Quantitation of Muscle Glycogen Synthesis in Normal Subjects and Subjects with Non-Insulin-Dependent Diabetes by ¹³C Nuclear Magnetic Resonance Spectroscopy

Subjects

We studied 11 subjects — 5 men with NIDDM (1 of whom was studied twice) and 6 healthy men matched to the diabetic subjects for age and weight. All subjects were within 20 percent of their ideal body weight according to the 1959 Metropolitan Life Insurance tables. The mean (±SD) weight in the diabetic and normal subjects was 82.1±8.6 kg and 79.6±5.6 kg, respectively; their mean ages were 62±7 and 55±10 years. The mean duration of diabetes, calculated from the first record of a plasma glucose concentration above 7.8 mmol per liter (140 mg per deciliter), was 14 years. All the diabetic subjects fulfilled the criteria for the diagnosis of diabetes mellitus as established by the National Diabetes Data Group.⁵ All of them had been treated with oral sulfonylurea agents; their medication was discontinued at least 10 days before the study. Their mean hemoglobin A_lc level was 11.3±1.5 percent (normal range, 4 to 8 percent), and their mean fasting plasma glucose concentration after the discontinuation of medication was 13.0±1.3 mmol per liter. No subject had a major disease other than diabetes mellitus or was taking other medications. None of the normal subjects had a family history of diabetes.

Informed consent was obtained from all subjects after the purpose, nature, and potential risks of the study were explained to them. The protocol was reviewed and approved by the Human Investigation Committee of the Yale University School of Medicine.

Experimental Protocol

All studies were begun at 8 a.m. after an overnight fast of 10 to 12 hours. On the evening before the study, all the diabetic subjects were admitted to the clinical research center and received an overnight variable infusion of insulin (0.6 to 1.5 pmol per kilogram of body weight per minute) through a Teflon catheter placed in an antecubital vein, in order to induce normoglycemia (≈5.5 mmol per liter). The induction of euglycemia allowed the diabetic and normal subjects to be studied while they had the same plasma glucose concentrations and responded to an identical increment above base line in the plasma glucose concentration. On the morning of the study a second Teflon catheter was inserted into the antecubital vein of the opposite arm to permit blood to be drawn. The normal subjects were admitted to the center at 8 a.m. after they had fasted for 12 to 14 hours, at which time an intravenous catheter was inserted in each arm.

NMR Spectroscopic Techniques

Throughout the study the subjects remained in the supine position within an NMR spectrometer (Biospec, 1-m bore, 2.1 T); the gastrocnemius muscle of the right leg was positioned within the homogeneous volume of the magnet, on top of ¹H—¹³C concentric surface coils (a Lucite plate 6 mm thick was placed between the coil and the leg). The leg was strapped into place with Velcro strips to minimize any motion during the experiment. A multislice gradient-echo image was obtained to confirm that virtually all of the NMR signal observed reflected underlying gastrocnemius muscle. The surface coils, used for both transmitting and receiving, consisted of an inner coil with a 9-cm diameter, which was used for ¹³C acquisition, and an outer coil with a 13-cm diameter, which was used for 1H acquisition and decoupling. The magnet was shimmed with the localized water signal. The loaded 90-degree pulse lengths were 150 μsec for ¹H and 100 μsec for ¹³C. The ¹³C radiofrequency pulse power during the ¹³C pulse was 400 W and was on for 0.2 percent of the cycle time. For ¹H decoupling, 10 W of continuous-wave energy at the C-1—proton frequency was applied with a 20 percent duty cycle. The total heat deposition did not exceed the guideline of the Food and Drug Administration (8 W per kilogram of tissue). ¹³C spectra were acquired with a 1—t—1 sequence, which was tailored to excite the 100-ppm region and to minimize the 30-ppm lipid region. Each pulse was set to be 45 degrees (≈75 μsec) at the center of the coil. The pulse angle at the coil center was determined from a 2-cm sphere containing [¹³C]formic acid (99 percent ¹³C enrichment) as a standard. The standard was also used to correct for any changes in the quality (Q) factor of the surface coil or system sensitivity. The interpulse delay was 320 μsec. The repetition time of 80 msec was optimized for the C-1 glycogen signal, for which the T₁ component was 80 msec and the T₂ component was 11 msec. Each spectrum required 11,250 scans and 15.5 minutes of signal accumulation to be visualized. The signal was then zero-filled to 4 K and apodized with a 40-Hz gaussian-exponential function before Fourier transformation. The other conditions under which the spectra were acquired have been described previously.⁴

Blood samples were obtained every 30 minutes, deproteinized, and analyzed by ¹H NMR at 360 MHz (Bruker NMR spectrometer, Billerica, Mass.) to determine the degree of enrichment in the C-1 position of plasma glucose at 360 MHz, as described elsewhere.⁶

Hyperglycemic—Hyperinsulinemic Clamp Procedure

Hyperglycemia—hyperinsulinemia was induced with the insulin—glucose clamp technique.⁷ To inhibit endogenous insulin secretion, an infusion of somatostatin (0.1 μg per kilogram per minute) was initiated five minutes before the start of the glucose—insulin infusion in both the normal and diabetic subjects. At time zero, insulin (Humulin, Eli Lilly, Indianapolis) was administered in a priming and continuous infusion of 240 pmol per square meter of body-surface area per minute to raise the plasma insulin concentration acutely and maintain it at approximately 400 pmol per liter. At the same time a variable priming infusion of glucose was begun so that the plasma glucose concentration could be raised acutely and maintained 5 mmol above base line for 120 minutes. Throughout the study, the plasma glucose concentration was measured every five minutes and a variable infusion of a [1–¹³C]glucose solution (1.11 M, 20 percent ¹³C enrichment) was periodically adjusted to maintain the desired hyperglycemic plateau. Under these conditions —constant hyperglycemia and hyperinsulinemia — hepatic glucose production is completely suppressed,⁸ and all the infused glucose is taken up by tissue cells, except a minor amount excreted in the urine. The latter amount was quantitated, and the rate of excretion was subtracted from the rate of glucose infusion. Therefore, the mean glucose infusion rate, minus urinary glucose excretion, serves as a measure of the total amount of glucose metabolized.

Measurement of Respiratory Exchange

Continuous indirect calorimetry was performed during the hyperglycemic—hyperinsulinemic clamp studies at 40 to 60 minutes and at 100 to 120 minutes, as previously described.² The nonprotein respiratory quotient was obtained from the tables of Lusk, according to which the respiratory quotient for 100 percent oxidation of fat is 0.707 and that for the oxidation of carbohydrates is 1.00.⁹ The amount of glucose disposed of by means other than oxidation was calculated by subtracting the amount of glucose oxidized from the total amount of glucose infused (minus the small amount of glucose excreted in the urine).

Analytical Procedures

Plasma glucose was measured every five minutes (Beckman glucose analyzer, Fullerton, Calif.). The ¹³C enrichment of plasma glucose was determined every 15 minutes by gas chromatography—mass spectrometry of the pentacetate derivatives of plasma glucose after deproteinization and deionization as previously described.¹⁰ Plasma immunoreactive insulin was measured every 20 to 30 minutes as previously described.¹¹

Calculations

The basal muscle glycogen concentration in each subject was calculated by comparing the signal intensity for basal C-1 glycogen with the signal intensity obtained from a cast, individually shaped to match the subject's leg, filled with a 4 percent (wt/vol) solution of glycogen in 50 mM potassium chloride. At the time of both measurements, the accuracy of the spectrometer was checked by measuring against the formic acid standard, and minor corrections (<10 percent) were made. Any systematic deviation from this method of quantitation would underestimate the actual glycogen concentration, because the quantitation is based on the assumption that 100 percent of the leg volume observed by the surface coil is muscle tissue. The reproducibility of the measurement of basal glycogen concentration has been established from repeated measurements in normal subjects to be within 10 percent of the actual value.³ The increment in muscle glycogen concentration during each 15.5-minute interval was calculated from the equation presented below. The 15.5-minute increment of the C-1—glycogen peak intensity, (ΔGly), was divided by the intensity of the basal glycogen peak, (Gly₀). Multiplying this ratio by the basal glycogen concentration, [Gly₀], and the percentage of ¹³C that occurs naturally (1.1 percent) and dividing by the ¹³C plasma glucose enrichment above the natural abundance of the isotope, (PG), measured during that interval gave the increment in glycogen concentration, [ΔGly]: Each increment was then added to the previous concentration, and the slope calculated by linear regression analysis to yield the rate of glycogen synthesis.

All values are expressed as means ±SE. All statistical calculations were performed with a CLINFO computer system. The two study groups were compared by the unpaired two-tailed t-test.

Figure 1. Figure 1. Plasma Glucose Concentrations and ¹³C Plasma Glucose Enrichment Values in Normal Subjects (Open Circles) and Subjects with NIDDM (Solid Circles) before and during the Hyperglycemic—Hyperinsulinemic Clamp Study.

Values are means ±SE. APE denotes atom percent excess.

Figure 1 shows the changes in both the mean plasma glucose concentration and the mean ¹³C plasma glucose enrichment in both the normal and diabetic subjects. The mean initial glucose concentration (top panel) was 5.2±0.1 mmol per liter in the normal subjects and 5.9±0.4 mmol per liter in the subjects with NIDDM. After the start of the glucose infusion, the plasma glucose concentration increased rapidly and reached a plateau value of approximately 10.6 mmol per liter in both groups in 15 minutes; the mean concentration recorded from 15 to 120 minutes was 11.0±0.3 and 10.8±0.4 mmol per liter in the normal and diabetic subjects, respectively. The level of ¹³C plasma glucose enrichment (bottom panel) increased rapidly after the start of the [1–¹³C]glucose infusion and approached a semiplateau of about 18 percent atom percent excess (APE) by 80 minutes in both groups. To determine how much of the ¹³C isotope was in the C-1 position of plasma glucose, ¹H NMR was performed on plasma samples at minutes 30, 60, and 90 of the glucose infusion. The mean C-1 plasma glucose enrichment values (all subjects) were 13.4±0.4 APE at 30 minutes, 15.9±0.5 APE at 60 minutes, and 17.6±0.5 APE at 90 minutes. These values corresponded very well to those determined by gas chromatography—mass spectrometry at the same time points in the same subjects: 13.7±0.5 APE at 30 minutes, 16.3±0.5 APE at 60 minutes, and 17.4±0.5 APE at 90 minutes. This close correspondence implied that all the administered isotope remained in the C-1 position of glucose and was not scrambled to other positions. This finding is in agreement with our previous observation and was not unexpected, since the hyperglycemic—hyperinsulinemic milieu would suppress hepatic glucose production.^{6 , 8}

The mean initial plasma insulin concentrations were 36±6 pmol per liter in the normal subjects and 60±12 pmol per liter in the diabetic subjects. After the start of the combined infusion of glucose, insulin, and somatostatin, the plasma insulin concentration quickly rose and within 10 minutes reached a plateau value of 402±54 pmol per liter in the normal subjects and 372±30 pmol per liter in the diabetic subjects.

Figure 2 shows the course of the decoupled ¹³C NMR spectra of leg glycogen in a representative normal subject before the [1–¹³C]glucose infusion (bottom spectrum) and every 15.5 minutes thereafter. The C-1 peak of glycogen was clearly delineated in the base-line spectrum at 100.4 ppm; it routinely yielded a minimal signal-to-noise ratio of 25:1. The basal concentrations of glycogen were significantly lower in the diabetic than in the normal subjects (39±6 vs. 73±11 mmol per liter; P<0.01). The C-1 glycogen peak increased progressively in all subjects during the next 120 minutes of [1–¹³C]glucose infusion.

Figure 3. Figure 3. Incremental Changes from Base Line in the Muscle Glycogen Concentration in the Normal (Open Circles) and Diabetic Subjects (Solid Circles) during the Hyperglycemic—Hyperinsulinemic Clamp Study.

Values are means ±SE, expressed in terms of the wet weight of muscle tissue.

Table 1. Table 1. Glucose Metabolism and Muscle Glycogen Synthesis in Normal and Diabetic Subjects during Combined Hyperglycemia—Hyperinsulinemia.*

Figure 3 shows the incremental changes in the glycogen concentration over the base-line values. Despite almost identical concentrations of insulin and glucose in the two study groups, the mean rate of glycogen synthesis in the diabetic subjects was about 60 percent lower than that in the normal subjects. The mean rate of glycogen synthesis from 40 to 120 minutes was calculated to be 183±39 μmol-glucosyl units per kilogram of muscle (wet weight) per minute in the normal subjects and 78±28 μmol-glucosyl units per kilogram per minute in the diabetic subjects (P<0.05). During the same period the rate of total-body glucose uptake was 51±3 μmol per kilogram of body weight per minute in the normal subjects and 30±4 μmol in the diabetic subjects (P<0.005) (Table 1). The rate of oxidative glucose metabolism, determined by indirect calorimetry, was 9±1 μmol per kilogram of body weight per minute in the normal subjects and 8±2 μmol per kilogram per minute in the diabetic subjects (P not significant). Subtracting the rate of oxidative glucose metabolism from the rate of total-body glucose uptake in both groups yielded a rate of nonoxidative glucose metabolism of 42±4 μmol per kilogram per minute in the normal subjects and 22±4 μmol per kilogram per minute in the diabetic subjects (P<0.005).

There was a lag period before glycogen synthesis began in either group (Fig. 3). When the linear portion of each curve was extrapolated to the time (abscissa), the mean lag period was 12±5 minutes in the normal subjects and 35±6 minutes in the diabetic subjects (P<0.01).

Figure 4. Figure 4. Muscle Glycogen Synthesis in Relation to Nonoxidative Glucose Metabolism during the Hyperglycemic—Hyperinsulinemic Clamp Study.

Each normal subject is represented by an open circle, and each subject with NIDDM by a solid circle; the third and fourth solid circles along the abscissa represent the results of two studies in the same diabetic subject.

Figure 4 shows the relations between the rate of glycogen synthesis as determined by ¹³C NMR in both study groups during the hyperglycemic—hyperinsulinemic clamp study and the corresponding rate of nonoxidative glucose metabolism. The rate of glycogen synthesis correlated very well with the rate of nonoxidative glucose metabolism as determined by indirect calorimetry (r = 0.89, P<0.001).

After glucose ingestion, the maintenance of normal glucose homeostasis depends on three events that must occur in a coordinated fashion: the secretion of insulin by the pancreas, the suppression of hepatic glucose production, and the stimulation of glucose uptake by the liver and peripheral tissues, primarily muscles. However, the precise intracellular fate of the glucose that is taken up by muscle was uncertain when we started this study.¹

Previous studies using indirect calorimetry in combination with femoral-vein catheterization and the euglycemic—insulin clamp technique have shown that nonoxidative glucose metabolism is the major intracellular fate of glucose taken up by muscle in normal subjects.² The nonoxidative pathway becomes even more dominant during concurrent hyperglycemia and hyperinsulinemia.^{12 , 13} Nonoxidative glucose disposal includes the conversion of glucose to glycogen, or fat, as well as anaerobic glycolysis. Because the amount of lactate released from muscle in response to insulin is small,^{14 , 15} and because little glucose is believed to be converted to lipid,^{16 , 17} glycogen synthesis was assumed to be the major metabolic pathway responsible for glucose disposal in muscle.¹ However, direct confirmation of this assumption in vivo was lacking. Extrapolating the results of measurement of glycogen during repetitive biopsies of the quadriceps femoris muscle to the whole body, Hultman and coworkers^{18 19 20} concluded that over half of an infused glucose load in normal subjects was taken up by muscle and converted to glycogen. However, the experimental conditions employed by these investigators were quite unphysiologic, in that the mean blood glucose concentration reached a peak of 20 mmol per liter. More recently, Bogardus and his associates^{21 22 23 24} and others^{25 26 27} attempted to examine the role of glycogen formation in overall glucose disposal, using the euglycemic—insulin clamp technique. Although the plasma insulin concentration reached pharmacologic elevations (>6 nmol per liter), a consistent increase in muscle glycogen content could not be demonstrated. However, a positive correlation between glycogen synthase activity and nonoxidative glucose uptake was noted.^{23 , 25 , 26} On the basis of this observation, these workers concluded that glycogen formation represents the primary pathway of nonoxidative glucose disposal in humans. The failure to detect a significant increase in muscle glycogen concentration, even in the presence of marked hyperinsulinemia, is not surprising, in view of the measured rate of total-body glucose uptake. Even if all glucose infused was deposited in muscle as glycogen, the increment in muscle glycogen concentration (5 to 10 mmol per liter) would be at the limit of detectability of the assay for glycogen.

In this study we used ¹³C NMR spectroscopic techniques to obtain continuous quantitative information about the rate of muscle glycogen synthesis in vivo. Since the NMR studies were performed in combination with indirect calorimetry, the rate of whole-body nonoxidative glucose disposal could be compared with the rate of muscle glycogen formation. As shown in Figure 4, the rate of glycogen formation closely paralleled the rate of nonoxidative glucose disposal. The rate of synthesis of whole-body muscle glycogen in each subject was calculated by extrapolation from the measured rate of muscle glycogen synthesis, and the result was compared with the value for glucose metabolism. Estimates of the fraction of body mass that is muscle in men of the ages of those we studied range from as high as 38 percent, determined by dissecting cadavers,²⁸ to 26 percent of total body weight, determined by measuring neutron activation, thought to be the more accurate method.²⁹ When we used the latter value, we found that the rate of muscle glycogen synthesis accounted for approximately 90 percent of whole-body glucose metabolism and all of the nonoxidative glucose disposal in the normal subjects. These results provide direct evidence that glycogen synthesis represents the primary pathway for nonoxidative glucose disposal in normal subjects. Moreover, since nonoxidative glucose uptake accounted for most of total-body glucose metabolism under the conditions of our study, it follows that muscle glycogen synthesis was the major pathway of overall glucose metabolism. These data are consistent with previous observations of DeFronzo et al.² and Bjorntorp and Sjostrom,¹⁶ who suggested that the splanchnic bed and adipose tissue are minor sites of glucose disposal under similar conditions. However, it is possible that if glucose had been infused for longer periods, more glucose could have been converted into fat by induction of hepatic lipogenesis.

Our results provide direct experimental evidence that glycogen synthesis is in fact impaired in persons with NIDDM. The rate of glycogen formation during the linear period (from 40 to 120 minutes) in our diabetic subjects decreased by 60 percent as compared with the rate in our normal subjects. Assuming that 26 percent of the body of each of these subjects is muscle, glycogen formation accounted for approximately 90 percent of nonoxidative glucose disposal and approximately 70 percent of whole-body glucose metabolism (Table 1). It is therefore clear that impaired glycogen synthesis is the major intracellular metabolic defect responsible for the decrease in nonoxidative and whole-body glucose metabolism in subjects with NIDDM. It has been suggested that the muscle glycogen concentration itself is an important determinant of glycogen synthase activity.^{20 , 30 , 31} An increase in the intracellular glycogen content has been shown to inhibit glycogen synthase activity and the ability of insulin to promote glycogen synthesis in both humans and animals.^{20 , 30 , 31} However, this explanation cannot explain the defect in glycogen synthesis that we found, since the mean basal muscle glycogen concentration in the diabetic subjects was in fact lower than that in the normal subjects. A similar reduction in muscle glycogen concentrations in patients with NIDDM was reported by Roch-Norlund and colleagues.^{32 , 33}

In addition to decreased steady-state rates of muscle glycogen synthesis, there was a longer lag period before the start of glycogen synthesis in our diabetic subjects (35±6 minutes) than in our normal subjects (12±5 minutes). Our observation of a longer lag period in our diabetic subjects is consistent with that of a previous study,¹⁵ in which measurements obtained by catheterization of the femoral artery and vein showed that the ability of insulin to stimulate glucose uptake in the leg was markedly delayed in subjects with Type II diabetes in whom euglycemic—hyperinsulinemic conditions had been induced. Our results strongly suggest that the delay in glucose uptake in the leg in the previous studies was due to a defect in muscle glycogen synthesis. However, the present studies do not pinpoint the defect, which conceivably could occur anywhere between glucose delivery to muscle and glycogen synthase.

We conclude that under hyperglycemic—hyperinsulinemic conditions, muscle glycogen synthesis is the major pathway of glucose disposal in both normal and diabetic subjects. Furthermore, defective muscle glycogen synthesis has a predominant role in impairing glucose metabolism in NIDDM.