Hypoglycemia is the leading limiting factor in glycemic management of insulin-treated diabetes. Skeletal muscle is the predominant site of insulin-mediated glucose disposal. Our study used a crossover design to test to what extent insulin-induced hypoglycemia affects glucose uptake in skeletal muscle and whether hypoglycemia counterregulation modulates insulin and catecholamine signaling and glycogen synthase activity in skeletal muscle. Nine healthy volunteers were examined on three randomized study days: 1) hyperinsulinemic hypoglycemia (bolus insulin), 2) hyperinsulinemic euglycemia (bolus insulin and glucose infusion), and 3) saline control with skeletal muscle biopsies taken just before, 30 min after, and 75 min after insulin/saline injection. During hypoglycemia, glucose levels reached a nadir of ∼2.0 mmol/L, and epinephrine rose to ∼900 pg/mL. Hypoglycemia impaired insulin-stimulated glucose disposal and glucose clearance in skeletal muscle, whereas insulin signaling in glucose transport was unaffected by hypoglycemia. Insulin-stimulated glycogen synthase activity was completely ablated during hyperinsulinemic hypoglycemia, and catecholamine signaling via cAMP-dependent protein kinase and phosphorylation of inhibiting sites on glycogen synthase all increased.
Introduction
Hypoglycemia is the most common side effect of insulin-treated diabetes and the main factor precluding optimal glucose management (1). Hypoglycemia is defined by the Whipple triad: symptoms related to hypoglycemia, low plasma glucose measured at the time of symptoms, and relief of symptoms when glucose levels are raised to normal. Furthermore, hypoglycemia is differentiated based on the severity of symptoms (or lack thereof) and glucose levels: asymptomatic hypoglycemia, which presents with no symptoms but plasma glucose <3.9 mmol/L (<70 mg/dL); symptomatic hypoglycemia, which shows symptoms and plasma glucose <3.9 mmol/L (<70 mg/dL); and severe hypoglycemia, which is an event requiring aid from another person and low plasma glucose (2). Insulin-induced hypoglycemia occurs rapidly (nadir plasma glucose within 30 min) (3). In healthy subjects, hypoglycemia leads to a decrease in insulin and increases in circulating levels of the counterregulatory hormones glucagon, norepinephrine, epinephrine, growth hormone (GH), and cortisol (4). Catecholamines and glucagon inhibit glycolysis and increase glycogenolysis as well as gluconeogenesis via increased adenylyl cyclase and cAMP-dependent protein kinase (PKA) activity (5). The combined effects of these hormones effectively and rapidly increase endogenous glucose production during hypoglycemia (6).
Insulin-stimulated glucose disposal is primarily governed by skeletal muscle (7). Intracellular signaling in this process is initiated by insulin binding to its receptor and the subsequent signaling through the insulin receptor substrate 1/phosphoinositide 3-kinase signaling pathway. This activates Akt and stimulates phosphorylation of AS160 and TBC1D1 (8). This stimulates translocation of GLUT4 to the cell surface and facilitates transport of glucose into the muscle. Once glucose is taken up it is metabolized by either oxidative or nonoxidative pathways (6).
The effects of insulin are inhibited during hypoglycemia (9), plausibly because of the action of counterregulatory hormones and a reduced glucose concentration gradient (10). Catecholamines stimulate G protein–coupled β-receptors, which are present on skeletal muscle, and thereby activate PKA (11). Increased PKA activity can inhibit insulin action and may thereby suppress glucose uptake during hypoglycemia (12). Epinephrine administration to isolated rat skeletal muscle decreases insulin-stimulated glucose uptake by decreasing glycogen synthase (GS) activity, leading to accumulation of glucose-6-phosphate (G6P) and subsequent inhibition of hexokinase activity (13). In humans, epinephrine administration in adrenalectomized subjects during exercise decreases glucose disposal (14). However, catecholamine-mediated signaling in skeletal muscle during acute hypoglycemia has not been determined. Similar to β-receptors, GH receptors are expressed on skeletal muscle, and GH stimulation potently inhibits insulin-stimulated glucose transport after latency through unknown mechanisms (15). The aim of this study was to investigate insulin action in skeletal muscle during acute hypoglycemia under the hypothesis that intracellular insulin action is impaired during this condition.
Research Design and Methods
The study protocol was approved by the local scientific ethics committee (1-10-72-113-13) and registered at clinicaltrials.gov (NCT01919788). The study was conducted in accordance with the Helsinki Declaration II, and all subjects gave oral and written informed consent to participate.
The design is a randomized crossover study of 3 days’ duration, previously described in detail (16). Briefly, each subject was allocated to 1) control (CTR) (bolus, 2 mL NaCl i.v.); 2) hyperinsulinemic hypoglycemia (HH) (bolus insulin, 0.1 unit Actrapid/kg in 2 mL NaCl i.v.); and 3) hyperinsulinemic euglycemia (HE) (bolus insulin, 0.1 unit Actrapid/kg in 2 mL NaCl i.v. and i.v. glucose infusion 20% [varying amounts]). The study days were separated by at least 21 days; they commenced at 9.00 a.m. (t = 0 min) and ended at 10.45 a.m. (t = 105 min).
Our primary outcome for this study was forearm glucose disposal. Secondary outcomes were glucose oxidation rates, insulin signaling in glucose transport, glycogen synthesis in skeletal muscle, and energy expenditure (EE).
Muscle biopsies from the vastus lateralis of the quadriceps muscle were obtained under sterile conditions 15 min after local anesthesia (lidocaine 1%, 10 mL) was applied. Biopsies were obtained from the right leg (separated by >10 cm) just before t = 0 min and at t = 75 min, and from the left leg at t = 30 min. Biopsies were immediately snap-frozen in liquid nitrogen and subsequently stored at −80°C until analyzed. All samples were homogenized twice for 30 s at 5,000 rpm in a buffer containing 50 mmol/L HEPES, 20 mmol/L NaF, 2 mmol/L NaOV, 5 mmol/L EDTA, 5 mmol/L nicotinamide, 137 mmol/L NaCl, 10 mmol/L Na4P2O7, 1 mmol/L MgCl2, 1 mmol/L CaCl2, 0.01 mmol/L trichostatin A, Halt Protease Inhibitor Cocktail (100X), 1% NP-40, and 10% glycerol in a Precellys Homogenizer (Bertin Instruments, Montigny-le-Bretonneux, France). Then samples were rotated at 4°C for 30 min before being centrifuged at 14,000g for 20 min. The resulting supernatant was collected.
Western blotting was performed using 4–15% Criterion XT Bis-Tris gels (Bio-Rad, Hercules, CA). The following primary antibodies were used AKT, pAKTser473, pAKTthr308, AS160, pAS160ser642, pAS160ser588, GS, pGSser641 (site 3a), glycogen synthase kinase-3 (GSK-3), pGSK-3α+β, mammalian target of rapamycin (mTOR), pmTORser2448, AMPK, pAMPKthr172, 4EBP1, nonphosphorylated thr46 4EBP1, PKA phospho-substrate (catalog nos. 4691, 9271, 9275, 5676, 4288, 8730, 3886, 3891, 5676, 9331, 2972, 5536, 2532, 2531, 9644, 4923, 9624, respectively; all from Cell Signaling Technology); acetyl-CoA carboxylase (ACC), pACCser79, GLUT4 (streptavidin; Upstate 07-303, 07-1404) (all from EMD Millipore, Darmstadt, Germany); and GSK-3β (610202; BD Biosciences). Antibodies detecting phosphorylated GS (pGS) ser7+10 (site2+2a) were raised against peptide PLSRSL(Sx)MS(Sx)LPGLED (residues 1–16 of rat GS), and pGS ser697 (site1a) and pGS ser710 (site1b) were raised against the peptides CEWPRRASpSCTSSTG (692–703 residues of human GS) and CSGSKRNSpVDTATS (704–716 residues of human GS); the specificity of these antibodies have been reported previously (17,18). Phosphorylation levels are expressed as a ratio with the targeted protein, measured on the same membrane. Stain-free protein technology was used to control for equal loading (19). GLUT4 and PKA phospho-substrate levels are expressed as a ratio with the protein content quantified by stain-free protein technology. PKA phospho-substrate antibody was used as a measure of PKA activation through the detection of PKA-phosphorylation on all detectable proteins >15 kDa. This antibody detects proteins containing a phospho-serine/threonine residue with arginine at the −3 position, which is a bona fide PKA motif (20). Glycogen phosphorylase (GP) a and b antibodies were used as previously described (21). Proteins were visualized and quantified by enhanced chemiluminescence after incubation with horseradish peroxidase–conjugated antirabbit/antimouse secondary antibodies using Image Lab 5.0 software (Bio-Rad). Protein measurements are expressed as the ratio with the mean CTR value at t = 0 min relative units (RU).
Muscle glycogen concentration was determined as glycosyl units after acid hydrolysis of muscle lysate using the fluorometric method (22). GS activity was measured using a modified version (18) of the method described by Thomas et al. (23). The activity was measured at 37°C, and the mixture contained uridine diphospho-glucose at a concentration of 1.7 mmol/L and G6P at concentrations of 0.02 mmol/L (zero), 0.167 mmol/L (low), and 8.0 mmol/L (high). GS activity at a G6P concentration of (Zero/High) × 100 is termed I-Form, whereas activity at a G6P concentration of (Low/High) × 100 is termed fractional velocity (FV).
Biochemical parameters were measured using various methods: plasma glucose was measured with the oxidase method (YSI 2300 STAT Plus; YSI Life Sciences, Yellow Springs, OH); serum insulin, with ELISA (Dako Denmark A/S, Glostrup, Denmark); plasma glucagon, with a radioimmunoassay kit (EMD Millipore); plasma epinephrine and norepinephrine, with electrochemical detection after high-performance liquid chromatography (9); serum GH, with chemiluminescence technology (IDS-iSYS Multi-Discipline Automated System; Immunodiagnostic Systems Nordic S/A, Copenhagen, Denmark); and C-peptide, with ELISA (ALPCO, Salem, NH).
Indirect calorimetry (Deltatrac monitor; Dantes Instrumentarium, Helsinki, Finland) was used from t = 90 to 105 min in order to determine total EE, the respiratory quotient, and glucose oxidation rates (24). Protein oxidation was set at 12% of EE (25).
Forearm blood flow (FBF) was measured using strain-gauge plethysmography (26) and forearm glucose disposal was assessed using the product of FBF and arteriovenous differences in glucose concentrations. Arterialization of venous blood was obtained as described previously (16). When looking at differences in glucose disposal using arteriovenous differences × FBF, possible differences in glucose disposal due to differences in glucose gradient across the cell membrane (glucose concentration) are not accounted for. To account for these differences in glucose levels, plasma glucose clearance in mL plasma ⋅ 100 mL forearm−1 ⋅ min−1 was calculated as follows: glucose disposal (mmol ⋅ 100 mL forearm−1 ⋅ min−1) ÷ arterial glucose concentration (mmol/L) × 1,000.
Statistical analyses were performed using a repeated-measures mixed model with visit order, visit number, time, intervention, and the interaction between intervention and time as factors (Stata 13.0; StataCorp, College Station, TX). When only one daily measure was applied, a repeated-measures mixed model was used, with visit order, visit number, and intervention as factors. Unequal correlations and SEs, when detected, were accounted for in the analysis. The model was validated by inspecting quantile-quantile plots of the residuals and scatterplots of the predicted versus the fitted values. If residuals were not normally distributed, logarithmic transformation was performed. P < 0.05 was considered statistically significant. Data in graphs show geometric means with SEMs. Numerical data are geometric means with 95% CIs.
Results
Subject Characteristics
Information about the study subjects has in part been published previously in a report of a study describing lipid metabolism and adipose tissue signaling during acute insulin-induced hypoglycemia (16). The subject characteristics are summarized in Table 1.
Variable . | Value . |
---|---|
Age (years) | 22 (18–27)* |
BMI (kg/m2) | 23.6 (21.7–26.1)* |
Smoking (cigarettes/day) | 0 |
Alcohol (units/week) | 3.7 (0–7.4) |
Systolic blood pressure (mmHg) | 124 (117–131) |
Diastolic blood pressure (mmHg) | 75 (71–80) |
Heart rate (bpm) | 61 (53–68) |
Exercise (h/week) | 4.6 (3.0–6.1) |
Glucose (mmol/L) | 5.0 (4.6–5.5) |
HbA1c (mmol/mol) | 30 (28–32) |
HbA1c (%) | 4.9 (4.7–5.1) |
C-peptide (nmol/L) | 0.63 (0.46–0.80) |
TSH (10−3 IU/L) | 1.93 (1.42–2.43) |
CRP (mg/L) | 0.8 (0.5–1.0) |
Creatinine (μmol/L) | 87 (77–96) |
Hemoglobin (mmol/L) | 9.3 (9.0–9.6) |
HOMA (%S) | 152 (105–200) |
HOMA (%B) | 80 (56–103) |
Variable . | Value . |
---|---|
Age (years) | 22 (18–27)* |
BMI (kg/m2) | 23.6 (21.7–26.1)* |
Smoking (cigarettes/day) | 0 |
Alcohol (units/week) | 3.7 (0–7.4) |
Systolic blood pressure (mmHg) | 124 (117–131) |
Diastolic blood pressure (mmHg) | 75 (71–80) |
Heart rate (bpm) | 61 (53–68) |
Exercise (h/week) | 4.6 (3.0–6.1) |
Glucose (mmol/L) | 5.0 (4.6–5.5) |
HbA1c (mmol/mol) | 30 (28–32) |
HbA1c (%) | 4.9 (4.7–5.1) |
C-peptide (nmol/L) | 0.63 (0.46–0.80) |
TSH (10−3 IU/L) | 1.93 (1.42–2.43) |
CRP (mg/L) | 0.8 (0.5–1.0) |
Creatinine (μmol/L) | 87 (77–96) |
Hemoglobin (mmol/L) | 9.3 (9.0–9.6) |
HOMA (%S) | 152 (105–200) |
HOMA (%B) | 80 (56–103) |
Data are mean (95% CI) or median (range), indicated by *. Most of these data have been published previously (16). HOMA estimates steady-state β-cell function (%B) and insulin sensitivity (%S), as percentages of a normal reference population. CRP, C-reactive protein; TSH, thyrotropin.
Glucose Disposal in Skeletal Muscle
Plasma glucose and insulin levels have been published previously in detail (16). Comparable peak insulin levels (∼900 pmol/L) were reached during both HH and HE (P = 0.87) at t = 15 min. Insulin levels remained slightly elevated during HE throughout the rest of the study (overall P < 0.001) (Fig. 1A). This was likely because of a mild overdose of glucose during HE, leading to increased endogenous insulin production, which was confirmed by overall C-peptide levels ∼2.7 times higher during HE than during HH (P < 0.001) and ∼1.3 times higher than the CTR (P < 0.001).
FBF showed no difference among interventions at any time point. Forearm glucose disposal was markedly larger (∼2.5 times overall) during HE than during HH (overall P < 0.001) (Fig. 1C). These differences were already present at t = 20 min.
Plasma glucose clearance (Fig. 1E) was reduced ∼35% during HH compared with HE (overall P < 0.01). Differences were evident at t = 40 min (P = 0.01), t = 60 min (P = 0.04), and t = 80 min (P < 0.01), thus confirming reduced insulin-stimulated glucose uptake in skeletal muscle during hypoglycemia.
Insulin Signaling in GLUT4 Translocation and Protein Synthesis
Insulin increased phosphorylation of Aktser473 (Fig. 2A) and Aktthr308 (Fig. 2B) 30 min after injection by ∼10-fold compared with the saline control. No differences were found between the hypoglycemic and euglycemic insulin-stimulated conditions. These findings were associated with increased signaling at the downstream target AS160 at thr642 (Fig. 2C) and ser588 (Fig. 2D) at the same time point during both HE and HH. Akt phosphorylations at ser473 and thr308 were still elevated, but only about fourfold, 75 min after insulin injection during HH and HE conditions compared with the saline control. Phosphorylation on AS160ser588 remained elevated during both HH and HE, but thr642 phosphorylation was increased during HE only. We did not detect any differences in Akt, AS160, or GLUT4 protein levels among all interventions.
Insulin stimulates protein synthesis through activation of the mTOR pathway (27). Insulin increased phosphorylation of mTOR at ser2448 ∼2.5-fold 30 min after injection during both HE and HH compared with CTR (both P < 0.001) (Fig. 2E) but remained stable throughout CTR. At t = 75 min, the increased mTOR phosphorylation during HH was ∼75% higher than that of the CTR (P < 0.001) and ∼35% higher than that during HE (P < 0.001). Similar regulation by insulin was evident at the downstream target protein 4EBP1 (Fig. 2F).
Counterregulatory Hormones and Signaling
Hypoglycemia profoundly increased circulating levels of epinephrine, norepinephrine, GH, and glucagon. For clarity, epinephrine and glucagon, which have been published previously (16), are described here.
Epinephrine levels were low and comparable between HE and CTR at all time points (P = 0.14) (Fig. 3A). During HH, epinephrine levels rose rapidly to a peak level of 891 pg/mL (560–1,221 pg/mL) at t = 30 min. Overall comparison revealed differences between both HH and CTR (P < 0.001) as well as between HH and HE (P < 0.001).
Norepinephrine quickly rose during HH to a peak level of 387 pg/mL (349–425 pg/mL) at t = 30 min, and was clearly increased compared with both HE (P < 0.001) and CTR (P < 0.001) (Fig. 3B). No overall difference was detected between CTR and HE (P = 0.07).
Glucagon levels increased during HH to a peak level of 133 pg/mL (119–147 pg/mL) at t = 45 min and were restored to CTR levels at t = 90 and 105 min (P > 0.05) (Fig. 3C). During HE, glucagon levels were suppressed compared with those of the CTR (P < 0.001). Differences between these two study days were evident from t = 30 min. Glucagon levels were stable during CTR, with an overall mean of 52 pg/mL (47–56 pg/mL).
GH was comparable between HE and CTR at all times (Fig. 3D). During HH, GH levels increased from t = 45 min and were elevated throughout the study day, with a peak level of 15.6 μg/L (6.3–38.7 μg/L) at t = 60 min. Overall comparison between interventions revealed increased GH levels during HH, HH and CTR (P = 0.001), and HH and HE (P < 0.001).
Activity of phospho-PKA substrate, as a surrogate marker of catecholamine stimulation, increased by ∼75% at t = 30 min during HH compared with both CTR (P < 0.001) and HE (P < 0.001) (Fig. 3E). At t = 75 min, PKA substrate phosphorylation decreased to only ∼10% more than that during CTR (not significant), and to ∼30% above HE levels (P < 0.01).
No differences in AMPKthr172 phosphorylation were detected between the interventions (Fig. 3F), nor did we detect any effect of the interventions on phosphorylation of the downstream target ACC (data not shown).
Regulation of Glycogen Synthesis and Breakdown, Total Glycogen, Glucose Oxidation, and EE
GS Activity
Insulin potently increased GS FV by ∼35% during HE at both t = 30 and 75 min compared with CTR values (P < 0.001) (Fig. 4A). Hypoglycemia totally abolished the effect of insulin on GS activity, and thus no differences were detected when comparing HH with CTR at t = 30 or 75 min.
I-Form.
The same pattern was seen with I-Form, which was increased by ∼75% at t = 30 min (P < 0.001) and by ∼65% at t = 75 min (P < 0.001) compared with CTR (Fig. 4B). Again, hypoglycemia completely impaired GS activity when I-Form was used, and no differences were detected when comparing HH with CTR.
These findings confirm that storage of glucose as glycogen in muscle tissue was inhibited by hypoglycemia. To unfold possible explanatory mechanisms for this decrease in GS activity, we also investigated known regulatory phosphorylation sites on GS and the upstream GSK-3.
GS Phosphorylations
Site 3a (ser641).
Insulin induced the activation of dephosphorylation on site 3a, leading to ∼30% reduced phosphorylation during HE (P < 0.001) (Fig. 4C), whereas the combined effects of insulin and hypoglycemia during HH decreased phosphorylation on site 3a by ∼13% (not significant) compared with CTR at t = 30 min. At t = 75 min, the insulin effects during HE were a little smaller, and phosphorylation on site 3a was decreased ∼25% (P < 0.01) compared with CTR. The effects of insulin were totally abrogated by hypoglycemia at t = 75 min when comparing HH with CTR.
Sites 2+2a (ser7+ser10).
Insulin decreased phosphorylation on site 2+2a by ∼20% (not significant) at t = 30 min and by ∼45% at t = 75 min (P = 0.001) during HE compared with CTR (Fig. 4D). The effects of hypoglycemia totally abolished the effects of insulin on GS site 2+2a, and thus increased phosphorylation by ∼90% (P < 0.001) at t = 30 min and by ∼60% (P < 0.01) at t = 75 min when comparing HH with CTR. Comparison of HH and HE revealed a ∼2.5-fold increase in phosphorylations on site2+2a during HH at both t = 30 and 75 min (both P < 0.001).
Sites 1a and 1b.
Phosphorylation on sites 1a and 1b revealed no statistically significant interaction (time × intervention; P = 0.09 and 0.06, respectively) (Fig. 4E and F). Thus the results were analyzed accordingly. A main effect of the intervention between HH and CTR was found regarding both sites 1a and 1b, with ∼25% increased site 1a phosphorylation (P = 0.01) and ∼15% increased site 1b phosphorylation (P < 0.05) during HH.
GSK-3 Phosphorylations (α and β).
Insulin increased phosphorylation of GSK-3α ∼40% (P < 0.001) and of GSK-3β ∼25% (P < 0.01) during both HE and HH compared with CTR at t = 30 min (Fig. 5A and B). No statistically significant differences in either GSK-3α or GSK-3β phosphorylation were detected when comparing HE with HH at t = 30 or 75 min. At t = 75 min, GSK-3α phosphorylation had normalized during HH, whereas GSK-3β phosphorylation was increased ∼20% compared with that of the CTR (P = 0.04). No statistically significant differences in GSK-3β phosphorylation were detected when comparing HE with both HH and CTR.
GPa and GPb
GP is present in two forms, GPa and GPb (Fig. 5C and D). They differ in phosphorylation, and GPb is converted into GPa by phosphorylation at two sites (28).
GPa.
No statistical significant interaction (time × intervention; P = 0.3) or main effect of any intervention (P > 0.05) was observed.
GPb.
At t = 30 min, GPb decreased during CTR and increased during HH, revealing an ∼75% increase in GPb levels during HH compared with CTR (P = 0.003). Levels were stable during HE and not statistically significantly different from those during either the CTR or HH at t = 30 and 75 min (P > 0.05). Total glycogen in skeletal muscle biopsies was not statistically significantly different between any time points among the interventions (Fig. 5E).
Glucose oxidation was determined using indirect calorimetry. Oxidation rates were at their lowest during CTR (669 kcal/day) and highest during HE (1,278 kcal/day) (P < 0,001). During HH (924 kcal/day), glucose oxidation were between that of CTR (not significant) and HE (P = 0.07). EE increased during hypoglycemia to 1,826 kcal/day, compared with 1,717 kcal/day during HE (P < 0.01). EE during CTR was 1,762 kcal/day; this was not statistically different from EE during HE or HH.
Discussion
The key finding of this study is that insulin action in skeletal muscle is profoundly inhibited during acute symptomatic hypoglycemia in humans. This was determined using a design that reflects a clinically relevant situation. Our study therefore allows for direct comparisons between acute insulin action during euglycemia and hypoglycemia. By contrast, previous investigations of hypoglycemia used insulin-clamp conditions (9,29–32) under which glucose levels are reduced for longer periods. The clamp conditions allow for steady-state estimates of glucose disposal, but the condition is far from the acute hypoglycemia experienced by insulin-treated patients. A previous investigation comparing hypoglycemia induced by either a 10-min or a 12-h insulin infusion found a substantial reduction of glucose disposal during prolonged, but not acute, hypoglycemia (33). This is in contrast to the findings in the current study, but several differences in the models could explain the discrepancies. Most important, the level of hypoglycemia was more pronounced in this study, in which glucose levels reached a nadir of ∼2.0 mmol/L, compared with ∼2.8 mmol/L in the previous study. The lower glucose levels in the current study were associated with a substantially higher epinephrine response at ∼900 pg/mL, compared with a peak <400 pg/mL in the previous investigation. A similar but less pronounced pattern was observed when comparing norepinephrine levels between the two studies. Furthermore, it is interesting that the increase in adrenaline concentration was much larger than the limited increase in noradrenaline, which indicates an adrenal effect rather than regulation by the sympathetic nervous system. This may implicate a role of β2 adrenergic receptors for counterregulation in muscles.
Because of the above-mentioned differences, our results do not necessarily conflict with existing data, but rather demonstrate the mechanisms involved during a more pronounced counterregulatory response.
The threshold for autonomic symptoms to hypoglycemia is ∼3.2 mmol/L; for neuroglycopenic symptoms it is ∼2.8 mmol/L (4). Population-based data demonstrate that 30–40% of individuals with type 1 diabetes mellitus experience between one and three episodes of severe (acquiring assistance from another person) hypoglycemia each year (34), indicating that, in a clinical setting, glucose levels <2.8 mmol/L (and probably even much lower) often occur in type 1 diabetes mellitus. Our data therefore demonstrate that decreased glucose disposal in skeletal muscle, in addition to increased endogenous glucose production, can be an important factor contributing to glucose counterregulation during symptomatic hypoglycemia.
When circulating glucose levels are within the physiological range and metabolism is normally regulated, transport across the cell membrane is the limiting factor for glucose uptake in skeletal muscle (35). During insulin stimulation, glucose is primarily disposed in skeletal muscle by facilitated transport through GLUT4 (36). Under these conditions, the amount of GLUT4 in the cell membrane and the gradient of glucose into the cell determine uptake. Therefore the decreased glucose gradient during HH contributes significantly to the decreased forearm glucose disposal observed in this and other studies (10). Km for GLUT4 is ∼4.3 mmol/L (37), and glucose clearance, defined as the ratio between glucose disposal and circulating glucose levels, can be assumed to be constant (33). In this study, glucose clearance was significantly reduced during hypoglycemia, and this could indicate reduced GLUT4 content in the cell membrane. We were not able to measure GLUT4 translocation in skeletal muscle biopsies in this study, but as expected, the interventions did not affect the total amount of GLUT4 protein. Nor did we observe impaired activation of signaling intermediates regulating GLUT4 trafficking 30 min after insulin stimulation, when glucose disposal peaked. In fact, phosphorylation of Akt and AS160 tended to be higher during HH than during HE, although these differences did not reach statistical significance. At t = 75 min, when glucose levels were nearly normalized, phosphorylation of AS160 at thr642 was lower during HH than during HE. This phosphorylation site on AS160 has consistently been shown to regulate glucose uptake (38), but the lower phosphorylation level at this time point is unlikely to explain differences in glucose disposal during acute hypoglycemia. It may instead be caused by the slightly elevated insulin levels during HE. Therefore, a lower glucose gradient through the plasma membrane during HH decreases glucose transport during these conditions. In addition, reduced GLUT4 content in the cell membrane cannot be excluded as a mechanism for decreased glucose clearance during symptomatic hypoglycemia, but it is not a consequence of reduced insulin signaling in GLUT4 translocation.
Under nonphysiological conditions in mouse models, glucose metabolism, and not transport across the cell membrane, can become the limiting factor for glucose uptake in skeletal muscle (39). Similar mechanisms may reduce glucose clearance during hypoglycemia in situations where insulin-stimulated GLUT4 translocation to the cell membrane is normal. Unlike the subtle effects of hypoglycemia on insulin signaling in glucose transport, we found profound effects of acute hypoglycemia on important regulators of glycogen synthesis. GS is the key enzyme responsible for glycogen synthesis in muscle tissue, and under physiological conditions, GS activity is stimulated by insulin (40). Reduced glycogen synthesis can impair glucose uptake through mechanisms originally described by Randle et al. (41) and may contribute to reduced glucose clearance during hypoglycemia. Impaired glycogen synthesis could be caused by the observed effects of hypoglycemia on several regulating phosphorylation sites on GS. Insulin signals to GS via activation of Akt and subsequent phosphorylation of GSK-3. This reduces the activity of GSK-3 and inhibits phosphorylation, especially of the inactivating site 3a on GS (42,43). Hypoglycemia did not affect insulin-stimulated Akt and GSK-3 phosphorylation; instead, we found evidence of direct inhibition of GS activity by catecholamine signaling. These findings are supported by the data from a previous study investigating the effects of insulin and adrenaline on Akt, GSK-3, GS activity, and phosphorylation (44). PKA directly inhibits GS activity via phosphorylation of sites 1a, 1b, and 2 (45). In addition, PKA possibly indirectly increases phosphorylation on site 3a via inhibition of the GM protein phosphatase 1 complex (12). The increased GS phosphorylations on sites 3a and 2+2a during hypoglycemia were therefore likely caused by catecholamine-stimulated PKA activity. This was further supported by the trends toward inhibition of phosphorylation at the sites 1a and 1b, although these differences failed to achieve statistical significance (P = 0.09 and 0.06, respectively).
Hypoglycemia not only decreased GS activity, it also decreased glucose oxidation despite an increase in total EE. In agreement with the reduced glucose oxidation, we did not find evidence of measurable glycogen degradation, and GPa, the active form of GP, was not increased during hypoglycemia (46). We were not able to measure all important regulators of glycogen metabolism, including protein phosphatase I complexes and G6P, in our biopsy material (46). However, our data suggest that glycogenolysis alone was not able to compensate for the reduced glucose uptake at the time point when the biopsy was obtained. This indicates that lipids become the primary metabolic substrate during hypoglycemia, which is supported by our previously published observations of the same subjects (16). Under physiological conditions, insulin reduces free fatty acid (FFA) levels. During hypoglycemia, counterregulatory hormones abrogate insulin inhibition, and lipolysis in adipose tissue is potently stimulated, leading to increased lipid availability (16). An increase in circulating FFAs induces insulin resistance in skeletal muscle in a dose-dependent manner, but this occurs only hours after FFA levels increase (47). Activation of lipolysis in adipose tissue by catecholamines has been suggested to have protein-sparing/anabolic effects in skeletal muscle (11). In rat skeletal muscle, this has been shown to be associated with catecholamine-induced Akt and mTOR activation (11). Our data extend these findings to human skeletal muscle and indicate that catecholamines may affect protein synthesis via mTOR signaling. It is therefore likely that the increase in lipid oxidation affects protein synthesis, but lipid-induced insulin resistance is unlikely to be involved in the acute response to hypoglycemia.
This study examined the effects of hypoglycemia in healthy volunteers, and this ensured homogeneous counterregulatory hormone responses. Unlike healthy individuals, subjects with diabetes treated with insulin often have impaired glucagon responses to hypoglycemia. In such cases, the catecholamine response is crucial in attempts to restore circulating glucose levels (48). The pronounced increase in catecholamine signaling to GS and the associated inhibition of glycogen synthesis observed in this study therefore demonstrate mechanisms that are likely intact among patients with impaired glucagon responses. This could indicate that treatment with β-receptor antagonists may dampen counterregulation in skeletal muscle, in addition to the well-known shift of glycemic thresholds for symptoms toward lower plasma glucose concentrations (49). The preserved glucagon response in healthy subjects is expected to increase endogenous glucose production, but it is not expected to directly target skeletal muscle because of the absence of glucagon receptors in this tissue (50). It is therefore likely that our data regarding insulin action in skeletal muscle from healthy volunteers can be extrapolated to patients with diabetes—even those with long disease duration and impaired glucagon response.
In conclusion, we found that acute symptomatic hypoglycemia decreases insulin-stimulated glucose disposal in skeletal muscle. These findings are not associated with decreased insulin signaling in glucose transport. Instead, insulin-stimulated glycogen synthesis is potently inhibited. This is explained, at least in part, by increased counterregulatory catecholamine levels leading to increased PKA signaling and to subsequent GS inactivation.
Clinical trial reg. no. NCT01919788, clinicaltrials.gov.
Article Information
Acknowledgments. The authors thank A. Mengel, K. Nyborg Rasmussen, E. Søgaard Hornemann, and K. Mathiassen, of Medical Research Laboratories; H. Zibrandtsen of the Research Laboratories for Biochemical Pathology; L. Pedersen of the Department of Endocrinology and Internal Medicine, Aarhus University Hospital, Aarhus, Denmark. The authors also thank B. Bolmgren and J.B. Birk of the Section of Molecular Physiology, Department of Nutrition, Exercise and Sports, Faculty of Science, University of Copenhagen, for their excellent technical assistance.
Funding. GPa and GPb antibodies were raised in sheep and donated by Yu-Chiang Lai, MRC Protein Phosphorylation and Ubiquitylation Unit, University of Dundee. This study was supported by Aarhus University, the Novo Nordisk Foundation, the Beckett Foundation, the KETO Study Group/Danish Agency for Science Technology and Innovation (grant no. 0603–00479 [to N.M.]), and the Danish Medical Research Council.
Duality of Interest. No potential conflicts of interest relevant to this article were reported.
Author Contributions. T.S.V. recruited the participants, conducted the trial, and performed the statistical analyses. T.S.V., M.H.V., U.K., N.M., and N.J. designed the study. T.S.V., J.R.H., J.F.P.W., M.V.S., N.M., and N.J. collected the data. T.S.V. and N.J. wrote the manuscript. M.H.V., U.K., J.R.H., J.F.P.W., M.V.S., and N.M. reviewed and edited the manuscript. All authors approved the final version of the manuscript. N.J. is the guarantor of this work and, as such, takes responsibility for the integrity of the data and the accuracy of the data analysis.