not enhanced in response to acute exercise either before or after training.

In conclusion, although exercise training increases insulin‐stimulated glucose uptake in skeletal muscle at rest, we show that it is not enhanced further by training in the acutely exercised leg. This leads to a diminished insulin‐sensitizing effect of a single bout of exercise by exercise training. Prior exercised muscle thus displays a similar absolute level of glucose uptake during insulin stimulation before and after training. Whether this reflects an upper limit for muscle insulin sensitivity or a strong dependency on the absolute muscle glycogen levels remains to be seen. Our data suggest that a reduced AMPKγ₃ expression with exercise training may contribute to a lesser AMPKα₂β₂γ₃ mediated signalling following a single bout of exercise. We propose this to be responsible for the reduced phosphorylation of TBC1D4 and thus the reduced ability of acute exercise to enhance insulin‐stimulated muscle glucose uptake in the trained state. Together with previous evidence in humans and rodents, these findings contribute to the idea that AMPK is an important regulator of insulin action in skeletal muscle.

Additional information

Competing interests

The authors declare that they have no competing interests.

Author contributions

The experiments were performed at the section of Molecular Physiology, Department of Nutrition, Exercise and Sports, University of Copenhagen, Denmark. DES and JFPW were responsible for the conception and design of the research. DES, NBJ, KAS, BK, EAR and JFPW performed the experiments. DES, NBJ and JBB performed the analyses. DES, NBJ, JBB, KAS, BK, EAR and JFPW interpreted the results. DES and JFPW drafted the manuscript. DES, NBJ, JBB, KAS, BK, EAR and JFPW edited and revised manuscript. All authors read and approved the final version submitted for publication. JFPW is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis

Funding

This study was supported by grants from the Danish Council for Independent Research Medical Sciences (FSS 6110‐00498B); The Novo Nordisk Foundation (NNF16OC0023046, NNF17OC0027224), The Lundbeck Foundation (R266‐2017‐4358); and the research program (2016) ‘Physical Activity and Nutrition for Improvement of Health’ funded by the University of Copenhagen Excellence Program for Interdisciplinary Research. DES was supported by a research grant from the Danish Diabetes Academy funded by the Novo Nordisk Foundation. KAS was supported by a postdoctoral research grant from the Council for Independent Research/ Medicine (4092‐00309).

Biography

Dorte E. Steenberg received her master's degree in human physiology from the Department of Nutrition, Exercise and Sports, UCPH. Her master thesis focused on the effects of acute exercise on AMPK signalling in skeletal muscle fibres. Subsequently, she continued into the fascinating world of science as a PhD student investigating the effects of acute exercise on insulin sensitivity under different conditions in humans. One aspect of this is whether training status affects the insulin‐sensitizing effect of acute exercise, as investigated in the present study. She considers that this will provide new knowledge to the important ongoing question of ‘how physical activity improves insulin sensitivity and health’.