Peroxisome Proliferator-activated Receptor-α Regulates Lipid Homeostasis, but Is Not Associated with Obesity

Considerable controversy exists in determining the role of peroxisome proliferator-activated receptor-α (PPARα) in obesity. Two purebred congenic strains of PPARα-null mice were developed to study the role of this receptor in modulating lipid transport and storage. Weight gain and average body weight in wild-type and PPARα-null mice on either an Sv/129 or a C57BL/6N background were not markedly different between genotypes from 3 to 9 months of age. However, gonadal adipose stores were significantly greater in both strains of male and female PPARα-null mice. Hepatic accumulation of lipids was greater in both strains and sexes of PPARα-null mice compared with wild-type controls. Administration of the peroxisome proliferator WY-14643 caused hepatomegaly, alterations in mRNAs encoding proteins that regulate lipid metabolism, and reduced serum triglycerides in a PPARα-dependent mechanism. Constitutive differences in serum cholesterol and triglycerides in PPARα-null mice were found between genetic backgrounds. Results from this work establish that PPARα is a critical modulator of lipid homeostasis in two congenic mouse lines. This study demonstrates that disruption of the murine gene encoding PPARα results in significant alterations in constitutive serum, hepatic, and adipose tissue lipid metabolism. However, an overt, obese phenotype in either of the two congenic strains was not observed. In contrast to earlier published work, this study establishes that PPARα is not associated with obesity in mice.

Peroxisome proliferators are a diverse class of compounds that include commercially used plasticizers (e.g. phthalates), industrial solvents (e.g. trichloroethylene), herbicides (e.g. lactofen), hypolipidemic drugs (e.g. fibrates), naturally occurring chemicals (e.g. phenyl acetate), and hormones (e.g. dehydroepiandrosterone sulfate) (1,2). Administration of peroxisome proliferators to rodents results in numerous hepatic alterations, including an increase in the number and size of peroxisomes; hepatomegaly; increased expression of genes encoding peroxisomal, mitochondrial, and microsomal fatty acid-metabolizing enzymes; and subsequent modulation of lipid homeostasis characterized by increased oxidation of fatty acids, decreased serum lipids, and reduced adipose stores (1). All of these effects are mediated by PPAR␣ 1 since PPAR␣-null mice are refractory to these changes when administered the prototypical peroxisome proliferator WY-14643 (3)(4)(5). In addition to modulation of lipid metabolism induced by peroxisome proliferators, a central role for PPAR␣ in lipid homeostasis during periods of fasting and in response to dietary fatty acids has also been established (6 -9). Thus, it is clear that PPAR␣ regulates lipid homeostasis in response to treatment with peroxisome proliferators, dietary fatty acids, and possibly endogenous fatty acids released during fasting.
The PPAR␣-null mouse was generated to identify PPAR␣dependent regulation induced by a variety of stimuli. Most of the early reports for this mouse line used mice with a mixed genetic background (C57BL/6N ϫ Sv/129) (3, 4, 9 -12). After the initial production (3), the PPAR␣-null mouse was subsequently backcrossed at the National Institutes of Health to obtain a pure Sv/129 line. The Sv/129 line of PPAR␣-null mice has been used extensively by many research groups to demonstrate that alterations induced by PPAR␣ activation require PPAR␣ (5,6,. There are a number of recent studies (8, 39 -46) that used PPAR␣-null mice on a C57BL/6 background that were generated from several rounds of backcrossing with an unidentified substrain of the C57BL/6 mouse line to the original mixed genetic background PPAR␣-null mice in an independent laboratory (42). However, due to the strategy used to generate PPAR␣-null mice (3), backcrossing to the C57BL/6N background requires backcrossing mice at least 10 generations to obtain a fully congenic mouse line (47).
The construction of the PPAR␣-null mouse used recombinant DNA and cells from two strains of mice, Sv/129 Jae and C57BL/6N (3). For the PPAR␣-null mouse line, the Sv/129 mouse was the source of the genomic DNA library used to construct a targeting vector and the embryonic stem cells used for transfection of a targeting vector, whereas the C57BL/6N mouse (NIH substrain) was the source of donor blastocysts used for microinjecting the heterozygous embryonic stem cells.
* The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18  Thus, the F 1 offspring from mating the chimeric mice generated by this approach were not congenic, but contained the genetic background of both Sv/129 and C57BL/6N mice. Although many published phenotypes for the PPAR␣-null mouse have been reported that have significant influence on lipid metabolism, many of these reports focused on mice that were either of mixed genetic background or congenic Sv/129 mice. In this work, the phenotypic characterization of lipid metabolism in wild-type or PPAR␣-null mice on either a pure Sv/129 or C57BL/6N genetic background was performed in both male and female mice to determine if the phenotype is consistent between congenic mouse lines.

Generation of Purebred Mice
Sv/129 (Jae Substrain)-Male chimeric mice for the targeted PPAR␣ allele (3) were bred with Sv/129 Jae females since this line is the same genotype as the embryonic stem cells used to generate the chimeric mice. The heterozygous F 1 agouti offspring from this breeding were subsequently crossed using brother-sister matings to obtain F 2 purebred wild-type or PPAR␣-null mice. The homozygous F 2 wild-type or PPAR␣-null mice were used to generate F 3 homozygotes, which were then randomly assigned to breeding cages to establish a larger colony of mice to perform experiments. The Sv/129 mice used for this work were from the F 6 generation of mice from this colony.
C57BL/6N (NIH Substrain)-The male chimeras described above were mated with purebred C57BL/6N females to obtain F 1 offspring. The heterozygous F 1 agouti offspring from this breeding were then backcrossed with purebred C57BL/6N mice (either heterozygous male ϫ wild-type female or heterozygous female ϫ wild-type male). The heterozygous F 2 offspring with black coat color were then removed and backcrossed with either male or female wild-type mice, and this process was continued until the F 10 generation of mice was obtained. Heterozygous F 10 mice were then crossed to produce homozygous wild-type or PPAR␣-null mice, and the homozygous F 11 mice were randomly distributed to make a breeding colony of mice to obtain F 12 mice for phenotypic analysis. Assessment of Body, Liver, and Adipose Weights 6 -8-week-old male or female PPAR␣ ϩ/ϩ or PPAR␣ Ϫ/Ϫ mice on either a C57BL/6N (F 12 generation) or an Sv/129 (F 6 generation) background were housed four to five animals per cage in a temperature-and lightcontrolled environment (T ϭ 25°C, 12-h light/12-h dark cycle). Mice were weighed every month for 9 months. Cohorts of mice were killed at the age of 12-14 weeks or 9 months by overexposure to carbon dioxide. Blood was collected by cardiac puncture for isolation of serum. Serum analysis of lipids and lipoproteins was performed as described below. Liver and gonadal fat pads were removed, weighed, snap-frozen, and stored at Ϫ80°C until further analysis. An additional section of liver was fixed in phosphate-buffered formaldehyde for analysis of liver lipid accumulation as previously described (48).

Feeding Experiments
10 -12-week-old male or female PPAR␣ ϩ/ϩ or PPAR␣ Ϫ/Ϫ mice on either a pure C57BL/6N (F 12 generation) or an Sv/129 (F 6 generation) background were housed three to five animals per cage as described above. Mice from both strains were fed either a control diet or one containing 0.1% WY-14643 for 7 days. Mice were killed by overexposure to carbon dioxide, and livers were removed, weighed, and snap-frozen until further use. Serum was obtained from whole blood collected from individual mice and used fresh for analysis of serum lipids and lipoproteins. Gonadal adipose was removed, and the weight was recorded for each mouse.

Lipid and Lipoprotein Measurements
Serum lipids (cholesterol and triglycerides) and high density lipoprotein cholesterol were measured as previously described (48).

RESULTS
Constitutive Phenotype-Monthly body weight measurements revealed small differences in average body weight between wild-type and PPAR␣-null mice on either the Sv/129 or C57BL/6N background (Fig. 1). Body weight was significantly higher in male PPAR␣-null mice on an Sv/129 background compared with the respective wild-type controls at 3-4 months of age (Fig. 1). Although average body weight tended to be higher in male and female PPAR␣-null mice on both genetic backgrounds, these differences were not statistically different (Fig. 1). Liver weights were similar between PPAR␣-null and wild-type mice of both sexes compared with the respective controls (Tables I and II). Although liver weights were not significantly different between genotypes, hepatic accumulation of lipids was considerably higher in the livers of male PPAR␣-null mice of both strains after 6 months (Fig. 2). Similar results were observed with female mice (data not shown). PPAR␣-null mice had significantly larger gonadal adipose stores than the respective wild-type controls, and this effect was slightly more pronounced in female PPAR␣-null mice compared with male mice (Tables I and II). Although internal adipose stores were significantly greater in PPAR␣-null mice than in controls, the overall sizes of 7-8-month-old male and female wild-type and PPAR␣-null mice were not markedly different on either an Sv/129 or a C57BL/6N background (Fig. 3).
Serum concentrations of cholesterol and high density lipoprotein cholesterol were significantly higher in 9-month-old purebred Sv/129 PPAR␣-null mice than in wild-type controls (Table II). This effect was observed in both male and female mice, with no apparent difference in the magnitude of these effects (Table II). Serum levels of triglycerides were similar in Sv/129 PPAR␣-null and wild-type mice (Table II). Serum concentrations of cholesterol and high density lipoprotein cholesterol were similar in 9-month-old purebred C57BL/6N PPAR␣-null mice and wild-type controls (Table II). This was observed in both male and female mice (Table II). In contrast to Sv/129 mice, serum levels of triglycerides were significantly higher in both male and female PPAR␣-null mice compared with the respective wild-type controls (Table II).
Constitutive hepatic levels of mRNAs encoding mitochondrial fatty acid-metabolizing enzymes (very long chain and long chain acyl-CoA dehydrogenases) were significantly lower in both C57BL/6N and Sv/129 PPAR␣-null mice of both sexes compared with wild-type controls (Fig. 4), consistent with previous results (5). Constitutive hepatic levels of apoC-III were not different between genotypes or sexes in either the C57BL/6N or Sv/129 mouse strain (Fig. 4). Similarly, constitutive hepatic levels of mRNA encoding PPAR␥ were not different a Significantly different from wild-type controls (p Ͻ 0.05). b Significantly different from control and WY-14643-treated wild-type controls (p Ͻ 0.05).
WY-14643 Feeding Experiment-Liver weight was significantly higher in male and female wild-type mice fed WY-14643 compared with controls in both C57BL/6N and Sv/129 mice, and this effect was not different between strains (Table I). In contrast, liver weight was not different between male and female null mice compared with controls, and again there was no difference in this effect between C57BL/6N and Sv/129 mice (Table I). Consistent with previous studies, gonadal adipose stores were significantly lower in male and female wild-type mice fed WY-14643 for 1 week compared with controls, and this effect was not found in either strain of PPAR␣-null mice fed WY-14643 (Table I). Administration of WY-14643 to mice caused a significant decrease in serum triglycerides in both strains of purebred wild-type mice compared with untreated controls (Table II).
Hepatic levels of mRNAs encoding acyl-CoA oxidase; bifunctional enzyme; 3-ketoacyl-CoA thiolase; cytochrome P450 4A1; and very long chain, long chain, and medium chain acyl-CoA dehydrogenases were higher in wild-type mice fed WY-14643 than in controls, and these effects were not different between wild-type C57BL/6N and Sv/129 mice of both sexes (Fig. 4). The PPAR␣-null mice were refractory to increased levels of these mRNAs, and there was no difference in this effect between strains (Fig. 4). Hepatic mRNA for apoC-III was reduced in wild-type mice fed WY-14643 compared with controls (Fig. 4), and this effect was absent in both strains and sexes of the PPAR␣-null mice. DISCUSSION The original phenotypic assessment of PPAR␣-null mice on a mixed genetic background (C57BL/6N ϫ Sv/129) provided strong in vivo evidence that PPAR␣ mediates the pleiotropic response to peroxisome proliferators, including hepatomegaly, peroxisome proliferation, and induction of genes encoding peroxisomal and microsomal lipid-metabolizing enzymes (3). Although constitutive expression of peroxisomal and microsomal lipid-metabolizing enzymes was not influenced by targeted disruption of the PPAR␣ gene, hepatic accumulation of lipids was described in PPAR␣-null mice, suggesting that constitutive lipid homeostasis is altered in the absence of a functional PPAR␣ (3). Evidence that constitutive gene expression is altered in PPAR␣-null mice on an Sv/129 background was provided by the report that mRNAs encoding mitochondrial fatty acid-metabolizing enzymes are reduced compared with wildtype mice, whereas constitutive expression of mRNAs encoding peroxisomal and microsomal fatty acid-metabolizing enzymes is unaffected (5). This study also confirmed that many of the observations made in mixed background PPAR␣-null mice are consistently found in purebred Sv/129 mice, including an absence of peroxisome proliferator-induced hepatomegaly and induction of mRNAs encoding peroxisomal and microsomal lipidmetabolizing enzymes (5). This suggests that hepatic lipid accumulation found in PPAR␣-null mice may be the result of reduced mitochondrial fatty acid oxidation. Results from the present study confirm and extend this characterization by demonstrating that male and female C57BL/6N PPAR␣-null mice are refractory to the pleiotropic response induced by peroxisome proliferators and that constitutive hepatic lipid accumulation occurs as previously described. Furthermore, this work demonstrates that this response is similar between male and female PPAR␣-null mice on either a pure Sv/129 or C57BL/6N genetic background.
Serum lipids in mixed background PPAR␣-null mice were also reported to be altered compared with wild-type controls. PPAR␣-null mice on a mixed genetic background exhibit significantly higher serum levels of cholesterol, in particular high density lipoprotein cholesterol, compared with wild-type controls (4). Similar results were found in this study in both male and female PPAR␣-null mice on a pure Sv/129 genetic background, consistent with the observations made in mixed background mice. In contrast, higher levels of serum cholesterol were not found, whereas serum levels of triglycerides were significantly higher than controls in both male and female PPAR␣-null mice on a C57BL/6N background. These results suggest that the genetic background of the PPAR␣-null mouse can significantly influence serum lipid biochemistry, likely through interactions with other genes. The mechanisms underlying this difference are unclear. Nevertheless, purebred Sv/ 129 and C57BL/6N PPAR␣-null mice provide unique tools for studies investigating the role of altered serum cholesterol and triglycerides in the etiology of atherosclerosis. The C57BL/6 mouse strain is better suited for evaluating the mechanisms contributing to atherosclerosis since atherosclerotic plaques can be induced by feeding a high fat diet (49,50). Thus, the PPAR␣-null mouse line on a C57BL/6N genetic background may be well suited for this purpose since constitutively higher levels of lipids are a known risk factor for this disease (51).
As PPAR␣-null mice exhibit significant lipid accumulation that may be due in part to impaired mitochondrial oxidation of fatty acids, it is not surprising that adipose stores are significantly greater in this mouse line as well. Although it is clear from these results that purebred PPAR␣-null mice on a pure Sv/129 or C57BL/6N genetic background have larger stores of adipose and accumulate lipids in the liver, differences in body weight are not of sufficient magnitude to be indicative of an obese phenotype. In the original mixed background PPAR␣null mouse line, it was noted that adipose stores were significantly greater than controls with little difference in overall body weight (52). Similar reports of PPAR␣-null mice on an Sv/129 background are consistent with this observation in that large differences in body weight were not found even in male mice that are Ͼ1-year-old (26,27). Conflicting reports suggest that this phenotype may be influenced by other factors, including diet and genetics.
Costet et al. (42) provided evidence suggesting that the PPAR␣-null mouse may be a useful model to study obesity and that this phenotype is more prevalent in female mice than in male mice. In contrast to results presented in the present study, these investigators reported that body weight of PPAR␣null mice is significantly greater than controls in both sexes after 7 months of age. Consistent with previous work (4) and the present study, alterations in serum lipids, adipose stores, and hepatic lipid accumulation were also detected in PPAR␣- null mice compared with controls (42). The difference in body weight between male and female PPAR␣-null mice was attributed in part to differences in hepatic PPAR␥ expression and differences in hepatic lipid accumulation (42); however, these changes were not detected in the present study. It is critical to emphasize that the genetic background of the PPAR␣-null mice used for the analysis performed by Costet et al. (42) is unclear, as the substrain of the C57BL/6 mouse used for backcrossing was not identified, and the extent of backcrossing described (Ͻ10 generations) theoretically would not result in a congenic line of mice. Thus, the congenic control C57BL/6 mice of unknown substrain used for controls are likely inappropriate and may have resulted in incorrect comparisons. Indeed, significant differences in the functional properties of another xenobiotic receptor (aryl hydrocarbon receptor) are known to exist between C57BL/6N and C57BL/6J mouse lines (53,54), demonstrating the importance of backcrossing mice with the identical line used for blastocyst transfer in this case. Differences in control mouse chow may also have contributed to the difference in body weight observed in PPAR␣-null mice between the pres-ent study and that of Costet et al. (42), although the percentage of fat was similar (4.5%), suggesting that the genetic background is more likely a confounding variable in this work.
That dietary fatty acids may influence the phenotype of PPAR␣-null mice is also suggested by another report showing that purebred Sv/129 PPAR␣-null mice have larger adipose stores than controls (29). In contrast to data presented in this study and that of Costet et al. (42), these investigators reported that gonadal adipose stores and average body weight were greater in male PPAR␣-null mice compared with female PPAR␣-null mice (29). Although increased adipose stores and body weight in PPAR␣-null mice are consistent with this work, the fact that male PPAR␣-null mice on an Sv/129 background were reported to have larger adipose stores than female mice (29) illustrates how significant variation can occur between laboratories using an identical mouse line. The most likely explanation for this difference is the source of fat used for the control diet, which can significantly influence lipid metabolism in these mice (7).
Given the conflicting accounts of phenotypes for the PPAR␣-  4. Northern blot analysis of hepatic mRNAs encoding peroxisomal, microsomal, and mitochondrial fatty acid-metabolizing enzymes; apolipoproteins; or PPAR␥. Shown are wild-type (ϩ/ϩ) or PPAR␣-null (Ϫ/Ϫ) mice on either an Sv/129 (A) or a C57BL/6N (B) genetic background. Con, control; WY, WY-14643; AXO, acyl-CoA oxidase; BIEN, bifunctional enzyme; THIOL, 3-ketoacyl-CoA thiolase; CYP4A1, cytochrome P450 4A1; VLCAD, LCAD, and MCAD, very long chain, long chain, and medium chain acyl-CoA dehydrogenase, respectively. null mouse lines with respect to obesity, it is critical that investigators indicate the source of fat used for control and experimental diets in the future and the strain of congenic mouse used for analysis. This study provides details of the backcrossing performed at the National Institutes of Health with the original line of mice, which to date has been the sole source for distribution of PPAR␣-null mice to independent investigators.