Fatty acid homeostasis and induction of lipid regulatory genes in skeletal muscles of peroxisome proliferator-activated receptor (PPAR) alpha knock-out mice. Evidence for compensatory regulation by PPAR delta.

Ablation of peroxisome proliferator activated receptor (PPAR) alpha, a lipid-activated transcription factor that regulates expression of beta-oxidative genes, results in profound metabolic abnormalities in liver and heart. In the present study we used PPAR alpha knockout (KO) mice to determine whether this transcription factor is essential for regulating fuel metabolism in skeletal muscle. When animals were challenged with exhaustive exercise or starvation, KO mice exhibited lower serum levels of glucose, lactate, and ketones and higher nonesterified fatty acids than wild type (WT) littermates. During exercise, KO mice exhausted earlier than WT and exhibited greater rates of glycogen depletion in liver but not skeletal muscle. Fatty acid oxidative capacity was similar between muscles of WT and KO when animals were fed and only 28% lower in KO muscles when animals were starved. Exercise-induced regulation and starvation-induced regulation of pyruvate-dehydrogenase kinase 4 and uncoupling protein 3, two classical and robustly responsive PPAR alpha target genes, were similar between WT and KO in skeletal muscle but markedly different between genotypes in heart. Real time quantitative PCR analyses showed that unlike in liver and heart, in mouse skeletal muscle PPAR delta is severalfold more abundant than either PPAR alpha or PPAR gamma. In both human and rodent myocytes, the highly selective PPAR delta agonist GW742 increased fatty acid oxidation about 2-fold and induced expression of several lipid regulatory genes, including pyruvate-dehydrogenase kinase 4 and uncoupling protein 3, responses that were similar to those elicited by the PPAR alpha agonist GW647. These results show redundancy in the functions of PPARs alpha and delta as transcriptional regulators of fatty acid homeostasis and suggest that in skeletal muscle high levels of the delta-subtype can compensate for deficiency of PPAR alpha.

Peroxisome proliferator activated receptors (PPARs) 1 ␣, ␦, and ␥ comprise a family of nuclear hormone receptors that regulate systemic fatty acid metabolism via ligand-dependent transcriptional activation of target genes (1). Strong evidence indicates that their endogenous ligands consist of fatty acids and/or lipid metabolites and that they function to mediate adaptive metabolic responses to changes in systemic fuel availability (1,2). PPAR␣, which is expressed most abundantly in tissues that are characterized by high rates of fatty acid oxidation (FAO), is considered the primary subtype that mediates lipid-induced activation of FAO genes (3). This premise is based largely on studies of PPAR␣ knockout (KO) mice, which, compared with wild type (WT) littermates, exhibit low rates of ␤-oxidation and abnormal accumulation of neutral lipids in both cardiac and hepatic tissues (4,5). The metabolic phenotype of KO mice is associated with decreased expression of FAO genes and failure of liver and heart to induce ␤-oxidative pathways in response to physiological or pharmacological perturbations in lipid metabolism (4 -6). Taken together, these studies indicate that, at least in rodents, PPAR␣ plays an essential role in maintaining lipid homeostasis in liver and heart by modulating both constitutive and inducible expression of genes that regulate fatty acid catabolism.
Skeletal muscle is also a major site of fatty acid catabolism in mammals. Similar to other oxidative organs, mRNA expression of muscle genes that promote selective utilization of lipid substrates is augmented during physiological states that are associated with increased systemic delivery of free fatty acids, such as exercise and starvation (7)(8)(9)(10). Furthermore, many of these same muscle genes are also up-regulated by in vivo administration of PPAR␣-selective drugs (11,12), prompting widespread speculation that fatty acid-induced activation of PPAR␣ plays a critical role in mediating the adaptive response of muscle to starvation and exercise. PPAR␣ protein expression is increased by exercise training (13,14) and induced during myocyte differentiation, coincident with increased ␤-oxidative capacity (15). Additionally, we recently showed that PPAR␣ regulates fatty acid utilization and mRNA expression of several FAO genes in primary human skeletal muscle cells (HSkMC) (15).
We hypothesized that deficiency of PPAR␣ might have profound consequences on skeletal muscle fuel metabolism and gene regulation. Pyruvate-dehydrogenase kinase 4 (PDHK4) and uncoupling protein 3 (UCP3) represent two muscle target genes that are robustly induced by PPAR␣ agonists (11,12), starvation (8,10), and exercise (7,9) and that are proposed to play key roles in mediating adaptive adjustments in muscle substrate selection. To investigate the requirement for PPAR␣ in regulating muscle fatty acid metabolism and transcriptional activation of PDHK4 and UCP3, we evaluated metabolic and gene regulatory responses to 24 h of starvation and endurance exercise, as well as in vitro FAO, in muscles from PPAR␣ KO mice and WT littermates. Contrary to our hypothesis, here we report that skeletal muscles from KO mice exhibited only minor changes in fatty acid homeostasis and, moreover, that neither constitutive nor inducible mRNA expression of known PPAR␣ target genes was negatively affected by its absence. We also found that skeletal muscle expressed high levels of PPAR␦ and that in both rodent and human skeletal muscle cells, activation of the ␦-subtype increased FAO as well as mRNA levels of several classical PPAR␣ target genes. These results, which are the first to show significant overlap in the functions of PPARs ␣ and ␦, indicate that both subtypes play important roles in mediating lipid-induced regulation of ␤-oxidative pathways. We propose that high levels of PPAR␦ might compensate for the lack of PPAR␣ in the skeletal muscles of KO mice.

EXPERIMENTAL PROCEDURES
Materials-Bovine serum albumin, carnitine, and sodium oleate were from Sigma. Fetal bovine serum and Hanks' balanced salt solution were from Invitrogen. Heat-inactivated horse serum was from Hyclone (Logan, UT). Dulbecco's low glucose modified Eagle's medium was from Invitrogen. Human skeletal muscle cell SingleQuots were from Bio-Whittaker Inc. (Walkersville, MD). Biocoat tissue culture plates were from BD PharMingen. PCR reagents were from PerkinElmer Life Sciences. GW742, GW647, and GW929 were obtained from Dr. Peter Brown at GlaxoSmithKline (16,17).
Animals-All of the protocols were approved by the Institutional Animal Care and Use Committee at East Carolina University. 4 -5month-old female homozygous PPAR␣ KO mice and aged matched WT littermates on an SV/129 background were obtained from a colony maintained at the Chemical Industry Institute of Toxicology (Research Triangle Park, NC). The original breeding pairs were obtained from Dr. F. Gonzalez (National Institutes of Health, Bethesda, MD) (18). The animals were housed under controlled temperature and lighting (20 -22°C; 12-h light-dark cycle) with free access to food and water. For starvation experiments, the food was removed at 10:00 a.m. the morning prior to collection of blood and tissue samples, which were harvested 24 h later along with samples from fed/rested and exercised mice.
Exercise Protocol-The mice were accustomed to the treadmill (Stanhope SAT 2000) with 5-min bouts of mild exercise (8 m/min) on three consecutive days prior to the experiment. The exercise test consisted of an incremental treadmill regimen during which mice ran at 10 m/min for the first 60 min. The speed was then increased 0.75 m/min at 15-min intervals, up to 13 m/min at 2 h. Mice in the 2-h exercise group were removed and immediately sacrificed for tissue and blood collection. The remaining mice continued running, and the treadmill speed was increased 1.0 m/min at 15-min intervals until the animals could no longer sustain the exercise. The exhausted mice were removed and immediately sacrificed for tissue and blood collection.
In Vitro Fatty Acid Oxidation Studies-Intact soleus and extensor digitorum longus muscles from the fed mice were removed between 9:00 and 11:00 a.m. under anesthesia (100 mg/kg ketamine and 10 mg/kg xylazine), cleaned free of adipose and connective tissue, and added to 1.0 ml of a modified Krebs-Ringer bicarbonate buffer containing 1.0 mM [1-14 C]oleate (1.0 Ci/ml), 1.0 mM carnitine, and 1.0% bovine serum albumin as described previously (19). After 90 min the incubation medium was transferred to new dishes, and the oleate oxidation rates (nmol/mg muscle weight/h) were determined by measuring production of 14 C-labeled acid-soluble metabolites (ASM), a measure of tricarbox-ylic acid cycle intermediates and acetyl esters (incomplete oxidation) (20), and [ 14 C]CO 2 (complete oxidation) trapped in 200 l of 1 N sodium hydroxide using a modified 48-well microtiter plate (Costar, Cambridge, MA) (21).
Mixed gastrocnemius muscles were removed from mice that were fed or starved for 24 h. Approximately 90 mg of tissue was minced thoroughly with scissors in 300 l of a modified sucrose-EDTA medium containing 250 mM sucrose, 1 mM EDTA, 10 mM Tris-HCl, pH 7.4, as described previously (22). The reactions were initiated by adding 40 l of whole homogenates to 160 l of the incubation buffer, pH 7.4, yielding final concentrations of 0.2 mM oleate ([1-14 C]oleate at 0.5 Ci/ml), 100 mM sucrose, 10 mM Tris-HCl, 5 mM potassium phosphate, 80 mM potassium chloride, 1 mM magnesium chloride, 2 mM L-carnitine, 0.1 mM malate, 2 mM ATP, 0.05 mM coenzyme A, 1 mM dithiothreitol, 0.2 mM EDTA, and 0.5% bovine serum albumin. After 60 min at 37°C, the reactions were terminated by adding 50 l of 70% perchloric acid. The CO 2 and ASM produced during the incubation were measured as described previously (21).
Northern Analyses-RNA was isolated from quick frozen muscle samples using the TRIzol (Invitrogen) reagent according to the manufacturer's instructions as described previously (23). Total RNA (10 g/sample) was fractionated on a 1.25% agarose, 2 M formaldehyde gel and then electrotransferred to a Hybond N1 membrane (Amersham Biosciences). Following UV cross-linking, the filters were prehybridized for 30 min at 68°C in ExpressHyb (CLONTECH Laboratories, Inc., Palo Alto, CA) and then hybridized 2 h at 68°C in Express Hyb buffer with random primed [␣-32 P]dATP-labeled cDNA probes for PDHK4 (24) or UCP3 (9) synthesized with Strip-EZ DNA kit (Ambion, Austin, TX) and purified with a G-50 Nick column (Amersham Biosciences). After stripping and prehybridization, the blots were then hybridized sequentially with glyceraldehyde-3-phosphate dehydrogenase and ␤-actin for normalization. The Northern blots were visualized with a Phosphor-Imager and quantified with Imagequant software (Molecular Dynamics, Sunnyvale, CA).
Cultures of Primary Human Skeletal Muscle Cells and Rat L6 Cells-Protocols for use of human specimens were approved by the Institutional Review Board at East Carolina University. Muscle samples weighing ϳ50 mg were obtained from vastus lateralis by needle biopsy and immediately transferred to ice-cold Dulbecco's modified Eagle's medium and cleaned free of adipose and connective tissues. Satellite cells were isolated by trypsin digestion (25) preplated for 1-3 h in 3.0 ml of growth medium (15) on an uncoated T-25 tissue culture flask to remove fibroblasts and then transferred to a type I collagen-coated T25 flask for attachment. Myoblasts were grown at 37°C in a humidified atmosphere of 5% CO 2 and subcultured onto 6-, 12-, and 24-well type I collagen-coated plates as described previously (15). When HSkMC reached 80 -90% confluence, differentiation was induced by changing to low serum differentiation medium (DFM; Dulbecco's modified Eagle's medium, 2% horse serum, 4.0 mM glutamine, 50 mg/ml gentamycin) (15). The medium was changed every 2 days, and PPAR-selective compounds or Me 2 SO vehicle (0.1% v/v) were added to mature myotubes for 48 h on days 6 -7. GW647 (16) and GW742 (a close analog of GW501516 (17) with equivalent activity) exhibit 1000-fold selectivity for PPARs ␣ and ␦, respectively. Day 8 myotubes were harvested in 1.0 ml of TRIzol reagent for RNA extraction.
Early passages of rat L6 myoblasts (CRL-1458) from the American Tissue Culture Collection (Manassas, VA) were grown in Dulbecco's modified Eagle's medium with 10% fetal bovine serum, 4.0 mM glutamine, and 50 mg/ml gentamycin in a humidified incubator at 37°C with 5% CO 2 . The myoblasts were grown on 100-mm dishes to 50 -60% confluence and then subcultured onto 6-and 24-well plates for experiments. When the cells were 70% confluent, they were induced to differentiate into myotubes by changing to low serum DFM. By day 6, the cells were fully confluent and had differentiated into multinucleated, contracting myotubes. Day 8 myotubes were treated for 48 h with vehicle and PPAR-selective compounds and then harvested as described for HSkMC.
Determination of Fatty Acid Oxidation in Cultured Myocytes-Following treatments with vehicle or PPAR-selective compounds, the myotubes were incubated at 37°C in sealed 12-or 24-well plates containing 500 or 750 l of DFM plus 12.5 mM HEPES, 0.25% bovine serum albumin, 1.0 mM carnitine, 100 M sodium oleate, 50 g/ml gentamycin, and 1.0 Ci/ml [ 14 C]oleate (PerkinElmer Life Sciences ). After 3 h the incubation medium was transferred to new dishes and assayed for labeled oxidation products (CO 2 and ASM) (21). The cells were placed on ice, washed twice with phosphate-buffered saline, scraped into a 1.5-ml Eppendorf tube in two additions of 0.30 ml of 0.05% SDS lysis buffer, and then stored at Ϫ80°C. All of the assays were performed in triplicate.
Real Time Quantitative PCR-Total RNA was prepared using the TRIzol reagent, treated with DNase I (Ambion) and quantified using the RiboGreen RNA quantitation kit (Molecular Probes, Eugene, OR). RTQ-PCR was performed using an ABI PRISM 7700 sequence detection system instrument and software (PerkinElmer Life Sciences). Primer/ probe sets were designed using the manufacturer's software and the sequences available in the GenBank TM data base. The sequences for the rat genes were: CPT1, F: GCTTATTAAGAACACGAGCCAACAA, R: AGTTTGCGGCGATACATGATC, P: TTGGGAAACACCGTTCACGC-CA; PDHK4, F: TCTAACGTCGCCAGAATTAAAGC, R: GGAACGTAC-ACGATGTGGATTG, P: ACACAAGTCAATGGAAAATTTCCAGGCC-AA; UCP3, F: TGGCCTCCCCAGTGGAT, R: GGGCTTCGGTACCTGC-CT, P: TAAAGACCCGATACATGAACGCTCCC. The sequences for the mouse genes were: PPAR␣, F: ACGATGCTGTCCTCCTTGATG, R: GTGTGATAAAGCCATTGCCGT, P: ACAAAGACGGGATGCTGATCG-CG; PPAR␦, F: GCTGCTGCAGAAGATGGCA, R: CACTGCATCATCT-GGGCATG, P: ACCTGCGGCAGCTGGTCACTGA; PPAR␥, F: CCATT-CTGGCCCACCAAC, R: AATGCGAGTGGTCTTCCATCA, P: TCGGAA-TCAGCTCTGTGGACCTCTCC. The sequences used for analysis of human genes were published previously (15). RNA samples were normalized for comparison by determining 18 S rRNA levels by RTQ-PCR. PPAR subtype expression in mouse tissues was analyzed by generating six-point serial standard curves using total RNA from heart, liver, and skeletal muscle. Estimates of relative mRNA abundance (arbitrary units) were determined by the ratio of calculated units of RNA from each tissue. The calculations were made using the portion of the curve for which the plot of the log input amount versus the C T (cycle threshold) differences resulted in a slope of approximately 0, indicating that the amplicon efficiencies were approximately equal. Relative quantitation of PPAR target genes in cultured myocytes was calculated by using the 2⌬ CT formula, in which ⌬C T equals the difference between C T values for vehicle (Me 2 SO) and drug-treated cells. This formula was validated for each primer/probe set using six-point serial standard curves as described previously (24).
Statistics-Statistical analyses were performed using JMP Statistical Software (SAS, Cary, NC). Two-and three-way ANOVAs were performed using a standard least squares model to test both the main and interaction effects of genotype ϫ starvation or exercise ϫ time (where appropriate) on serum parameters and mRNA expression. In HSkMC and L6 myocytes differences in FAO rates and mRNA expression (C T values) were analyzed by Student's t test for paired data. (4), we found that the body weights of older KO mice (23.4 Ϯ 0.73 g) were greater than their WT littermates (21.7 Ϯ 0.98 g). The effects of PPAR␣ ablation on muscle fuel metabolism were first evaluated during in vivo studies that compared the metabolic responses to 24 h of starvation with that of endurance exercise. We found that when KO mice were challenged by 24 h of starvation, they exhibited increased serum nonesterified fatty acids (NEFA) and hypoketonemia (Fig. 1, A and B) but showed no changes in serum glycerol (not shown) compared with WT. These results are consistent with previous studies (4) and support the premise that ablation of PPAR␣ severely impairs the capacity of the liver to clear and oxidize fatty acids. Moreover, because starvation fully depletes liver glycogen levels ( Fig. 2), gluconeogenesis becomes the primary pathway for sustaining blood glucose. KO mice exhibited hypoglycemia, consistent with reduced liver gluconeogenic capacity, and lower serum lactate levels, suggesting increased dependence on this substrate as a gluconeogenic precursor (Fig. 1, C and D). These metabolic abnormalities observed in KO mice after 24 h of starvation were remarkably similar to those exhibited in response to exercise, and as predicted, time to exhaustion (180 Ϯ 35 versus 150 Ϯ 15 min; p Ͻ 0.05) and total distance run (1.80 versus 1.37 km; p Ͻ 0.01) were greater in WT compared with KO, respectively, indicating that KO mice are less tolerant of endurance exercise. Notably, depletion of liver rather than muscle glycogen corresponded with exhaustion in both groups ( Fig. 2). Moreover, estimated rates of liver glycogen depletion were ϳ50% greater in null compared with WT mice, whereas in contrast, muscle glycogen depletion rates were similar between genotypes. These findings suggested that, contrary to our prediction, PPAR␣ ablation might not increase the reliance of muscle on glucose substrate.

Metabolic Responses to Exercise and Starvation Are Altered in PPAR␣ KO Mice-Consistent with earlier reports
In Vitro Fatty Acid Oxidation Rates in Muscles of PPAR␣ KO Compared with WT Mice-To evaluate changes in muscle fatty acid oxidative capacity directly, we performed in vitro experiments using skeletal muscles comprised of red (soleus), white (extensor digitorum longus), or mixed (gastrocnemius) fiber types. In the fed state, rates of muscle FAO were similar between WT and null mice (Table I), regardless of the fiber composition, implying that PPAR␣ is not required to maintain constitutive activity of muscle FAO enzymes. When animals were starved for 24 h, FAO rates were 28% lower in muscle homogenates from KO compared with WT mice, but neutral lipid accumulation, analyzed by Oil Red O staining, was not different between genotypes (not shown). These results suggested that the metabolic consequences of PPAR␣ deficiency in skeletal muscle are much less severe than those previously shown in liver and heart (4,5,9).
PPAR␣-independent Regulation of Target Genes in Skeletal Muscle-Next, we evaluated both constitutive and inducible mRNA expression of classical PPAR␣ target genes in mixed quadriceps muscle. In fed/rested mice, mRNA levels of PDHK4 and UCP3 were similar in muscles from KO compared with WT mice (Fig. 3). In WT, exercise increased muscle mRNA levels of UCP3 and PDHK4 86 and 60%, respectively, whereas starvation increased expression of both genes five to eight times. Surprisingly, the exercise-and/or starvation-induced up-regulation of these genes was either similar or greater (UCP3) in muscles from KO mice, indicating that PPAR␣ is not required for these physiological responses. In comparison, PPAR␣ gene ablation attenuated constitutive and/or inducible expression of the same transcripts in heart (Fig. 4).
PPAR Subtype Expression in Fatty Acid Oxidative Tissues-Our results in Figs. 3 and 4, taken together with those from previous studies that investigated liver and heart, revealed tissue-specific differences in PPAR␣-dependent regulation of FAO and target gene expression. Using RTQ-PCR, we evaluated whether the requirement for PPAR␣ a might be associated with distinct expression patterns of the three PPAR subtypes in various oxidative tissues (Fig. 5). RTQ-PCR detected the nonfunctional ␣ transcript in tissues from KO mice because our mPPAR␣ primer/probes produce an amplicon from exon 7 that occurs before the targeted disruption in exon 8 (18). We found that, compared with skeletal muscle, PPAR␣ mRNA levels were 7-and 19-fold enriched in heart and liver, respectively, whereas PPAR␦ mRNA abundance was similar among oxidative tissues. However, in skeletal muscle, expression level of the ␦-subtype was severalfold greater than either PPAR␣ or PPAR␥. PPAR␦ expression decreased 68% in liver of KO, whereas PPAR␥, which was the least abundant subtype in all three tissues from WT mice, increased 2-3-fold in the heart and liver, but not the skeletal muscle, of KO mice.
The PPAR␦-selective Agonist GW742 Increases Fatty Acid Oxidation in Cultured Skeletal Muscle Cells-The present results, as well as those from other animal studies, indicate that at the mRNA level, PPAR␦ is the most abundant subtype in rodent skeletal muscle. In human skeletal muscle we have observed high levels of PPARs ␣ and ␦ and low levels of PPAR␥. 2 Taken together, these findings suggest that PPAR␦ might play a major role in regulating oxidative metabolism in both rodent and human skeletal muscle. To address this possibility we evaluated the effects of a novel PPAR␦ agonist, GW742, on FAO and gene expression in mature myotubes   (Fig. 6A). The maximal effect of the PPAR␦ agonist was nearly identical to that of the ␣ agonist GW647, whereas the PPAR␥ agonist GW929 did not effect FAO (Fig.   FIG. 3. Absence of PPAR␣ does not affect expression of lipid regulatory genes in quadriceps muscle. Total RNA from quadriceps (Quad) muscle of PPAR␣ KO mice and age-matched WT littermates was prepared as described under "Experimental Procedures." mRNA levels of PDHK4 (A) and UCP3 (B) in muscles of mice that were fed/rested (F), exercised for 2 h (E), and starved for 24 h (S) were quantified by Northern blot analysis. The values are the means Ϯ S.E. from 7-9 animals, and the differences between WT and KO mice ( ‡) and fed/rested versus exercised/starved mice (*) were analyzed by two-way ANOVA.

FIG. 4. Absence of PPAR␣ attenuates expression of lipid regulatory genes in heart.
Total RNA from heart of PPAR␣ KO mice and age-matched WT littermates was prepared as described under "Experimental Procedures." mRNA levels of PDHK4 (A) and UCP3 (B) in heart of mice that were fed/rested (F), exercised for 2 h (E), and starved for 24 h (S) were quantified by Northern blot analysis. The values are the means Ϯ S.E. from 6 -8 animals, and the differences between WT and KO mice ( ‡) and fed/rested versus exercised/starved mice (*) were analyzed by two-way ANOVA. 6B). Treating myocytes with maximal doses of GW647 and GW742 together did not produce an effect that was greater than either compound added alone (not shown), suggesting that PPARs ␣ and ␦ might activate similar target genes via the same regulatory element.
PPARs ␣ and ␦ Target Similar Lipid Regulatory Genes in Cultured Muscle Cells-We recently reported that in primary HSkMC, PPAR␣ activation increases mRNA expression of several genes that promote FAO (15). To investigate possible redundancy in skeletal muscle PPAR␣ and ␦ target genes, here we used RTQ-PCR to perform similar evaluations in human and rodent myotubes that were treated with agonists selective for either PPAR␦ (GW742) or PPAR␣ (GW647). The results in Table II show that PPAR␦-mediated increases in myotube mRNA expression of FAO genes were similar to that of PPAR␣. Most remarkably, activation of either ␣ or ␦ increased PDHK4 mRNA expression ϳ30-fold. Both treatments also increased mRNA expression of malonyl-CoA decarboxylase (MCD) and the muscle isoform of carnitine palmitoyltransferase 1 (CPT1), both of which encode for proteins that increase muscle fatty acid uptake and/or ␤-oxidation. Induction of these genes in drug-treated myotubes relative to vehicle controls was highly consistent among cells obtained from different subjects and similar in myotubes treated with GW647 compared with GW742. In contrast, changes in UCP3 mRNA expression were highly variable and significant only in cells that were treated with the PPAR␦ agonist.
Similar to HSkMC, in L6 myotubes both the ␣ (GW647) and ␦ (GW742) agonists increased mRNA abundance of PDHK4 ϳ24-fold (Table III). Both agonists also increased mCPT1 and UCP3, but notably, in L6 myocytes induction of these genes was severalfold more robust than in the HSkMC. Together, these results suggest that gene-and species-specific elements modulate the physiological induction of PPAR targets. In results not shown, the PPAR␥ agonist, GW929, did not affect the mRNA levels of any of the genes analyzed. DISCUSSION PPAR␣ is expressed most abundantly in highly oxidative tissues where it plays a key role in activating pathways of ␤-oxidation (2). Conversely, PPAR␥, which is expressed primarily in adipose tissue, promotes adipocyte differentiation and induces genes involved in lipogenesis (26). The function of PPAR␦, which is expressed ubiquitously, is less certain, although recent data implicate this subtype as a regulator of cholesterol metabolism (17) and adiposity (27). Skeletal muscle expresses all three PPAR subtypes, but their distinct roles in regulating muscle energy metabolism are unknown. In the present investigation we used PPAR␣ KO mice to determine whether the ␣-subtype is essential for maintaining skeletal muscle lipid homeostasis. The mice were challenged by exercise or 24 h of starvation, two physiological stresses that increase systemic delivery of fatty acid and induce adaptive changes in muscle fuel metabolism. In contrast to starvation, a long term stress in which lipid catabolism increases most markedly in the liver, exhaustive exercise represents a short term stress in which high rates of skeletal muscle FAO are critical for sustaining prolonged work. Although we found that KO mice were less tolerant of exercise, they still performed better than expected given our hypothesis that ablation of PPAR␣ would produce severe defects in muscle fatty acid oxidative capacity. The findings that glycogen depletion rates were greater in liver but not in muscle of KO mice and that in vitro muscle FAO rates were similar between genotypes suggested that exercise intolerance was due to metabolic perturbations in liver and heart rather than skeletal muscle. Interestingly, exhaustive exercise depleted muscle glycogen only 50%, whereas depletion of liver glycogen reached nearly 100%, corresponding with the onset of hypoglycemia in both genotypes (not shown). These data implicate liver fuel metabolism as a limiting factor during exhaustive, low intensity exercise.
Previous studies using KO mice have provided convincing evidence that, in rodents, PPAR␣ plays an obligatory role in regulating lipid homeostasis in liver and heart. In these oxidative tissues, absence of functional PPAR␣ decreases mitochondrial FAO ϳ70% (28 -30), a finding that is explained by reduced constitutive and/or inducible expression levels of several fatty acid catabolic enzymes including MCD (30) and PDHK4 (31). Furthermore, when KO mice are challenged by physiological or pharmacological perturbations in fatty acid metabolism, cardiac and hepatic tissues accumulate abnormally high amounts of neutral lipids (4,32). In contrast, the present study found that skeletal muscles of KO mice exhibited minor abnormalities in FAO only when animals were starved and showed no evidence of neutral lipid accumulation. KO muscles also maintained constitutive and inducible mRNA levels of the PPAR␣ target genes, PDHK4 and UCP3. Although exercise/ starvation induced stress hormones probably contribute to the physiological regulation of these genes, PPAR␣ is widely considered a primary mediator because administration of PPAR␣selective drugs, as well as treatments that elevate serum NEFA in fed animals, mimic the effects of starvation on UCP3 and PDHK4 gene expression (10,12,33). Surprisingly, starvation-induced up-regulation of muscle UCP3 was actually more robust in KO mice, which corresponded with higher serum NEFA. This result is consistent with previous studies describing a tight correlation between systemic NEFA and changes in skeletal muscle UCP3 mRNA levels (33); however, it is inconsistent with speculation that PPAR␣ is an essential mediator of these changes. In contrast, when we examined the same target genes in the heart, deficiency of PPAR␣ decreased the constitutive levels of UCP3 and diminished exercise-and starvationinduced increases in both UCP3 and PDHK4. Thus, importantly, these results are the first to show that skeletal and cardiac muscles, both of which are highly oxidative tissues, exhibit distinct requirements for PPAR␣.
Tissue-specific differences in the metabolic consequences of PPAR␣ deficiency might be related to the relative abundance of the three subtypes in various oxidative tissues. We found predominant expression of PPAR␣ in liver, whereas PPARs ␣ and ␦ were similarly abundant in heart. In contrast, PPAR␦ was the major subtype expressed in mouse quadriceps muscle. These results, obtained by RTQ-PCR, are similar to studies that used RNase protection assays to evaluate expression levels of the PPAR subtypes in mouse (34) and rat (35) tissues. Low levels of PPAR␥ in oxidative tissues from WT are consistent with the notion that this subtype does not play a major role in regulating FAO. On the contrary, increased PPAR␥ in liver and heart of KO mice is consistent with its function as an activator of lipogenesis and glycerolipid synthesis and, similar to other reports (36), implicates elevated ␥ levels as an early response in the progression of hepatic and cardiac lipotoxicity.
The results from several in vivo studies have implied a role for the PPARs in mediating fatty acid-induced regulation of muscle metabolism, but only recently has direct evidence emerged showing that PPAR-selective drugs induce expression of lipid regulatory genes in cultured skeletal myocytes (15,35,37,38). Still, there is little information on whether the three skeletal muscle PPAR subtypes target similar or distinct metabolic pathways. The results of our study using PPAR␣ KO mice suggested that PPAR␦ might play a role in regulating muscle lipid homeostasis. This hypothesis was supported by experiments in cultured myocytes showing that the highly selective PPAR␦ agonist, GW742, increased FAO and induced expression of several lipid regulatory genes. In both human and rat myotubes the metabolic and gene regulatory effects of the ␦ agonist were similar to those observed when cells were treated with the PPAR␣ agonist. Our results are consistent with recent studies showing that activation of PPAR␦ induces mRNA expression of UCP2 and UCP3 in human and L6 myotubes, respectively. Skeletal muscle UCPs, which are mitochondrial membrane proteins that uncouple oxidative phosphorylation, are proposed to facilitate FAO and/or protect muscle against damaging by-products of oxidative metabolism (39,40). Additionally, we found that PPAR␦ activation also increased mRNA expression of PDHK4, CPT1, and MCD. PDHK4 functions by phosphorylating and inactivating pyruvate dehydrogenase, the multienzyme complex that catalyzes the irreversible oxidation of pyruvate to acetyl-CoA (41). Increased expression and activity of PDHK4, which occurs in response to exercise (7), starvation (8), and a low carbohydrate diet (42), is thought to promote fatty acid oxidation by decreasing glucose-derived acetyl-CoA. CPT1 catalyzes the initial and rate-limiting step in the transport of fatty acid into mitochondria, whereas MCD disposes of the potent CPT1 inhibitor, malonyl-CoA. Thus, in the aggregate, activation of these genes increases catabolism of fatty acids by promoting their entry into the mitochondria and by relieving competition between carbohydrate and lipid-derived acetyl-CoA. Importantly, this study is the first to demonstrate that PPAR␦ is capable of inducing multiple pathways that cooperatively promote FAO, a function that is generally assigned to PPAR␣.
Because PPAR␦ binds to many of the same fatty acids and lipid metabolites as the other PPAR subtypes, it has been presumed to play a role in fatty acid homeostasis, although FIG. 6. Effects of PPAR agonists on fatty acid oxidation in cultured myocytes. After 6 days in DFM, mature rat L6 or primary human myotubes were treated for 48 h with vehicle (Me 2 SO) or 0.10 -100 nM of the PPAR␦ agonist GW742 (A) and maximal doses of GW647 (1.0 M), GW742 (10 nM), and GW929 (100 nM), agonists selective for PPARs ␣, ␦, or ␥, respectively (B). Day 8 myotubes were then incubated for 3 h at 37°C in DFM with 100 M [ 14 C]oleate and 0.25% bovine serum albumin. Oleate oxidation (nmol/g/h) was determined by measuring production of 14 C-labeled CO 2 and ASM as described under "Experimental Procedures." All of the assays were performed in triplicate, and the values are the means Ϯ S.E. from 3 subjects (human myocytes) or two experiments (L6 myocytes) (A) and 3-10 subjects (B) as indicated. The differences between vehicle and drug-treated cells were analyzed by Student's t test for paired data and one-way ANOVA. *, p Ͻ 0.05; **, p Ͻ 0.001 versus Me 2 SO-treated controls.  (17). Changes in serum lipid profiles were at least partly explained by induction of genes that promote reverse cholesterol transport, but our data suggest that increased skeletal muscle lipid uptake and catabolism might have contributed to the therapeutic effects of the treatment.
All three PPAR subtypes bind to DNA as obligate heterodimers with the retinoic acid receptor RXR, and the preferred DNA-binding site, referred to as the DR-1 motif, is the same for each of the PPAR/RXR heterodimer pairs (1). Because PPARs ␣ and ␦ bind to the same response element within a given target promoter, conceivably, full activation of these two distinct transcription factors, as occurs when the receptor binds to its chemically engineered agonist in vitro, might produce nearly identical responses. However, considerable data, including phenotypic differences between PPAR␣-and PPAR␦null mice (43), indicate that these transcription factors play distinct physiological roles in vivo. Our studies in KO mice showed that PPAR␣ is indispensable for UCP3 regulation in heart, despite high expression levels of PPAR␦, indicating that properties other than mRNA abundance contribute to the tissue-specific roles of these transcription factors. In addition to tissue distribution, specificity of function might be conferred by mechanisms that impart subtype-specific regulation of receptor activity, such as differences in their binding affinities for endogenous ligands (44,45). Transcriptional activity of PPARs ␣ and ␥ is also regulated by phosphorylation of their transcriptional activation domains (44, 46 -48) and by tissue-specific interactions with co-activators/repressors (47). Less is known about PPAR␦, but the activity of this subtype is likely to exhibit similar mechanisms of regulation.
In summary, the results from our experiments using cultured myocytes show overlap in the functions of PPARs ␣ and ␦ as transcriptional regulators of FAO. Taken together with previous studies, our data from PPAR␣ KO mice indicate that the essential and/or permissive roles of each subtype in controlling expression FAO genes in vivo depend on complex physiological interactions that exhibit tissue and target gene specificity. A full accounting of the distinct and cooperative roles of PPARs ␣ and ␦ in regulating lipid homeostasis, and perhaps other metabolic pathways, will require combined loss of function experiments in both cell culture and tissue-specific knockout mouse models. Information from such studies should have important implications for treating energy metabolic diseases such as obesity, hyperlipidemia, and type 2 diabetes.