1. Volume 5, Issue 6, Pages 426-437 (June 2007)
Hepatic Fibroblast Growth Factor 21 Is Regulated by PPARα and Is a Key Mediator of Hepatic Lipid Metabolism in Ketotic States 
Michael K. Badman, Pavlos Pissios, Adam R. Kennedy, George Koukos, Jeffrey S. Flier, Eleftheria Maratos-Flier 
Cell Metabolism 
Volume 5, Issue 6, Pages (June 2007)
DOI: /j.cmet
Copyright © 2007 Elsevier Inc. Terms and Conditions

2. Figure 1
FGF21 Is Upregulated in Livers of KD-Fed Mice Together with Transcripts for Other PPARα-Regulated Genes
(A) Output from microarray revealed significant induction of FGF21 in livers from mice fed a ketogenic diet (KD) for 30 days. (n = 3 per group; ∗p < 0.05.) In this and all other figures, error bars represent mean ± SEM.
(B) Confirmation of altered expression by qPCR analysis. Induction of FGF21 (normalized to cyclophilin) by KD occurred after feeding for 72 hr. Similar >25-fold induction was seen after 30 days of KD. (n = 6–8 per group; ∗p < 0.05.)
(C) GSEA heat map shows a significant enrichment of PPARα targets within transcripts upregulated by KD feeding (red) in KD versus chow conditions.
(D) qPCR confirmation of upregulation of PPARα targets. Multiple classical PPARα-responsive genes were induced (normalized to cyclophilin) by KD feeding, including uncoupling protein 2 (UCP2), fatty acid translocase (FAT/CD36), and the members of the ATP-binding cassette family ABCD2 and ABCB4. (n = 6–8 per group; ∗p < 0.05.)
Cell Metabolism 2007 5, DOI: ( /j.cmet )
Copyright © 2007 Elsevier Inc. Terms and Conditions

3. Figure 2
FGF21 Is Induced In Vivo by Fasting and In Vitro by PPARα Agonist and Conjugated Fatty Acids
(A) β-hydroxybutyrate levels increased significantly between 8 and 24 hr of fasting. In (A)–(G), n = 6 animals per group; ∗p < 0.05 versus control or t = 0.
(B) A significant increase in circulating nonesterified fatty acids (NEFAs) was seen after 24 hr of fasting.
(C) FGF21 induction (normalized to cyclophilin) paralleled the time course of increases in serum β-hydroxybutyrate and NEFAs. Significant elevation of FGF21 levels was noted only after 24 hr of fasting.
(D) After 48 hr of fasting, there was significant elevation of circulating β-hydroxybutyrate, which was rapidly suppressed upon refeeding to reach fed levels within 1 hr.
(E) Circulating NEFA levels following fasting and refeeding. The time course of NEFA clearance was also rapid, with NEFA returning to fed levels within 1 hr of refeeding.
(F) Expression of FGF21 (normalized to cyclophilin) following refeeding after a 48 hr fast. A rapid reduction in expression of FGF21 was noted after 1 hr of refeeding; fasting expression levels were attained 8 hr after refeeding.
(G) Serum FGF21 levels as assessed by radioimmunoassay (RIA) were elevated after a 48 hr fast and rapidly suppressed after chow refeeding, returning to fed levels after 8 hr.
(H) qPCR analysis of FGF21 expression (normalized to cyclophilin). In order to establish direct regulation of FGF21 by PPARα, isolated hepatocytes were serum starved and subsequently exposed to either 100 μM fenofibrate or vehicle for 2 hr. In (H) and (I), n = 4 culture dishes per group; ∗p < 0.05.
(I) qPCR analysis of FGF21 expression (normalized to cyclophilin) following exposure of isolated hepatocytes to BSA-conjugated fatty acids. No stimulation was noted with lauric acid (C12) or palmitic acid (C16). A modest elevation was seen with stearic acid (C18) and linoleic acid (C18:2). The greatest increase was seen in cells treated with oleic acid (C18:1).
Cell Metabolism 2007 5, DOI: ( /j.cmet )
Copyright © 2007 Elsevier Inc. Terms and Conditions

4. Figure 3 Effect of KD Feeding on PPARα Null Mice
(A) Expression of FGF21 (normalized to cyclophilin) in control and PPARα null fed, fasted, or KD-fed mice. PPARα null mice did not express FGF21 in the chow-fed state or following a 24 hr fast. In contrast, KD-fed PPARα null mice exhibited a substantial, though attenuated, induction in FGF21 expression. (n = 5 per group; ∗p < 0.05.)
(B) Circulating FGF21 levels paralleled hepatic gene expression. Fasting induced a significant increase in circulating levels only in wild-type mice, and increased circulating FGF21 was detectible in KD-fed PPARα null mice.
(C) After 24 hr of fasting, both control and PPARα null mice lost an equivalent amount of weight. The weight loss on KD feeding was less in wild-type animals than in PPARα null animals.
(D) Glucose decreased with both fasting and KD feeding in PPARα null mice.
(E) β-hydroxybutyrate failed to increase in fasted PPARα null mice. An increase was seen in both groups of animals when fed KD.
(F) No significant differences were noted in circulating triglycerides in the fasted or KD-fed states.
(G) While NEFA levels were the same in chow-fed or fasted wild-type animals, they were significantly elevated in fasted or KD-fed PPARα null mice.
(H) No significant changes in circulating cholesterol were observed.
(I) Gross appearance of liver (left lobe depicted) shows marked increase in size of the liver, reflecting fat accumulation in fasted and KD-fed PPARα null mice.
(J) Histological analysis showed no difference between control and PPARα null mice in the chow-fed state. Fasting increased fat accumulation in the liver of PPARα null mice. KD feeding caused accumulation of fat in a microvesicular pattern in control mice; there were marked changes in the pattern of fat accumulation in KD-fed PPARα null mice. Scale bar = 50 μm.
(K) Total hepatic triglyceride content of liver showed a marked increase in triglyceride accumulation in the livers of fasted or KD-fed PPARα null mice and paralleled gross changes.
Cell Metabolism 2007 5, DOI: ( /j.cmet )
Copyright © 2007 Elsevier Inc. Terms and Conditions

5. Figure 4 Effect of FGF21 Knockdown in the Ketotic State
(A) qPCR confirmation of changes in FGF21 mRNA (normalized to 18S). Hepatic FGF21 expression decreased by 70% following treatment with shFGF21. (n = 5 per group; ∗p < 0.05.)
(B) Analysis of circulating levels of FGF21 by RIA revealed a significant increase in circulating FGF21 due to KD feeding; circulating levels were reduced by treatment with shFGF21.
(C) Gross appearance of livers (left lobe depicted) from adenovirus-treated mice. The difference in appearance between chow-fed and KD-fed shLacZ-treated mice reflects the expected accumulation of lipid seen in noninfected animals. The fatty appearance of the liver was substantially more marked in mice injected with shFGF21. Scale bar = 5 mm.
(D) Histological analysis of adenovirus-treated mice revealed increased lipid accumulation in KD-fed mice; this was more marked in mice treated with shFGF21. Scale bar = 50 μm.
(E) Triglyceride content of liver. Triglyceride accumulation was increased in KD-fed mice, and there was a notable increase in total triglyceride accumulation in the livers of KD-fed mice treated with shFGF21.
Cell Metabolism 2007 5, DOI: ( /j.cmet )
Copyright © 2007 Elsevier Inc. Terms and Conditions

6. Figure 5 Effect of FGF21 Knockdown on Circulating Metabolites
(A) Serum from KD-fed mice injected with either shLacZ control or shFGF21. FGF21 knockdown was associated with onset of extremely lipemic serum.
(B–F) Significant elevations in triglycerides (B), cholesterol (C), and NEFAs (D) were seen in KD-fed mice treated with shFGF21 adenovirus compared to mice treated with control shLacZ virus. There were modest but statistically significant decreases in circulating β-hydroxybutyrate (E) and glucose (F) levels in KD-fed mice that trended lower than in chow-fed mice and were not altered by FGF21 knockdown. (n = 6–8 per group; ∗p < 0.05.)
(G) FPLC analysis shows a substantial increase in triglycerides in the chylomicron/VLDL fraction in KD-fed shFGF21-treated animals.
(H) Plasma triglycerides of KD-fed shFGF21-treated mice were substantially elevated after 4 days compared to control virus-treated mice.
(I) In KD-fed mice treated with control virus in which circulating triglycerides were low, treatment with the LPL inhibitor tyloxapol led to significant increase in circulating triglycerides (6-fold icnrease within 6 hr of injection) in the fed state. In mice receiving shFGF21 virus, plasma triglycerides were high, and tyloxapol did not change these levels. (Percent change from baseline shown.)
Cell Metabolism 2007 5, DOI: ( /j.cmet )
Copyright © 2007 Elsevier Inc. Terms and Conditions

7. Figure 6
Changes in Hepatic Expression of Genes Related to Lipid Metabolism
All gene expression was assessed by qPCR and normalized to 18S RNA.
(A–C) Expression of enzymes involved in fatty acid oxidation including acyl-CoA dehydrogenase very long chain (ACADVL; [A]), long chain (ACADL; [B]), and medium chain (ACADM; [C]) was elevated by KD feeding. FGF21 knockdown led to a return to chow-fed levels. (n = 6–8 per group; ∗p < 0.05.)
(D) Hydroxyacyl-CoA dehydrogenase (HADH) was significantly induced by KD; there was a modest but statistically significant decrease in expression in shFGF21-treated versus shLacZ-treated KD-fed animals.
(E and F) FGF21 knockdown led to a small but significant decrease in carnitine palmitoyltransferase 1a (CPT1a) and acyl-CoA oxidase (ACOX1) expression in KD-fed animals.
(G) Fatty acid translocase/CD36 was markedly induced by KD feeding, an effect partially reversed by FGF21 knockdown.
(H) Hydroxymethylglutaryl-CoA synthase 2 expression (HMGCS2) was significantly induced by KD feeding; treatment with shFGF21 returned expression to chow-fed levels.
(I) Treatment with shFGF21 produced a significant reduction in expression of 3-hydroxybutyrate dehydrogenase (BDH1).
Cell Metabolism 2007 5, DOI: ( /j.cmet )
Copyright © 2007 Elsevier Inc. Terms and Conditions

8. Figure 7 Schema Detailing Regulation and Potential Roles of FGF21
Both endogenous lipids and drugs activate PPARα, which interacts with the retinoid X receptor (RXR) and then binds to specific DNA response elements within the promoters of specific genes to regulate their expression. A number of genetic programs are stimulated by PPARα, including those responsible for β-oxidation and ketosis. FGF21 is a target for PPARα-mediated expression and is released from the hepatocyte. FGF21 may have autocrine or paracrine roles in the liver or endocrine effects upon distant targets such as adipose tissue. The mechanism by which FGF21 exerts actions on the hepatocyte is unclear, and the exact receptor or receptor complex for FGF21 has yet to be determined. Our loss-of-function studies demonstrate that FGF21 is required for the expression of a number of PPARα target genes. This may be due to an effect of FGF21 signaling on the PPARα signaling system or coregulatory molecules, or due to distinct effects on gene transcription.
Cell Metabolism 2007 5, DOI: ( /j.cmet )
Copyright © 2007 Elsevier Inc. Terms and Conditions