Fatty Acid Cycling in Human Hepatoma Cells and the Effects of Troglitazone*

Fatty acid cycling by chain shortening/elongation in the peroxisomes is an important source of fatty acids for membrane lipid synthesis. Its role in the homeostasis of nonessential fatty acids is poorly understood. We report here a study on the cycling of saturated fatty acids and the effects of troglitazone in HepG2 cells in culture using [U-13C]stearate or [U-13C]oleate and mass isotopomer analysis. HepG2 cells were grown in the presence of 0.7 mmol/liter [U-13C]stearate or [U-13C]oleate, and in the presence and absence of 50 μm troglitazone for 72 h. Fatty acids extracted from cell pellets after saponification were analyzed by gas chromatography/mass spectrometry. Peroxisomal β-oxidation of uniformly 13C-labeled stearate (C18:0) and oleate (C18:1) resulted in chain shortening and produced uniformly labeled palmitate (C16:0) and palmitoleate (C16:1). In untreated cells, 16% of C16:0 was derived from C18:0 and 26% of C16:1 from C18:1 by chain shortening. Such contributions were significantly increased by troglitazone to 23.6 and 36.6%, respectively (p < 0.001). Desaturation of stearate contributed 67% of the oleate, while reduction of oleate contributed little to stearate (2%). The desaturation of C18:0 to C18:1 was not affected by troglitazone. Our results demonstrated a high degree of recycling of C18:0 and C18:1 to C16:0 and C16:1 through chain shortening and desaturation. Chain shortening was accompanied by chain elongation in the synthesis of other long chain fatty acids. Troglitazone specifically increased recycling by peroxisomal β-oxidation of C18 to C16 fatty acids, and the interconversion of long chain fatty acids was associated with reduced de novo lipogenesis.

Fatty acid cycling by chain shortening/elongation in the peroxisomes is an important source of fatty acids for membrane lipid synthesis. Its role in the homeostasis of nonessential fatty acids is poorly understood. We report here a study on the cycling of saturated fatty acids and the effects of troglitazone in HepG2 cells in culture using [U- 13  Fatty acids extracted from cell pellets after saponification were analyzed by gas chromatography/mass spectrometry. Peroxisomal ␤-oxidation of uniformly 13 C-labeled stearate (C18:0) and oleate (C18:1) resulted in chain shortening and produced uniformly labeled palmitate (C16:0) and palmitoleate (C16:1). In untreated cells, 16% of C16:0 was derived from C18:0 and 26% of C16:1 from C18:1 by chain shortening. Such contributions were significantly increased by troglitazone to 23.6 and 36.6%, respectively (p < 0.001). Desaturation of stearate contributed 67% of the oleate, while reduction of oleate contributed little to stearate (2%). The desaturation of C18:0 to C18:1 was not affected by troglitazone. Our results demonstrated a high degree of recycling of C18:0 and C18:1 to C16:0 and C16:1 through chain shortening and desaturation. Chain shortening was accompanied by chain elongation in the synthesis of other long chain fatty acids. Troglitazone specifically increased recycling by peroxisomal ␤-oxidation of C18 to C16 fatty acids, and the interconversion of long chain fatty acids was associated with reduced de novo lipogenesis.
The peroxisomes and the mitochondria are two separate fatty acid ␤-oxidation systems having distinct roles in fatty acids catabolism, energy production, and substrate cycling within the cell. The ␤-oxidation system of the peroxisomes, unlike that of the mitochondria, is not coupled to oxidative phosphorylation and is an important source of acetyl (2-carbon) units for the synthesis of long chain fatty acids by chain elongation (1). Fatty acid cycling of polyunsaturated fatty acids in the peroxisomes has been shown to play an important role in the metabolism of essential fatty acids (2). The role of fatty acid cycling by chain shortening/elongation of saturated fatty acids is not well known. Because of recycling of label and the lack of proper isotopic methods, the study of chain shortening/elongation of nonessential fatty acids has been difficult.
Recently, we have developed stable isotope methods for the study of essential and nonessential fatty acid metabolism using uniformly labeled compounds and mass spectrometry (3,4). For example, chain shortening of [U-13 C]stearate produces palmitate with a mass shift of 16 daltons due to 13 C carbons, and the elongation of [U-13 C]stearate produces arachidate (C20:0) and behenate (C22:0) with a characteristic mass shift of 18 daltons. Thus, chain shortening and elongation can be measured by the formation of these unique isotopomer species. We report here a study of chain shortening and elongation of stearate (C18:0) and the role of activation of peroxisome oxidation with troglitazone, a peroxisome proliferator-activated receptor (PPAR␥) 1 ligand, on stearate metabolism in HepG2 cells in culture using uniformly 13 C-labeled stearate and oleate and mass isotopomer analysis.

MATERIALS AND METHODS
Tissue Culture-Human hepatoma cell line HepG2 was obtained from the American Type Culture Collection (ATCC, Rockville, MD) and it was grown in 75-ml flasks in Dulbecco's modified Eagle's medium augmented with 10% fetal bovine serum (5). When the cells were ϳ50% confluent (ϳ2.5 ϫ 10 6 cells/flask), the medium was removed, the cells were washed with phosphate-buffered saline, and the appropriate medium containing U-13 C-fatty acids was added as described below to begin the experiment. The incubation lasted 72 h with changes of fresh medium daily.
Isotopes and Drugs-[U-13 C]Stearic acid and [U-13 C]oleic acid were obtained from Martek Biosciences (Columbia, MD) as their sodium salts. They were dissolved in warm water and added separately to the culture medium at a concentration of 0.7 mmol/liter. Troglitazone was obtained from Park Davis Pharmaceuticals (Ann Arbor, MI). It was dissolved in dimethyl sulfoxide (Me 2 SO) and added to the appropriate flasks to a final concentration of 50 M. The same volume of Me 2 SO was added to the flasks that did not contain troglitazone. Each incubation condition was performed in triplicate, and each analysis was also run in triplicate. During the experiment, HepG2 cells doubled to approximately 0.5 to 1 ϫ 10 7 cells per plate, and remained 80 -90% viable after harvest.
Extraction of Lipids from the Cell Pellet-Fatty acids were extracted according to the method described by Lowenstein et al. (6). The cell pellet was saponified with 1 ml of 30% KOH:ethanol (v:v, 1:1) at 70°C overnight. Neutral lipids were first removed with petroleum ether extraction. The solution containing the saponified fatty acids was then acidified, and palmitate and other fatty acids were recovered with another petroleum ether extraction. Fatty acids were methylated with 0.5 N HCl in methanol (Supelco, Bellfonte, PA) for GC/MS analysis (7).
GC/MS Analyses-Fatty acids were analyzed as their methyl esters. Palmitate, palmitoleate, stearate, and oleate were separated on HP5840A GC with a 3-foot SP2330 glass column using temperature programming. The GC conditions were as follows. Helium flow rate was 20 ml/min and the initial temperature was held at 180°C for 1 min, and * This work was supported by National Institutes of Health United States Public Health Service Grants PO1-CA 42710, MO1-RR 00425, and RO1-DK46353. 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.
§ To whom correspondence should be addressed: Harbor UCLA Research and Education Institute, 1124 W. Carson St., Torrance, CA 90502. Tel.: 310-222-6729; Fax: 310-533-0627; E-mail: lee@gcrc. humc.edu. then the oven temperature was programmed to increase at 3°C/min to a final temperature of 210°C. Under these GC conditions, the retention times for palmitate, palmitoleate, stearate, and oleate were 3.1, 3.7, 5.1, and 5.7 min, respectively. The temperature of the GC to mass spectrometer interface was maintained at 275°C and the source temperature at 200°C. Mass spectra were obtained on the HP5985 mass spectrometer using electron ionization at (Ϫ70 eV) and selected ion monitoring. Arachidate (C20:0) and behenate (C22:0) were analyzed as their methyl ester using a HP5973 mass spectrometer/HP6890 GC. Fatty acids were separated on a Bpx70 column (25-m length, 220-m diameter, 0.25-m film thickness from SGE Incorporated (Austin, TX). The oven temperature was programmed as follows: initial temperature 160°C for 1 min, then programmed at a rate of 5°C/min to 230°C. The split ratio was 15:1. Retention times for C20:0 and C22:0 were 6.18 and 8.09 min, respectively. Ion clusters monitored for methyl ester of C20:0 were m/z 325-330 and 342-347; and m/z 352-358 and 370 -377 for C22:0. Electron ionization spectra of "unlabeled" C20:0 and C22:0 methyl esters showed base peaks (molecular ions) at m/z 326 and 354. The corresponding m ϩ 18 peaks were found at clusters around m/z 344 and 372.
Data Analysis-Mass spectra were acquired in the traditional "normalized" 3 format. For the purpose of determining enrichment and fractional uptake and conversion, the mass spectra were expressed as molar fractions, i.e. the fraction of molecules with a particular mass. This is achieved by dividing the intensity of each peak by the sum of intensities of all relevant peaks of that compound. The expression of spectral data as molar fraction allows the determination of average mass (or average molecular weight) by the sum of the products of molar fraction and mass over the relevant mass range (8). 4 Since the chance of [ 13 C]acetyl-CoA condensing to form uniformly labeled fatty acids is almost zero, the sum of individual fractions of ions in the clusters corresponding to the uniformly labeled fatty acids gives the fraction of molecules converted by chain elongation or shortening. 5 Spectral data can further be processed to provide information on the distribution of labeled mass isotopomer (m i ) and molar enrichment (ME) using the method of Lee et al. (9) that corrects for the contribution of derivatizing agent and 13 C natural abundance to the mass isotopomer distribution of the compound of interest. The resultant mass isotopomer distribution represents the fraction of molecules containing 0, 1, 2, 3, . . . 13 C substitutions and is expressed as a fraction of the total number of molecules. The observed number of 13 C atoms incorporated per molecule is the ME. ME is the stable isotope counterpart of specific activity of radioisotopes and is expressed in units of atoms of isotope per molecule. It is calculated from the mass isotopomer distribution (m i ) by the formula ⌺ m i ϫ n i (10), where n i is the number of isotope substitutions of the isotopomer species. Data reduction and regression analyses were performed using the computer software Excel® (version 5.0).
Determination of Precursor Enrichment and de Novo Synthesis-Finally, the enrichment of the acetyl-CoA pool from ␤-oxidation of uniformly labeled fatty acids can be determined from the distribution of mass isotopomers of palmitate. De novo synthesis of palmitate produces palmitate with 2, 4, or 6 13 C atoms (m 2 , m 4 , and m 6 ). The distribution of these mass isotopomers has previously been shown to conform to that of a binomial distribution (10,11). Thus, the acetyl-CoA enrichment can be obtained from the consecutive mass isotopomer ratio m 4 /m 2 according to the formula: m 4 /m 2 ϭ (n Ϫ 1)/2 p/q or 3.5 p/q, where n is 8, the number of acetyl units in palmitate, p is the enrichment of [1,2-13 C]acetyl-CoA, and q the unenriched acetyl-CoA. Once the precursor enrichment is determined, fractional synthesis can be calculated by dividing the observed to the predicted mass isotopomer (m 2 ) fraction (10).

Enrichment of Labeled Precursors-
The spectra of uniformly labeled and "unlabeled" stearate and oleate are shown in Fig. 1. The molecular ion for saturated fatty acid is the methyl ester itself. The distribution of m ϩ 1, m ϩ 2, . . . isotopomers follows that of a binomial distribution reflecting the natural abundance of 13 C. The spectrum of oleate does not follow that of a simple binomial distribution. The spectrum of oleate is more complex than that of stearate showing the loss of -OCH 3 and -HOCH 3 6 from the methyl group. The molecular ions of stea-2 A molecule is "unlabeled" when it is made up of atoms of natural isotopes, i.e. with 13 C, 2 H, and 18 O in their natural abundances. "Unlabeled" and "unenriched" are used interchangeably. 3 The common practice is to normalize the intensities of all peaks to the base peak or the peak with the most intensity, which is set to 100%. 4 Fig. 5. The presence of m ϩ 18 suggests that the substrate for chain elongation can be C16:0, C18:0, C20:0, and C22:0 creating mixed products with different degrees of labeling. The isotopomer pattern due to incorporation of labeled acetyl units in arachidate and behenate could not be explained by a simple chain elongation model.
Effects of Troglitazone on Fatty Acid Metabolism-The contributions of uptake of labeled fatty acids and their conversion by chain shortening/elongation to the fatty acids of HepG2 cells are shown in Fig. 6. When HepG2 cells were supplied with 700 M of stearate or oleate, the uptake of medium fatty acids accounted for 93 and 84% of the cellular stearate and oleate suggesting suppression of de novo synthesis of these fatty acids. The incorporation of medium fatty acids was not affected by troglitazone (Fig. 6A). Under the same incubation medium, 16% of the palmitate (C16:0) was derived from [U-13 C]stearate and 26% of C16:1 from [U-13 C]oleate by chain shortening. Such contributions were significantly increased by troglitazone to 23.6 and 36.6%, respectively (p Ͻ 0.001) (Fig. 6B). Chain elongation was much more active than chain shortening when cells were supplied with additional stearate. Chain elongation accounted for 88.3% of arachidate and 63.7% of behenate molecules. These fractions were reduced in the presence of troglitazone to 84.0% and 54.2% respectively (p Ͻ 0.001).
Desaturation of stearate contributed 67% of the oleate (Fig.  7A), while reduction of oleate contributed little to stearate (2%) (Fig. 7B). The formation of C16:1 from C18:0 and that of C16:0 from C18:1 requires the obligatory step of chain shortening. The effect of troglitazone on chain shortening was thus propagated onto the differences in the contribution of desaturation/ reduction of these compounds between control and troglitazone treatment. The interconversions of C18:0 and C18:1 by desatu- ration/reduction were not affected by troglitazone.
The ␤-oxidation of labeled fatty acids generate labeled acetyl-CoA, which can be recycled in de novo lipogenesis or chain elongation. In the case of de novo synthesis of palmitate, the incorporation of labeled acetyl-CoA produced isotopomers with even number of 13 C atoms (m 2 , m 4 , and m 6 ) ( Table I). From the consecutive mass isotopomer ratio, we determined the precursor enrichment of acetyl-CoA to be 6.15% (i.e. 6.15% of the acetyl units were [1,[2][3][4][5][6][7][8][9][10][11][12][13] C]acetyl-CoA) from the oxidation of [U-13 C]stearate and 4.12% from [U-13 C]oleate. The precursor enrichments were almost doubled under the influence of troglitazone to 10.5 and 10.4%, respectively (p Ͻ 0.001). It should be added that precursor enrichment can also be determined from m ϩ 20 to m ϩ 18 ratios in arachidate. Since these isotopomers are derived from the addition of one unlabeled or labeled acetyl unit. This ratio is an approximation of tracer/ tracee (p/q) ratio of the acetyl-units for chain elongation. We found that the precursor enrichments as determined from arachidate were much higher than those estimated from the palmitate isotopomer ratios (Fig. 5 and Table I) suggesting that chain elongation and de novo lipogenesis may have different precursor pools within the cells. Despite the increased in precursor enrichment, the total enrichments in palmitate under troglitazone treatment were less than those of palmitate from the untreated cells. Thus, the fraction of palmitate (not counting those from chain shortening) from de novo synthesis was decreased significantly from 41.25-33.18% to 22.0 -15.3% by troglitazone treatment. Allowing for the contribution from chain shortening, de novo lipogenesis contributed to 34.75% of the total palmitate pool when cells were incubated with 0. 7 mM stearate, and 16.28% when incubated with 0.7 mM oleate. Troglitazone reduced these fractional contributions to 25.35 and 9.7%, respectively.

DISCUSSION
Supplementation of diet with fatty acids is known to regulate endogenous synthesis of lipids. Transcriptional regulation of lipogenic enzymes, hepatic fatty acid synthase, malic enzyme, and glucose-6-phosphate dehydrogenase, in hepatocytes by fatty acids has been extensively studied (12)(13)(14). The effects on enzymes of de novo lipogenesis appeared to be specific for polyunsaturated fatty acids of the n-6 and n-3 families, and saturated and monunsaturated fatty acids did not have the same inhibitory effects on these lipogenic enzymes (15). Fatty acids can also be recycled by chain shortening/elongation of other fatty acids. The existence of such a fatty acid cycling system has been well documented for essential fatty acid metabolism. Fatty acid interconversion requires the participation of peroxisomal, microsomal and endoplasmic reticulum systems involving enzymes of ␤-oxidation, desaturases, isomerases and reductases. The function of some of these enzymes are often inferred from their precursor-product relationship, and the substrate specificity of these enzymes have not been well characterized (2). We demonstrated here that such a system also exists for the non-essential fatty acids. However, the enzymes involved in the reactions of the cycling of nonessential fatty acids remain to be elucidated. The present study exam- FIG. 4. Mass spectra of methyl esters of palmitate, palmitoleate, and stearate isolated from HepG2 cells incubated with [U-13 C]oleate. The base peaks for palmitate, palmitoleate, and stearate were m/z 270, 236, and 298, respectively. There was little formation of palmitate from oleate, which requires chain shortening and reduction. A substantial percent of palmitoleate molecules with mass shift of 16 daltons (m ϩ 16) was produced by chain shortening of [U-13 C]oleate. The lack of reductase action was also seen in the conversion of oleate to stearate (bottom panel). ined the impact of stearate and oleate supplementation on fatty acid cycling and de novo synthesis of nonessential fatty acids. With supplementation of 0.7 mM of stearate or oleate, we observed a number of very specific effects on the synthesis of a series of C16, C18, C20, and C22 fatty acids. When stearate and oleate were provided in relative excess, the uptake of these fatty acids accounted for over 80% of these fatty acids found in HepG2 cells. The conversion of these fatty acids by chain shortening and desaturation were also important in the production of palmitate and palmitoleate and oleate. By inference, de novo lipogenesis of these fatty acids was suppressed. De novo lipogenesis of palmitate in HepG2 cells was previously shown to be   13 C]oleate were used to determine precursor enrichment and fractional synthesis of palmitate. The ion clusters around m/z 270 corresponding to unlabeled palmitate and partially labeled palmitate were first processed to give mass isotopomer distribution due to 13 C incorporation. The (m ϩ 16) ion clusters corresponding to the uptake of uniformly labeled fatty acids were not used. The process corrects for the natural abundance of 13 C and the contribution of the methyl group giving m 0 representing the unenriched palmitate (12). The m 0 fraction of natural palmitate is 0. Results are presented as molar fractions and values are means and standard deviations of triplicate incubations. De novo synthesis of palmitate resulted in the formation of even-numbered isotopomers (m 2 and m 4 ). Almost no odd-numbered isotopomers were detected, suggesting very little recycling through the citric acid cycle. 13  80% for the same period of incubation (5). When supplied with stearate and oleate, de novo lipogenesis was suppressed to about 40%. Troglitazone is a thiazolidinedione compound which is known to bind with the PPAR␥. The binding of troglitazone to PPAR␥ stimulates peroxisome proliferation, and induces the expression of a number of genes regulated by the peroxisome proliferator response elements (16,17). Among these genes are the acyl-CoA oxidase and dihydroxyacetone phosphate acyl transferase, which are enzymes of lipid oxidation and biosynthesis (18,19). The effect of troglitazone on lipid metabolism was previously studied in hepatocytes isolated from troglitazone treated rats (20,21). Mitochondrial ␤-oxidation as measured by the release of radioactive CO 2 or the release of acid soluble product (ketone bodies) from 14 C-labeled palmitate or oleate was reduced by troglitazone treatment. However, conflicting observations were reported by Shimabukuro et al. (22) showing an increase in mitochondrial ␤-oxidation of [ 3 H]palmitate in islets of troglitazone treated Zucker diabetic fatty rats. The discrepancies in these observations may be attributed to the difference in the isotopic methods used or in tissue specific responses. Troglitazone was shown to noncompetitively inhibit mitochondrial and microsomal acyl-CoA synthase of rat hepatocytes (20) and decrease the mRNA of glycerol-3-phosphate acyltransferase and acyl-CoA synthase mRNA content in islets of Zucker diabetic fatty rats (21). Subsequently, the esterification of fatty acid to triglycerides and the triglyceride content in cells were inhibited by troglitazone treatment. In the present study, we showed that troglitazone increased chain shortening of C18 fatty acids thus peroxisomal ␤-oxidation of these fatty acids in HepG2 cells in culture. The increased ␤-oxidation resulted in higher 13 C enrichment of the acetyl precursor pools both for de novo lipogenesis and chain elongation. However, de novo lipogenesis of palmitate as well as the synthesis of arachidate and behenate by chain elongation were significantly inhibited by troglitazone treatment. The action of stearoyl-CoA desaturase activity contributed significantly to the formation of mono-unsaturated fatty acids from palmitate and stearate. The desaturation of saturated fatty acids to mono-unsaturated fatty acids was not affected by troglitazone.
Mitochondria and peroxisomes are the main cellular systems for fatty acids ␤-oxidation. Mitochondrial ␤-oxidation which is coupled to oxidative phosphorylation results in the complete breakdown of the fatty acid to acetyl-CoA, CO 2 , high energy phosphate bonds and reducing equivalents. Peroxisomal ␤-oxidation system on the other hand is not coupled to oxidative phosphorylation, and produces acetyl-CoA and the fatty acids of shorter chain lengths. The action of peroxisomal ␤-oxidation system is responsible for chain shortening/elongation of polyunsaturated fatty acids, and for the recycling of acetyl (2carbon) units and essential fatty acids in the homeostasis of these essential fatty acids (2). The substantial amount of palmitate and palmitoleate formed by chain shortening in our experiments suggests that the peroxisomal system also plays a significant role in the ␤-oxidation of long chain saturated and mono-unsaturated fatty acids, just as in the case of the very long chain fatty acid (23). This process was stimulated by the peroxisome proliferator troglitazone.
Chain shortening/elongation by peroxisomal ␤-oxidation characteristically allows energy to be stored or used without significant changes in the number of fatty acid molecules. Furthermore, chain shortening/elongation is a less "futile" process involving the recycling of 1-3 acetyl units as compared with eight via mitochondrial ␤-oxidation and de novo synthesis of palmitate. Peroxisome ␤-oxidation is potentially an energy saving system for the interconversion of fatty acids needed for membrane lipid synthesis. It is conceivable that these effects of troglitazone on oxidation, interconversion, and synthesis saturated fatty acids play a major role in cellular energy metabolism, and membrane lipid composition and turnover. However, the mechanism relating these effects to its "insulin-sensitizing" effect in the treatment of non-insulin dependent diabetes is not known and deserves further investigation.