Peroxisome Proliferator-activated Receptor α Deficiency Abolishes the Response of Lipogenic Gene Expression to Re-feeding

The mRNA expression of lipogenic genes Scd-1 and Fas is regulated partly by the insulin-sensitive transcription factor SREBP-1c and liver X receptor α (LXRα). Compared with normal mice, the increase in the mRNA expression of hepatic Scd-1, Fas, and Srebp-1c was severely attenuated in peroxisome proliferator-activated receptor α (PPARα)-deficient mice during the transition from the starved to the re-fed states. The concentration of the membrane-bound form of SREBP-1c was also lower in the livers of the PPARα-deficient mice during re-feeding but there was little difference in the concentration of the active, nuclear form, or in the abundance of Insig-2a mRNA. The response of plasma insulin to starvation and re-feeding was normal in the PPARα-deficient mice. Rat hepatocytes transfected with an adenovirus encoding a dominant negative form of PPARα were resistant to the stimulatory effects of insulin on Fas and Scd-1 mRNA expression in vitro. When LXRα was activated in vivo by inclusion of a non-steroidal ligand in the diet, the expression of the mRNA for hepatic Srebp-1c, Fas, and Scd-1 was increased severalfold in mice of both genotypes and resistance associated with PPARα deficiency was abolished during re-feeding. However, although re-feeding the LXRα ligand induced the immature form of SREBP-1c equally in the livers of both genotypes, the concentration of the nuclear form remained relatively low in the livers of the PPARα-deficient mice. We conclude that intact PPARα is required to mediate the response of Scd-1 and Fas gene expression to insulin and that this is normally achieved directly by activation of LXRα.

Insulin-mediated regulation of hepatic lipid metabolism is central to the efficient control of plasma lipid concentration and the avoidance of excessive lipemia. Insulin resistance and its associated hyperinsulinemia are of common occurrence in metabolic diseases such as obesity and Type 2 diabetes (1), both of which are also associated with excessive hepatic synthesis of fatty acids and triacylglycerol (2). Both of these pathways are insulin-dependent processes. Hepatic overproduction of these lipids reflects, both in experimental animals (3) and in man (4), a selective hepatic insulin resistance in which lipid synthesis is stimulated by the accompanying hyperinsulinemia, whereas hepatic glucose output is not suppressed. A "worst case" metabolic profile results in which hyperglycemia co-exists with hyperlipidemia (3). The insulin-inducible transcription factor SREBP-1c (sterol regulatory element-binding protein-1c) 2 regulates the expression of a portfolio of glycolytic and lipogenic genes including fatty acid synthase (Fas) and stearoyl-CoA desaturase 1 (Scd-1) (5,6). SREBP-1c thus forms an integral part of the mechanism by which insulin co-ordinates hepatic glucose and lipid metabolism. Evidence is also emerging that implicates another transcription factor, peroxisome proliferatoractivated receptor ␣ (PPAR␣), in the regulation of SREBP-1c and its target genes. During starvation, fatty acids, rather than carbohydrates, are the major sources of fuel for body tissues, and PPAR␣ plays a major role in the up-regulation of genes encoding enzymes of fatty acid oxidation in the starved state (7,8).
It has been shown that activation of PPAR␣ induces transcription of Scd-1 (9) and other lipogenic genes (10). PPAR␣ deficiency also abolishes the natural diurnal periodicity of Fas mRNA and de novo lipogenesis (11) in the fed state. These processes are normally insulin-dependent and the diurnal changes result from the periodicity of plasma insulin concentrations, which reflect the daily pattern of food intake. Although SREBP-1c forms part of the signaling mechanism by which insulin stimulates lipogenesis (12,13), details of the signaling pathway remain unclear. It appears, however, that insulin may act by enhancing the production of a natural ligand for another transcription factor, LXR␣, which then activates transcription of Srebp-1c, one of its target genes (14). However, LXR␣ ligands activate lipogenic genes such as Fas and Scd-1 in mice lacking SREBP-1c, suggestive of direct LXR␣-mediated activation in a non-SREBP-1c dependent manner (5). LXR␣ has also been shown to interact directly with the Fas promoter (15) and to * This work was supported by a Medical Research Council (MRC) Programme directly activate the transcription of Chrebp (carbohydrateresponsive element-binding protein), another lipogenic transcription factor, in a non-SREBP-1c manner (16). In the present work, we have sought to provide answers to three important questions regarding the role of PPAR␣ in the insulin regulation of lipogenic gene expression. First, does PPAR␣ deficiency interfere with the normal insulin-mediated up-regulation of lipogenic genes during the starved to re-fed transition? Second, if so, does this abnormality reflect defective insulin action resulting directly from the abolition of PPAR␣ at the level of the hepatocyte? Finally, is PPAR␣ involved in insulin signaling through LXR␣ and, if so, does this occur via SREBP-1c-dependent or non-SREBP-1c-dependent pathways?

EXPERIMENTAL PROCEDURES
Animals-PPAR␣-null mice bred onto a SV/129 genetic background were kindly provided by Dr. J. Peters and Dr. F. J. Gonzalez (National Institutes of Health, Bethesda, MD). Wildtype SV/129 mice were used as controls. All mice were male, and were used between the ages of 16 and 25 weeks. Animals were maintained in a temperature-controlled cabinet (22-24°C) on a 12-h light/12-h dark cycle (lights on: 15.00 h) and were fed a commercial diet (Special Diet Services, Witham, Essex, UK) containing (by weight), 4.3% fat (0.02% cholesterol), 51.2% carbohydrate (mainly starch), 22.3% protein, and 7.7% ash. When required, the diet was supplemented with TO-901317 (0.01% w/w) obtained from Sigma. In some experiments, mice of both genotypes were starved for 24 h (food removed at 09.00 h) and were then either killed or re-fed with the normal diet or with the TO-901317-supplemented diet for periods of 6, 12, or 24 h. Mice were killed by cervical dislocation at the end of each re-feeding period. PPAR␣ deficiency had no effect on food intake during either the dark phase or the light phase of the diurnal cycle (11), nor did dietary supplementation with TO-901317. Mean (Ϯ S.E.) values for food intake for 4 mice (g/24 h) were 16.8 Ϯ 2.1 (wild-type fed chow), 14.9 Ϯ 1.7 (wild-type fed TO-901317), 15.2 Ϯ 2.0 (PPAR␣-deficient fed chow), and 17.0 Ϯ 3.6 (PPAR␣-deficient fed TO-901317). All procedures performed in this study were approved by the Local Ethics Review Committee of the University of Oxford and were carried out under the appropriate Home Office (UK) personal and project licenses in accordance with the Home Office (UK) Animals (Scientific Procedures) Act 1986.
mRNA Expression-Livers were freeze-clamped immediately after killing the mice and ground to a powder under liquid nitrogen. mRNA concentrations were determined by reverse transcription followed by real-time PCR (10) using an ABI PRISM Sequence Detection System (PE Applied Biosystems, Foster City, CA) or a Rotor-Gene 2000 Real Time Cycler (Corbett Research, Sydney, NSW, Australia). Primers and probes for mouse Srebp-1c, Scd-1, Fas, Aco, Lxr␣, ␤-actin, and Insig-2a were prepared as described previously (11,17) and for glucokinase and Chrebp as given by Dentin et al. (18). Reactions were carried out in triplicate in 30 l of TaqMan Universal PCR Master Mix (11) with ␤-actin as internal standard. All values were related to a curve generated by a standard liver preparation and were corrected for ␤-actin mRNA content.

Preparation of Liver Fractions Containing Nuclei and
Microsomes-Liver microsomes were obtained using a modification of the method of Pawar et al. (19). Briefly, livers were excised and most of the blood was removed by rapidly perfusing with ice-cold phosphate-buffered saline containing protease inhibitors (pepstatin, 5 g/ml; leupeptin, 5 g/ml; aprotinin, 2 g/ml) via the vena cava. The perfused livers were placed in ice-cold 0.25 M sucrose solution containing protease inhibitors (18) and minced with scissors. The minced livers were homogenized in the same buffer (6 ml per 0.5 g) using a Potter-Elvehjem Teflon pestle. The homogenate was centrifuged at 500 ϫ g for 5 min to obtain a crude pellet containing nuclei and a supernatant fraction. A fraction containing microsomal membranes was obtained by re-centrifuging the 500 ϫ g supernatant at 105,000 ϫ g for 1 h at 4°C. The microsomal pellet was re-suspended in Hepes buffer pH 7.6 (20). Nuclei were prepared from the 500 ϫ g pellet obtained after the first centrifugation using the method of Gorski et al. (21). Briefly, the pellet was suspended in 2 M sucrose containing protease inhibitors (aprotinin, 2 g/ml; leupeptin, 2 g/ml; benzamide 1 mM; dithiothreitol, 1 mM; and phenylmethylsulfonyl fluoride, 1 mM) and layered over a 2 M sucrose cushion. The nuclear fraction was pelleted by centrifugation at 80,000 ϫ g for 35 min at 4°C. A nuclear extract containing mature SREBP-1c was obtained following lysis of the nuclear fraction and ammonium sulfate precipitation (21).
Western Blotting-Samples containing 20 g of protein from either the nuclear extract or the microsomal fraction were loaded onto a 12% polyacrylamide gel. Following electrophoresis and transfer to nitrocellulose paper, the resulting blots were probed with a mouse monoclonal anti-SREBP-1c antibody (IgG 2A4, Neomarkers, Fremont, CA). The second antibody was sheep anti-mouse IgG conjugated with horseradish peroxidase (Amersham Biosciences). SREBP-1c-containing bands were detected by ECL and visualized by radioautography. Band densities were determined by scanning with a Bio-Rad "Gel-Doc" scanner.
Hepatocyte Transfection-Rat hepatocytes were prepared and cultured as described previously (22). After 3 h in culture, during which time the cells formed a monolayer, the medium was removed and the hepatocyes were transfected with adenovirus expressing either green fluorescent protein (Adv-GFP) or a dominant negative form of PPAR␣ (Adv-null) (23). Control dishes contained the adenoviral storage buffer only. After 3 h, the virus-containing medium was removed and replaced with virus-free medium containing either 100 nM of the PPAR␣ agonist GW7647 (23), or 78 nM insulin. In view of the extremely rapid degradation of insulin by hepatocytes, high initial concentrations were required to ensure that insulin levels approximated to physiological values during most of the incubation period (24). Control culture dishes and those containing the PPAR␣ agonist alone received a much lower concentration of insulin (0.1 nM). Dishes were incubated for 18 h, the cells were harvested, and the RNA was extracted using RNAzol (Tel-Test, Inc., Friendswood, TX). mRNA concentrations were determined using reverse transcriptase-PCR in an identical manner to that described above for whole liver samples.
Other Methods-Plasma and tissue lipid concentrations were assayed as described previously (11). Liver microsomal SCD-1 activity was assayed as described by Kim et al. (25). Tissue and subcellular fraction protein contents were measured by the method of Lowry and colleagues (26). Plasma glucose was determined enzymically in duplicate in neutralized perchloric acid extracts (27). Plasma insulin concentrations were measured in duplicate using a commercial mouse insulin ELISA enzyme immunoassay (Mercodia AB, Uppsala, Sweden).

RESULTS
Re-feeding following Starvation Fails to Restore Srebp-lc, Fas, and Scd-1 mRNA Expression in PPAR␣-deficient Mice-When normal mice were deprived of food for 24 h, the mRNA expression of hepatic Srebp-lc declined, as expected (6) to about 25% of that observed in the fully fed state (Fig. 1A). Re-feeding the starved animals for 6 h reversed this decline and expression was restored to that observed in the fully fed state after re-feeding for 24 h. The response of Srebp-lc mRNA expression to starvation and re-feeding in the livers of the PPAR␣-deficient mice differed markedly from that observed in the wild-type. First, in the PPAR␣-deficient mice fed ad libitum, the expression of Srebp-lc was somewhat lower than that in the wildtype (Fig. 1A) as observed previously (11). This differential expression was retained after 24 h starvation (Fig. 1A). The most striking feature, however, was the almost complete lack of response of Srebp-lc mRNA expression to re-feeding in the livers of the PPAR␣-null mice. The expression of Fas mRNA in each genotype also showed a pattern similar to that of Srebp-1c in response to starvation and re-feeding. Thus, in the PPAR␣-deficient mice the response of Fas mRNA to re-feeding was blunted compared with that in the wild-type (Fig. 1B). The response of Scd-1 mRNA was somewhat more complex (Fig. 1C) and, in the wild-type, showed an increase in expression after starvation followed by a decline after re-feeding for 6 h. Expression was restored to that observed in the fully fed state after a total of 24 h re-feeding. Nevertheless, similar to that observed for Srebp-1c and Fas mRNAs, Scd-1 mRNA expression was also highly resistant in the PPAR␣-deficient mice to the normal stimulatory effects of re-feeding for 24 h. Thus, after 24 h, hepatic Scd-1 mRNA concentration in this genotype was less than 20% of that observed in the animals fed ad libitum. There were no significant abnormalities in the insulin response to starvation and re-feeding in the PPAR␣-deficient mice (Fig. 1D).
Adenovirus-mediated Attenuation of PPAR␣ Expression in Vitro Blunts the Response of Fas and Scd-1 mRNA to Insulin and to a PPAR␣ Ligand-Rat hepatocytes in primary culture were transfected with adenoviral constructs encoding either a dominant negative form of PPAR␣, a constitutive transcriptional co-repressor (Adv-PPAR␣-null), or GFP (Adv-GFP) (23). When compared with control hepatocytes that received the adenovirus storage buffer only, transfection with the Adv-GFP construct had no significant effect on the mRNA expression of any of the genes studied (see legend to Fig. 2). To establish if introduction of the dominant negative mutant was effective in attenuating PPAR␣ activity, hepatocytes were treated with a FIGURE 1. PPAR␣ deficiency attenuates the normal response of lipogenic genes in the liver to re-feeding following starvation with little effect on plasma insulin or glucose levels. Mice were fed a standard, low-fat diet ad libitum and killed at the mid-point of the dark phase of the light/dark cycle. Other similar mice had food withdrawn at mid-dark and were fasted for 24 h. Some of these mice were then killed. Others were re-fed the standard diet for periods of 6, 12, and 24 h and were killed at the end of each time period. All livers were freeze-clamped and mRNA abundance (A-C) was determined. SCD-1 activity was also measured in the microsomes of the same livers, portions of which were removed prior to freeze-clamping (F). Each point represents the mean Ϯ S.E. of 6 mice fed ad libitum and 3 mice at all other points. D and E, animals were terminally anesthetized and blood samples were withdrawn from the descending vena cava. Each point represents the mean Ϯ S.E. on plasma samples from 6 mice. In some cases the S.E. range lies within the area of the point. E, wild-type mice; F, PPAR␣-deficient mice. potent PPAR␣ ligand, GW7647, and induction of the mRNA for a known PPAR␣ target gene, acyl-CoA oxidase, was determined. Fig. 2A shows that, although GW7647 increased the expression of Aco mRNA about 10-fold in hepatocytes transfected with the construct encoding GFP, this increase was almost totally abolished in the hepatocytes transfected with the construct containing the dominant negative form of PPAR␣. Treatment of the Adv-GFP hepatocytes with the PPAR␣ ligand also led to a substantial induction of the mRNA encoding Fas, Scd-1, and Srebp-1c (Fig. 2), thus providing direct support in vitro for our observation that activation of PPAR␣ enhances hepatic de novo lipogenesis and lipogenic gene expression in vivo (10). Induction of Scd-1, and Fas mRNAs by GW7647 was significantly attenuated in the Adv-PPAR␣-null hepatocytes (Fig. 2, B and C). As expected, insulin enhanced the mRNA expression of all the lipogenic genes in the hepatocytes treated with the Adv-GFP construct. However, the stimulatory effect of insulin on the expression of Fas and Scd-1 mRNA was greatly reduced in the Adv-PPAR␣-null hepatocytes (Fig. 2, B and C). In contrast the response of SREBP-1c to insulin was not significantly attenuated by repression of the Ppar␣ gene (Fig. 2D).
PPAR␣ Deficiency Affects Neither the Expression nor the Exogenous Activation of LXR␣-Because the transcription of Srebp-lc, Fas, and Scd-1 is enhanced by activated LXR␣ (28) it was of interest to determine whether the effects of genotype, starvation, and re-feeding on the expression of these genes were accompanied by changes in the expression of Lxr␣ mRNA. In the livers of mice fed ad libitum, PPAR␣ deficiency did not affect Lxr␣ mRNA expression (Fig. 3). Starvation for 24 h increased expression significantly in the livers of the wild-type mice as noted previously and attributed to a fatty acid-mediated increase in LXR␣ transcription and mRNA stability (29). Expression also increased, to a similar extent, in the livers of the PPAR␣-deficient mice, suggesting that the effect of fatty acids on Lxr␣ was not mediated by PPAR␣. Re-feeding for a period as short as 6 h was sufficient to reduce the expression of Lxr␣ mRNA to concentrations similar to those observed in the livers of mice fed ad libitum. Once again, PPAR␣ deficiency had no effect on the extent of this decrease during the re-feeding period (Fig. 3).
To determine whether PPAR␣ deficiency affected the activation of hepatic LXR␣ by exogenous ligands in vivo, we re-fed a diet containing the potent non-steroidal LXR␣ ligand TO-901317 (30) to mice of both genotypes following a 24-h period of starvation. The effects of this treatment on the mRNA expression of the LXR␣ target genes Fas, Scd-1, and Srebp-1c were determined (Fig. 4). These results show that the patterns of recoveries of mRNA expression during re-feeding under these conditions were indistinguishable in the livers of both the wild-type and PPAR␣-deficient mice, in marked contrast to those observed following re-feeding of the un-supplemented chow diet (Fig. 1). Fig. 4 also shows that when the diet of the animals fed ad libitum was supplemented with TO-901317 the hepatic abundance of all the lipogenic mRNAs was increased 6 -8-fold when Intact PPAR␣ Is Required for the Lipogenic Response to Re-feeding FEBRUARY 22, 2008 • VOLUME 283 • NUMBER 8 compared with those on the un-supplemented diet (Fig. 1). This considerable increase probably resulted from the reinforcement of the direct stimulatory effect of LXR␣ activation (5,15,16) by an indirect effect operating by stimulation of SREBP-1c (5,6). LXR␣ activation also gave rise to other detailed changes in the response of Scd-1 and Fas to starvation and re-feeding. For instance, in the wild-type mice, Scd-1 mRNA expression decreased after starvation (Fig. 4C), in contrast to the increase observed after re-feeding with the un-supplemented diet (Fig. 1C). Furthermore, the "overshoot" of Fas mRNA expression observed after re-feeding the wild-type mice with the chow diet (Fig. 1B) was not observed when LXR␣ was activated in vivo during the re-feeding period. (Fig. 4B). These changes in lipogenic gene expression following activation of LXR␣ in vivo were not accompanied by any marked changes in the concentration of plasma glucose (Fig. 4E). However, plasma insulin concentrations were considerably decreased in wild-type mice but only marginally in the PPAR␣-deficient animals (compare Fig. 4D with Fig. 1D).
PPAR␣ Deficiency Decreases the mRNA Expression of Glucokinase (Gk) and Chrebp in the Fed and Re-fed States-In addition to their regulation by SREBP-1c, glucose directly regulates lipogenic and glycolytic gene function via changes in the expression of Chrebp (31) in a manner that is dependent upon the presence of Gk (18). In view of recent evidence that Chrebp is regulated by LXR␣ (16) in a glucose-dependent manner (32) we studied the effects of PPAR␣ deficiency on Chrebp and Gk mRNA expression and their responses to LXR␣ activation. When mice were fed a normal diet, Gk expression declined during starvation and increased following re-feeding in the livers of the wild-type mice (Fig. 5A). In the PPAR␣-deficient mice, Gk expression was lower than that in the wild-type mice in the fed state but it neither decreased upon starvation nor increased following refeeding (Fig. 5A). This pattern meant that Gk mRNA expression was considerably higher in the PPAR␣-deficient mice than in the wild-type in the starved state and suggested that glycolysis remained active following 24 h of starvation. The mRNA expression of Chrebp was also significantly lower in the livers of the knock-out mice in the re-fed state compared with that in the wild-type (Fig. 5C). However, Chrebp mRNA expression showed a consistent increase following starvation in the wild-type mice, which then declined following refeeding. This pattern differs from that described previously (18), possibly because the diet used in the present work contained a higher concentration of fat and a lower proportion of carbohydrate  than the high-carbohydrate, low-fat diet used in the previous study.
In the wild-type mice there was little, if any, stimulation of Gk mRNA expression resulting from activation of LXR␣ with dietary TO-901317 in the fed and re-fed states (Fig. 5B). Gk mRNA in the livers of the PPAR␣-deficient mice responded to a somewhat greater extent to dietary TO-901317 supplementation. Surprisingly, the greatest stimulatory effect of LXR␣ activation on Gk mRNA expression in both genotypes was observed following 24 h starvation (Fig. 5B), so that there was little effect of starvation or re-feeding on Gk mRNA abundance in mice of either genotype. Because there is no evidence that Gk is a direct LXR␣ target gene, any effect of LXR␣ activation must have arisen indirectly via SREBP-1c. Compared with Gk, Chrebp mRNA expression responded more vigorously to LXR␣ activation in the fed and re-fed states in both genotypes, so that there was a reversible reduction in Chrebp mRNA abundance upon starvation (Fig. 5D). There was no obvious difference in hepatic ChREBP mRNA abundance between genotypes in mice fed TO-901317 (Fig. 5D).
Activation of LXR␣ Modifies the Effects of PPAR␣ Deficiency on the Processing of SREBP-1c during Re-feeding-Western blotting of microsomal membranes from the livers of mice re-fed the chow diet showed a greater concentration of the immature form of SREBP-1c in the wild-type compared with the PPAR␣-deficient animals (Fig. 6A). This difference reflects the lower expression of Srebp-1c mRNA in the liver of the PPAR␣-deficient mice following re-feeding the chow diet (Fig. 1A). However, Western blotting of the nuclear extracts from the livers of each genotype (Fig. 6B) showed that these differences were not reflected in similar relative changes in the concentration of the active form of SREBP-1c. Although expression of mature SREBP-1c was significantly lower (p Ͻ 0.05) in the nuclei of the PPAR␣-deficient mice compared with the wild-type, this difference was relatively small in relation to the decrease in mRNA abundance. Li and colleagues (33) have also shown that PPAR␣ deficiency results in little or no difference in the nuclear abundance of SREBP-1c.
Insig-2a is a protein that prevents the activation of immature SREBP-1c (34,35) and, consistent with the small decrease in the nuclear concentration of the active SREBP-1c in the PPAR␣-deficient mice, there was only a small but significant (p Ͻ 0.05) difference in the mRNA expression of Insig-2a in the livers of the two genotypes following re-feeding after a 24-h period of starvation (Fig. 7). However, starvation gave rise to a large increase in Insig-2a mRNA expression in the livers of both genotypes compared with that observed when the mice were fed ad libitum. As observed previously (34) the increase in Insig-2a correlated with the decreased processing of SREBP-1c in the starved state.
When the non-steroidal LXR␣ ligand TO-901317 was included in the diet during the re-feeding period there was a large increase in the concentration of the microsomal, membrane-bound form of SREBP-1c compared with that observed following the re-feeding of chow only (Fig. 6A). In this case, however, there was no difference between the genotypes, reflecting the similarity in the expression of Srebp-1c mRNA in FIGURE 5. Effect of starvation and re-feeding on Gk and Chrebp mRNA levels. Groups of mice were treated in a similar manner to that described in the legend to Fig. 1, using either the standard diet or the standard diet supplemented with 0.01% of TO-901317. There were no significant differences in the expression of Gk or Chrebp at any of the 6-, 12-, and 24-h re-feeding points in mice of either genotype, so these values have been combined in the re-fed columns. Gray columns correspond to wild-type mice and black columns to PPAR␣-deficient mice. Each value represents the mean Ϯ S.E. of 3-8 mice in each group. Symbols * and ** indicate a significant (p Ͻ 0.05 and 0.01, respectively) effect of starvation, ‡ indicates a significant (p Ͻ 0.05) effect of re-feeding, and † indicates a significant difference (p Ͻ 0.05) between PPAR␣ knock-out and wild-type animals.
the livers of the two groups of mice after re-feeding with TO-901317 (Fig. 4A). However, blotting of the nuclear extracts from the mice fed the exogenous ligand showed a greater concentration of active SREBP-1c in the livers of the wild-type compared with the PPAR␣-deficient animals. Thus the processing of SREBP-1c during re-feeding with TO-901317 appeared to be retarded in the liver of the PPAR␣-deficient mice, compared with the wild-type, despite the fact that TO-901317 increased the concentration of the membranebound, immature form to the same extent in both genotypes.

PPAR␣ Deficiency Does Not Affect the Response of Plasma Glucose Concentration during Refeeding after Starvation-
There was no difference in the concentration of plasma glucose between the wild-type and PPAR␣-deficient mice fed ad libi-tum (Fig. 1E). Fasting for 24 h gave rise to a significant decline in plasma glucose in the wild-type mice and to an even greater extent in the PPAR␣-deficient mice (Fig. 1E). This relative hypoglycemia in the PPAR␣-deficient genotype may reflect either a lack of ATP for gluconeogenesis because of defective fatty acid oxidation (7), or a continued whole body reliance on glucose oxidation for energy that cannot be provided by utilization of fatty acids. The impaired up-regulation of PDK4 in the livers of starved PPAR␣-deficient mice (36) and the relatively high expression of glucokinase mRNA in these animals (Fig. 6A) suggests that glycolysis remains active, providing some support for the latter explanation. Re-feeding for only 6 h following starvation restored plasma glucose to pre-starvation levels in mice of both genotypes. There was no hyperglycemia in the PPAR␣-deficient mice compared with the normal mice suggesting that insulin effectively suppressed hepatic gluconeogenesis and hepatic glucose output in both genotypes.

DISCUSSION
The results presented here show that, in the liver, PPAR␣ deficiency affects the response of Srebp-1c, Fas, and Scd-1 mRNAs to re-feeding following starvation. The response pattern of the lipogenic genes in the PPAR␣-deficient mice differed in detail from each other. For instance, whereas refeeding for 24 h restored Fas mRNA concentration to the pre-starved level in the PPAR␣-deficient mice, this was not FIGURE 6. Activation of LXR␣ modifies the effects of PPAR␣ deficiency on SREBP-1c processing during re-feeding. Mice were starved for 24 h and either killed or re-fed the standard or the TO-901317-containing diet for 24 h. Livers were perfused with ice-cold phosphate-buffered saline (containing protease inhibitors), and microsomal membranes and nuclear extracts were obtained. Following SDS-PAGE and transfer to nitrocellulose, the blots were probed for SREBP-1c that was detected by ECL and radioautography. Band intensities were quantified by scanning. Each value represents the mean Ϯ S.E. relative to wild-type (WT) re-fed chow values of between 2 and 6 different preparations. Columns marked with * are significantly different (p Ͻ 0.05) from the values for the wild-type re-fed chow. Knock-out (KO) (PPAR-null) columns marked with † are significantly different (p Ͻ 0.05) from the corresponding wild-type value. FIGURE 7. Re-feeding following starvation rapidly suppresses the expression of Insig-2a mRNA in the livers of wild-type and PPAR␣deficient mice. Mice were treated in an identical manner to that described in the legend to Fig. 1 but were re-fed for only 6 h. Each column represents the mean Ϯ S.E. of the livers from 6 mice. Open bars, wild-type mice; filled bars, PPAR␣-deficient mice. Values in the columns for the refed mice marked * are significantly different (p Ͻ 0.01) from those for the starved mice. Values in the columns marked † denote a significant effect (p Ͻ 0.05) of PPAR␣ deficiency in the fed and re-fed states compared with the wild-type mice.
the case for Srebp-1c or Scd-1. Again, Scd-1 mRNA did not increase whatsoever even after 24 h re-feeding. Clearly, feeding for much longer than 24 h is required to restore Srebp-1c and Scd-1 mRNAs to the pre-starvation levels. These distinctive patterns presumably reflect the complexities of regulation of these genes, each of which is subject to control by its own distinct set of regulatory factors. Nevertheless, a common feature of each of these mRNA species to re-feeding in the PPAR␣deficient mice was the attenuated response compared with that observed in the wild-type, suggesting that PPAR␣ is normally responsible for transmitting lipogenic signals that mediate the transition from the starved to the fed state.
In the absence of PPAR␣, fasting results in hypoglycemia (7) (Fig. 1E) and relative hepatic steatosis (Fig. 8B), which might induce a counter-regulatory response, thereby antagonizing and delaying the re-expression of Srebp-1c. However, normoglycemia is restored in the PPAR␣-deficient mice following refeeding for 6 h (Fig. 1E), yet Srebp-1c expression is suppressed at this point and remains so for the next 18 h (Fig. 1A). Moreover, following re-feeding for 6 and 12 h, the relative steatosis was resolved in the PPAR␣-deficient mice (Fig. 8B) but lipogenic gene expression remained low. Thus there is no evidence that the suppressed response of hepatic lipogenic mRNAs to refeeding in PPAR␣-deficient mice is related to fasting hypoglycemia or steatosis.
We have previously shown that Ppar␥ mRNA expression is increased 2-fold in the livers of PPAR␣-deficient mice fed a low-fat chow diet compared with that in wild-type mice (11). However, even this relatively elevated expression of Ppar␥ remained less than 5% of the concentration of hepatic PPAR␣ in the intact mice. We do not consider that this small change in Ppar␥ expression would contribute significantly to the effects of PPAR␣ deficiency that we have described.
The attenuated response of lipogenic gene expression to refeeding in the livers of the PPAR␣-deficient mice raised two immediate questions. First, is the decreased mRNA expression of Srebp-1c in the PPAR␣-deficient mice accompanied by a decreased post-translational processing to give the nuclear, active form and, if so, does this account for the decreased expression of its targets, Scd-1 and Fas? Second, are the different responses of the two genotypes to re-feeding a direct result of the differing PPAR␣ status at the level of the hepatocyte and, if so does this reflect differences in the efficiency of insulin signaling during the re-feeding process? To address the first question, we compared the concentrations of the nuclear and microsomal forms of SREBP-1c in the livers of the two genotypes (Fig. 6). Although PPAR␣ deficiency was associated with a decreased microsomal, inactive form, consistent with the decreased mRNA (Fig. 1A), this decline was not accompanied by a correspondingly large decrease in the concentration of the nuclear, active form. Previous work using isolated hepatocytes (37) has also shown that the expression of Srebp-1c mRNA can be uncoupled from its post-translational processing by activation of LXR␣. Thus, irrespective of whether insulin is directly involved, it would appear unlikely that, despite the decreased abundance of Srebp-1c mRNA, the decreased Fas and Scd-1 mRNA expression during refeeding in PPAR␣-deficient mice resulted exclusively from a lack of active SREBP-1c.
To answer the second question, PPAR␣ expression was abolished in cultured hepatocytes in vitro by transfection of an adenoviral construct encoding a dominant negative form of PPAR␣ (Adv-PPAR␣-null hepatocytes) (Fig. 2). Possible nonspecific effects of transcription were accounted for by comparing the response of hepatocyes transfected with a similar adenoviral construct encoding GFP. The mutant PPAR␣ harbored twin substitutions in the ligand binding domain (23) and repressed endogenous wild-type activity as shown in the present work by abolition of the normal response of Aco, a PPAR␣ target gene, to the potent PPAR␣ ligand, GW7647 ( Fig. 2A). PPAR␣ deficiency in these transfected hepatocytes also abolished the normal stimulatory effects of the GW7647 ligand on the mRNA expression of Scd-1, Fas, and Srebp-1c. To determine whether a lack of insulin response contributed to the FIGURE 8. Effect of starvation and re-feeding on plasma and hepatic lipid concentrations. Mice were treated in a manner identical to that described in the legend to Fig. 1. Plasma samples were assayed for non-esterified fatty acids (NEFA) and liver samples were assayed for triacylglycerol (TAG). Each point represents the mean Ϯ S.E. of 3 mice in each group. E, wild-type mice; F, PPAR␣-deficient mice.
insensitivity of lipogenic genes to re-feeding in the livers of the PPAR␣-deficient mice, the Adv-GFP and Adv-PPAR␣-null hepatocytes were challenged with insulin in vitro for 18 h. As expected, insulin stimulated the mRNA expression of all three lipogenic genes in the control and Adv-GFP hepatocytes. The response of Fas and Scd-1 mRNA to insulin was blunted in the Adv-PPAR␣-null hepatocytes suggesting that the lack of response of these genes to refeeding in the PPAR␣-deficient mice was due, at least in part, to a disturbance in insulin signaling. The response of Srebp-1c to insulin was not, however, attenuated significantly in the Adv-PPAR␣-null hepatocytes suggesting that the lack of a response in vivo resulted from a factor(s) other than defective insulin signaling. Although there is no evidence for such an action, we cannot entirely discount the possibility that the adenovirus nonspecifically interfered with signaling between PPAR␣ and Srebp-1c.
Insulin forms only a part of the complex integrated response responsible for transmitting information by which the body regulates the physiological transformation between the starved and fed states. Our experiments in vitro cannot, of course, completely rule out the possibility that during the transition between the starved and the fed states in vivo, PPAR␣ deficiency compromises signaling by messengers other than insulin, such as counter-regulatory hormones like glucocorticoids and glucagon. As mentioned above, another possible factor is abnormal fatty acid metabolism, although we were unable to demonstrate any significant elevation of plasma NEFA in the PPAR␣-deficient compared with the wild-type mice after 6 h of re-feeding (Fig. 8A). Nevertheless, the possibility still remains that PPAR␣ deficiency may also confer an insensitivity of lipogenic gene expression to the changes in lipid metabolism characteristic of the transition from the starved to the fed state.
The mechanism by which defective PPAR␣ may interfere with the insulin regulation of Scd-1 and Fas remains obscure. However, it has been shown recently that insulin is capable of activating LXR␣, probably by increasing the endogenous provision of an activating LXR␣ ligand (14). Furthermore, in addition to their transcriptional activation by SREBP-1c, the transcription of both Scd-1 and Fas is mediated by activation of LXR␣ in a non-SREBP-1c-dependent manner (5,15,16). It is possible, therefore, that at least in part, insulin normally regulates FAS and SCD-1 by ligand-mediated activation of LXR␣ independent of SREBP-1c. If this is the case, then the insulinsignaling pathway may be disturbed under conditions of PPAR␣ deficiency by a lack of endogenous LXR␣-ligand. To test this idea, we investigated whether the defects in insulin signaling in the PPAR␣-null mice could be rescued by providing excess of an exogenous LXR␣ ligand during the re-feeding process. Fig. 4 shows that when the potent LXR␣ ligand TO-901317 (30) was included in the diet during re-feeding after starvation, the responses of Fas and Scd-1 mRNA in the two genotypes were virtually superimposable. Thus the original PPAR␣ dependence of insulin action was abolished under conditions in which the concentration of an LXR␣ ligand was not limiting. It appears, therefore, that PPAR␣ may be required for the normal, insulin-dependent production of an endogenous LXR␣ ligand. The almost identical patterns of Lxr␣ mRNA expression in the livers of the wild-type and PPAR␣-deficient mice during the fed/fasted/re-fed transitions (Fig. 3) showed that differences in the mRNA expression of Lxr␣ itself were not responsible for the differing sensitivities of lipogenic gene expression. Fig. 3 also shows that, in the livers of both genotypes, Lxr␣ mRNA expression increased in starvation. This effect has also been shown previously in the livers of starved wild-type animals, and has been ascribed to a PPAR␣-dependent mechanism activated by the starvation-induced increase in fatty acids (29). The present results suggest that, if fatty acids are involved in the increased LXR␣ mRNA expression, then this effect is independent of PPAR␣. Fig. 4 shows that the presence of TO-901317 during the refeeding period also abolished the inhibitory effects of PPAR␣ deficiency on the mRNA expression of Srebp-1c. Under these conditions, there was now no difference in Srebp-1c mRNA expression between the two genotypes. Consistent with this, the LXR␣-ligand also increased the microsomal form of SREBP-1c to an equal extent in the livers of both genotypes (Fig.  6). Despite this, however, the concentration of the active, nuclear form of SREBP-1c was decreased in the PPAR␣-deficient mice, compared with the wild-type. Activation of LXR␣ has been shown to increase Srebp-1c transcription but decrease processing in hepatocytes from normal rats (37). Under these conditions, processing was restored by insulin. In the present work, PPAR␣ deficiency may, therefore, interfere with the insulin-dependent processing of SREBP-1c in the presence of high concentrations of an LXR␣-ligand. The relative lack of nuclear SREBP-1c in the livers of the TO-901317-treated PPAR␣-deficient mice makes it unlikely that the increased Fas and Scd-1 mRNA expression occurs exclusively through the SREBP-1c pathway. In all likelihood, expression of these genes is increased by direct LXR␣-mediated activation of Fas and Scd-1, or by their activation by Chrebp, a recently discovered LXR␣ target (16). Consistent with this, in the present work, Chrebp mRNA expression increased by ϳ3-fold in the fed state when LXR␣ was activated in vivo by TO-901317. When mice were fed the normal, un-supplemented diet, Chrebp mRNA expression was decreased by PPAR␣ deficiency in all nutritional states, similar to that observed with Srebp-1c, Fas, and Scd-1. However, activation of LXR␣ in vivo was unable to completely restore Chrebp mRNA expression during the re-feeding period although the difference between the wild-type and PPAR␣-deficient mice was no longer significant (Fig. 5).
The role of PPAR␣ in the development of insulin resistance is controversial. On the one hand, stimulation of PPAR␣ activity by administration of potent ligands such as WY 14,643 in animal models of obesity has been reported to decrease hepatic triacylglycerol concentrations and increase glucose tolerance (38). PPAR␣ is also required to mediate the lipid-lowering effects of hyperleptinemia in liver and adipose tissue (39). On the other hand, PPAR␣ activation gives rise to an increase in the expression of TRB-3 mRNA, which encodes a protein that decreases insulin sensitivity by inhibition of insulin signaling via Akt/PKB (40). PPAR␣ deficiency has also been reported to protect against insulin resistance induced by fat feeding (41), although this conclusion has recently been challenged (42). Finally, although the present work has concentrated on the effects of PPAR␣ deficiency on lipogenic gene mRNA expres-sion, it should be recognized that the overall lipogenic response to re-feeding following starvation occurs at other levels of regulation, particularly those that involve post-translational effects. These, often more rapid responses, can contribute to the overall changes in whole body metabolism. However, we have studied the enzymic activity of SCD-1, which can be rapidly regulated at the level of protein stability (43), and found that the patterns of change in activity in the livers of both genotypes closely paralled those of Scd-1 mRNA expression in all nutritional states (Fig. 1F), as did those of SCD-1 protein expression as measured by Western blotting (results not shown).
In conclusion, therefore, it appears that intact PPAR␣ forms part of the signaling mechanism by which insulin activates the LXR␣-mediated transcription of Fas and Scd-1 during re-feeding. That an exogenous non-steroidal ligand of LXR␣ is effective in rescuing PPAR␣ deficiency suggests that intact PPAR␣ is somehow involved in the generation of an endogenous LXR␣ ligand. Why a transcription factor, whose primary role is the coordination of fatty acid oxidation in the starved state, is required to facilitate fatty acid synthesis during the starved/fed transition is not clear. Nevertheless, it has been shown previously that PPAR␣ activation stimulates the transcription of Scd-1 (9) and other lipogenic genes (10) and it was suggested that the physiological purpose of this response is to compensate for the oxidative removal of structurally important fatty acids during the previous period of starvation (10). This interpretation was supported by the observation that abolition of PPAR␣mediated fatty acid oxidation during the period of low food intake in the diurnal cycle also abolished the insulin-mediated increase in lipogenic gene expression and lipogenic flux in the liver when food intake increased naturally during the diurnal cycle in vivo (11). Interestingly, a recent report has documented a role for newly synthesized fatty acids in the activation of PPAR␣ (44). Such a role, together with the present results, provides evidence for a symbiotic interdependence of PPAR␣ signaling and newly synthesized fatty acids in the coordinated regulation of hepatic lipid metabolism. We suggest that the physiological purpose of this relationship is to ensure the correct balance between de novo lipogenesis and fatty acid oxidation in the liver, consistent with the energy requirements when food intake of the animal fluctuates over a period of time (45).