13 C and 31 P NMR Studies on the Effects of Increased Plasma Free Fatty Acids on Intramuscular Glucose Metabolism in the Awake Rat*

The effects of increased plasma free fatty acids (FFA) on insulin-dependent whole body glucose disposal, skeletal muscle glycolysis, glycogen synthesis, pyruvate ver- sus FFA/ketone oxidation, and glucose 6-phosphate (Glu-6-P) were investigated in the awake rat. A control group (glycerol-infused) and high plasma FFA group (Liposyn-infused) were clamped at euglycemia ( (cid:59) 6 m M )-hyperinsulinemia (10 milliunits/kg/min) throughout the experiment (180–240 min). In the initial experiment, 13 C NMR was used to observe [1- 13 C]glucose incorporation into [1- 13 C]glycogen in the rat hindlimb for glycogen synthesis calculations and into [3- 13 C]lactate and [3- 13 C]alanine for glycolytic flux calculations. These experiments were followed by 31 P NMR measurements of Glu-6-P changes under identical conditions of the initial experiment. Plasma FFA concentrations were 2.25 (cid:54) 0.36 and 0.20 (cid:54) 0.03 m M in the high plasma FFA and control groups respectively ( p < 0.0005). Glucose infu- sion rates (G inf ) decreased significantly in the Liposyn- infused rats (29.5 (cid:54) 0.7 and 27.2 (cid:54) 1.2 mg/kg/min for control and high plasma FFA group, respectively, at 15 min to 30.7 (cid:54) 2.3 and 17.7 (cid:54) 1.3 mg/kg/min, respectively, Clamps— Euglycemic-hyperinsuline-mic (10 milliunits/kg/min, Humulin Regular, Lilly) clamps were per- formed using [1- 13 C]glucose (99% enriched, 20% w/v, Cambridge Isotope MA) infused simultaneously with glycerol (0.425 mg/min) or Liposyn II (1:3 v/v saline at 39 (cid:109) l/min) plus heparin (0.0975 IU/min) at 2.5 min after the commencement of the primed continuous insulin infusion. Plasma glucose concentrations were clamped (cid:59) m M . All clamps lasted for 180–240 min. High plasma FFA group clamps were generally 240 min, because of the longer period required for label turnover in lactate and alanine. Blood samples were drawn during base-line NMR measurement, at 7.5 and 15 min and every 15 min thereafter for immediate assessment of plasma glucose and lactate concentrations.

Increased levels of plasma free fatty acids (FFA) 1 are prevalent in people with non-insulin-dependent diabetes mellitus and may play an important role in mediating the insulin resistance associated with this disease. A decrease in insulin-dependent glucose uptake by increased FFA was originally observed in vitro in heart and diaphragm muscle by Randle et al. (1). This phenomenon was postulated to be a result of pyruvate dehydrogenase (PDH) enzyme allosteric control by increased intramitochondrial acetyl-CoA/CoA and NADH/NAD ϩ ratios that activates pyruvate dehydrogenase kinase enzyme which subsequently deactivates the PDH enzyme complex (2). Intramitochondrial citrate also increases under conditions of increased FFA oxidation and decreased flux through isocitrate dehydrogenase. Citrate is a potent allosteric effector of phosphofructokinase (3) and therefore inhibits this step which will increase glucose 6-phosphate (Glu-6-P). Increased Glu-6-P will allosterically inhibit hexokinase (4). Therefore, increased intracellular glucose will diminish the glucose gradient, and glucose transport into the muscle cell will also be inhibited. This metabolic phenomenon, also known as the glucose-free fatty acid cycle, has more recently come into question in skeletal muscle as conflicting results have been presented.
Previous studies in rat have shown that an increase in plasma FFA concentration caused an increase (5), decrease (6 -8), or no change (9,10) in skeletal muscle or whole body glucose uptake. Discrepancies within these previous findings may in large be due to the vastly different techniques used to elucidate skeletal muscle and/or whole body glucose uptake.
Most in vivo studies (6,(11)(12)(13)(14)(15) that have examined the effects of increased plasma FFA on whole body glucose uptake have relied on indirect calorimetry or measurements of plasma tritium derived from [3-3 H]glucose to provide indirect measurements of glycolysis and fat oxidation (indirect calorimetry) and reflect average whole body glycolytic and FFA oxidation measurements. Due to the quantitatively important contribution of other organs to whole body glycolysis (brain) and whole body FFA oxidation (liver and heart), rates of glycolysis and FFA oxidation must be measured directly in skeletal muscle to accurately assess the effects of FFA on these processes in this tissue.
To address these issues, we have developed a method using in vivo 13 C NMR to directly measure rates of glycolysis and glycogen synthesis simultaneously in skeletal muscle of awake rats, and additional measurements in skeletal muscle tissue extracts were used to calculate relative rates of pyruvate versus FFA/ketone oxidation. These measurements were combined with 31 P NMR temporal measurements of ⌬[Glu-6-P] to assess the mechanism for decreased insulin-dependent glucose uptake in rat muscle under conditions of euglycemia-hyperinsulinemia and increased plasma FFA.

EXPERIMENTAL PROCEDURES
Animals-Harlan Sprague Dawley rats (Charles River, Raleigh, NC) were maintained on standard rat chow (Ralston Purina Co., St. Louis, MO) and housed in an environmentally controlled room with a 12-h light/dark cycle. At weights between 250 and 300 g, rats were chronically catheterized in the right jugular vein and carotid artery. The catheters were externalized through an incision in the skin flap behind its head (16). The rats were allowed to recuperate after surgery for the indwelling catheters until they were at least of pre-operative weight (approximately 5-10 days). All rats were fasted 24 h before the infusion experiment. Rats (250 -350 g) were placed in a customized restraining tube that allowed their left hindlimb to be secured to the outside of the tube in a manner to limit free movement of the leg for NMR measurements. The rats were transiently anesthetized (Ͻ30 s) with a low dose (2.5 mg) of thiopental (Sigma) to place them in the restraining tube. Two groups were studied. The control group (n ϭ 15) was infused with glycerol to eliminate any contribution of glycerol from triglycerides to metabolic changes in the muscle, and the high plasma FFA group (n ϭ 14) was infused with Liposyn II (1:3 v/v with saline, Abbott), a 20% triglyceride emulsion combined with heparin (continuous infusion, 0.0975 IU/min). Heparin was used to activate lipoprotein lipase and thereby catalyze the hydrolysis of triglycerides. At the end of the in vivo NMR experiment, rats were anesthetized with thiopental (50 mg/kg). Superficial skin was rapidly removed from the left hindquarter followed by in situ freeze clamping of the gastrocnemius and quadriceps muscles. Rats were euthanized with a lethal dose of thiopental.
Euglycemic-Hyperinsulinemic Clamps-Euglycemic-hyperinsulinemic (10 milliunits/kg/min, Humulin Regular, Lilly) clamps were performed using [1-13 C]glucose (99% enriched, 20% w/v, Cambridge Isotope Laboratories, Cambridge, MA) infused simultaneously with glycerol (0.425 mg/min) or Liposyn II (1:3 v/v saline at 39 l/min) plus heparin (0.0975 IU/min) at 2.5 min after the commencement of the primed continuous insulin infusion. Plasma glucose concentrations were clamped ϳ6 mM. All clamps lasted for 180 -240 min. High plasma FFA group clamps were generally 240 min, because of the longer period required for label turnover in lactate and alanine. Blood samples were drawn during base-line NMR measurement, at 7.5 and 15 min and every 15 min thereafter for immediate assessment of plasma glucose and lactate concentrations.
In Vivo NMR Spectroscopy-All in vivo NMR experiments were performed on a Bruker Biospec 7.0 tesla system (horizontal 22-cm diameter bore magnet). In the initial experiment (control: n ϭ 8, high plasma FFA: n ϭ 9), 13 C observe/ 1 H decouple NMR was performed using dual concentric surface coils. The outer 1 H coil (30 mm) was tuned to 300.68 MHz, and the inner 13 C coil (18 mm) was tuned to 75.65 MHz. The radio frequency isolation between the two coils was 43 db. The rat hindlimb was positioned over the 13 C coil and placed in magnet isocenter.
Global 1 H shimming was followed by localized shimming with a STEAM sequence (17) over a 1 ϫ 2 ϫ 2-cm volume of the leg to optimize magnetic field homogeneity over the volume observed by the 13 C surface coil. Water line widths of 35-45 Hz were obtained. 1 H-decoupled 13 C NMR spectroscopy was performed in the following manner. An initial frequency-selective sinc pulse (20 ms) set on the low field side of the methylene carbon of lipids at 30 ppm was immediately followed by a nonselective hard pulse (approximately 70°flip angle, 5 mm from surface coil). The sinc pulse power was adjusted to eliminate most of the signal in that region. Broadband 1 H Waltz-16 decoupling was applied during acquisition, and additional nuclear Overhauser effect was achieved using low power decoupling (0.4 watts) during the relaxation delay (TR ϭ 0.5 s, SW ϭ 20 KHz, 4K data). A 15-min base-line spectrum was followed by subsequent 15-min acquisitions throughout the duration of the experiment. All data were processed using a Gaussian filter followed by Fourier transformation. All spectra were base line-subtracted prior to peak analysis, and the (␣) [1-13 C]glucose peak was set to 93.0 ppm. The C-1 glycogen peak (100.5 ppm) was integrated over the downfield half and multiplied by 2 to limit overlapping contribution from ␤-glucose (96.8 ppm). C-3 lactate (21.0 ppm) and C-3 alanine (16.9 ppm) were clearly resolved for peak integration. The signal to noise ratio (S/N) of the C-3 lactate and C-3 alanine peaks in the final spectrum as determined by peak to peak noise amplitude were 11.0 Ϯ 3.6 and 14.8 Ϯ 2.8, respectively.
In the experiments using 31 P NMR (control: n ϭ 7, high plasma FFA: n ϭ 5), the 31 P observe 1 H shim coils were of the same design and geometry as the 13 C/ 1 H coil assembly. Shimming was followed in the same manner as described above. Direct 1 H-decoupled 31 P NMR was performed at 121.72 MHz. A hard pulse (45°flip angle) was optimized 5 mm from the surface coil (TR ϭ 1.4 s, SW ϭ 5 KHz, 4K data). Broadband 1 H Waltz-16 decoupling was applied during acquisition, and frequency centered on 6-H 2 of Glu-6-P. Data were accumulated into 15-min acquisitions as was done with the carbon data and processed with Gaussian filtering and Fourier transformation. In two experiments, fully relaxed base-line spectra (TR ϭ 30 s, NS ϭ 64) were obtained for saturation correction factors. The phosphocreatine peak was set to 0 ppm and base line manually corrected. A standard solution containing phosphocreatine, P i , ␣-glycerol phosphate, IMP, and Glu-6-P at pH ϭ 7 was run to determine peak chemical shift assignments. ␤-ATP was integrated and used as the internal concentration standard (7.2 mM) (18). All spectra were integrated over a frequency window of 7.13-7.36 ppm which corresponds to the downfield side of Glu-6-P and then multiplied by 2. This was done to minimize error in the measurement resulting from ␣-glycerol phosphate peak overlap at 6.92 ppm (19). The S/N for the basal Glu-6-P peak was determined to be 4.3 Ϯ 0.2.
Tissue Extract Analysis-Muscle tissue extracts were prepared for high field NMR analysis by homogenizing approximately 1 g of combined quadriceps and gastrocnemius muscle with a variable high speed electric homogenizer after sample was placed in a vortex tube filled with 0.9% perchloric acid (3 v/w) and 100 l of 1 N sodium formate. The homogenization was performed over ice to keep the sample cold. After homogenization, the sample was centrifuged at 4°C for 10 min (3400 rpm). The supernatant was extracted and the pellet saved for glycogen enrichment measurements. Minimal additional [1-13 C]glycogen was detected in the supernatant as determined by high field 13 C NMR but was not different between groups. KOH (4 N, 0.675 v/w) was added to the supernatant to precipitate excess chlorate ions. The sample was centrifuged once more at 4°C for 15 min (3400 rpm). The supernatant was extracted, and 500 mM phosphate buffer, pH ϭ 7, was added to neutralize the sample. The sample was dried in a speed-vac (Savant, Farmingdale, NY) overnight, and 1 ml of D 2 O was added to the dried powder before placing it in a 5-mm NMR tube for NMR analysis at 8.4 tesla (Bruker WB-360 NMR spectrometer). Proton observed-carbon enhanced spectroscopy was performed on tissue extract samples for fractional enrichment calculations (20). The broadband 13 C inversion pulse frequency was placed between C-4 glutamate and C-3 alanine (ϳ26.2 ppm) and turned on during alternate transients with raw data separated into two data sets providing spectra with heteronuclear coupled spins inverted (spectrum B) and non-inverted (spectrum A). The fractional enrichments (APE) of glutamate, lactate, and alanine were calculated from their respective resonances in spectrum A and B as follows in Equation 1: Spectra were acquired with TR ϭ 6 s, NS ϭ 512, 16K data, and broadband carbon decoupling. The 4-H 2 glutamate triplet that appears at 2.33 ppm has overlapping signal contribution from malate and ␤-hydroxybutyrate. Malate should be similarly enriched at steady state and is low in concentration, and negligible ␤-hydroxybutyrate was detected at 1.19 ppm (4-H 3 ). Therefore, these co-resonating signals are expected to have little effect on the glutamate enrichment calculation. For quantitation, a correction factor was calculated when a TR ϭ 19 s was used. Glutamate, glutamine, and alanine were quantitated by comparing signal intensity with a known internal concentration standard (sodium formate) that was added during the extraction procedure. High pressure liquid chromatography was used to validate amino acid concentrations measured by NMR (mean 3.8% error). Intramuscular lactate concentration was calculated by extrapolation of in vivo NMR data (C-3 lactate and C-3 alanine) after correcting for T 1 differences. Analytical Procedures-Plasma glucose concentrations were measured by the glucose oxidase method (Glucose Analyzer II; Beckman Instruments, Fullerton, CA). Plasma immunoreactive free insulin was measured with a double antibody radioimmunoassay technique (Linco Research Inc., St. Charles, MO) (21). 13 C enrichment of plasma glucose and alanine was determined by gas chromatography-mass spectroscopy using a Hewlett-Packard 5890 gas chromatograph (HP-1 capillary column, 12 m ϫ 0.2 mm ϫ 0.33 mm film thickness) interfaced to a Hewlett-Packard 5971 A mass selective detector operating in the positive chemical ionization mode with methane as a reagent gas.
100-ml aliquots of plasma from each time point were deproteinized with equal amounts of Ba(OH) 2 and ZnS0 4 . The samples were centrifuged, and the supernatant was run over a cation exchange chromatography column (AG50W-X8 100 -200 mesh Bio-Rad). Both amino acid and glucose fractions were dried using a speed-vac. Glucose was derivatized for gas chromatography-mass spectroscopy analysis as described previously (22), and amino acids were derivatized similar to Leimer et al. (23).
Selective ion analysis was as described previously (22). In addition, ions with m/z ϭ 242, 243, and 244 were detected for alanine. Plasma lactate fractional enrichments were measured by 1 H NMR.
Glycogen 13 C fractional enrichments were determined using the precipitated glycogen from the initial muscle perchloric acid extraction. The precipitate was placed in microdialysis tubing (6 -8000 molecular weight cut-off, Spectra/Por, Spectrum, Houston, TX) and placed in 5 liters of deionized water. The glycogen pellet was dialyzed for approximately 24 h before it was digested to free glucose with amyloglucosidase in phthalate buffer (50 mM, pH ϭ 4.5). The sample was centrifuged, and supernatant was processed as described above for [ 13 C]glucose fractional enrichment measurement. Absolute glycogen concentrations were measured on a separate portion of muscle as described previously (24).
Plasma free fatty acids were determined using an acyl-CoA oxidasebased colorimetric kit (WAKO NEFA-C, WAKO Pure Chemical Industries, Osaka, Japan). Plasma lactate concentrations were measured by a 2300 STAT PLUS lactate analyzer (Yellow Springs Instrument Co., Yellow Springs, OH).
Whole Body Glucose Disposal Calculation-Whole body glucose disposal rates (mg/kg/min) were calculated as the sum of the glucose infusion rate (variable) and hepatic glucose production rate (HGP) at 180 min when the plasma glucose enrichment was at steady state. HGP was calculated as described before (25), and using plasma [ 13 C]glucose APE measurements as a marker of hepatic glucose production.
Glycogen Synthesis Rate Calculation-The incremental change in C-1 glycogen peak intensity from [1-13 C]glucose incorporation was measured at 100.5 ppm. Incremental plasma glucose 13 C fractional enrichment as well as final glycogen 13 C enrichment and concentrations were used to back-extrapolate the glycogen concentration (mol/g which represents mol of glucosyl units/g muscle wet weight) at each measured time point to base line as described by Bloch et al. (26). Rates may be underestimated (Ͻ1%) due to minimal loss of 13 C-labeled glycogen in the tissue extraction procedure as discussed above. Glycogen synthesis rates were determined using a linear regression analysis over the individual time point glycogen concentrations.
Glycolytic Flux (V Gly ) Calculations-Metabolic steady state conditions were assumed to create a model for calculating carbon flux through the glycolytic pathway into the intermediate triose pool of lactate, pyruvate, and alanine. Differential equations were developed from steady state rate equations (27). These equations may be solved for defining V Gly . Equation 2 was used to calculate V Gly : where [Lac] ϭ muscle lactate concentration (nmol/g tissue), [Pyr] ϭ muscle pyruvate concentration (nmol/g tissue), [Ala] ϭ muscle alanine concentration (nmol/g tissue), [Lac * ] ϭ 3-13 C-labeled muscle lactate concentration (nmol/g tissue), [Pyr * ] ϭ 3-13 C-labeled muscle pyruvate concentration (nmol/g tissue), [Ala * ] ϭ 3-13 C-labeled muscle alanine concentration (nmol/g tissue), a ϭ label dilution factor resulting from unlabeled substrate contribution via glycogenolysis and pentose phosphate pathway, Glc * /Glc ϭ [1-13 C]glucose fractional enrichment, and V Gly ϭ glycolytic flux (nmol/g tissue/min). In this experiment, 13 C label incorporation from glucose into lactate and alanine in the hindlimb muscles was observed by 13 C NMR as an indirect marker of pyruvate labeling that was not detectable by 13 C NMR in vivo due to its low concentration (10). [3-13 C]Lactate and [3-13 C]alanine NMR signals were determined to be predominantly intramuscular as concluded from 13 C NMR measurements of the muscle extracts when assuming the [1-13 C]glucose NMR signal was entirely of extracellular origin. Label incorporation into lactate and alanine is a qualitative indicator of glycolytic flux (V Gly ), which under steady state metabolic conditions and negligible anaerobic glycolysis equals the flux leaving the [alanine ϩ pyruvate ϩ lactate] pool through pyruvate dehydrogenase (V PDH ) and pyruvate carboxylase (V pc ) (Fig. 1). Label turnover in lactate and alanine was measured until there was no longer an increase in NMR signal from these metabolites. The label turnover curve was normalized to the glucose fractional enrichment (Glc * /Glc) which was not stable through the initial 60 min of the experiment. Therefore, at each time point the combined lactate, pyruvate, and alanine pool fractional enrichment [Lac * ϩ Pyr * ϩ Ala * ]/[Lac ϩ Pyr ϩ Ala] was divided by the plasma glucose fractional enrichment and a dilution factor (a) which accounts for unlabeled glucose oxidation via glycogenolysis and pentose phosphate pathway where 1-13 C label is lost. The label turnover rate constants were calculated for alanine and lactate separately using a single exponential least squares fit analysis (Simfit, W. G. Bardsley, University of Manchester, UK), and the turnover rates were determined to be indistinguishable, consistent with alanine and lactate being in rapid exchange with intracellular pyruvate. The lactate and alanine fractional enrichments at the end of the experiment were equal. Therefore, a single metabolic compartment model was used, and it was assumed that cytosolic Pyr * ϭ mitochondrial Pyr * . Basal intramuscular lactate and alanine concentrations were measured in muscle tissue extracts in a separate group of rats to compare with the metabolite concentrations at the end of the euglycemic-hyperinsulinemic clamp experiment. The concentrations were not significantly different, so the pool concentration ([Ala ϩ Pyr ϩ Lac]) was assumed to be constant (steady state) throughout the experiment. Lactate and alanine concentrations were quantitated by NMR, and an intracellular pyruvate concentration of 0.15 mol/g (10) was used for the calculation of V Gly . Plasma alanine and lactate fractional enrichments were also measured to assess tissue alanine and lactate exchange with blood. They were determined to follow a similar label FIG. 1. Schematic of skeletal muscle metabolite labeling following [1-13 C]glucose precursor infusion. 13 C label from glucose becomes incorporated into [1-13 C]glycogen and [3-13 C]pyruvate which can be reduced to [3-13 C]lactate or converted to [3-13 C]alanine via aminotransferase reaction. The rate at which substrate enters the [lactate ϩ pyruvate ϩ alanine] pool is 2 ϫ the glycolytic rate (2V Gly ). Under steady state metabolic conditions 2V Gly is equal to pyruvate dehydrogenase (V PDH ) ϩ pyruvate carboxylase (V pc ) flux. Subsequent labeling of [4-13 C]glutamate/-glutamine occurs from label entering the tricarboxylic acid (TCA) cycle via pyruvate dehydrogenase. Label entering the tricarboxylic acid cycle via pyruvate carboxylase will label C-3 oxaloacetate (OAA). V ppp ϭ pentose phosphate pathway flux (stippled due to loss of label). turnover as the intramuscular lactate and alanine label turnover, so label dilution from blood was excluded from Equation 2 for the calculation of V Gly . Due to the large contribution of skeletal muscle to total body weight and high proportion of insulin-stimulated GLUT4 transporters, skeletal muscle has been measured to account for the majority (ϳ88%) of insulin-stimulated whole body glucose uptake in man (28). Therefore, although the cell membrane transport mechanisms for lactate (29) and alanine (30) have been characterized and influx and efflux activities are high for both substrates, the plasma-labeled lactate and alanine turnover under euglycemic-hyperinsulinemic conditions should be dictated by the skeletal muscle-labeled lactate and alanine turnover rather than by label turnover of these pools in smaller organs (e.g. heart) that have higher glycolytic activity than skeletal muscle.
Statistical Analysis-All data are reported as the mean Ϯ S.E. Student's two-tailed t test was performed on data to determine significance at a minimum p Յ 0.05 threshold.

RESULTS
Euglycemic-hyperinsulinemic Clamp-In both control and high plasma FFA groups, the plasma glucose concentrations were clamped at approximately 6 mM, and there was no significant difference between the groups (Fig. 2A). Plasma insulin concentrations in both groups started at 102 Ϯ 20 and 130 Ϯ 38 pM and increased to a level of 1230 Ϯ 174 and 1356 Ϯ 84 pM in the control and high plasma FFA groups, respectively, at 60 min and were clamped throughout the experiment (Fig. 2B).
Basal plasma free fatty acid concentrations were 0.91 Ϯ 0.11 and 0.96 Ϯ 0.19 mM in the control and high plasma FFA group, respectively (Fig. 2C). At the first measurement (60 min), the plasma FFA concentrations were 0.28 Ϯ 0.03 and 2.25 Ϯ 0.29 mM in the control and high plasma FFA group (p Ͻ 0.005) and remained there for the duration of the study.
Plasma lactate concentrations increased rapidly in the control group from 0.86 Ϯ 0.15 mM at base line to 1.42 Ϯ 0.22 mM at 15 min (Fig. 2D) but less rapidly in the high plasma FFA group (0.92 Ϯ 0.07 mM at base line to 1.45 Ϯ 0.24 at 90 min).
Glucose infusion rates (G inf ) were 29.5 Ϯ 0.7 and 27.2 Ϯ 1.2 mg/kg/min at 15 min for control and high plasma FFA groups, respectively. While the G inf in the high plasma FFA group continued to decrease throughout the experiment to 17.7 Ϯ 1.3 mg/kg/min at 240 min (Fig. 2E), it remained relatively stable in the control group throughout the experiment (30.7 Ϯ 2.3 mg/ kg/min at 180 min, p Ͻ 0.002). The whole body glucose disposal is defined as the glucose infusion rate plus HGP. The HGP rate has been shown to increase when glycerol was infused simultaneously with a euglycemic-hyperinsulinemic clamp over controls (31). The glycerol infusion rate chosen in our control group was set to match the plasma glycerol concentration produced by lipolysis during the triglyceride infusion, and no significant differences in estimated hepatic glucose production rates between the control and high plasma FFA groups were calculated (7.3 Ϯ 0.7 and 7.6 Ϯ 0.6 mg/kg/min, respectively) at 180 min when the plasma glucose enrichment was at steady state. At 240 min, the whole body glucose disposal rate in the high plasma FFA group was 63% of the control group at 180 min (23.9 Ϯ 1.8 versus 37.9 Ϯ 2.8 mg/kg/min, p Ͻ 0.002).
Glycogen Synthesis Rates-Twenty four-hour fasted basal muscle glycogen concentrations were 23.7 Ϯ 2.5 and 23.8 Ϯ 2.0 mol/g for control and high plasma FFA groups, respectively. Fig. 3 illustrates a typical time course for the relative in vivo net glycogen production in a series of 15-min 13 C NMR collected spectra. At base line, no basal glycogen was detected by NMR due to lack of sensitivity with only 1.1% 13 C natural abundance. Once the [1-13 C]glucose infusion was begun, the ␤and ␣-anomer of glucose was observed at 96.8 and 93.0 ppm, respectively. [1-13 C]Glycogen appeared by 15 min at 100.5 ppm and continued to increase at a constant rate throughout the experiment in both groups yielding a net synthesis rate of 163 Ϯ 32 and 104 Ϯ 17 nmol/g/min for control and high plasma FFA groups, respectively (Fig. 4).
Glycolytic Flux (V Gly ) Calculations-A series of base linesubtracted 13  nine turnover are illustrated in Fig. 5. Peaks appearing at 16.9 and 21.0 ppm are indicative of C-3 alanine and C-3 lactate, respectively. The peaks were integrated and normalized for plasma glucose fractional enrichments before fitting the label turnover data (Fig. 6A). Line fitting of the data was done with R 2 Ն 0.90. There was a distinct difference in rate of turnover with the high plasma FFA group being much slower than the control group (1.44 Ϯ 0.25 ϫ 10 Ϫ2 versus 3.35 Ϯ 0.30 ϫ 10 Ϫ2 min Ϫ1 , p Ͻ 0.0005). Therefore, an additional 60 min experimental time was required for NMR detection of lactate and alanine label turnover in the high plasma FFA group. The lactate and alanine concentrations were measured from muscle tissue extracts, and the concentrations were determined to be 1.51 Ϯ 0.09 and 1.88 Ϯ 0.12 mol/g, respectively, for the control group and 1.27 Ϯ 0.16 and 1.67 Ϯ 0.21 mol/g for the high plasma FFA group. The value of (a) used in Equation 2 that corresponds to the label dilution factor resulting from unlabeled substrate contribution via glycogenolysis and pentose phosphate pathway was 0.68 Ϯ 0.03 and 0.60 Ϯ 0.02 in the control and high plasma FFA groups, respectively (p Ͻ 0.02). V Gly (Fig. 6B) was significantly lower in the high plasma FFA group (19.5 Ϯ 3.6 and 57.9 Ϯ 8.0 nmol/g/min in the high plasma FFA and control groups respectively, p Ͻ 0.002).
Steady State Isotopic Enrichment-Label enrichments were calculated for lactate, alanine, and glutamate in tissue extracts. Table I  and 23.2 Ϯ 0.9 and 21.8 Ϯ 1.3% for high plasma FFA groups, respectively (p Ͻ 0.005). Assuming one pool of pyruvate labels both alanine and lactate, then the pyruvate enrichment must equal that of C-3 lactate and C-3 alanine at isotopic steady state. The C-4 glutamate fractional enrichment was 15.0 Ϯ 2.4 and 5.6 Ϯ 1.0% for the control and high plasma FFA groups, respectively (p Ͻ 0.005). The ratio of C-4 glutamate, which becomes labeled when label from C-3 pyruvate enters the tricarboxylic acid cycle via PDH (Fig. 1), to C-3 lactate reflects the percent of flux entry via PDH versus FFA and ketone oxidation under steady state metabolic conditions. We assume a single compartment model and cell type for this analysis. Therefore, the pyruvate pool in exchange with lactate and alanine will label glutamate in the mitochondria. C-4 Glu/C-3 Lac was 56.9 Ϯ 8.7 and 25.4 Ϯ 4.0% for control and high plasma FFA groups, respectively, reflecting an approximate 2-fold increase in the relative rate of intracellular FFA/ketone oxidation in the high plasma FFA versus control group.
Glucose 6-Phosphate-31 P NMR was used with the same experimental conditions as the 13 C NMR experiments. Fig. 7 contains a 15-min acquired base-line phosphorus spectrum processed with Gaussian filtering. Adenosine triphosphate, phosphocreatine, inorganic phosphate, and phosphomonoester peaks are visible. The peaks appearing at 7.13 and 6.92 ppm correspond to glucose 6-phosphate (18,19) and ␣-glycerol phosphate, respectively. The temporal ⌬[Glu-6-P] measurements for both control and high plasma groups are shown in Fig. 8. Basal [Glu-6-P] was 0.19 Ϯ 0.03 and 0.17 Ϯ 0.02 mol/g in control and high plasma FFA groups, respectively. A rapid increase in Glu-6-P in the control group of 0.04 Ϯ 0.01 mol/g occurred during the initial 30 min of the experiment. This was followed by a moderate additional increase of 0.02 mol/g before slightly decreasing at 150 -180 min. In the high plasma FFA group, the ⌬Glu-6-P concentration continued to increase throughout the experiment to 0.10 Ϯ 0.01 mol/g at 210 -240 min. Glu-6-P concentrations increased by 29.8 Ϯ 7.0 and 52.8 Ϯ 12.3% (p Ͻ 0.05) in the control and high plasma FFA groups, respectively, by 180 min. DISCUSSION Previous studies in rats have shown that an increase in plasma FFA concentration resulted in an increase (5), decrease (6 -8), or no change (9,10) in muscle glucose uptake. In our studies, whole body glucose disposal rates decreased from approximately 60 min on throughout the experiment in the high plasma FFA group (Fig. 2E). It is difficult to assess changes in skeletal muscle glucose uptake per se from changes in whole body glucose metabolism since the brain (32) accounts for a significant proportion of whole body glucose uptake under rest-  [1-13 C]glucose enrichment before data were fit to a single exponential curve to determine turnover rate constants (3.35 Ϯ 0.30 ϫ 10 Ϫ2 and 1.44 Ϯ 0.25 ϫ 10 Ϫ2 min Ϫ1 in control and high plasma FFA group respectively, p Ͻ 0.0005). B, skeletal muscle V Gly . Rates were 57.9 Ϯ 8.0 and 19.5 Ϯ 3.6 nmol/g/min in control (shaded) and high plasma FFA (black) group, respectively. *, p Ͻ 0.002, control versus high plasma FFA group.
ing conditions. Measurements of whole body glycolysis also suffer from the inability to reflect tissue-specific glycolytic measurements. Currently, most of these measurements are determined by indirect calorimetry (11,12,33,34) or measurements of tritiated water production from [3-3 H]glucose (13,25,35). The significant advantage of the NMR technique for glycolytic flux measurements is that it provides a direct assessment of intramuscular oxidative glycolytic metabolism.
Using this technique, we found the glycolytic flux decreased (67%) in the high plasma FFA versus control group (19.5 Ϯ 3.6 and 57.9 Ϯ 8.0 nmol/g/min, respectively). Kim et al. (6) estimated glycolysis in rat skeletal muscles as a difference between glucose uptake measured by 2-deoxyglucose uptake and glycogen synthesis. Their results also showed a significant reduction in glycolysis (41-50%) in Intralipid-infused versus control rats, although their absolute glycolytic fluxes were higher ranging from 67 to 316 nmol/g/min in various muscle types of the rat hindlimb at euglycemia and insulinemia similar to that in our experiments. Effects of oleate on rates of glycolysis in isolated soleus muscle following incubation with [ 14 C]glucose showed a similar trend (ϳ50% decrease) but in general were much lower (decreasing from ϳ12 to 7 nmol/g/min) (7) which may be attrib-uted to differences between the in vitro and in vivo preparations.
Pyruvate dehydrogenase flux (V PDH ) was estimated to be Ϸ2V Gly -V pc under steady state metabolic conditions assuming that net lactate efflux from muscle due to anaerobic glycolysis was negligible. Anaplerotic substrate contribution to the tricarboxylic acid cycle was determined to be minor with respect to citrate synthase flux under similar conditions in the heart (36). Therefore assuming negligible anaplerosis in muscle, our mean PDH flux was approximately 116 nmol/g/min in the control group which decreased by 66% in the high plasma FFA group. These data are in relative agreement with a 52% decrease in PDH activity measured in vitro by Kruszynska et al. (37) in rat gastrocnemius muscle after a euglycemic-hyperinsulinemic/ high plasma FFA clamp. PDH rates were 58 versus 28 nmol/g/ min in FFA versus control, respectively, as calculated from their 24-h fasted PDH Total and % PDH active data.
Furthermore when using PDH flux data in combination with steady state isotopic enrichments of lactate and glutamate in muscle extracts, one may calculate the combined FFA and ketone oxidation rates. We estimate that the rate of intramuscular FFA/ketone oxidation almost doubled in the high plasma FIG. 7. In vivo 31 P NMR spectra. 15 min acquired spectrum processed with Gaussian filtering. Peaks present are phosphocreatine set to 0 ppm, adenosine triphosphate (␥-ATP, ␣-ATP, ␤-ATP), inorganic phosphate (P i ), and phosphomonoester region. The expanded phosphomonoester region include peaks corresponding to glucose 6-phosphate (G-6-P) at 7.13 ppm, and ␣-glycerol phosphate at 6.92 ppm. FFA versus control group, and substrate entry via glycolysis into the tricarboxylic acid cycle accounted for 57% in controls and decreased to 25% (p Ͻ 0.005) in the high plasma FFA group.
Our in vivo NMR measurements were obtained on the rat hindlimb which is comprised of both type I and type II muscle fibers. Therefore, the calculated glycolytic flux, glycogen synthesis rate, and Glu-6-P concentration represent a weighted average of these respective measurements in the individual muscles. Although localized NMR spectroscopy of specific muscles is possible, the signal sensitivity is such that these measurements would be extremely difficult in the rat. The V Gly calculations were based upon steady state metabolic kinetic modeling as was used by Mason et al. (27) to measure tricarboxylic acid and glycolytic flux in brain. To perform these calculations, we assumed that the intramuscular lactate, pyruvate, and alanine concentrations did not change throughout the duration of the experiment and that glucose entering the [lactate ϩ pyruvate ϩ alanine] pool through glycolysis was at a steady state rate. In support of this assumption, we measured intramuscular [lactate ϩ alanine] before and after the euglycemic-hyperinsulinemic clamp and found them to be similar (3.17 Ϯ 0.49 versus 3.39 Ϯ 0.25 mol/g). Since steady state metabolic conditions were assumed to simplify the glycolytic flux calculation, V Gly was possibly overestimated in the high plasma FFA group, because the intramuscular glucose uptake as reflected by the glucose infusion rate was decreasing throughout the experiment. This would result in a slight overestimation in V Gly calculated from fitting the label turnover of lactate and alanine with emphasis on the earlier time points when glycolysis may be greater before the whole body glucose disposal initially begins to decrease. Nevertheless, Fig. 6A depicts a significant difference in lactate and alanine label turnover illustrating the ability of increased plasma FFA to inhibit V Gly in skeletal muscle.
Muscle glycogen synthesis rates tended to be lower, although not significantly, in the high plasma FFA versus control groups (104 Ϯ 17 and 163 Ϯ 32 nmol/g/min, respectively). There was no significant change in these rates during the time course of the experiment (Fig. 4). Therefore, the predominant mechanism by which increased concentrations of plasma FFA decrease muscle glucose uptake in rat is through inhibition of glycolysis. Li et al. (38) also showed no difference in 14 C label incorporation into glycogen in rat soleus or extensor digitorum longus muscle with the addition of palmitate. This technique was used to obtain similar results in rhesus sartorius muscle (39). In contrast Jenkins et al. (5) found an increase in total glucose label incorporation into glycogen in soleus and red and white gastrocnemius muscles in the rat hindquarter at 2 or 4 mM plasma FFA concentrations. Kim et al. (6) also observed significantly increased glucose label incorporation into glycogen in soleus, extensor digitorum longus, and tibialis anterior at physiological insulin concentrations and only in soleus at maximal insulin concentrations of rats receiving Intralipid infusion. The advantage of the 13 C NMR method over that of these previous studies is that it allows one to make temporal measurements of muscle glycogen synthesis as opposed to a single time point measurement.
To assess the effects of FFA on potential rate-controlling steps in muscle glucose metabolism, glucose 6-phosphate was measured by in vivo 31 P NMR (13). Glu-6-P concentrations increased 30% (p Ͻ 0.01) and 53% (p Ͻ 0.005) above basal values in the control and high plasma FFA groups, respectively, by the end of the euglycemic-hyperinsulinemic clamp (Fig. 8). It is interesting to note that while ⌬[Glu-6-P] increased rapidly by 30 min and slowed thereafter in the control group, it continued to increase throughout the experiment in the high plasma FFA group. These data lend support to the mechanism that Randle et al. (1) proposed to explain FFA inhibition of insulin-stimulated glucose uptake which predicts that Glu-6-P should increase as a result of inhibition of phosphofructokinase. If glucose transport/hexokinase were rate controlling for insulin-stimulated glucose uptake, then a decrease in Glu-6-P would be expected (11). Previous studies found an increase in FIG. 8. Net increase in intramuscular Glu-6-P concentration. 30 min time averaged measurements of ⌬Glu-6-P were made in control (E) and high plasma FFA (f) groups during a euglycemic-hyperinsulinemic clamp. *, p Ͻ 0.05, control versus high plasma FFA group at respective times.
We may analyze metabolic flux control (40) with knowledge of changes in whole body glucose disposal, glycogen synthesis, glycolysis, and Glu-6-P after an increase in plasma FFA. At the end of the experiment, the rate of whole body glucose disposal and glycolysis decreased by 37 and 67%, respectively, in the high plasma FFA versus control group. This was accompanied by a 60% increase in absolute ⌬[Glu-6-P] (0.05 versus 0.08 mol/g in control versus high plasma FFA at 150 -180 min) and no significant change in glycogen synthesis rate in the high plasma FFA versus control group. The ratio of glycolytic to total glucose disposal rates in muscle (V Gly /(V Gly ϩ V glycogen )) decreased 40% in the high plasma FFA versus control group. Thus, there is good correlation between the decreased flux through glycolysis as a fraction of total muscle glucose disposal (Ϫ40%) and decreased whole body glucose disposal (Ϫ37%). In addition, the Glu-6-P data provide substrate data required for flux control analysis. From the traditional notion of metabolic flux control, the enzyme whose activity is less sensitive to substrate change exerts greater control in a metabolic pathway, so phosphofructokinase activity controls Glu-6-P levels that subsequently exert allosteric control on hexokinase activity and glucose transport.
There are a number of studies performed in humans that suggest, as in animals, that increased plasma FFA does reduce whole body glucose disposal (11-14, 31, 34, 41). This phenomenon was also observed at the tissue-specific level by the arteriovenous balance technique in leg muscle (41) and by positron emission tomography used to non-invasively monitor glucose uptake in heart, arm, and femoral muscle (42). Some have shown that glucose uptake decreases only after 3-4 h (11-14). Boden et al. (13) measured fractional glycogen synthase activity and Glu-6-P in muscle biopsies at 0, 2, 4, and 6 h after a euglycemic-hyperinsulinemic clamp at three different levels of plasma FFA. They suggested that the reduced muscle glucose uptake may result from two independent defects, impairment of glycogen synthase activity at 750 M FFA (high) and a reduction of glucose transport/phosphorylation at 550 M FFA (medium), although no significant change in intramuscular Glu-6-P concentration as measured in biopsy samples was observed in the latter study. Using a similar NMR approach in humans, our group (11) has recently shown that raising plasma FFA concentrations by ϳ2 mM during a euglycemic-hyperinsulinemic clamp caused an ϳ50% decrease in rates of muscle glycogen synthesis which was accompanied by an ϳ50% reduction in Glu-6-P in the gastrocnemius muscle after approximately 4 h, when the glucose infusion rate initially decreased. In contrast to the rat data, the human data suggest that increased plasma FFA reduces muscle glucose uptake due to decreased glucose transport/phosphorylation activity. It is possible that the mechanism for reducing insulin-dependent skeletal muscle glucose uptake with increased plasma FFA may differ between species. The contribution of glycolysis to glucose disposal in muscle is greater in rat than in human, so the probability of the glucose-FFA cycle contribution to reducing insulin-dependent glucose uptake in muscle may be greater in rat than in human. However, it is also possible that the variation in protocol (i.e. length of fast, insulin concentration, length of Liposyn/glucose infusion, etc.) may have also contributed to these observed differences between rat and humans.
In conclusion, we have developed a method using in vivo 13 C NMR to directly measure rates of glycolysis and glycogen syn-thesis simultaneously in skeletal muscle of awake rats, and measurements in skeletal muscle tissue extracts provided relative rates of FFA/ketone versus pyruvate oxidation. Additionally, in vivo 31 P NMR was used to make temporal ⌬[Glu-6-P] measurements to assess the mechanism for decreased insulindependent glucose uptake in rat muscle under conditions of euglycemia-hyperinsulinemia and increased plasma FFA. The results are consistent with the mechanism proposed by Randle and colleagues (1) to explain FFA inhibition of insulin-stimulated muscle glucose uptake. This in vivo NMR approach is directly applicable to humans and should prove useful for examining the regulation of intramuscular glucose metabolism in normal and diabetic humans.