Unsaturated fatty acid regulation of peroxisome proliferator-activated receptor alpha activity in rat primary hepatocytes.

Peroxisome proliferator-activated receptors (PPARs alpha, beta, gamma1, and gamma2) are widely regarded as monitors of intracellular nonesterified fatty acid (NEFA) levels. As such, fatty acid binding to PPAR leads to changes in the transcription of many genes involved in lipid metabolism and storage. Although the composition of the intracellular NEFA pool is likely an important factor controlling PPAR activity, little information is available on factors affecting its composition. Accordingly, we have examined the effects of exogenous fatty acids on PPARalpha activity and NEFA pool composition in rat primary hepatocytes. Prior to the addition of fatty acids to primary hepatocytes, nonesterified unsaturated fatty acid levels are very low, representing </=0.5% of the total fatty acid in the cell. The relative abundance of putative PPARalpha ligands in the NEFA pool is 20:4n-6 = 18:2n-6 = 18:1n-9 > 22:6n-3 > 18:3n-3/6 = 20:5n-3. Of these fatty acids, only 20:5n-3 and 22:6n-3 consistently induced PPARalpha activity. Metabolic labeling of primary hepatocytes indicated that both 14C-18:1n-9 and 14C-20:5n-3 are rapidly assimilated into neutral and polar lipids. Although the addition of 18:1n-9 had no effect on NEFA pool composition, 20:5n-3 mass increased >15-fold within 90 min. Changes in NEFA pool 20:5n-3 mass correlated with dynamic changes in the PPARalpha-regulated transcript mRNACYP4A. Metabolic labeling also indicated that a significant fraction of 14C-20:5n-3 was elongated to 22:5n-3. Cells treated with 22:5n-3 or 22:6n-3 led to a significant accumulation of 20:5n-3 in the NEFA pool through a process that requires peroxisomal beta-oxidation and fatty acyl CoA thioesterase activity. Further analyses suggest that 20:5n-3 and 22:6n-3, but not 22:5n-3, are active ligands for PPARalpha. These studies suggest that basal fatty acid levels in the NEFA pool coupled with rates of fatty acid esterification, elongation, desaturation, peroxisomal beta-oxidation, and fatty acyl thioestease activity are important determinants controlling NEFA pool composition and PPARalpha activity.

esterified fatty acids (NEFA) levels. As such, fatty acid binding to PPAR leads to changes in the transcription of many genes involved in lipid metabolism and storage (1,2). Fatty acids enter cells through transporters, e.g. fatty acid transport proteins 2-5, and are bound by binding proteins (FABP), which may play a role in directing fatty acids to various intracellular compartments for metabolism and gene expression (3)(4)(5). Esterification of fatty acids into triglycerides, polar lipids, and cholesterol esters and their ␤-oxidation (mitochondrial and peroxisomal) requires conversion of fatty acids to fatty acyl CoA thioesters (6). Other pathways, like microsomal NADPH-dependent mono-oxidation and eicosanoid synthesis, utilize nonesterified fatty acids as substrates. These reactions are likely to influence cellular levels of activating ligands. In the case of PPAR␣ and PPAR␥, formation of eicosanoids may be important routes for receptor activation (2,(7)(8)(9).
Because PPARs are known to bind nonesterified fatty acids (1), it is reasonable to expect that the composition of the intracellular NEFA pool is an important determinant in the control of PPAR activity. The composition of the intracellular NEFA pool is affected by the concentration of exogenous fatty acids entering cells, their rate of removal via acyl CoA-synthetasedependent and -independent mechanisms, e.g. microsomal monooxygenation, and the return of NEFA or oxidized lipids to the NEFA pool as a result of lipid metabolism. In addition, fatty acid structure may also be an important determinant. For instance, 18-and 20-carbon N6-PUFA, but not N3-PUFA, are preferred substrates for cyclooxygenase-and lipoxygenase-dependent eicosanoid synthesis (10,11); Ն20-carbon PUFAs are poor substrates for cholesterol ester formation (12); 20:5n-3 is a poor substrate for diacylglycerol acyl transferase (13), the terminal step in formation of triglycerides; and Ͻ14and Ͼ20carbon fatty acids bind PPAR poorly (1). Hepatic parenchymal cells have little or no cyclooxygenase or lipoxygenase activity but have robust NADPH-dependent CYP monooxygenase activity (14). Certain acyl CoA synthetases are reported to display fatty acyl chain length selectivity (6). Clearly, the factors contributing to the composition of the intracellular NEFA pool are complex. Added to this, drugs, disease, and genetic background are likely to affect lipid metabolism and NEFA pool composition, which in turn affects fatty acid-regulated nuclear receptor activity.
In an effort to better understand how cellular metabolism contributes to the control of fatty acid-regulated transcription factor activity, we have taken advantage of previous findings for the fatty acid regulation of PPAR␣ in primary rat hepatocytes. In vivo feeding studies have shown that when calories supplied as saturated and monounsaturated fat are Յ20% of total calories, hepatic PPAR␣-regulated transcripts, like acyl CoA oxidase and cytochrome P-450 -4A (CYP4A) are marginally affected (15,16). Substituting fish oil, a rich source of 20and 22-carbon N3-PUFA, leads to a pronounced induction of enzymes involved in lipid oxidation. Eicosapentaenoic acid (20: 5n-3), a known PPAR␣ ligand (1), induces acyl CoA oxidase and CYP4A mRNA levels in primary hepatocytes and activates a PPAR␣-reporter gene. Moreover, PUFA induction of these transcripts requires a functional PPAR␣. However, other PPAR␣ ligands, like 18:1n-9 and 20:4n-6, have little effect on PPAR␣ activity or PPAR␣-regulated genes in cultured primary hepatocytes (16).
The differential effects of putative fatty acid ligands on PPAR␣ activity and hepatic gene expression raises the question of the role hepatic metabolism plays in the control of PPAR␣ activity. Accordingly, we have carried out a detailed analysis of fatty acid regulation of PPAR␣ as well as fatty acid metabolism in primary hepatocytes. These studies will document dynamic changes in the intracellular NEFA pool composition and effects on PPAR␣-regulated hepatic gene expression. In addition, our studies will provide evidence for ␤-oxidation products of 22-carbon fatty acids in the NEFA pool and their effects on PPAR␣ activity.

Primary Hepatocytes and Transfections-Male
Sprague-Dawley rats were maintained on a Tek-Lad chow diet ad lib and were used for primary hepatocyte preparation (17). For metabolic labeling and RNA studies, the cells were plated onto 50-or 100-mm Primaria plastic dishes at 3 ϫ 10 6 or 10 7 cells/plate, respectively, in Williams E with 10 mM lactate, 10 nM dexamethasone, 100 nM insulin, and 10% fetal calf serum. After a 4 -6-h attachment period, the medium was changed to a serum-free medium, Williams E with 10 mM lactate, 10 nM dexamethasone, and 100 nM insulin.
For transfection studies, the cells were plated in the same medium onto 6-well Primaria plastic dishes at 10 6 cells/well. The cells were transfected in this serum-free medium using Lipofectin (6 l/g DNA) or LipofectAMINE 2000 (1.3 l/g DNA) (Invitrogen) as described (14,18,19). pM-rPPAR␣-LBD and TKMH100x4-Luc were previously described (20). pM-rPPAR␣-LBD is a fusion of the Gal4-DNA-binding domain fused to the ligand-binding domain of rPPAR␣. The TK-MH100x4-Luc reporter contains four binding sites for the Gal4-DNAbinding domain. phRG-Luc was obtained from Promega (Madison, WI) and served as a control for transfection efficiency as well as nonspecific effects on promoter activity.
The medium was changed the next morning to Williams E with 25 mM glucose, 10 nM dexamethasone, 100 nM insulin, fatty acid (NuChek Prep, Elysian, MN), and bovine serum albumin (BSA to fatty acid ratio was 1:5) or the PPAR␣ agonist, WY14,643 (Chemsyn Science Laboratories, Lenexa, KS). After 24 h of treatment, the cells were harvested for luciferase assays. Each treatment involves triplicate samples, and each study was repeated at least twice. The results are expressed as relative luciferase activity: firefly luciferase activity/Renilla luciferase activity.
RNA Isolation and Northern Analysis-Primary hepatocytes were plated onto 100-mm primary plates at 10 7 cells/plate and treated as described above. RNA was extracted from rat primary hepatocytes using Triazol (Invitrogen) and separated electrophoretically in denaturating agarose gels, transferred to nitrocellulose, and probed with 32 P-cDNA for CYP4A (16,19). Levels of hybridization were quantified using a Molecular Dynamics Phosphoimager 820 (Amersham Biosciences, Piscataway, NJ).
Fatty Acid Metabolism-Primary hepatocytes were plated in the same medium as described above but onto 50-mm Primaria plastic dishes at 3 ϫ 10 6 cells/plate. The ratio of culture medium to cell number was maintained constant for the different plating conditions. The cells were treated with different fatty acids (see figure legends for details) essentially as described for gene expression studies. The cells were incubated overnight in serum-free medium prior to fatty acid treatment. For metabolic labeling studies, the cells were treated with 14 Clabeled fatty acids in 3 ml of medium containing 250 M fatty acid (0.5 Ci, 1.7 Ci/mol) (20). 14 C-Labeled fatty acids were purchased from (PerkinElmer Life Sciences). Nonradioactive fatty acids were pur-chased from NuChek Prep. Fatty acid-free BSA (Roche Applied Science) was included at a ratio of fatty acid to BSA of 5:1.
After fatty acid treatment, the medium was collected, the cells were washed one time with phosphate-buffered saline and 40 M BSA, washed one time with phosphate-buffered saline, and resuspended in 500 l of 40% methanol (20). This washing method minimizes contamination of cellular lipids with unincorporated free fatty acids. The methanol extract of cells was acidified with HCl to 0.25 N, and lipids were extracted with chloroform:methanol (2:1) containing 1 mM butylated hydroxytoluene (BHT). The protein and aqueous phases were re-extracted with chloroform. The organic phases were pooled, dried under nitrogen, resuspended in chloroform and 1 mM BHT, and stored at Ϫ80°C. 14 C-Labeled lipid extracts were further fractionated by thin layer chromatography (LK6D Silica G 60A; Whatman) and developed in hexane:diethyl ether:acetic acid (90:30:1). The distribution of 14 C-fatty acids in various lipid fractions was visualized by exposure of the TLC to a phosphorimaging screen (Amersham Biosciences), and the levels of radioactivity were quantified. The location of lipids was compared with authentic standards for triacylglycerol, diacylglycerol, cholesterol ester, fatty acids, fatty acid (wax) esters, and glycerol-and sphino-phospholipids (Avanti Polar Lipids). The uptake of 14 C-fatty acids into cells and the organic fraction was quantified by scintillation counting. The depletion of 14 C-labeled fatty acids from medium was quantified by scintillation counting and TLC followed by phosphorimaging analysis as described above.
The NEFA fraction in total cellular lipids was fractionated on aminopropyl columns (Alltech Associates, Deerfield, IL) (20). Lipid extracts in chloroform and 1 mM BHT were applied to amino-propyl columns (100 mg) and washed extensively with chloroform:isopropanol (2:1) to remove neutral lipids. NEFA were eluted with diethyl ether and 2% acetic acid. Phospholipids were retained on the column. The diethyl ether, 2% acetic acid fraction was dried under nitrogen, resuspended in methanol and 10 M BHT, and used directly for RP-HPLC fractionation and quantification of unsaturated fatty acids. The fractional recovery of NEFA from whole cell lipid extracts was Ն95%.
RP-HPLC Analysis of Unsaturated Fatty Acids-The total extracted lipids were saponified (0.4 N KOH in 80% methanol for 1 h at 50°C), neutralized, extracted in diethyl ether and 1% acetic acid, dried, and resuspended in methanol and 0.1 mM BHT for RP-HPLC analysis (reverse phase C18 column; Symmetry Shield, 2487 UV detector set to 192 nm with a 600 Controller; Waters Corp., Milford, MA). A linear gradient of 22 to 100% acetonitrile and 0.1% acetic acid over 40 min was used to fractionate unsaturated fatty acids (20). Verification and quantification of unsaturated fatty acids by RP-HPLC used authentic fatty acid standards (NuChek Prep) and Win-flow Radio HPLC software (IN/US Systems, Inc, Tampa, FL). The identity of specific fatty acids was verified by gas chromatography/mass spectrometry at the mass spectrometry facility at Michigan State University.

RESULTS
Effect of Unsaturated Fatty Acids on PPAR␣ Activity in Primary Hepatocytes-To examine the effect of unsaturated fatty acids on PPAR␣ activity in primary hepatocytes, the cells were transfected with a chimeric receptor composed of the PPAR␣ ligand-binding domain (LBD) fused to a Gal4-DNA-binding domain. Co-transfection with the TKMH100x4Luc reporter allows for an examination of changes in PPAR␣ activity without recruitment of its heterodimer partner, retinoid X receptor, a prospective target for fatty acid control (21).
A dose response analysis reveals the difference in effect of two prospective ligands on PPAR␣ activity (Fig. 1B). Increasing the dose of 20:5n-3 significantly induced PPAR␣ activity (ϳ8fold at 1 mM), whereas a comparable dose of 18:1n-9 had no effect. These studies confirm previous findings on the differen-tial effects of these putative ligands on PPAR␣ activity in primary hepatocytes (15,16). These studies also indicate that the difference in effect of these two fatty acids on PPAR␣ activity cannot be explained on the basis of treatment dose alone. In addition, by using the Gal4-PPAR␣-LBD chimeric receptor, these studies indicate that 22-carbon PUFAs are activators of PPAR␣ in primary hepatocytes and exclude a role for retinoid X receptor heterodimerization for this action.
In the studies to follow, we will first examine the metabolic basis for the differential effect of 18:1n-9 and 20:5n-3 on PPAR␣ activity. Because structural studies suggest that fatty acids with Ͼ20 carbons are poor ligands (1), the second part of the study will determine whether the 22-carbon PUFA effect on PPAR␣ activity is due to its conversion to an active ligand, e.g. 20:5n-3.
Analysis of the Hepatocellular Unsaturated Fatty Acid Mass prior to Fatty Acid Treatment-Numerous reports with a variety of cell lines indicate that a broad array of fatty acids affect PPAR activity (2). In fact, our studies with the rat hepatoma cell line, FTO2B, indicates that, with the exception of 22:5n-3, all of the fatty acids examined in Fig. 1 induce PPAR␣ Ն2-fold. 2 We hypothesized that the difference in fatty acid responsiveness of rat primary hepatocytes and established cell lines is related to the distribution of fatty acids in the NEFA pool. Accordingly, the mass of hepatocellular fatty acids in the total saponified lipid fraction and in the NEFA pool was examined. Our analysis focused only on unsaturated fatty acids (Fig. 2).
For comparison, 18:1n-9 is the predominant fatty acid in the total saponified (480 nmol/mg protein) and NEFA (2.2 nmol/mg protein) fractions in FTO2B cells. All other unsaturated fatty acids are very low: Ͻ50 nmol/mg protein in the saponified fraction and Ͻ0.2 nmol/mg protein in the NEFA fraction. 2 The profile of fatty acids in FTO2B cells parallels the distribution of fatty acids in the fetal calf serum used to maintain the cells. Thus, FTO2B cells display a very different fatty acid profile from primary hepatocytes.
Metabolic Labeling of Primary Hepatocytes-Because of the striking difference in effect of 18:1n-9 and 20:5n-3 on PPAR␣ activity and the Ͼ100-fold difference in mass of 18:1n-9 and Primary rat hepatocytes were transfected with pMN-PPAR␣-LBD and MH-TK-Luc; phRG-luc was used to correct for transfection efficiency and to assess nonspecific effects of treatments on promoter activity. A, effect of various unsaturated fatty acids on PPAR␣ activity in primary hepatocytes. After an overnight transfection period, the cells were treated without or with various fatty acids at 250 M each for 24 h. The cells were harvested for protein and luciferase assays. The results are reported as the relative luciferase activity (RLA, firefly luciferase activity/Renilla luciferase activity). The results are expressed as the means Ϯ S.D. of three separate studies with triplicate samples per group. The insert for PPAR␣ binding data is taken from Ref. (1). B, dose response of 18:1n-9 and 20:5n-3 on PPAR␣ activity. Primary hepatocytes were transfected as above but treated with either 18:1n-9 or 20:5n-3 at the doses specified in the figure. The fatty acid to BSA ratio was maintained at 5:1 throughout this study. After a 24-h treatment period, the cells were harvested for luciferase assays. The results are reported as the relative luciferase activity, as described above; means Ϯ S.D., two separate experiments with triplicate samples per group. 20:5n-3, we examined the metabolism of these fatty acids in primary hepatocytes. Primary hepatocytes were treated with 14 C-labeled fatty acids at 250 M for 1.5, 6, and 24 h. Within 6 h of treatment, nearly Ն80% of each fatty acid was cleared from the medium; by 24 h of treatment essentially all exogenously added fatty acids were cleared from the medium (not shown). A difference in fatty acid clearance and assimilation into the organic extracts was observed only at the 1.5-h time point (Fig.  3); 140 and 230 nmol/mg protein of fatty acid accumulated in cells receiving 18:1n-9 and 20:5n-3, respectively. By 6 h, the mass of exogenous 18:1n-9 exceeded 20:5n-3 by ϳ15%, and by 24 h, there was no difference.
Primary hepatocytes assimilate a fraction of exogenous fatty acids into complex lipids, which are then packaged into lipoprotein particles (very low density lipoprotein) and released to the medium. The amount of exogenous fatty acid appearing in the medium as triacylglycerol and cholesterol ester at the 1.5-and 6-h time points was not different between the two fatty acids and represented 1 and 5% of the total exogenous fatty acid added to hepatocytes. By 24 h, levels of 14 C-fatty acid appearing in complex lipids increased to 15% for 18:1n-9-treated cells and 9% for 20:5n-3-treated cells (not shown). Eicosapentaenoic acid (20:5n-3) has a modest repressive effect on release of complex lipids from primary hepatocytes, a finding consistent with the well known effect of fish oils on hepatic very low density lipoprotein production and secretion (25,26).
Because intracellular lipids are the likely regulators of PPARs, we focused on the distribution of exogenous fatty acids in various intracellular lipid fractions. The 14 C-fatty acids were distributed to five fractions: triacylglycerols (ϳ80%) Ͼ polar lipids (ϳ10 -15%) Ͼ cholesterol esters (variable) Ͼ diacylglycerol (ϳ3%) (Fig. 3) Ͼ NEFA (Fig. 4). In contrast to in vitro studies (13), these results indicate that both 18:1n-9 and 20: 5n-3 are good substrates for di-and tri-acylglycerol formation. 14 C-Fatty acids recovered as cholesterol esters increased over time. In vitro studies have suggested that 20-carbon unsatur-ated fatty acids are poor substrates for cholesterol ester synthesis (12). At the 1.5-and 6-h time points, Ͼ50% less 20:5n-3 is assimilated into cholesterol ester than 18:1n-9. By 24 h, however, this difference was not apparent. This may be due to the fact that by 24 h, 20:5n-3, but not 18:1n-9, attenuates secretion of cholesterol ester and triglycerides to the medium as very low density lipoprotein, leading to retention of cholesterol esters and triglycerides in cells.
The fraction of fatty acid recovered as NEFA was measured by two methods (Fig. 4). The TLC method examines only the level of 14 C-fatty acid recovered as NEFA and reflects the mass of the exogenous fatty acid. The HPLC method measures total mass of NEFA. The approach allows us to describe the effect of exogenous fatty acids on NEFA pool composition and provides an indication of the flux of the fatty acid through the NEFA pool The fraction of 14 C-fatty acid recovered as NEFA is very low, representing ϳ1% of the total 14 C-fatty acid in organic extracts. The addition of 14 C-18:1n-9 to cells results in an accumulation of 18:1n-9 to ϳ1 nmol/mg protein between 1.5 and 6 h; this value was sustained up to 24 h. Interestingly, the total mass of 18:1n-9 in the NEFA pool remained unchanged (ϳ1 nmol/mg protein) throughout the 24-h treatment period. Thus, the 18: 1n-9 in the NEFA pool is composed predominantly of exogenous 18:1n-9. The lack of change in 18:1n-9 mass in the NEFA pool indicates that 18:1n-9 is rapidly esterified.
The addition of 14 C-20:5n-3 to primary hepatocytes also increased to nearly 1 nmol/mg protein by 90 min but declined to ϳ0.5 nmol/mg protein by 24 h. Like 18:1n-9, the exogenous 20:5n-3 represents the major fraction of 20:5n-3 in the NEFA fraction. In contrast to 18:1n-9, 20:5n-3 mass in the NEFA pool is very low prior to addition of fatty acids to cells. Thus, the addition of 20:5n-3 to hepatocytes leads to a 17-fold increase in mass within 90 min. Fig. 4C illustrates how fatty acid treatment perturbs the mass of unsaturated fatty acid in the NEFA pool. This figure illustrates the masses of 18:1n-9 (white bars) and 20:5n-3 (black bars) and the sum of other nonesterified unsaturated fatty acids (gray bars) in the NEFA pool. These other unsaturated fatty acids remained relatively constant at ϳ1 nmol/mg protein following the addition of either 18:1n-9 or 20:5n-3 to cells. Although the addition of 18:1n-9 does not perturb the total mass of unsaturated fatty acids, the addition of 20:5n-3 leads to an approximately 25% increase in the total nonesterified unsaturated fatty acid mass within 90 min. By 24 h, the total mass of all nonesterified unsaturated fatty acids declines to pretreatment levels (ϳ2 nmol/mg protein) by 24 h.
The total mass of 20:5n-3 incorporated into the saponified lipid fraction increased from 5.5 to ϳ347 nmol/mg protein within 1.5 h of addition of 20:5n-3 to cells (not shown). The mass increase of 20:5n-3 in the NEFA pool represents ϳ0.28% of the total cellular 20:5n-3. This fraction is comparable with the level of 20:5n-3 in the NEFA pool relative to total cellular 20:5n-3 prior to fatty acid treatment (Fig. 2). Clearly, the major fraction of cellular 20:5n-3 is esterified and assimilated into complex lipids (Fig. 3). Because 18:1n-9 did not increase in the NEFA pool argues against contamination of these extracts with medium lipids. Whether the apparent enrichment of the NEFA pool is due to different rates of fatty acyl CoA formation or the return of 20:5n-3 to the NEFA pool by metabolic events remains unresolved.

Mass of 20:5n-3 in the NEFA Pool, and Not the Formation of Eicosanoids, Correlates with Induction of the PPAR␣-regulated
Transcript, mRNA CYP4A -To determine whether the changes in intracellular nonesterified 20:5n-3 (Fig. 4) correlated with effects on gene expression, we measured mRNA CYP4A , a  Fig. 1. BSA was included at 50 M. Primary hepatocytes were plated on to 50-mm Primaria plates with 3 ϫ 10 6 cells/plate and received 3 ml of medium. The ratio of medium and fatty acids to cells was identical to that used in Fig. 1. At the times indicated, the cells were harvested for protein determination and lipid extraction. The lipid extracts were fractionated by TLC and developed in hexane: diethyl ether: acetic acid (90:30:1). After separation, the TLC plates were dried and exposed to a phosphorimaging screen, and the levels of radioactivity were quantified. This method provides a measure of 14 Cfatty acid, based on the specific activity of the fatty acid used for treatment. 14  PPAR␣-regulated transcript (15,16). Addition of 20:5n-3 to primary hepatocytes induced a prompt rise (2-fold within 6 h) in mRNA CYP4A , followed by a 50% drop by 24 h (Fig. 5). The 20:5n-3-mediated induction of mRNA CYP4A was blocked by the inhibitor of transcription, actinomycin D (not shown). The decline in mRNA CYP4A after the 6-h time point likely represents diminished stimulation of transcription coupled with enhanced mRNA CYP4A turnover. Vehicle and 18:1n-9-treated cells displayed a decline in mRNA CYP4A over the 24-h treatment period. The profile of the 20:5n-3 effect on mRNA CYP4A parallels the change in intracellular nonesterified 20:5n-3 (Fig. 4, B and C). Overall, the net gain in mRNA CYP4A following 20:5n-3 addition is ϳ2-fold, which is comparable with the change in CYP4A protein (not shown). Taken together, these results reveal dynamic changes in 20:5n-3 within the NEFA pool, which correlate well to PPAR␣-regulated hepatic gene transcription.
Eicosanoids, like leukotriene B4, have been reported to be PPAR␣ ligands (2,(7)(8)(9). Previous studies suggested that inhibitors of cyclooxygenase and lipoxygenase had little impact on fatty acid-regulated hepatocyte gene expression (14). The addition of 20:5n-3 to primary hepatocytes leads to changes in levels of oxidized lipids in the NEFA pool (not shown). NADPHdependent microsomal fatty acid oxidation represents a likely route for the generation of these oxidized lipids. To determine whether this pathway contributes to the 20:5n-3 control of gene expression, hepatocytes were treated with the inhibitor of microsomal oxidation, diethyldithiocarbamate (DDC) (27). Studies with isolated rat hepatic microsomal preparations indicated that DDC is a robust inhibitor of NADPH-dependent oxidation of both 20:4n-6 and 20:5n-3. 2 Arachidonic acid typically has a  4. Change of 18:1n-9 and 20:5n-3 in the NEFA pool following fatty acid challenge. Quantitation of 18:1n-9 (A) and 20:5n-3 (B) mass in the NEFA lipid fraction following treatment of primary hepatocytes with 14 C-18:1n-9 or 14 C-20:5n-3. The hepatocytes were treated with 18:1n-9 or 20:5n-3, the total lipids were separated by TLC, and the NEFA fraction was quantified as described in the legend to Fig. 3 (closed circles, solid line). This method quantifies the mass of the exogenous ( 14 C-fatty acid). NEFA were also quantified by first fractionating total lipids on aminopropyl columns followed by RP-HPLC (closed squares, dashed line). The results are expressed as fatty acid mass (nmol/mg protein) and are representative of two separate studies. The mass levels of NEFA prior to fatty acid treatment are shown in marginal effect on mRNA CYP4A in primary hepatocytes (15,16) and PPAR␣ activity (Fig. 1). The combined treatment of 20: 4n-6 and DDC induced CYP4A mRNA ϳ50% (Fig. 6). Treatment with 20:5n-3 alone induced the mRNA CYP4A ϳ2.5; with DDC added, CYP4A mRNA increased nearly 4-fold. Similar effects were seen with 1-aminobenozotriazole, another inhibitor of microsomal oxidation (not shown). Thus, interference with microsomal fatty acid metabolism is sufficient to impact levels of this PPAR␣ regulated transcript. These studies suggest that the generation of eicosanoids is not required for the 20:5n-3-mediated activation of PPAR␣ in primary hepatocytes. In addition, these results also indicate that microsomal 20:4n-6 and 20:5n-3 metabolism may be important for degrading PPAR␣ ligands in liver.

20-Carbon, but Not 18-Carbon, Unsaturated Fatty Acids Are Elongated in Cultured Primary
Hepatocytes-In addition to the assimilation of exogenous fatty acids into complex lipids and oxidation, fatty acyl CoA thioesters serve as substrates for malonyl CoA-dependent elongation and NADPH-dependent desaturation. These modified fatty acids then serve as substrates for esterification into complex lipids or ␤-oxidation in mitochondria and peroxisomes. To assess these transformations, saponified lipids from the 14 C-fatty acid labeling studies were fractionated by RP-HPLC (Fig. 7). As 14 C-18:1n-9 accumulates in cells, Ͻ3% of 14 C-18:1n-9 is elongated to a 20-carbon fatty acid. 18:2n-6 and 18:3n-3 were also poorly elongated to 20-carbon fatty acids in primary hepatocytes (not shown). Thus, Ͼ97% of the 18-carbon fatty acid supplied to hepatocytes enters complex lipids as the 18-carbon fatty acid.
In contrast, 14 C-20:5n-3 is elongated to 22:5n-3 in primary hepatocytes over a 24-h period (Fig. 7B). The fraction accumulating as 22:5n-3 increases linearly with time, reaching ϳ30% of the total 14 C-fatty acid in cells by 24 h. Similar studies with 20:4n-6 revealed ϳ25% of its conversion to 22:4n-6 (not shown). Some studies have indicated that 22:5n-3 is elongated to 24: 5n-3, but its appearance is transient, reflecting ␤-oxidation. However, no evidence of desaturation of any 18-or 20-carbon fatty acid was detected in these metabolic labeling studies. Thus, 14 C-22:6n-3 is not generated in primary hepatocytes treated with 14 C-18:3n-3 (not shown) or 14 C-20:5n-3 (Fig. 7B). Consistent with this observation is the lack of increase in total cellular 22:6n-3 mass following 20:5n-3 administration to primary hepatocytes. The failure to generate 22:6n-3 from 18-and 20-carbon precursors is likely due to a decline in ⌬ 5 and ⌬ 6desaturase activity when liver is explanted for primary hepatocyte culture. The reason for this decline is unknown.
22-Carbon unsaturated fatty acids have been reported to undergo peroxisomal ␤-oxidation to form 20-carbon fatty acyl CoA thioesters through a process called retroconversion (28,29). These products of ␤-oxidation are typically esterified into neutral and polar lipids (29). We considered the possibility that the differential effect of 22:5n-3 and 22:6n-3 on PPAR␣ activity was due to 22:6n-3 being a preferred substrate for the generation of 20:5n-3 in the NEFA pool. This metabolic scheme is illustrated in Fig. 9A. Accordingly, the levels of 20:5n-3, 22: 5n-3, and 22:6n-3 in the total saponified lipid and NEFA fractions were measured (Fig. 9, B and C). The basal level of 20:5n-3 in the total saponified lipid fraction was 10 nmol/mg protein, and the addition of 20:5n-3 (at 500 M) to cells increased 20:5n-3 to 397 nmol/mg protein within 6 h (Fig. 9B). thioesterase activity converts this inactive ligand to an active ligand (Fig. 9A). DISCUSSION We have examined the role hepatic metabolism plays in the control of cellular levels of certain natural ligands for PPAR␣. Although previous studies have suggested that oxidized lipids are prospective ligands for PPAR␣ and PPAR␥ (eicosanoids) (2,(7)(8)(9), such metabolic routes do not apply to hepatic parenchymal cells because of the absence of robust cyclooxygenase or lipoxygenase activity in these cells (10). In fact, our studies argue against a requirement for the generation of oxidized lipids, i.e. eicosanoids, to activate PPAR␣ in hepatocytes (Fig.  6). Instead, the presence of certain 20-and 22-carbon PUFA in the intracellular NEFA pool appear to represent major determinants controlling PPAR␣ activity. The new information reported here includes: 1) the rapidity and magnitude of change in NEFA pool composition following fatty acid challenge; 2) dynamic changes in PPAR␣ activity and mRNA CYP4A abundance following the treatment of primary hepatocytes with 18-, 20-, and 22-carbon PUFA; and 3) the identification of several biochemical reactions likely to be important for regulating cellular levels of PPAR ligands, i.e. microsomal fatty acid oxidation (Fig. 6) and elongation (Fig. 7), and peroxisomal ␤-oxidation and fatty acyl CoA thioesterase activity (Figs. 8 and 9). Together, these findings provide a biochemical basis for under-standing the differential effects of 18:1n-9, 20:5n-3, 22:5n-3, and 22:6n-3 on PPAR␣ activity in primary rat hepatocytes.
The composition of the NEFA pool is influenced not only by exogenous fatty acids but also by endogenous metabolic events.
Finding that the addition of either 22:5n-3 or 22:6n-3 to primary hepatocytes leads to the accumulation of 20:5n-3 in the NEFA pool implicates a role for two peroxisomal functions. 22and 24-carbon PUFA are preferentially ␤-oxidized in the peroxisome resulting in a reduction in chain length by 2 or 4 carbons (29). The resulting fatty acyl CoA thioesters must be hydrolyzed by a thioesterase for 20:5n-3 to appear in the NEFA pool (Fig. 9). Previous studies with acyl CoA oxidase null mice suggested that the peroxisome was important for regulating PPAR␣ ligands (30). Our studies extend this observation to include the peroxisome as a key organelle for controlling cellular levels of 20-and 22-carbon PUFA ligands for PPAR␣.
Increased abundance of 20:5n-3 and 22:6n-3 in the NEFA pool correlates well with PPAR␣ activation in primary hepatocytes. Whether this same mechanism accounts for the dietary fatty acid regulation of PPAR␣ in hepatic and nonhepatic tissues in vivo remains unresolved. Some insight into the NEFA pool composition can be obtained by analysis of fatty acids co-isolated with FABP. Native L-FABP isolated from the livers of rats maintained on standard lab chow were found to contain endogenous fatty acids of various chain lengths, i.e. C 16 -22 ; PUFA represented 44% of the total. Although L-FABP binds 20:5n-3, 22:5n-3,and 22:6n-3 with affinities ranging from ϳ50 to 250 nM (31), only 22:6n-3 was associated with L-FABP at 1.9 mol % (24). The low abundance of 20:5n-3 and 22:5n-3 relative to 22:6n-3 in liver probably accounts for this distribution of L-FABP-associated fatty acids (22,23) (Fig. 2). Interestingly, the relative distribution of L-FABP-bound fatty acids reported by Murphy et al. (24), is remarkably similar to the composition of the NEFA pool derived from male rats maintained on a TekLad chow diet (Fig. 2). Nevertheless, feeding humans or rats fish oil, a rich source of n3-PUFA, is reported to increase 20:5n-3, 22:5n-3, and 22:6n-3 in blood (32) and liver (23), respectively. Because PPAR␣ is required for n3-PUFA effects on hepatic CYP4A and acyl CoA oxidase induction (16), we predict that future studies will describe the enrichment of the NEFA pool and L-FABP with 20-and 22-carbon PUFA following fish oil feeding.
In conclusion, we have used PPAR␣ in rat primary hepatocytes as a model system to evaluate the role metabolism plays in the control of transcription factor activity. Our studies support the concept that dynamic changes in NEFA pool composition trigger PPAR␣ activation and induce PPAR␣-regulated gene transcription. Several key biochemical reactions appear to be important for controlling the hepatocyte levels of PPAR␣ ligands, including acyl CoA synthetase and thioesterase activities, fatty acid elongation, and desaturase activities as well as peroxisomal ␤-oxidation. Other nuclear receptors have been implicated as targets for fatty acid control, including hepatic nuclear factor-4 (␣ and ␥), liver X receptor ␣, retinoid-related orphan receptor, and retinoid X receptor (20,(33)(34)(35)(36). The mechanisms described here may be important for controlling cellular levels of ligands regulating these other nuclear receptors.