Alterations in Lipoprotein Metabolism in Peroxisome Proliferator-activated Receptor α-deficient Mice*

The peroxisome proliferator-activated receptor-α (PPARα) controls gene expression in response to a diverse class of compounds collectively referred to as peroxisome proliferators. Whereas most known peroxisome proliferators are of exogenous origin and include hypolipidemic drugs and other industrial chemicals, several endogenous PPARα activators have been identified such as fatty acids and steroids. The latter finding and the fact that PPARα modulates target genes encoding enzymes involved in lipid metabolism suggest a role for PPARα in lipid metabolism. This was investigated in the PPARα-deficient mouse model. Basal levels of total serum cholesterol, high density lipoprotein cholesterol, hepatic apolipoprotein A-I mRNA, and serum apolipoprotein A-I in PPARα-deficient mice are significantly higher compared with wild-type controls. Treatment with the fibrate Wy 14,643 decreased apoA-I serum levels and hepatic mRNA levels in wild-type mice, whereas no effect was detected in the PPARα-deficient mice. Administration of the fibrate Wy 14,643 to wild-type mice results in marked depression of hepatic apolipoprotein C-III mRNA and serum triglycerides compared with untreated controls. In contrast, PPARα-deficient mice were unaffected by Wy 14,643 treatment. These studies demonstrate that PPARα modulates basal levels of serum cholesterol, in particular high density lipoprotein cholesterol, and establish that fibrate-induced modulation in hepatic apolipoprotein A-I, C-III mRNA, and serum triglycerides observed in wild-type mice is mediated by PPARα.

Peroxisome proliferator-activated receptors (PPARs) 1 are a subfamily of the nuclear hormone receptor gene family. There are three distinct PPARs, termed ␣, ␦ (also called ␤, NUC-1), and ␥, each encoded by a separate gene and showing a distinct tissue distribution (for review, see Refs. [1][2][3][4][5]. Activated PPARs heterodimerize with another nuclear receptor, RXR and alter transcription of target genes after binding to specific response elements or PPREs. PPREs consist of a direct repeat of the nuclear receptor hexameric DNA recognition motif spaced by one nucleotide. Numerous PPAR target genes have been identified (for review, see Ref. 4). Since they are activated by various fatty acid metabolites as well as several drugs used in the treatment of metabolic disorders, PPARs can be considered as key messengers that modulate nutritional, pharmacological, and metabolic stimuli into changes in gene expression. PPARs were initially considered orphan receptors, since no direct binding of the various activators to the receptors could be demonstrated. However, PPAR␣ has recently been shown to bind and be activated by leukotriene B4 and fibrates (6). In addition, prostaglandin J2 derivatives and the antidiabetic thiazolidinediones have been shown to be natural and synthetic ligands for PPAR␥, respectively (7)(8)(9).
PPAR␣ was the first PPAR to be identified (10), and it is expressed primarily in tissues that have a high level of fatty acid catabolism such as liver (11). In the liver, PPAR␣ modulates oxidation of fatty acids and detoxification of several xenobiotic compounds. Numerous studies have demonstrated that several genes encoding enzymes involved in metabolic pathways, such as ␤and -oxidation, contain a PPRE in their promoter region and are under transcriptional control of PPAR␣ (reviewed in Ref. 4). Consistent with this observation, PPAR␣ knockout mice, which are apparently healthy under basal conditions, are not able to induce genes involved in ␤and -oxidation when treated with compounds that activate PPAR␣ (12). Fatty acid oxidation pathways have diverse roles in physiology, extending from a role in lipid metabolism strictu sensu to a role in the metabolism of various lipid mediators and signaling factors (6).
In addition to its role inand ␤-oxidation pathways, PPAR␣ was suggested to be important in the control of extracellular lipid metabolism (for review, see Refs. [1][2][3][4][5]. For example, PPAR␣ activators such as fibrates have an important role in the control of HDL cholesterol levels. Gene expression of apoA-I and apoA-II, the major apolipoproteins in HDL, is controlled by PPAR␣. Thus, these protagonists of reverse cholesterol transport (13), a protective pathway against coronary artery disease (14), can be modulated by PPAR␣ (reviewed in Refs. [3][4][5]. Besides the role of PPARs in HDL metabolism, they also affect metabolism of triglyceride-rich lipoprotein particles (reviewed in Refs. [3][4][5]. In fact, activation of either PPAR␣ and/or PPAR␥ has pronounced triglyceride-lowering effects in animals and man due to effects on both clearance and production rates of triglyceride-rich lipoproteins.
Given the effects of treatment with PPAR␣ activators on lipoprotein metabolism, the goal of this study was to analyze in a more direct fashion whether PPAR␣ is involved in the regulation of lipoprotein metabolism. Thus, we examined lipopro-tein metabolism in PPAR␣-deficient mice. These data unequivocally demonstrate that PPAR␣ has an important regulatory role in lipid metabolism.
Animals-For all experiments, we used male mice, 10 -16 weeks of age, 20 -35 g, F4 C57BL/6N X Sv/129 homozygous wild-type (ϩ/ϩ) or knockout (Ϫ/Ϫ) for PPAR␣ (12). Mice from both genotypes were fed either the control or Wy 14,643 diet ad libitum for 14 days. At the end of the experiments, animals were weighed and euthanized by overexposure to carbon dioxide. Blood was collected, and serum was separated and used within 1 week for analysis of lipids, lipoproteins, and apolipoproteins. The liver was removed immediately, weighed, and frozen in liquid nitrogen and stored at Ϫ80°C until further analysis.
Lipid, Apolipoprotein, and Lipoprotein Measurements-Serum lipids (cholesterol and triglycerides) and HDL cholesterol were determined by enzymatic assay adapted to microtiter plates using commercially available reagents (Boehringer Mannheim). Serum HDL cholesterol content was determined after precipitation of apoB-containing lipoproteins with phosphotungstic acid/mg (Boehringer Mannheim). Serum levels of mouse apoA-I and apoA-II were measured by an immunonephelometric assay using specific polyclonal antibodies.
The distribution of lipoproteins in serum from mice was analyzed by nondenaturing discontinuous gradient polyacrylamide gel electrophoresis (Lipofilm kit, Sebia, Issy-les-Moulineaux, France) according to the manufacturer's instructions.
HDL lipoprotein fraction (d ϭ 1.063-1.21 g/ml) was isolated by sequential ultracentrifugation as described (15). The HDL fraction was assayed for its protein (16) and lipid (cholesterol, triglyceride, and phospholipid) content. HDL size was determined on polyacrylamide gradient gels (4 -20%, Novex, San Diego, CA). 10 g of HDL protein were loaded, and the electrophoresis was performed in a Novex apparatus at 125 V for 13 h in 0.025 M Tris, 0.192 M glycine, pH 8.3. Gels were stained with Coomassie Brilliant Blue R-250. Proteins in the high molecular mass calibration mixture (Pharmacia Biotech Inc.) were used as calibrating proteins on these gels.
Lipoprotein cholesterol profiles were obtained by fast protein liquid chromatography as described (17). This system allows separation of the three major lipoprotein classes; VLDL, low density lipoproteins, and HDL. Cholesterol concentrations were determined in the eluted fractions. Accumulated data were analyzed by the Millenium 20/0 program (Waters).
RNA Analysis-RNA was isolated from liver by the acid guanidinium thiocyanate/phenol/chloroform method (18). Northern and dot blot analysis of total cellular RNA was performed as described (19). Rat apoA-I, apoA-II, apoA-IV, and apoC-III cDNAs were used as probes (20,21). GAPDH and 36B4 (22, 23) (encoding the human acidic ribosomal phosphoprotein PO (23)) were used as control probes. All probes were labeled by random primed labeling (Boehringer Mannheim). Filters were hybridized with 1.5 ϫ 10 6 cpm/ml of each probe as described (21). Autoradiograms were analyzed by quantitative scanning densitometry (Bio-Rad GS670 Densitometer) as described (21). RNA expression data of the various apolipoproteins were corrected for the expression of a control probe. An arbitrary value of 100 was assigned to the average of the wild-type (ϩ/ϩ) untreated animals for each experiment.

PPAR␣ (Ϫ/Ϫ) Mice Have Elevated Levels of HDL Particles-
Compared with control (ϩ/ϩ) mice, serum total and HDL cholesterol concentrations were significantly higher in PPAR␣ (Ϫ/Ϫ) mice (Table I). Total serum cholesterol and serum HDL cholesterol levels in the PPAR␣ (Ϫ/Ϫ) mice were 64 and 63% higher, respectively, than values in control PPAR␣ (ϩ/ϩ) mice. Plasma triglyceride values were not significantly different between the two groups.
Separation of the different lipoprotein fractions by electrophoresis of lipostained samples from PPAR␣ (Ϫ/Ϫ) and (ϩ/ϩ) mice confirmed a robust increase in HDL concentrations in PPAR␣ (Ϫ/Ϫ) serum (data not shown). When staining intensity of 100% was arbitrarily attributed to the HDL band of the PPAR␣ (ϩ/ϩ) animals, the relative staining intensity of the HDL from the PPAR␣ (Ϫ/Ϫ) animals was 197%.
Serum Lipoproteins in PPAR␣ (Ϫ/Ϫ) Mice-To analyze in more detail the distribution of lipoproteins in plasma from the PPAR␣ (Ϫ/Ϫ) and (ϩ/ϩ) mice, aliquots from pooled serum from these animals were subjected to gel filtration chromatography. The cholesterol profiles showed a striking increase in HDL levels as depicted in Fig. 1. This increase is consistent with the increase in HDL cholesterol levels measured by the phosphotungstic acid/mg precipitation technique (Table I) and by lipofilm electrophoresis. Cholesterol concentrations from each of the lipoprotein fractions were further measured. As expected, almost all of the cholesterol was distributed in HDL in both genotypes of mice (85.9 Ϯ 3.5% versus 88.9 Ϯ 0.6% in PPAR␣ (ϩ/ϩ) and (Ϫ/Ϫ), respectively). In addition, the data confirmed a large increase in HDL cholesterol concentrations in PPAR␣ (Ϫ/Ϫ) mice compared with PPAR␣ (ϩ/ϩ) mice (137.1 Ϯ 9.7 versus 78.9 Ϯ 10.2 mg/dl, respectively).
HDL from pooled serum samples from PPAR␣ (Ϫ/Ϫ) and (ϩ/ϩ) mice was isolated by density equilibrium ultracentrifugation, and the composition was determined. HDL particles from PPAR␣ (ϩ/ϩ) were not different in their composition compared with PPAR␣ (Ϫ/Ϫ) mice (Table II). Since the particle size distribution of HDL was comparable and showed a homogeneous population of HDL particles in both groups of animals ( Fig. 2), we can conclude that the increase in HDL cholesterol is essentially due to an increase in the amount of circulating HDL lipoproteins.
The Increase in HDL Is Due to an Increase in the HDL Apolipoprotein apoA-I-To verify whether the observed increase in HDL particle number and HDL cholesterol was associated with a concomittant increase in the major HDL apolipoproteins, we determined serum apoA-I and apoA-II concentrations in PPAR␣ (Ϫ/Ϫ) and (ϩ/ϩ) mice by a nephelometric assay (Fig. 3, A and B). Serum apoA-I was significantly greater in the PPAR␣ (Ϫ/Ϫ) mice compared with (ϩ/ϩ) mice (136 Ϯ 46 versus 96 Ϯ 19 mg/dl; p Ͻ 0.01). The increase in serum apoA-I levels was due to an increase in hepatic apoA-I mRNA levels observed in PPAR␣ (Ϫ/Ϫ) mice compared with control (ϩ/ϩ) mice (Figs. 3A and 4). ApoA-II concentrations tended also to be higher in PPAR␣ (Ϫ/Ϫ) mice relative to the (ϩ/ϩ) controls, but the difference was not statistically significant (64 Ϯ 27 versus 54 Ϯ 15 mg/dl) (Fig. 3B). Similarly, hepatic apoA-II mRNA had a tendency to be more elevated in the (Ϫ/Ϫ) animals (Figs. 3B and 4).
ApoA-I Levels Are Not Controlled by Fibrates in PPAR␣ (Ϫ/Ϫ) mice-Previous work from our laboratories and others has shown that rodent serum apoA-I and hepatic mRNA levels are down-regulated by treatment with fibrate hypolipidemic drugs. Since fibrates are potent activators and ligands of PPAR␣, we examined this regulation in PPAR␣ (Ϫ/Ϫ) mice. A significant decrease in both hepatic apoA-I mRNA levels (p Ͻ 0.05) and serum apoA-I (p Ͻ 0.01) was observed in (ϩ/ϩ) mice after treatment with Wy 14,643. However, administration of Wy 14,643 to PPAR␣ (Ϫ/Ϫ) animals resulted in no significant change in hepatic apolipoprotein A-I mRNA and serum apoA-I levels (Figs. 3A and 4). Absence of a regulatory effect of fibrates on apoA-I protein and mRNA levels in Wy 14,643 fed PPAR␣ (Ϫ/Ϫ) mice suggests and implies PPAR␣ unequivocally as a major determinant of HDL metabolism. ApoA-II serum levels increased after fibrate treatment in PPAR␣ (ϩ/ϩ) mice (p Ͻ 0.01) whereas they decreased in PPAR␣ (Ϫ/Ϫ) mice (p Ͻ 0.01) (Fig. 3B). The increase in apoA-II levels was accompanied by an increase in hepatic apoA-II mRNA levels in PPAR␣ (ϩ/ϩ) mice (Figs. 3B and 4). No change in hepatic apoA-II mRNA levels was present in PPAR␣ (Ϫ/Ϫ) mice (Figs. 3B and 4). (Table I), previous studies have shown that administration of fibrates to both rodents and humans results in lower triglyceride concentrations and lower hepatic apoC-III mRNA (21,24,25). In addition, the apoC-III gene has a PPRE and plays a critical role in the control of triglyceride metabolism. Combined, these observations suggest that fibrate regulation of triglyceride metabolism occurs through a PPAR␣-dependent pathway. Consistent with this idea, PPAR␣ (ϩ/ϩ) mice fed the Wy 14,643 diet had significantly lower serum triglyceride concentration compared with untreated control (ϩ/ϩ) mice (p Ͻ 0.001) (Fig. 5A). In contrast, the characteristic lowering of triglyceride levels in response to fibrate administration was not observed in the PPAR␣ (Ϫ/Ϫ) mice fed the Wy 14,643 diet compared with controls (Fig. 5A).

Fibrate Treatment Reveals Abnormalities in Triglyceride and apoC-III Metabolism in PPAR␣ (Ϫ/Ϫ) Mice-Although no major difference in basal serum triglyceride levels between PPAR␣ (ϩ/ϩ) and (Ϫ/Ϫ) mice was detected
Consistent with the hypothesis that PPAR␣ has a major role in regulating apoC-III levels accompanying major changes in triglyceride metabolism, we observed a decrease in liver apoC-III mRNA levels in PPAR␣ (ϩ/ϩ) mice fed the Wy 14,643 diet (p Ͻ 0.001) (Fig. 5, A and B). In contrast, liver apoC-III mRNA levels were not affected in the PPAR␣ (Ϫ/Ϫ) mice fed the Wy 14,643 diet compared with untreated controls (Fig. 5, A and B). DISCUSSION Since peroxisome proliferators induce altered expression of genes encoding peroxisomal ␤-oxidation enzymes, microsomal -oxidation enzymes, and apolipoproteins, a role in lipid homeostasis for these compounds can be hypothesized. This idea is further supported by the known triglyceride-lowering effects of fibrates, a commonly used class of hypolipidemic agents. Recently it was shown that several peroxisome proliferators including fibrates and a fatty acid derivative are capable of activating PPAR␣. To better delineate a role for PPAR␣ in lipid homeostasis, we examined lipid and lipoprotein metabolism in PPAR␣-deficient mice.   2. Size distribution of HDL particles in PPAR␣ (؉/؉) and PPAR␣ (؊/؊) mice. HDL particles were isolated by density equilibrium ultracentrifugation and electrophoresed in a gradient polyacrylamide gel. The gel was stained for protein with Coomassie Brilliant Blue R-250. Lane 1 contains molecular weight markers. Lanes 2 and 3 represent HDL profiles from PPAR␣ (Ϫ/Ϫ) and PPAR␣ (ϩ/ϩ) mice, respectively.
Serum lipid and lipoprotein parameters were markedly altered in PPAR␣ (Ϫ/Ϫ) mice compared with PPAR␣ (ϩ/ϩ) mice. Most striking was the consistent increase in total and HDL cholesterol levels in the PPAR␣ (Ϫ/Ϫ) mice. This increase in HDL cholesterol levels was associated with a parallel elevation in serum apoA-I levels, which was associated with higher levels of hepatic apoA-I mRNA levels. Despite the higher level of serum HDL cholesterol, the composition and the size distribution of HDL particles was not different between the two genotypes. This suggests that the increase in serum HDL cholesterol in the PPAR␣ (Ϫ/Ϫ) mice was essentially due to an increase in the amount of circulating HDL lipoproteins. Another strong argument supporting an active role of PPAR␣ in the control of HDL metabolism was provided by the absence of a similar regulatory effect of fibrate lipid-lowering drugs on liver apoA-I and apoA-II mRNA and serum apoA-I and apoA-II levels in PPAR␣ (Ϫ/Ϫ) animals. In fact, apoA-I mRNA levels were, as reported previously (20,26), decreased after fibrate treatment in PPAR␣ (ϩ/ϩ) animals, whereas no regulation was observed in the (Ϫ/Ϫ) mice. Similar to previous work in humans (27), apoA-II mRNA and serum levels were only increased by fibrates in PPAR␣ (ϩ/ϩ), but not in (Ϫ/Ϫ) mice. Combined, these observations clearly establish PPAR␣ as a key regulatory factor in HDL metabolism.
The pathways through which PPAR␣ and its activators/ligands alter apolipoprotein expression and HDL metabolism is an active area of research. There is some confusion in this area that stems from differences between human and rodent apoA-I gene regulation by fibrates. In rats, serum HDL cholesterol and hepatic apoA-I mRNA levels are typically down-regulated by fibrates (20,26,28), whereas these parameters increase to a variable extent in humans (26,29,30). The mechanisms by which fibrates exert an overall negative effect on the minimal promoter of the mouse apoA-I gene are as of yet not defined.The present results, which demonstrate an increase in liver apoA-I mRNA levels under basal conditions in PPAR␣ (Ϫ/Ϫ) mice and show no regulatory effect of fibrates in these mice, are consistent with a negative regulatory effect of PPAR activators on rodent apoA-I expression (20,26). This indicates that PPAR␣ is one of the key players in determining liver apoA-I expression.
Human liver apoA-II expression is stimulated by PPAR␣. In fact, PPAR induces apoA-II expression through interaction with a PPRE located in the apoA-II J site (27). The absence in PPAR␣ (Ϫ/Ϫ) mice of increased apoA-II mRNA and serum levels, typically observed in PPAR␣ (ϩ/ϩ) mice after fibrate treatment, indicates that regulatory mechanisms similar to those for the human apoA-II gene must exist in mice. The regulatory function of PPAR␣ on gene expression of these two major HDL apolipoproteins may have important clinical implications in man. Since apoA-I and apoA-II are major determinants of HDL metabolism, alterations in their gene expression could significantly affect reverse cholesterol transport pathway, which seems to protect against coronary atherosclerosis.
PPAR␣ (ϩ/ϩ) mice fed the 0.1% Wy 14,643 diet have significantly lower hepatic apoC-III mRNA levels and triglyceride concentration compared with control untreated (ϩ/ϩ) mice. This is consistent with previous reports showing a similar reduction in apoC-III levels in rats after treatment with hypolipidemic fibrates (21,24,25). In contrast, PPAR␣ (Ϫ/Ϫ) mice fed Wy 14,643 did not exhibit the prototypical response to a peroxisome proliferator since hepatic apoC-III mRNA and triglyceride levels were not different compared with control (ϩ/ϩ) mice. These observations demonstrate that Wy 14,643-induced reductions in hepatic apoC-III mRNA and triglyceride metabolism are mediated by PPAR␣ and hence establish a role for PPAR␣ in the control of VLDL and triglyceride metabolism in addition to its role in HDL metabolism described above.
Our observations are also consistent with previous studies, which suggested that activation of PPAR␣ results in lower serum triglyceride levels in animals and man. Steady state triglyceride levels are dependent on two pathways, endogenous synthesis and tissue clearance. Production of triglycerides in the liver is controlled in large part by substrate (fatty acids) availability, whereas tissue clearance of triglycerides is dependent on lipoprotein lipase activity and apolipoprotein C-III. Fibrates can affect both of these processes (reviewed in Ref. 4). For example, fibrates have been shown to stimulate lipolysis and clearance of triglycerides due to decreased transcription and production of apoC-III (21,24,25), an apolipoprotein that limits tissue clearance of triglyceride. The effect of peroxisome proliferators on apoC-III expression may involve competition by PPAR for binding to a cis-acting sequence on the apoC-III promoter as well as a direct repression of hepatic nuclear factor-4 expression (25). In addition to apoC-III-mediated effects on the clearance of triglyceride-rich lipoproteins induced by PPAR␣, this receptor also influences production rates of these particles. Fibrate activation of PPAR␣ in the liver stimulates fatty acid uptake and conversion to acyl-CoA derivatives by the induction of the genes coding for the fatty acid transporter protein 2 and acyl-CoA synthase (31)(32)(33). The resulting acyl-CoA derivatives in hepatocytes are then more efficiently oxidized by induction of fatty acid ␤-oxidation pathways in peroxisomes and mitochondria (see above). Besides modulating ␤-oxidation of fatty acids, PPAR␣ activation can also inhibit de novo fatty acid synthesis that contributes to decreased triglyceride synthesis and VLDL production (reviewed in Ref. 4). Hence, both enhanced catabolism of triglyceride-rich particles as well as reduced secretion of VLDL particles are mechanisms that contribute to the hypolipidemic effect of PPAR␣ activation.
Our results clearly show that PPAR␣ modulates lipid metabolism in a homeostatic mechanism since mice lacking functional PPAR␣ have higher levels of serum HDL. In addition, PPAR␣ has an important role in mediating the effects of fibrates on apolipoprotein, HDL, and triglyceride metabolism. Our observations in the PPAR␣-deficient mouse warrant further studies to delineate the role of normal and abnormal PPAR activity in man. It may be especially interesting to evaluate genetic linkage in the PPAR␣ gene from patients with abnormalities in lipid metabolism such as hypoalphalipoproteinemia, hypertriglyceridemia, or combined hyperlipidemia. When existent, statistically significant differences between treated and untreated animals of the same genotype, as well as between wild-type and deficient mice are indicated by asterisks (Mann-Whitney test, * p Ͻ 0.05; ** p Ͻ 0.01; *** p Ͻ 0.001). B, representative Northern blot showing the differences in apoC-III and acyl-CoA oxidase mRNA levels in PPAR␣ (ϩ/ϩ) and (Ϫ/Ϫ) mice. As a control for loading efficiency, the Northern blots were rehybridized with a GAPDH probe.