Systemic and Local Adrenergic Regulation of Muscle Glucose Utilization During Hypoglycemia in Healthy Subjects

We prospectively studied 12 healthy adult subjects (9 men, 3 women; age 27.3 ± 4.7 years, BMI 24.5 ± 4.5 kg/m², mean ± SD) on no medications. The Institutional Human Use Committee of the University of Iowa approved the studies, and written informed consent was obtained from all subjects before the study.

An insulin bolus and constant infusion were administered through a catheter placed in the antecubital fossa in the nondominant arm to produce and maintain plasma insulin at 40 mU · m⁻² · min⁻¹ for 120 min. For the first 60 min, arterial plasma glucose concentrations were maintained at euglycemic levels using a variable 20% dextrose infusion; arterial plasma glucose levels were then lowered to 50 mg/dl for the next 60 min. Regular insulin (Humulin; Eli Lilly, Indianapolis, IN), diluted in saline with 1 ml of the subject’s blood, was infused intravenously using a syringe pump (Bard Medsystems, North Reading, MS); dextrose was infused using an infusion pump (Flo-Gard 6200; Travenol Laboratories, Deerfield, IL). Plasma arterial and venous glucose concentrations were determined at 5-min intervals during a 25-min baseline period before the clamp and during the clamp.

Subjects were admitted to the Clinical Research Center at the University of Iowa and studied in the Human Cardiovascular Physiology Laboratory at similar times each morning on 5 separate study days after an overnight fast. Intravenous catheters were placed in each arm. The catheter in the dominant arm was placed in the antecubital fossa with the tip nonpalpable so that blood sampled from this catheter reflected forearm drainage (28). The catheter in the nondominant arm was used to infuse insulin and 20% dextrose. A 27-gauge steel needle attached to 16-gauge epidural catheter was placed in the brachial artery of the dominant study arm to infuse propranolol, phentolamine, or saline all at a rate of 1 ml/min. This method of intra-arterial infusion does not interfere with blood flow measurements or changes (29). Venous occlusion plethysmography (EC4; Hokanson, Bellevue, WA) with two plethysmographic cuffs and a mercury-in-silastic strain gauge was used to measure FBF in the dominant arm. In the nondominant arm, a radial artery catheter was placed for blood sampling and continuous arterial pressure measurement. Heart rate was measured continuously by electrocardiography.

During the entire study, arterial pressure, heart rate respirations, and FBF were recorded for 5 min of every 10 min. Just before and after each 5-min measurement, arterial and venous blood was drawn for plasma insulin and glucose levels. Five minutes of baseline measurements were recorded before the initiation of intravenous propranolol (IV PROP, 80 μg/min), intravenous phentolamine (IV PHEN, 500 μg/min) (2), intra-arterial propranolol (IA PROP, 25 μg/min), and intra-arterial phentolamine (IA PHEN, 12 μg/min per 100 ml of forearm tissue), and saline (SAL), as randomly assigned. This dose of IA PHEN decreases forearm vasoconstriction during norepinephrine infusion from 67 to 15% without affecting systemic blood pressure or heart rate (30). The dose of IA PROP blocks isoproterenol induced vasodilation without systemic effects (C.A.S., unpublished data). The infusions were started 20 min before beginning insulin. The hyperinsulinemic clamp was then initiated and maintained for 120 min with 60 min of euglycemia followed by 60 min of hypoglycemia. Just before the clamp, at 60 min, and at 120 min, arterial and venous blood was drawn for measurement of plasma lactate and free fatty acid (FFA) concentrations; venous blood was drawn for measurement of plasma catecholamines.

Subjects verbally completed a symptom questionnaire at the end of the baseline period before the clamp and just before the completion of euglycemia and hypoglycemia (31). Symptoms were assigned as adrenergic, neuroglycopenic, or cholinergic as assigned by Towler et al. (32).

Electrocardiogram, FBF, respirations, and arterial pressure were measured simultaneously with a computerized data acquisition system (MacLab; AD Instruments, Grand Junction, CO) and Macintosh Quadra 950 Computer (Apple Computer, Cupentino, CA). FBF was measured as milliliters per minute per 100 ml of forearm volume. FGU was measured using the Frick principle (FGU = FBF [arterial glucose − venous glucose (AV-diff)] [1 − hematocrit]).

Plasma glucose was measured immediately using one of two YSI 2300 Stat Glucose Analyzers (Yellow Springs Instruments, Yellow Springs, OH). Arterial and venous samples were alternately measured on each machine; lactate was measured on one of the analyzers that also had a lactate probe. Catecholamines were determined by adding 1 ml centrifuged plasma to a glass extraction vial containing 20 mg acid-washed alumina (Bioanalytical Systems, West Lafayette, IN), 20 μl of a solution containing the internal standard (3,4 dihydroxy-benzylamine in 0.01 N HCl), 1 ml phosphate buffer (0.l mol/l [pH 7.0] plus 0.05 mol/l EDTA) and l ml Tris buffer (1.5 mol/l [pH 8.6] plus 0.05 mol/l NaEDTA). After immediate gentle shaking for 10 min, the alumina was allowed to settle and the supernatant was aspirated to waste. After two washes with water, catecholamines were eluted from the alumina with 200 μl of 4% acetic acid. After centrifugal microfiltration using individual 0.2-μm regenerated cellulose membranes, each sample was chromatographed on a Catecholamine Column (Keystone Scientific, Bellefonte, PA; 3-μ particle size, 100 × 4.6 mm, reverse-phase, C-18 ODS, 10% carbon, end-capped) using a mobile phase of 75 mmol/l monobasic sodium phosphate, 0.12 mmol/l NaEDTA, 10 mmol/l citric acid, 15% acetonitrile, 10% methanol, and 1.5 mmol/l sodium dodecyl sulfate as the ion pairing agent. The catecholamines were detected with a Coulochem II Dual Potentiostat Electrochemical Detector (ESA, Chelmsford, MA). Peaks were quantified on a Shimadzu CR5-A integrator. A standard curve for extracted catecholamines (0, 125, 250, 500, 750, 1,000, 1,500, and 2,000 pg of each catecholamine) was prepared using “blank” human plasma (dialyzed to remove endogenous catecholamines), and linear regression analysis was used to determine sample plasma concentrations. The assay has interassay and intra-assay coefficients of variation of 3.4 and 3.1%, respectively, and a lower limit of detection of 25 pg/ml. Plasma FFAs were measured by an enzymatic colorimetric method (Wako NEFA C Kit; Biochemical Diagnostics, Edgewood, NY). The low-, middle-, and high-range coefficients of variation were 2.7, 1.1, and 1.1%, respectively. Plasma insulin was measured by double-antibody radioimmunoassay with interassay and intra-assay coefficients of 9.4 and 5.3%, respectively.

Differences in measured parameters were determined using a one-way repeated measures ANOVA of all data. Planned contrasts were used to determine differences between groups and for changes between the end of euglycemia and hypoglycemia. Multiple linear regression analysis was used to assess correlations between subjects. Statistical significance was defined as P < 0.05. Results are reported as mean ± SE.

Arterial plasma glucose levels did not differ at any time between the studies. At the end of hypoglycemia, mean arterial plasma glucose levels were 2.7 ± 0.1 mmol/l for SAL, 2.8 ± 0.1 mmol/l for IA PROP, 2.9 ± 0.1 mmol/l for IV PROP, 2.8 ± 0.1 mmol/l for IA PHEN, and 3.0 ± 0.1 mmol/l for IV PHEN. Venous plasma insulin levels did not differ at any time among the three studies (Fig. 1).

ANOVA revealed that heart rate varied over time (F_14–462 = 6.13; P < 0.001) and that there were differences between studies (time by study interaction, F_56–462 = 6.33; P < 0.001). IA PROP, IV PROP, IA PHEN, and IV PHEN alone did not alter heart rate. Heart rate did not change during euglycemic hyperinsulinemia with SAL, IA PROP, or IV PROP but was increased at the end of 60 min of euglycemic hyperinsulinemia with both IA PHEN (P = 0.003) and IV PHEN (P < 0.001). Heart rate tended to fall during hypoglycemia with IA PROP (P = 0.057) and IV PROP (P = 0.072) but did not change with SAL. Heart rate was increased further at the end of hypoglycemia for IV PHEN (P < 0.001; Table 1).

ANOVA also revealed that mean arterial pressure varied over time (F_14–462 = 5.01; P < 0.001) and that there were differences between studies (time by study interaction, F_56–462 = 8.31; P < 0.001). During the initial infusion alone period, mean arterial pressure fell with IV PHEN (P < 0.001). No additional changes occurred during the first 60 min with euglycemic hyperinsulinemia during any session. Mean arterial pressure significantly increased during hypoglycemia with both IA PROP and IV PROP (P = 0.001 and P = 0.003, respectively) but tended to fall with SAL (P = 0.077). Mean arterial pressure fell significantly at the end of hypoglycemia with both IA PHEN and IV PHEN studies (P < 0.001).

FBF (Fig. 2) significantly changed with time (F_12–462 = 33.8; P < 0.001), and these changes differed between sessions (time by session interaction, F_56–462 = 12.0; P < 0.001). FBF did not change with SAL, IV PROP, or IV PHEN but decreased 20 min after starting IA PROP (P = 0.009) and increased after 20 min of IA PHEN (P < 0.001). FBF did not change during 60-min euglycemic insulin infusion with SAL, IA PROP, or IV PROP but increased at the end of euglycemic hyperinsulinemia with IA PHEN (P = 0.049) and with IV PHEN (P = 0.048). FBF at the end of hypoglycemia was increased compared with the end of euglycemia with SAL (P < 0.001), IA PHEN (P < 0.001), and IV PHEN (P = 0.003). There was no change during hypoglycemia with IA PROP or IV PROP.

AV-diff (Fig. 3) also changed significantly with time (F_12–462 = 31.3; P < 0.001), and these changes differed between sessions (time by session interaction, F_56–462 = 1.43; P = 0.026). AV-diff did not change with any of the infusions alone and increased, as expected, during euglycemic insulin infusion with SAL (P = 0.010), IA PROP (P = 0.005), IV PROP (P = 0.001), and IV PHEN (P = 0.014). The increase in AV-diff during euglycemic hyperinsulinemia with IA PHEN did not quite reach statistical significance (P = 0.060). At the end of hypoglycemia (120 min), AV-diff was lower than at the end of euglycemia (60 min; P < 0.001) during SAL, IA PHEN (P = 0.007), and IV PHEN (P = 0.006). AV-diff at the end of hypoglycemia and euglycemia tended to be different with IA PROP (P = 0.084) but not with IV PROP.

FGU varied with time (F_12–462 = 23.0; P < 0.001), but the changes over time between sessions were not different. Baseline FGU was not altered by any of the infusions. Again, as expected, FGU increased during 60 min of euglycemic insulin infusion with IA SAL (P = 0.051), IA PROP (P = 0.018), IV PROP (P = 0.022), IA PHEN (P = 0.001), and IV PHEN (P = 0.007). FGU at the end of hypoglycemia with SAL tended to fall (P = 0.079) as a result of the fall in glucose extraction, but this was counterbalanced by an increase in glucose delivery. The nonsignificant decreases in FBF and AV-diff during hypoglycemia with IA PROP led to a significant decrease in FGU at the end of hypoglycemia (P = 0.015). FGU was unchanged at the end of hypoglycemia with IV PROP because glucose extraction was unchanged and glucose delivery fell only slightly. FGU at the end of hypoglycemia was not different from that at the end of euglycemia for IA PHEN and IV PHEN because the decreases in glucose extraction were offset by the increase in glucose delivery (Fig. 4).

Arterial plasma FFA levels changed with time (time by session interaction, F_2–64 = 46.8; P < 0.001), and the changes differed between sessions (F_8–64 = 3.49; P = 0.002). Arterial plasma FFA levels fell during euglycemic hyperinsulinemia for all five sessions and fell further during hypoglycemia with IV PROP (P = 0.027) but increased during hypoglycemia with IA PHEN (P = 0.032) and IV PHEN (P = 0.008). No change was seen during hypoglycemia with IA PROP or SAL. Arterial FFA (AFFA) levels at the end of hypoglycemia with IV PROP were lower than after IA PROP (P = 0.022), SAL (P = 0.044), or IV PHEN (P = 0.003), and levels were higher at the end of hypoglycemia with IV PHEN than with SAL (P = 0.035). Net forearm fat uptake did not change during euglycemia hyperinsulinemia or with hypoglycemia and did not differ between subjects (Table 2).

Forearm lactate production increased during euglycemic hyperinsulinemia during all sessions (P < 0.02); it did not increase further during hypoglycemia with IA PROP, IV PROP, or IV PHEN but did with SAL (P = 0.005) and IA PHEN (P < 0.001).

Plasma epinephrine increased during hypoglycemia with all sessions (P < 0.05). Levels were significantly higher at the end of hypoglycemia with IA PROP and IV PROP than with SAL. Plasma norepinephrine levels increase during euglycemic hyperinsulinemic clamp with IA PHEN and IV PHEN (P < 0.05) and during hypoglycemia for the other three studies.

For all studies, AV-diff at the end of hypoglycemia was negatively correlated with both AFFA level (AFFA, r = −0.41, P = 0.012; Fig. 5) and FBF (r = −0.55, P < 0.001). Multiple linear regression analysis including both variables indicated that the equation AV-diff = 0.70 − 0.028 · FBF − 0.23 · AFFA accounts for 33% of the variance in AV-diff. The relationship to FBF was statistically significant (P = 0.002). The relationship with AFFA approached significance (P = 0.065).

Adrenergic symptom scores (Fig. 6) significantly increased from euglycemia to hypoglycemia with IV PHEN (P < 0.001) but not with IV PROP or SAL. Adrenergic symptom scores at the end of hypoglycemia were higher with IV PHEN than with IV PROP (P = 0.003) or SAL (P = 0.009). Neuroglycopenic symptoms followed a similar pattern with a significant increase only during hypoglycemia with IV PHEN (P = 0.001). Again, neuroglycopenic symptom scores with IV PHEN were higher at the end of hypoglycemia than with IV PROP (P = 0.031) or SAL (P = 0.008). The latter two did not differ. Cholinergic symptoms increased during hypoglycemia with IV PHEN (P < 0.001) and IV PROP (P < 0.001) but not during SAL. Cholinergic symptoms at the end of hypoglycemia tended to be higher with IV PHEN than with SAL (P = 0.054). Cholinergic symptoms at the end of hypoglycemia with IV PROP were not different from the other studies.

Epinephrine induces peripheral insulin resistance and thus decreases peripheral glucose uptake for up to 6 h after short-term epinephrine infusion (33,34). One possible mechanism for this is through increased FFAs, which provide an alternative fuel and competitively inhibit glucose uptake (35,36). Glucose infusion rates during insulin clamps are inversely related to fat oxidation rates in control subjects (37) and patients with type 2 diabetes (38). The increase in FFAs provides an alternative fuel to glucose for muscle utilization and competitively inhibits muscle glucose utilization (34,37). FFAs have also been shown to alter cell membrane glucose transport (39).

Epinephrine may also have a direct muscular action (40,41). Intra-arterial epinephrine infusions have been shown to decrease FGU (27) through several possible mechanisms. These include decreased intrinsic activity of GLUT4 glucose transporters at the cell membrane (10) and increased muscle glycogenolysis and intracellular glucose-6-phosphate, which inhibits cellular glucose transport (42,43).

Our results demonstrate that during hypoglycemia, the main local effect of increased β-adrenergic activity and thus epinephrine is to increase blood flow and glucose delivery to the tissues. In normal situations, this vasodilatory effect is partially balanced by an increase in vasoconstrictor α-adrenergic activity as indicated by the differences in FBF between hypoglycemia with SAL and hypoglycemia with IA PHEN. Because glucose extraction decreased during hypoglycemia with IA PROP and was unchanged during hypoglycemia with IV PROP, β-adrenergic system activation during hypoglycemia primarily causes muscular insulin resistance through its systemic effects. The differences in AFFA levels between IA PROP and IV PROP at the end of the two studies and the correlation between AFFA levels and AV-diff support the hypothesis that the systemic effect is exerted through increased lipolysis. Last, the lack of relationship between forearm FFA and glucose uptake suggests that the effect is mediated independent of fat utilization and is likely related to FFAs altering membrane glucose transport (39).

Microdialysis studies have indicated that epinephrine has a dual effect on lipolysis in subcutaneous tissues. It stimulates lipolysis primarily through β-adrenergic receptors and inhibits lipolysis primarily through α2-adrenergic receptors (44,45). Our results during hypoglycemia agree with these findings because AFFA levels fell during hypoglycemia with IV PROP but increased dramatically during hypoglycemia with IV PHEN and unopposed β-adrenergic activity.

From our data, there seems to be little difference between the local and systemic effects of α-adrenergic activation on muscular glucose utilization because the pattern changes in FBF, AV-diff, and FGU were very similar. This cannot be said conclusively because the intra-arterial phentolamine, at the dose used, clearly had a systemic effect and local effect. This is demonstrated by the similarities in FFAs during hypoglycemia with IA PHEN and IV PHEN. The differences in heart rate and blood pressure indicate that the systemic effect was less with IA PHEN than with IV PHEN. Conversely, the local effect of IV PHEN was less than that of IA PHEN because FBF was greater with the latter. It is clear, however, that the main effect of α-adrenergic activation during hypoglycemia is to prevent excess vasodilation. The increased glucose delivery during hypoglycemia with α-adrenergic blockade is offset by a decrease in glucose extraction. This decrease likely is a compensatory response to the increases in flow and is due to increased AFFA concentrations.

Of particular interest during α-adrenergic blockade is that there is a marked increase in adrenergic and neuroglycopenic symptoms during hypoglycemia with IV PHEN compared with SAL or IA PROP. Although our data allow us to speculate only regarding the cause, one possibility is that the increase is due to decreased glucose delivery to the brain secondary to the marked fall in blood pressure. This effect thus would be similar to that of caffeine, which has been shown to decrease middle cerebral artery flow velocity and increase autonomic and neuroglycopenic symptoms during hypoglycemia in patients with type 1 diabetes (46). This study also indicated that caffeine increased the epinephrine response to hypoglycemia. We did not see differences in epinephrine response between hypoglycemia with IA PHEN and SAL; however, the changes in catecholamine in our study were clearly altered by the blocking agents and thus do not reflect what might happen with decreased brain blood flow during hypoglycemia under other conditions. Thus, the α-adrenergic inhibition of excessive vasodilation maintains normal blood pressure and likely helps sustain brain blood flow. The increased cholinergic symptoms with both IV PHEN and IV PROP may be due to unopposed cholinergic activity.

The major limitation of our study involves the use of venous occlusion plethysmography for measurement of FBF. With this technique, direct comparisons between different sessions are not appropriate. Because FBF was used in determining FGU, the variability in FBF likely explains the lack of significant differences between sessions. Comparisons within sessions are valid, however (28), and thus the comparison of responses between sessions is valid.

These results help discern some of the complex local and systemic interactions by which sympathetic activation protects the body and brain from the adverse effects of hypoglycemia. β-Adrenergic activation decreases muscular glucose extraction primarily by increasing FFAs, whereas α-adrenergic activation prevents excessive β-induced vasodilation.

These studies were supported by a Juvenile Diabetes Foundations Research Grant (R.P.H.) and The National Institute of Health Clinical Research Center Grant RR59. J.M.D. is a recipient of an American College of Clinical Pharmacy Research Institute Cardiovascular Fellowship. B.G.P. is a Sleep Academic Awardee of the National Institutes of Health and is supported by NIH grants HL-14388 and HL-65176.

We thank Cynthia Broderick for help with sample and data analysis and the nurses of the Clinical Research Center for help in the care of the patients.