Colberg, Sheri R., James M. Hagberg, Steve D. McCole, Joseph M. Zmuda, Paul D. Thompson, and David E. Kelley. Utilization of glycogen but not plasma glucose is reduced in individuals with NIDDM during mild-intensity exercise. J. Appl. Physiol. 81(4): 2027–2033, 1996.—To test the hypothesis that substrate utilization during mild-intensity exercise differs in non-insulin-dependent diabetes mellitus (NIDDM) compared with nondiabetic subjects, seven lean healthy subjects (L), seven obese healthy subjects (O), and seven individuals with NIDDM were studied during 40 min of mild-intensity cycling (40% of peak O2 uptake). Systemic utilization of plasma glucose (Glc Rd) was determined by using isotope dilution methods. Gas exchange was measured to determine rates of carbohydrate (CHO) and lipid oxidation. During exercise, when CHO oxidation was greater than Glc Rd, the net oxidation of glycogen was calculated as the difference: CHO oxidation − Glc Rd. During mild-intensity cycling, the respiratory exchange ratio was similar across groups (0.87 ± 0.02, 0.85 ± 0.02, and 0.86 ± 0.01 in L, O, and NIDDM subjects, respectively), and CHO oxidation accounted for one-half of total energy expenditure during exercise. Glc Rd increased during exercise and was greatest in subjects with NIDDM (3.0 ± 0.2, 2.9 ± 0.2, and 4.5 ± 0.4 ml ⋅ kg−1 ⋅ min−1in L, O, and NIDDM subjects, respectively,P < 0.05), yet Glc Rd was less than CHO oxidation during exercise, indicating net oxidation of glycogen. Glycogen oxidation was greater in L and O than in NIDDM subjects (3.4 ± 1.0, 2.5 ± 0.9, and 1.7 ± 0.8 ml ⋅ kg−1 ⋅ min−1;P < 0.05). In summary, during mild-intensity exercise, NIDDM subjects have an increased Glc Rd and a decreased oxidation of muscle glycogen.
- non-insulin-dependent diabetes mellitus
- glucose uptake
- muscle glycogen
- gas exchange
mild- to moderate-intensity exercise is the form often recommended and chosen by patients with non-insulin-dependent diabetes mellitus (NIDDM) (9). Although substrate metabolism during exercise has been extensively investigated in healthy physically trained individuals (21, 24, 29), less is known about patterns of substrate utilization during exercise in individuals with NIDDM (12, 16, 19, 22, 29). Abnormalities of glucose and lipid metabolism are present at rest in individuals with NIDDM (17, 18) and could affect substrate metabolism during exercise. Fasting hyperinsulinemia and hyperglycemia in NIDDM might potentiate glucose utilization (glucose Rd) during exercise, although insulin resistance of skeletal muscle could have opposing effects. Patients with NIDDM have reduced uptake and oxidation of free fatty acids (FFA) by skeletal muscle at rest (18) and therefore could manifest a similar defect during mild-intensity exercise. Also, sedentary individuals have greater utilization of carbohydrate relative to fat during exercise than do trained subjects (5, 20, 29).
An indication that greater glucose Rd may occur in NIDDM is the greater fall in plasma glucose in NIDDM compared with nondiabetic subjects at moderate-intensity exercise (12, 22, 25). This has been attributed to an attenuated response of hepatic glucose production in NIDDM (25). Recent leg balance studies indicate, however, that during moderate-intensity cycling (60% of maximal O2 uptake) subjects with NIDDM have higher rates of glucose Rd by skeletal muscle than do healthy control subject volunteers (19).
The present study was undertaken to examine glucose Rd and FFA during mild-intensity exercise and to test the hypothesis that, in obesity and NIDDM, glucose Rd is increased whereas oxidation of fat is diminished. Substrate utilization was determined at rest, during 40 min of mild-intensity cycling, and during postexercise recovery. Isotope dilution methods were used to assess the glucose Rd and FFA Rd, and gas exchange was measured to examine substrate oxidation.
Individuals with NIDDM and lean (L) and obese (O) nondiabetic subjects were recruited for this study; nondiabetic subjects had normal glucose tolerance. The clinical characteristics of the subjects are shown in Table 1. All of the subjects were untrained, and peak O2 uptake (V˙o 2 peak), which was similar across groups, is consistent with a sedentary status and prior studies of subjects with NIDDM (26). L, O, and NIDDM subjects were matched for age. The groups had an equivalent gender distribution. NIDDM subjects were moderately obese, although O nondiabetic subjects had a higher body mass index (kg/m2) and fat mass (kg). Fat-free mass (FFM) also tended to be greatest in O subjects, but the difference among groups was not significant. NIDDM volunteers had fasting hyperglycemia (10.7 ± 1.7 mM) and elevated glycosylated hemoglobin (11.0 ± 1.0%), whereas L and O subjects had normal values. Volunteers with indications of ischemic heart disease, peripheral vascular disease, or marked dyslipidemia (triglyceride >3 mmol/l) were excluded. All subjects had normal hematologic, renal, and hepatic function. NIDDM volunteers who had proteinuria, evidence of more than mild retinopathy, symptomatic neuropathy, or who were receiving insulin therapy were excluded. The protocol was approved by the University of Pittsburgh Institutional Review Board, and informed consent was obtained from each subject.
Body composition was determined by using dual-energy X-ray absorptiometry. Transverse scans were obtained by using a Lunar model DPX-L total body scanner (Lunar Radiation, Madison, WI) for measurement of fat and lean tissue mass. Pixels of mainly soft tissue were used to calculate the ratio (R value) of mass attenuation coefficients at 38 and 70 keV. The proportion of fat and lean tissue in pixels of mainly soft tissue was calculated and averaged for the total body, trunk, and limbs by using software version 1.3Z for analysis.
V˙o 2 peak was determined during a progressive protocol on a cycle ergometer (Bosch ERG 601, Electra-Med, Flint, MI). Initial work rates were 0 and 20 W for women and men, respectively. Resistance was increased by 20 W every 2 min until subjects were unable to continue exercise. During this test, subjects breathed through a two-way non-rebreathing valve (Hans Rudolph, Kansas City, MO). Expired O2 and CO2 fractions were determined by a Marquette MGA-1100 mass spectrometer (Marquette Electronics, Milwaukee, WI). Ventilation was determined with an Interface Associates flow turbine (Interface Associates, Laguna Niguel, CA) connected to a 5-liter mixing chamber (Rayfield RMC-1, Rayfield Equipment, Waitsfield, VT). O2 uptake (V˙o 2) was calculated every 30 s; V˙o 2 peak was the highest V˙o 2 achieved during the test. Regression analysis from this test was used to calculate the target V˙o 2 (40% ofV˙o 2 peak) for the subsequent metabolic studies during mild-intensity exercise. Heart rate and blood pressure were recorded at the end of every stage of the V˙o 2 peak trial.
Mild-intensity exercise protocol.
Exercise studies were performed at the Preventive Cardiology Applied Exercise Physiology Laboratory at the University of Pittsburgh. Subjects were instructed to ingest at least 200 g of carbohydrate daily for the 3 days preceding a study and not to exercise on the day before a study. NIDDM subjects on sulfonylurea medications stopped this medication at least 5 days before an exercise study (none of the NIDDM subjects were treated with metformin). Subjects fasted overnight and were admitted to the University of Pittsburgh General Clinical Research Center at 7 a.m. A venous catheter was placed in a forearm vein and a primed (20 μCi) continuous (0.20 μCi/min) infusion of [3-3H]glucose (New England Nuclear, Boston, MA) was started, with an allowance of 3 h for isotope equilibration before basal (resting) measurements of the rates of glucose appearance (Ra) and Rd. To measure rates of FFA Ra and Rd, a continuous (0.3 μCi/min) infusion of [9,10-3H]palmitate was begun, with 1 h for isotope equilibration (14). A catheter for blood sampling was placed in the retrograde direction in a vein on the dorsum of the hand, and a heating pad was used to obtain arterialized samples. Metabolic measurements were started 30 min before exercise. With subjects sitting, blood samples were obtained for determination of plasma glucose and glucose radioactivity, FFA and palmitate radioactivity, whole blood lactate and glycerol, and plasma levels of insulin and norepinephrine. Samples were obtained at −20-, −10-, and 0-min time points. Resting gas exchange was determined by collecting expired air while subjects breathed through a low-resistance breathing valve (Daniels, Cortland, NY) connected by spiral tubing to neoprene meteorological balloons. Expired air was analyzed for O2 and CO2 content with a Marquette MGA-1100 mass spectrometer, and volumes were determined with a chain-compensated Tissot spirometer. Expired air was collected for a total of 12 min during seated rest on the cycle. At the start of exercise, resistance on the cycle ergometer was set so that subjects would work at 40% of their predeterminedV˙o 2 peak while pedaling at 60 revolutions/min.V˙o 2 was checked and the workload adjusted as necessary during the first 10 min of exercise in order for subjects to achieve their target level of exertion. At the start of cycling, infusion rates of both isotopes were doubled to maintain stable specific activities (8, 15) and were later reduced to basal rates at completion of exercise and throughout the 40 min of postexercise recovery. During exercise, blood and expired air were sampled every 10 min (10, 20, 30, and 40 min), with sampling continuing during postexercise recovery (50, 60, 70, and 80 min). Heart rate was recorded every 10 min.
Plasma glucose was measured with a glucose oxidase system (YSI glucose analyzer, Yellow Springs Instruments, Yellow Springs, OH). Blood for lactate, alanine, and glycerol analyses was mixed with chilled 7% perchloric acid, and, after centrifugation, supernatant was kept at −70°C until analysis (6). Plasma FFA was determined by using an enzymatic method (NEFAC kit; Wako Chemicals, Dallas, TX). Plasma insulin was measured by radioimmunoassay. Plasma norepinephrine was measured by a high-performance liquid chromatography method with electrochemical detection. For determination of [3-3H]glucose radioactivity, plasma was deproteinized with barium sulfate and zinc hydroxide, supernatant was dried under vacuum, residue was reconstituted in 0.5 ml of distilled water, and radioactivity was measured with liquid scintillation spectrometry. Organic extraction of plasma was performed to obtain palmitate radioactivity, with separation and collection of the palmitate fraction by isocratic reverse-phase high-performance liquid chromatography; this fraction was counted with liquid scintillation spectrometry, as previously described (18). As previously demonstrated, aqueous and organic extraction of plasma quantitatively separates tritiated palmitate and glucose (18).
Plasma glucose and palmitate Ra and Rd were calculated by using non-steady-state equations (8, 14); in NIDDM subjects, glucose Rd was adjusted for urinary loss of glucose. Rates of carbohydrate (CHO) oxidation and lipid oxidation were calculated by using the equations of Frayn (10), with protein oxidation estimated on the basis of measurement of urinary nitrogen loss. During exercise, when rates of CHO oxidation exceeded simultaneous rates of glucose Rd, the net oxidation of glycogen was estimated as the difference between CHO oxidation and glucose Rd, as previously described in well-trained healthy L subjects (24).
Data are expressed as means ± SE. Time (rest, exercise, and recovery) and group (L, O, and NIDDM) comparisons were made by using analyses of variance. A value of P ≤ 0.05 was considered statistically significant.
Substrate and hormonal responses at rest and during exercise.
During basal conditions of the study, NIDDM subjects had fasting hyperglycemia (10.7 ± 1.7 mM) compared with euglycemia in L and O nondiabetic subjects (4.7 ± 0.2 and 4.7 ± 0.2 for L and O subjects, respectively). No significant group differences existed in fasting concentrations of blood lactate or plasma FFA, as shown in Table 2. Fasting concentrations of plasma insulin were lower in L subjects (P < 0.05) but similar in O and NIDDM subjects. Plasma norepinephrine concentrations were similar across groups during basal conditions.
During 40 min of mild-intensity cycling exercise, plasma glucose did not change significantly from resting values in any of the subject groups. Blood lactate concentrations increased moderately during exercise, from 0.95 ± 0.04 to 1.15 ± 0.04 mmol/l, with no significant group differences, as shown in Table 2. During exercise, plasma FFA concentrations did not change from resting values and were not different across groups. Concentrations of plasma insulin also remained stable during exercise, but norepinephrine concentrations increased significantly (P < 0.01) and similarly in L, O, and NIDDM subjects.
Plasma glucose concentrations remained stable during the 40 min of recovery after exercise, while blood lactate concentrations decreased to basal levels. In contrast, concentrations of plasma FFA were higher during recovery after exercise than at baseline or during exercise, and plasma FFA during recovery were greater in O and NIDDM subjects than in L subjects (P < 0.05). Plasma insulin levels also increased significantly during recovery and were higher in O and NIDDM subjects than in L subjects (P < 0.05). Plasma concentrations of norepinephrine, which had increased during exercise, returned to basal levels during recovery.
Energy expenditure and substrate oxidation.
L, O, and NIDDM subjects had similar values forV˙o 2 peak, and these values are consistent with an untrained status (Table3). TheV˙o 2 during exercise was similar across groups. Expressed as a percentage ofV˙o 2 peak, the value during exercise was 42 ± 2%, which was the target intensity of exercise and was similar across groups. Rates of energy expenditure during exercise were fourfold higher than during basal conditions (P < 0.01). During basal conditions, the respiratory exchange ratio (RER), CHO oxidation, and lipid oxidation were equivalent in L, O, and NIDDM subjects. At rest, lipid oxidation accounted for approximately two-thirds of energy production in all groups. During mild-intensity exercise, rates of lipid oxidation were double the resting rates, but the increase in CHO oxidation was substantially greater, increasing 10-fold above resting rates. This pattern of lipid and glucose oxidation was similar in L, O, and NIDDM subjects. In L, O, and NIDDM subjects, respectively, lipid oxidation accounted for 37 ± 6, 47 ± 5, and 44 ± 3% of energy expenditure during exercise (nonsignificant across groups). The relative contribution of lipid oxidation to energy expenditure during exercise was less than during basal conditions (P < 0.05), as reflected in an increase of RER during exercise to 0.86 ± 0.02 (P < 0.01 vs. basal). RER was similar across subject groups during exercise. During recovery after exercise, these parameters of gas exchange returned to basal values.
Utilization of glucose, glycogen, and FFA.
During basal conditions, glucose Ra was greater in NIDDM (P < 0.05) than in L or O subjects (1.85 ± 0.15, 1.88 ± 0.18, and 2.81 ± 0.32 mg ⋅ kg−1 ⋅ min−1in L, O, and NIDDM, respectively). During exercise, glucose Ra increased by ∼50% and remained higher in NIDDM subjects (2.95 ± 0.17, 2.90 ± 0.17, and 4.15 ± 0.45 mg ⋅ kg−1 ⋅ min−1, in L, O, and NIDDM, respectively). Rates of Rd for plasma glucose were nearly identical to rates of glucose Ra, as would be anticipated from the stability of plasma glucose concentrations during exercise. Plasma glucose Rd during exercise was greater in NIDDM subjects (P < 0.05) than in L and O subjects (3.03 ± 0.17, 2.91 ± 0.21, and 4.28 ± 0.44 mg ⋅ kg−1 ⋅ min−1in L, O, and NIDDM, respectively).
Across all groups, rates of plasma glucose Rd were lower than rates of carbohydrate oxidation, as shown in Fig 1. These findings that glucose Rd was less than corresponding rates of CHO oxidation indicate oxidation of glycogen during exercise. Rates of glycogen oxidation, whether expressed per weight of FFM (Fig. 1) or as a percentage of overall energy expenditure (Fig.2), were lowest in NIDDM subjects (P < 0.05 vs. L or O), whereas differences between L and O subjects did not achieve significance (P = 0.10). Thus, at similar rates of CHO oxidation during mild-intensity exercise, subjects with NIDDM oxidized a greater amount of plasma glucose and a lesser amount of glycogen than did nondiabetic individuals.
During recovery after exercise, glucose Ra decreased to values that were slightly and significantly lower than had been present during baseline conditions (P < 0.05 for all groups). The disparity between glucose Ra in NIDDM and nondiabetic subjects was also less during recovery than at baseline (1.52 ± 0.09, 1.50 ± 0.11, and 1.95 ± 0.18 mg ⋅ kg−1 ⋅ min−1in L, O, and NIDDM). During this time, rates of glucose Rd exceeded rates of overall carbohydrate oxidation (0.54 ± 0.14, 0.66 ± 0.20, and 0.78 ± 0.15 mg ⋅ kg−1 ⋅ min−1in L, O, and NIDDM, respectively).
During basal conditions, rates of palmitate utilization were similar in L, O, and NIDDM subjects (1.91 ± 0.53, 2.46 ± 0.69, and 2.14 ± 0.23 μmol ⋅ kg FFM−1 ⋅ min−1, respectively). These rates increased during mild-intensity exercise (P < 0.01) but remained similar across groups (2.44 ± 0.81, 3.35 ± 0.60, and 2.67 ± 0.40 μmol ⋅ kg FFM−1 ⋅ min−1for L, O, and NIDDM, respectively). During recovery after exercise, rates of palmitate utilization decreased in all groups (P < 0.05) but were lower in L subjects (1.04 ± 0.21, 2.49 ± 0.59, and 1.74 ± 0.41 μmol ⋅ kg FFM−1 ⋅ min−1in L, O, and NIDDM subjects, respectively;P < 0.05).
Regular exercise is regarded as one of the cornerstones in the management of patients with NIDDM. In part, this recommendation for exercise in patients with NIDDM is aimed at improving insulin sensitivity (2, 3, 7). The effect of exercise to improve insulin sensitivity is largely attributable to short-lived (i.e., 1–2 days) effects of each session rather than long-term effects of training per se (21), especially if there are not concomitant changes in weight (27). From the perspective, therefore, of potential short-term effects of exercise, it is important to determine patterns of substrate utilization during an exercise session. For example, a single bout of strenuous (glycogen-depleting) exercise can improve insulin sensitivity and lower hepatic glucose production in NIDDM (7). Recently, our group has reported that rates of glucose utilization during moderate-intensity exercise are similar in previously sedentary NIDDM and O nondiabetic subjects (16). This is consistent with an earlier report by Minuk et al. (22), who noted a greater decline of plasma glucose in NIDDM than in nondiabetic subjects during moderate-intensity exercise and attributed this to an attenuated stimulation of hepatic glucose production. Increased glucose utilization during exercise in NIDDM compared with nondiabetic subjects could also be a factor in the more pronounced change in plasma glucose during exercise in NIDDM. Martin et al. (19) examined glucose Rd by skeletal muscle during moderate-intensity exercise (60% of maximal O2 uptake) by measuring arteriovenous balance across the leg and found higher rates in NIDDM compared with nondiabetic subjects. Thus, despite the insulin resistance that characterizes skeletal muscle glucose metabolism during resting conditions in NIDDM, prior studies indicate that exercise-induced glucose utilization is either not reduced or even increased in NIDDM. The current study was undertaken to examine the effects of NIDDM on substrate utilization during mild-intensity exercise. Although mild-intensity exercise is the form most commonly chosen by patients with NIDDM (9), there have not been published data regarding patterns of substrate utilization in patients with NIDDM during mild-intensity exercise. Braun et al. (3) have recently reported that low-intensity exercise is effective in improving insulin sensitivity in NIDDM, complementing other studies that have indicated that moderate- to high-intensity exercise improves insulin sensitivity in NIDDM (2, 7).
In the present study, subjects with NIDDM were found to have higher rates for glucose Rd during mild-intensity exercise than comparably sedentary L and O glucose-tolerant subjects. Thus our results are concordant with those of Martin et al. (19). In the present study, rates of CHO oxidation and lipid oxidation during exercise were, however, similar in NIDDM and nondiabetic subjects. Also, rates of CHO oxidation exceeded simultaneous rates for the glucose Rd during mild-intensity exercise. Because glucose Rd during mild-intensity exercise was less than CHO oxidation, this indicates net oxidation of glycogen as an additional contribution to CHO oxidation. The net oxidation of muscle glycogen can be estimated as the difference between CHO oxidation determined by indirect calorimetry and the glucose Rd determined by isotope dilution methods (24). Using this approach, we found that subjects with NIDDM, at equivalent rates of energy expenditure and CHO oxidation, oxidized a lesser amount of glycogen and had a proportionately greater glucose Rd during mild-intensity exercise than did nondiabetic subjects. We cannot discern with this approach whether there were group differences in rates of glycogenesis and glycogenolysis during exercise but only that the results indicate less net oxidation of glycogen in NIDDM. Whether the lower net use of glycogen in NIDDM was solely compensatory to greater uptake of plasma glucose or whether these were additional effects related to altered metabolism of muscle glycogen in NIDDM also cannot be addressed with the present study.
Hyperglycemia is probably of primary importance in accounting for increased uptake of plasma glucose during exercise in NIDDM. At rest, hyperglycemia in NIDDM partially compensates for insulin resistance (17). If contraction-induced translocation of the glucose transporter isoform GLUT-4 and glucose transport are normal in insulin-resistant human skeletal muscle, as has been found in a rodent model of insulin resistance (4), then mass action effects of hyperglycemia may be accentuated during exercise. Of note, insulin-resistant human skeletal muscle has normal glucose transport in response to hypoxia (1), which connotes that impairment of glucose transport in NIDDM may be selective for insulin stimulation. The clinical physiology described by the present study, as well as the findings of Martin et al. (19), suggests that, during exercise, glucose transport in NIDDM is not impeded and, in fact, may even be enhanced by concomitant hyperglycemia. The present study was undertaken as a single session of exercise in untrained subjects. Because exercise training can increase the expression of GLUT-4 in skeletal muscle (11), it would be of interest to determine what impact training-induced increases of GLUT-4 might have on glucose utilization during exercise in individuals with NIDDM.
In healthy, physically trained L individuals, utilization of plasma FFA can account for the majority of energy production during sustained mild-intensity exercise (24). This is not the case in sedentary subjects, who seem to rely less on fat oxidation (20, 29). In the present study, we hypothesized that untrained individuals with NIDDM would have still lower rates of fat oxidation during mild-intensity exercise than comparably sedentary L and O nondiabetic subjects. This hypothesis was based on lower rates of FFA utilization by skeletal muscle during resting conditions in NIDDM (18). However, rates of fat oxidation during mild-intensity exercise were similar in NIDDM and nondiabetic subjects. Rates of systemic palmitate utilization were also similar across groups. Thus, with regard to utilization of lipid in NIDDM, the abnormalities found at rest do not appear to be manifest during exercise. In comparing L women, O women with upper body fat distribution and O women with lower body fat distribution, among whom there are differences in resting rates of lipolysis (14) and FFA utilization by skeletal muscle (6), Kanaley et al. (15) found similar rates of palmitate utilization and overall lipid oxidation during mild-intensity exercise, which is consistent with the present findings. Several caveats temper complete rejection of hypotheses concerning impaired FFA utilization by skeletal muscle during exercise in NIDDM. One caveat is that we did not directly assess skeletal muscle uptake of palmitate, which would have required limb balance methodology. Also, the metabolic rate of FFA taken up by muscle cannot be ascertained with the isotope methods used, and, on the basis of systemic gas exchange, we cannot distinguish between the oxidation of plasma FFA vs. that of muscle triglyceride.
The low fractional contribution of fat oxidation (less than one-half of overall energy expenditure during mild-intensity exercise) was similar across groups and is consistent with the closely matched sedentary status of the subjects. Factors limiting fat oxidation during exercise in sedentary individuals may include lower oxidative capacity of skeletal muscle (28) and reduced skeletal muscle capillary density. The capacity of untrained individuals to utilize plasma FFA during exercise can be improved by exercise training (20, 29). Among nondiabetic subjects, just 10 days of daily exercise improve lipid oxidation in previously sedentary subjects (23). It is uncertain whether subjects with NIDDM would respond to training in a similar manner. This is a question of clinical importance given the adverse impact of adiposity and elevated plasma FFA and lipoproteins in NIDDM. Bogardus et al. (2) found that, after weight loss and exercise training, subjects with NIDDM did increase their rates of resting lipid oxidation. However, on the basis of the present findings that glucose uptake is increased in NIDDM during exercise and recognizing that hyperglycemia is not usually rectified by exercise training in NIDDM, we postulate that subjects with NIDDM might not be able to as effectively augment lipid utilization during exercise training as can nondiabetic individuals.
In summary, during mild-intensity exercise, oxidation of CHO and lipid is similar in L and O sedentary nondiabetic subjects and O individuals with NIDDM. Rates of glucose Rd are increased during exercise in NIDDM, although still less than rates of CHO oxidation during exercise. This indicates that the contribution to energy expenditure from the net oxidation of muscle glycogen is reduced in NIDDM compared with nondiabetic individuals. The findings of increased glucose Rd during exercise in NIDDM suggest that contraction-mediated glucose transport and oxidation are not impaired in NIDDM despite the severe insulin resistance that characterizes glucose homeostasis during resting conditions.
This project was supported by a Department of Veterans Affairs Merit Award and by the University of Pittsburgh General Clinical Research Center (5MO1RR00056).
Address for reprint requests: D. E. Kelley, Division of Endocrinology and Metabolism, E-1140 Biomedical Science Tower, Univ. of Pittsburgh, Pittsburgh, PA 15261.
- Copyright © 1996 the American Physiological Society