Metabolism and Structure of Triacylglycerols in Rat Epididymal Fat Pad Adipocytes Determined by 13C Nuclear Magnetic Resonance*

Carbon-13 nuclear magnetic resonance (NMR) meth- ods have been applied to a study of the structure and metabolism of the triacylglycerols from rat epididymal fat pad adipocytes. Complete NMR signal assignments are provided for adipocytes, the extracted triacyl- glycerols, and methyl esters of the derived fatty acids. 13C NMR yielded rapid, nondestructive, quantitative analysis of the amounts of unsaturation of the fatty acyl chains; in cells from rats given ad Zibitum access to a standard laboratory diet the predominant fatty acids were found to be palmitate (29.9%), oleate (27.9%), and linoleate (34.1%). These results agreed with gas chromatographic separation of the derived methyl esters of the extracted lipids. Lipid dynamics were examined in situ and showed a substantial re-striction of motion of glyceride-glycerol as compared with free glycerol; the nuclear magnetic spin-lattice relaxation times for free glycerol of 2.52 f 0.12 (C1,3) and 4.37 f 0.21 (C2) s decreased to 0.15 f 0.009 and 0.21 f 0.013 s, respectively, upon esterification. Seg- mental motion of the chains, monitored by relaxation time measurements, increased progressively from the a-carbon (nTl = 0.70 s) to the methyl ends of the chains (nTl = 9.63). The incorporation of C-13-labeled substrates ([l-'3C]glucose and [3-13C]lactate) into the Simultaneous lipogenesis and were were the labeling tern versus triacylglycerols significantly different from that found using Our indicate a

Metabolism and Structure of Triacylglycerols in Rat Epididymal Fat Pad Adipocytes Determined by 13C Nuclear Magnetic Resonance* (Received for publication, August 23,1985) Laurel 0. Sillerud Carbon-13 nuclear magnetic resonance (NMR) methods have been applied to a study of the structure and metabolism of the triacylglycerols from rat epididymal fat pad adipocytes. Complete NMR signal assignments are provided for adipocytes, the extracted triacylglycerols, and methyl esters of the derived fatty acids. 13C NMR yielded rapid, nondestructive, quantitative analysis of the amounts of unsaturation of the fatty acyl chains; in cells from rats given ad Zibitum access to a standard laboratory diet the predominant fatty acids were found to be palmitate (29.9%), oleate (27.9%), and linoleate (34.1%). These results agreed with gas chromatographic separation of the derived methyl esters of the extracted lipids. Lipid dynamics were examined in situ and showed a substantial restriction of motion of glyceride-glycerol as compared with free glycerol; the nuclear magnetic spin-lattice relaxation times for free glycerol of 2.52 f 0.12 (C1, 3) and 4.37 f 0.21 (C2) s decreased to 0.15 f 0.009 and 0.21 f 0.013 s, respectively, upon esterification. Segmental motion of the chains, monitored by relaxation time measurements, increased progressively from the a-carbon (nTl = 0.70 s) to the methyl ends of the chains (nTl = 9.63). The incorporation of C-13-labeled substrates ([l-'3C]glucose and [3-13C]lactate) into the glycerol moiety of triacylglycerols was monitored in real time, in the presence of insulin. Lactate (10 mM) inhibited the incorporation of glucose (5.5 mM) into glyceride-glycerol. Lipolysis at the natural abundance level of I3C was measured in the presence of 10 p~ isoproterenol. Simultaneous lipogenesis and lipolysis were found to occur in situ and were measured with the aid of [l-13C]glucose and isoproterenol; the labeling pattern of medium glycerol versus extracted triacylglycerols was significantly different from that found using natural abundance glucose. Our results indicate that 13C NMR is a useful new method for the real-time monitoring of lipid structure and metabolism in vivo.
Adipose tissue constitutes an important site for the synthesis, storage, and mobilization of triacylglycerols. Its involvement in the metabolism of lipid, carbohydrate, and other substrates has implicated major roles for this tissue in a variety of normal and abnormal physiological events, including the etiologies of cardiovascular disorders and adult onset diabetes mellitus (1). The metabolism of adipose tissue is regulated by hormones which alter both lipogenesis and lipol-* This work was performed under the auspices of the Department of Energy. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked ''advertisement'' in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. ysis. Adipose tissue metabolism has been extensively studied in the past with the aid of radioactive tracers in cell suspensions, perfused fat pads, and in animals. The well-known dangers associated with radioactivity have limited the applications of radioisotope technology in man. Consequently, it would be desirable to find a method for the noninvasive examination of the hormonal regulation of adipose tissue metabolism in normal and pathological states in uivo.
Carbon-13 nuclear magnetic resonance (NMRl) (2) methods offer, for the first time, the means by which metabolic regulation may be examined in adipose tissue in vivo, in real time, in a noninvasive manner, without the use of radioactive materials. In principle, it should be possible to apply these NMR methods to the study of lipid metabolism in adipocyte suspensions, perfused fat pads, and eventually, man. We have chosen to initially determine the applicability of 13C NMR to the study of triacylglycerol metabolism in adipocytes,from the epididymal fat pad of the rat as a model system. It has previously been shown that 13C NMR spectra from tissue lipids can be obtained in uivo from the rat head, abdomen, and hind leg, human arm (3), and the liver of the rat (4).
We report here the basic 13C NMR data needed to establish the biochemical utility of the triacylglycerol resonances from adipocyte suspensions. These data include the lipid chemical shift assignments, fatty acyl chain compositions, spin-lattice relaxation times, line widths, and nuclear Overhauser enhancements. In addition, we demonstrate that 13C NMR can be utilized to follow lipid metabolism in real time in situ within the adipocyte without the need for lengthy chemical extractions. The 13C NMR method is shown to provide unique information with respect to substrate cycling, the integrity of chemical bonds, and lipid composition and dynamics. Finally, we show how the use of 13C-labeled substrates can provide information on the hormonal regulation of metabolism in the intact adipocyte. MATERIALS  The pulse angle was optimized according to the spin-lattice relaxation times of the slowest relaxing signal of interest (5). This signal was usually from C1 of glucose (Tl = 1.28 s).' Spin-lattice relaxation times for the triacylglycerols in both isolated adipocytes and chloroform extracts were measured by means of an inversion recovery sequence (180"-~-90"). The relaxation times and standard deviations were determined from 3-parameter fits to the data using the spectrometer system software (DISNMRP, 830701) and a convergence limit of 1 X IO4. Nuclear Overhauser enhancements were measured by using gated proton decoupling. All NMR experiments reported here were done at a temperature of 310.0 & 0.2 K, measured during acquisition with the aid of a Luxtron (Mt. View, CA) model lOOOB fluorescence thermometer. All chemical shifts were measured by the spectrometer system software by setting the signal of the 0-1 carbon of the triacylglycerol alkyl chain to 6 = 23.45 ppm with respect to external Me& (6); to correct the shifts in CDCls to an internal Me4Si scale, subtract 0.74 ppm from those shown in Table I.
Natural abundance 13C NMR spectra of the isolated fat cells were obtained from suspensions containing 50-70% cells by volume in medium 199 supplemented with 10% fetal bovine serum and 5.5 mM glucose. Because adipocytes have a density (-0.7 g/ml) substantially lower than that of water, they will rapidly rise to the top of the suspension. In order to ensure that the cells remained within the radiofrequency coil of the NMR probe throughout the experiment we used a sample volume (1.8 ml) that just filled the observation coil. The 10-mm NMR tubes were siliconized prior to use to prevent cells from sticking to the glass walls.
A similar concentration of adipocytes was used in experiments designed to monitor the hormone stimulation of triacylglycerol synthesis and degradation. The Tl-corrected (2) integrals of the C1 a and C1,6 resonances of [l-13C]glucose and the ratio, r, of the corrected integrals of C1,3 to C2 for the triacylglycerols were determined to probe the hormone effect. The metabolic enrichment, e = n (r -2), where n is the natural abundance of 13C (l.l%), gives the per cent of 13C above natural background. When 13C glucose was used for the substrate, prior washings of the cells with medium containing no natural abundance glucose were performed.
(95%:5%) on the surface of the NMR sample. Due to the ease of fat The cells were oxygenated by blowing water-saturated 02:C02 cell breakage and the resulting formation of a triacylglycerol layer on the surface of the sample, stirring of the cell suspension was abandoned prior to the initiation of the experiments reported here, and neither a field-frequency lock nor sample spinning were used.
Preparation of Fat Cells-Rats were placed in a C02-saturated desiccator for 2-4 min. The epididymal fat pads were removed and rinsed in phosphate-buffered saline, pH 7.40, and cut into small pieces weighing about 10 mg. The pieces were transferred to a 125ml polypropylene Erlenmeyer flask containing 30 ml of Parker's medium 199 with 10% fetal bovine serum, 5.5 mM glucose, and collagenase (380 pg/ml, Millipore Corp., see "Results"). The suspension of tissue was incubated at 37 "C for 1 h in a New Brunswick Gyrotory water bath/shaker (New Brunswick Scientific Co., Inc., Edison, NJ). Several collagenases were tested for the known adverse effects on hormone responsiveness of contaminating phospholipases and proteases (7)(8)(9). Suitability of collagenases was determined by insulin stimulation of adipocyte [14C]glucose incorporation into l4COZ and triacylglycerols using standard radioisotopic procedures (24). The Millipore collagenase (lot 41H108) gave the best insulin responses and was used throughout the experiments reported here. The isolated adipocytes and the remaining tissue fragments were poured through nylon mesh (Small Parts, Inc., Miami, FL) with a pore size of 500 pm into a 50-ml graduated polypropylene centrifuge tube, The isolated adipocytes were allowed to float for 10 min, and the infranatant containing the medium and stromal vascular cells was removed by aspiration. The cell layer was washed 3 times with 20 ml of medium (37 "C) without collagenase. Washed cells were resuspended in the desired volume of medium 199.
Incubation of Fat Cells-The cell suspension was gently swirled to ensure homogeneity, and 1.0-ml aliquots were transferred to 25-ml polypropylene Erlenmeyer flasks containing medium 199, hormones, and subtrates to a total sample volume of 2.0-4.0 ml. The flasks were covered with Parafilm, placed in a Gyrotory water/shaker bath at 37 "C, and swirled gently (60 rpm) and incubated for various times.
The number of fat cells in each incubation flask was determined by the following method. The cell suspension (50 pl in a 5-ml polypropylene tube) was diluted with 300 pl of phosphate-buffered saline and 150 pl of a 10% acridine orange solution. Immediately after staining, cell numbers were determined using a Zeiss fluorescence microscope and a standard hemocytometer (American Optics, Inc.). Cell suspensions contained approximately 2 X lo6 cells/ml. Cell size was also estimated by measuring the diameters of 300 cells using a calibrated reticule. Mean volumes were calculated from the measured diameters and the formula for the volume of a sphere. Cells fell within the volume range 130-870 pl, with a mean cell volume of 452 pl.
Determination of the Effect of Hormones on [U-'4ClGlucose Metabolism and Lipoly~is-[U-~~C]Glucose was added to the incubation flask containing 3.0 ml of the rat cell suspension to give a final specific activity of 0.1 pCi/pmol of glucose. The final insulin concentration in the incubation flask was lo-' M (1.52 microunit/ml). This concentration gave the maximum effect of insulin on ["C]glucose incorporation into triacylglycerol by the isolated fat cells during a 3h incubation at 37 "C. At the end of this time, a 0.5-ml aliquot of the cell suspension was removed during gentle swirling of the flask. The reaction was terminated by mixing the cell aliquot with 1.0 ml of chloroform in a 15-ml centrifuge tube. The triacylglycerols were extracted (see below) by vigorous Vortex mixing. The chloroform layer was washed 3 times with 0.5 ml of a 10% NaCl solution. A 0.5ml aliquot of the chloroform layer was air dried, and the radioactivity determined using standard scintillation techniques. The quantity of ["C]glucose carbon converted to triacylglycerol was calculated from the initial specific activity of glucose in the medium and the quantity of radioactivity in the product.
To medsure the lipolytic effect of epinephrine M) and other lipolytic hormones and compounds, incubations were performed as above, but in the presence of added hormones. A 0.5-ml aliquot was removed as described above and allowed to stand for 5-10 min at room temperature to allow the cells to float. The infranatant was gently aspirated, and glycerol in the medium was measured by means of the chromotropic acid assay (10). Data are expressed as a function of adipocyte number, and, in some cases, triacylglycerol concentration.
Isolation of Trincylglycerols from the Isolated Fat Cells-The triacylglycerols in 0.5 ml of a 50% cell suspension in a 15-ml centrifuge tube were extracted 4 times with 1.0-ml portions of chloroform by agitation on a Vortex mixer. The first extract gave 85% of the total, the second 12%, the third 3%, and the fourth a negligible amount of the total triacylglycerol. The combined chloroform extracts were washed with 0.1 M KC1 (2 ml), dried, and evaporated under a stream of N,. Lipid analysis indicated that 0.5 ml of a 50% cell suspension gave 74 mg of triacylglycerols.
Isolation of Fatty Acids from Fat Cell Triacylglycerols-The fat cell triacylglycerols from above were dissolved in 10 p1 of chloroform/ toluene (2:8, v/v) and mixed with 20 ml of anhydrous methanolic HCl (5%) by the method of Eisen et al. (11) with the following modifications. Five ml of 2,2-dimethoxypropane was added, and the mixture was stirred in a capped flask overnight. Water (20 ml) was added, and the fatty acid methyl esters were extracted three times with 20ml aliquots of petroleum ether. The combined extract was dried with anhydrous ammonium sulfate and sodium bicarbonate (41) and evaporated with a stream of Nz.
The methyl esters of the fatty acids were separated by gas chromatography with a Hewlett-Packard model 3380A on an EGSS column (6 feet X 1/8 inch) at 200 "C. Identification of each fatty acid was done by comparing the retention time to that of a known fatty acid methyl ester. The amount of each fatty acid was determined from an integration of the peaks.  (Table I), free fatty acids, and fatty acid methyl esters (Table 11) follow    trum of the extracted triacylglycerols dissolved in chloroform (Fig. 1B) is qualitatively similar to that for these compounds in situ (see also Figs. 2B and 3B). The spectrum of the adipocytes can be accounted for solely by the triacylglycerols (14); no other cellular constituents were observed in the natural abundance 13C NMR spectrum. 13C NMR signals from the membranes or the cytoplasm of these cells were not expected to be found because these latter components constitute less than 5% of the cell mass. Important information about the types and sites of unsaturation of the fatty acyl chains of the triacylglycerols can be rapidly obtained from the NMR spectrum. Characteristic signals from double bonded carbons appear in two places in the spectrum, near 130 pprn (Fig. 3) and in the methylene region (Fig. 2). The olefinic region of the spectrum is dominated by two major groups of resonances separated by about 1.9 ppm (Fig. 3). This separation arises from the magnetic   Table 11).

Chemical Shift Assignments-Assignments for the
The resonances from methylene carbons adjacent to olefinic sites in the fatty acyl chains are shifted due to changes in their shielding arising from their olefinic neighbors. The signal at 23.35 ppm arises from carbon nuclei in the penultimate position (15) of the acyl chain of linoleate (Fig. 2 A ) . The peak at 26.35 ppm is indicative of an allylic carbon in a cisdiene system; this could arise from either linoleate or linolen- ate (16,17). The resonance at 27.92 ppm is characteristic of a cis-allylic carbon (18,29); this could arise from either oleate or palmitoleate. The signal at 32.32 ppm is uniquely assigned to the w-2 carbon of a dienoic fatty acid (20).
Assignments for the main methylene envelope (the region near 30 ppm in Fig. 2 A ) are based on an analysis of the corresponding region of the spectra of the fatty acids and methyl esters (Table 11) having 12-18 carbons and 0-3 double bonds. The signals at 29.89 and 29.99 ppm were assigned to C4 and 0-3 carbon in oleate, respectively; this scheme is supported by studies of the substituent shifts on saturated carbons (13). The signal at 30.12 ppm was tentatively assigned to C5 and C6. These carbons are shifted 0.30 ppm upfield with respect to their field positions in the corresponding alkanes (13). A tentative assignment of the peak at 30.35 ppm can be made to carbons which are at the w-4 position. The remaining intense peak at 30.47 ppm can be assigned .to methylene carbons which are free from magnetic effects from double bonds, carbonyls, or terminal methyl groups.
Certain features of these assignments have particular relevance with regard to the usage of 13C NMR for metabolic studies. Upon esterification, the signals from glycerol shift upfield sufficiently (C1,3, -0.8; C2, -3.1 ppm) so that free and bound glycerol can be easily resolved. This implies that the integrity of the ester bond in triacylglycerols can be monitored by means of I3C NMR. The methyl carbon of lactate resonates between the two signals from the ultimate and penultimate carbons of the fatty acyl chains (Table I).
The carboxyl carbons of fatty acids shift upfield by 6.3 ppm upon methyl ester formation (Table 11), and an additional 1.3-2.6 ppm upon esterification with glycerol (Table I). This shift of the carboxyl resonances offers another way to monitor the integrity of the ester bond by I3C NMR.
Nuclear resonances from carbons near the glycerol moiety dycerols in Adipocytes are influenced by the dielectric constant of the medium (Table  I). For example, there is a 1.2-1.4 ppm shift of the carboxyl carbon resonances of the triacylglycerols when they are extracted from the adipocyte and dissolved in chloroform. A similar effect of the solvent on the resonance frequencies of the carboxyl carbons of phospholipids has been interpreted as the result of solvent competition for intermolecular hydrogen bonds (21). We are able to identify resonances from the 9 major fatty acyl chain lengths in the 13C NMR spectrum (Fig. 1B) of a chloroform extract of the fat pads from rats given ad libitum access to a standard laboratory diet. The amounts and saturation patterns of the various fatty acyl chain lengths were measured from the I3C NMR spectra, and independently, by means of gas chromatography (Tables I1 and 111). The amount of saturation was estimated from the difference between the integrals of C-3 and the total integrals of the unsaturated carbons. For the four fatty acyl chain types which give rise to unique 13C nuclear resonances (16:0, 181, 182, and 183) the mol % determined by NMR (Table 111) agrees well with that found using GC analysis. The NMR determination is much faster (10 min) than the GC method which relies on the quantitative hydrolysis of the esters and their subsequent quantitative methylation (2-3 days). The average difference between the mol % determined by the two methods is 2 mol %. In the case of linoleic acid (18:2) we could determine its amount in the triacylglycerol mixture in three independent ways from the 13C NMR spectrum (Figs. 2 and 3, and Table  IV). Integrals of the signals at 23.35 ppm (w-1), 26.35 ppm (Cll), and 32.29 ppm (w-2) (Fig. 2B) from linoleic acid carbons were used (Table IV) to calculate a mol percentage of 34.1 & 1.8%. This determination agrees within the error limits quoted with that found by GC analysis.
The results of both the GC and the NMR unsaturation analyses are in agreement that the major fatty acids found in these adipocytes are palmitic, oleic, and linoleic (Table 111).   integrals of the unsaturated carbons. See Table IV. Lipid Dynamics-Carbon-13 NMR provides additional information with respect to the properties of triacylglycerols; the spectral characteristics are sensitive to the dynamical state of the lipids and their components. The spin-lattice ( Tl), spin-spin (T2) relaxation times, and the nuclear Overhauser enhancements (~+ 1 ) are all determined by the rotational correlation times for the various nuclei. The carbon nuclear spin relaxation mechanism is dominated by direct dipole interaction with the covalently attached protons for all triacylglycerol I3C-nuclei except the fatty acyl carboxyls.
As we proceed down the methylene chain from the carboxyl carbon toward the terminal methyl group, the number of directly bound protons is constant; consequently, we can interpret the spin-lattice relaxation times directly in terms of the mobility of the various carbon nuclear segments. The data ( Table I)  There is a mobility gradient down the chain, with the shortest rotational correlation time appearing at the terminal methyl. The terminal methyl group has a rotational correlation time which is about 30-fold shorter than that found for the exterior glycerol carbons (C1,3). The spin-lattice relaxation times were found to be 2.15 f 0.27 times longer for the extracted triacylglycerols dissolved in CDC13 than for these molecules in situ. This is understood from the known dependence of rotational correlation time on viscosity (22); the interior of the adipocyte is pure lipid with a viscosity much higher than that of CDCl,. The nuclear Overhauser enhancements are 1.6 f 0.3 times larger in CDCl, (Table I).
The I3C nuclear magnetic relaxation parameters for long fatty acyl chains cannot be explained in terms of single isotropic rotational correlational times; the spin-lattice, spinspin, and Overhauser factors do not fit any isotropically reorienting model. The molecular motion is characterized by at least two correlation times, one describing overall isotropic tumbling and one relating to the rate of internal motion about carbon-carbon bonds (23) of the molecules. The data in Table  I for the triacylglycerols shows that anisotropic motion is also characteristic of these molecules, both in situ and in CDC13. The spin-lattice relaxation times increase as the methyl end of the chains is approached, which is in contrast to the position independence of the relaxation times that would be expected for isotropic tumbling. The fact that the spin-lattice relaxation times and the nuclear Overhauser enhancements measured in CDC13,are a constant multiple of the data found for the triacylglycerols in situ indicates that the motion of these molecules is the same in the two environments; there is an overall decrease in the correlation times on going to the CDCls.
The resolution of the triacylglycerol I3C NMR spectrum is greater when these molecules are dissolved in CDCL as compared with the in situ spectrum. The widths of the resonances were measured in both states and at two different magnetic field strengths (2.0 and 7.05 tesla); the protonated carbons gave signal widths of 3.0 f 0.2 and 6.8 f 1.4 Hz, respectively, in CDCb and 3.8 f 0.5 and 48-77 Hz, respectively, in situ. The larger widths found at the higher field are due in part to inhomogeneous broadening by the particulate nature of the cellular sample and partly due to chemical shift anisiotropy in these long bulky molecules, particularly for the triacylglycerols in the viscous lipid droplet within the adipocyte. At the lower field (2 tesla) the carbonyls give resonance widths of 1.7 f 0.5 Hz in either state; we expect that these molecules would have resonance widths of around 1 Hz based on their mass of about 1000 daltons. At 7 tesla, on the other hand, the width of the carbonyl resonances changes from 6.9 k 0.5 to almost 25 Hz when one compares the extracted molecules in CDC1, to those in situ within the adipocyte. These widths in situ at 7 tesla scale approximately as the square of the field, lending support to the proposal of chemical shift anisotropy as an important line-broadening mechanism for the lipid signals from within the adipocyte.
The measurement of the relaxation parameters for the triacylglycerols has a value beyond the study of segmental motion in lipids. These parameters must be known if peak integrals are to be converted to amounts of 13C so that NMR can be used as a quantitative detector for carbon-13 isotopic enrichment. In situations where there is no 13C recycling or scrambling into all molecular sites, certain sites will remain at the natural abundance level of 1.1% 13C, These latter sites can be used, once corrected for saturation and Overhauser effects, as internal standards for resonance signals from those sites in the biomolecules which do incorporate 13C as a result of metabolic flow. We can use 13C NMR in this way to measure the synthesis of chemical amounts of metabolites.
Insulin Stimulation of [1-'3C]Glucose Incorporation into Triacylglycerols-A key advantage of the use of I3C NMR in the study of the metabolism of adipose tissue is the fact that the natural abundance spectrum of the tissue contains resonances from many distinct carbon nuclei (Fig. 1, Table I). In particular, the resonances from the glycerol moiety of the triacylglycerols appear in a window removed from those of the other carbons and nuclei (Fig. 3). In the glycolytic pathway responsible for the formation of glycerol 3-phosphate, the 13C label added as C1 of glucose appears as C3 of the glycerol, while no 13C flows directly to the C2 position of the glycerol. Since these two glycerol positions give rise to distinct NMR signals, one can separately determine the time course of the intensities of these signals and, therefore, use the C2 signal as a constant internal control during the extended NMR time courses. Furthermore, the C2 signal has an integral equal to that of a single carbon nucleus in a triacylglycerol at natural abundance so that it acts as a standard for determining the absolute incorporation of 13C into the glycerol carbon 3. The chemical shifts of free glycerol and the glycerol moiety of triacylglycerols are also distinct so that one can monitor the formation and breakdown of triacylglycerols during lipogenesis and lipolysis from an examination of the time dependence of the 13C NMR spectra of adipocytes.
One of the classical effects of insulin on adipose tissue is to promote the incorporation of glucose into the glycerol moiety of triacylglycerols. In our incubations with [U-'4C]glucose, insulin at a concentration of 10 nM caused a 5.5-fold increase in the rate of ['4C]glucose incorporation into triacylglycerols; the rate in the presence of insulin was found to be 450 nmol/ lo6 cells/h. These results indicated that it would be feasible to monitor with 13C NMR the flow of 13C-labeled glucose into the triacylglycerol pool of adipocytes, since an average adipocyte contains from 1 to 10 nmol of lipid, and this rate would then correspond to a change in the NMR signal of from 0.3 to 3.0%/h. Changes of this magnitude (-1 pmol/sample/h) can be measured given the very high signal to noise ratio (-1001 for glycerol C2 in 1 min) obtained from the adipocyte suspensions.
In the presence of a physiological concentration of insulin (10 nM) the incorporation of glucose into the glycerol moiety of adipocyte triacylglycerols is stimulated. In order to determine the efficacy of 13C NMR as applied to a study of this process we incubated adipocytes with insulin (10 nM) and [ 1-13C (go%)]-D-glucose. The 13C NMR spectrum of the adipocyte system (1.8 ml of a 50% cell suspension) was taken and stored on the disk in 10-20-min blocks. The results (Fig. 4) show that there is a significant decrease in the 13C NMR signal from the added [l-13C]glucose over the course of this 3-h experiment. At the same time, we observed no change in the integral of the signal from carbon 2 of the triacylglycerol. The constancy of this internal control signal indicates that our adipocyte suspension was fully contained within the NMR coils and that sample flotation out of the sensitive volume was not a problem.
In order to demonstrate that the NMR method is capable of detecting chemical amounts of triacylglycerol synthesis against a large natural abundance 13C background, we incubated adipocytes for 18 h with insulin (10 nM) and  (1,3), 6.189 X lo-' for TAG (2), and adipocytes. The control ratio is 2 because 13C is randomly distributed in nature and there are 2 carbons contributing to the C1,3 signal as opposed to only 1 for the C2 resonance. These data also serve as a good illustration of the use of one of the triacylglycerol signals as internal controls so that very accurate measurements of the amount of 13C at a labeled site can be made.
Natural Abundance 13C NMR Monitoring of Lipolysk-Hormonal regulation of metabolic pathways often takes the form of opposing effectors. It is well known that insulin promotes lipogenesis while the adrenergic effectors such as epinephrine and isoproterenol stimulate lipolysis. We have already shown that 13C NMR can monitor lipogenesis. Our results show that during incubations of adipocytes with 10 p~ isoproterenol we could also monitor lipolysis in real time (Figs. 5 and 6). Noteworthy is the decrease with time of the glyceride-glycerol C1,3 signal over the course of the experiment, while essentially no change occurs in the control penultimate carbon resonance. Similar control results (data not shown) were obtained when the time dependence of the glycerol C2 signal was examined. These data provide values for the rate of lipolysis of 300 nmol of glycerol/106 cells/min. The success of this method depends on the fact that there are substantial shifts of several of the carbon resonances when the fatty ester bond is broken. As the triacylglycerol is degraded, the glyceride-glycerol signal decreases and, if enough is formed, a signal from free glycerol appears, either in the cells or from the medium. Adipocytes lack glycerol kinase so that the free glycerol formed during lipolysis cannot be reesterified and is given up to the medium.
The free carboxyl carbons from the cleaved fatty acids will also resonate in a different and distinct position of the 13C NMR spectrum from those ester bound into triacylglycerols (Tables I and 11). Exploitation of these spectral properties was accomplished by monitoring isoproterenol-stimulated lipolysis at natural abundance in adipocytes incubated with 10 ~L M hormone for 192 h. This extensive time was chosen in order to provide for substantial chemical-scale hydrolysis of the triacylglycerols in the adipocytes. The results (Fig. 6) show that the bound and free fatty acids are easily resolved in the 13C NMR spectrum of the extracted triacylglycerols and that 37.4 f 1.4% of the lipids were cleaved over this time span. Chemical analyses of the incubation media and extracted triacylglycerols indicated that 38.6% of the pre-existing lipids were cleaved over this time period; the agreement between these two different methods of measuring lipolysis is excellent.
Lactate Inhibition of [l-13C]Glucose Metabolism-In the present set of experiments in which adipocytes were incubated simultaneously with [ 3-l3C] -L-lactate and [ 1-13C] -D-glucose, we monitored both of these labeled sites simultaneously, in real time. The lactate methyl carbon resonates at 20.9 ppm and is resolved from the w-1 (23.45 ppm) and w (14.72 ppm) carbons of the triacylglycerols, while the glucose C1 resonances are found around 95 ppm, far away from any competing resonances from the lipids (Fig. 1, Table I).
Substrate interactions are indeed of some interest in adipose tissue with respect to metabolic disorders, such as diabetes and obesity. Using radio-labeled procedures, lactate, in particular, has been shown by us (24) to exert a potential glucose-sparing effect on adipocyte metabolism by virtue of its preferential utilization over glucose for glycerogenesis, when both are present in the medium. Given the excellent resolution available from the 13C NMR spectrum of adipocytes, we sought to measure the rates of glucose and lactate utilization in the presence of insulin (10 nM). The results (Fig. 7) show that lactate is a good substrate for the glycerol portion of triacylglycerols. The lactate was consumed at a rate of about 4.2%/h or 220 nmol/106 cells/h, while 13C from lactate appeared in the glyceride-glycerol at a rate of 2.3%/h or 121 nmol/106 cells/h. The consumption of glucose was significantly decreased in this experiment by the presence of the lactate; glucose only accounts for approximately onequarter of the amount of 13C incorporated into lipids. We observed a slight upward trend to the glycerol C2 signal; the rate was about 6-fold smaller than that found for carbons C1 and C3. Lacate and glucose consumption was about 60% greater than the rate of incorporation of 13C into glycerideglycerol; the excess lactate-and glucose-derived 13C was lost to the atmosphere as 13C02. Subsequent experiments using hyamine as a carbon dioxide trap and 13C NMR have demonstrated significant 13C labeling of the evolved COz (data not shown).
Substrate Cycling during [l -13C]Glucose Metabolism in Adipocytes-We have shown that 13C NMR can monitor both major directions of flow in the metabolism of triacylglycerols: lipolysis and lipogenesis. In this regard, we sought to deter- mine whether NMR could be utilized to determine the origin of the carbons of glycerol during lipolysis in the presence of glucose. Adipocytes were incubated in the presence of either 9.4 mM [l-'3C]glucose or natural abundance glucose, with half of the samples receiving 10 p~ isoproterenol and the other half only saline control additions. At the end of the incubations, after all the glucose had been consumed, the medium and the cells were separated and the triacylglycerols were extracted from the cells. In the absence of labeled glucose, the C-13 NMR spectrum of the medium and triacylglycerols (Fig.  8) shows a ratio of glycerol signal intensities of r = 2.018 f 0.056 ( n = 18), while in the .presence of labeled glucose this ratio increased to 3.02 2 0.06 for both the glycerol and the triacylglycerols; isoproterenol had no effect on this ratio. The presence of the hormone did result in a %fold stimulation of the amount of glycerol formed during the incubations and in the appearance of labeled lactate in the medium (Fig. 8).
These results indicate that the carbons of glycerol in the medium originated from medium glucose, rather than from the intrinsic triacylglycerols. The glucose in the medium was taken up, cleaved in glycolysis, formed into glyceride-glycerol, and cleaved in lipolysis to form glycerol. The 13C enrichment at glycerol carbons 1,3 was found to be e = 1.122 zt 0.055%.

DISCUSSION
Of primary interest in this study was the establishment of the efficacy and accuracy of 13C NMR as applied to the quantitative determination of the metabolic parameters of isolated rat epididymal adipocytes. From the results it is clear that nuclear resonance methods can be applied in many unique ways in the study of metabolic molecular biotransformations in adipocytes. The resolution, sensitivity, and specificity of the carbon NMR spectrum are well suited to the examination of the various factors which regulate the flow of metabolites through the glycolytic, lipogenic, and lipolytic pathways in adipocytes in vitro. Advances in surface coil NMR technology (3) will eventually enable the examination of these properties in uivo. Strengths of this approach include the large I3C chemical shift range (coupled with the assignments of the 13C NMR spectrum), the sensitivity of the shifts to the formation and degradation of covalent bonds, and the ability of the method to give structural information with respect to the naturally occurring fatty acyl chain composition and to the dynamic aspects of lipids in situ. The accuracy and precision of 13C NMR are sufficient to permit quantitative metabolic measurements to be made in cell suspensions. Extrapolation of this technique to the in vivo situation as has been done with the liver (3,25,26) and heart (27) will undoubtedly lead to important new insights.