PPARδ activation attenuates hepatic steatosis in Ldlr-/- mice by enhanced fat oxidation, reduced lipogenesis, and improved insulin sensitivity.

PPARδ regulates systemic lipid homeostasis and inflammation, but its role in hepatic lipid metabolism remains unclear. Here, we examine whether intervening with a selective PPARδ agonist corrects hepatic steatosis induced by a high-fat, cholesterol-containing (HFHC) diet. Ldlr−/− mice were fed a chow or HFHC diet (42% fat, 0.2% cholesterol) for 4 weeks. For an additional 8 weeks, the HFHC group was fed HFHC or HFHC plus GW1516 (3 mg/kg/day). GW1516-intervention significantly attenuated liver TG accumulation by induction of FA β-oxidation and attenuation of FA synthesis. In primary mouse hepatocytes, GW1516 treatment stimulated AMP-activated protein kinase (AMPK) and acetyl-CoA carboxylase (ACC) phosphorylation in WT hepatocytes, but not AMPKβ1−/− hepatocytes. However, FA oxidation was only partially reduced in AMPKβ1−/− hepatocytes, suggesting an AMPK-independent contribution to the GW1516 effect. Similarly, PPARδ-mediated attenuation of FA synthesis was partially due to AMPK activation, as GW1516 reduced lipogenesis in WT hepatocytes but not AMPKβ1−/− hepatocytes. HFHC-fed animals were hyperinsulinemic and exhibited selective hepatic insulin resistance, which contributed to elevated fasting FA synthesis and hyperglycemia. GW1516 intervention normalized fasting hyperinsulinemia and selective hepatic insulin resistance and attenuated fasting FA synthesis and hyperglycemia. The HFHC diet polarized the liver toward a proinflammatory M1 state, which was reversed by GW1516 intervention. Thus, PPARδ agonist treatment inhibits the progression of preestablished hepatic steatosis.

with GW1516. Additionally, in db / db mice injected with adenoviral PPARD (ad PPARD ), liver lipid droplets were reduced coincident with decreased SREBP-1c processing, suggesting attenuated lipogenesis ( 23 ). Despite reports suggesting reduced hepatic steatosis, other studies have demonstrated that PPAR ␦ activation exerts either no effect ( 24 ) or induces liver TG accumulation ( 19,25 ). In db / db mice, GW1516 treatment resulted in increased hepatic TG as a result of direct transcriptional activation of acetyl-CoA carboxylase ␤ ( Acc2 ), and increased lipogenic gene expression ( 19 ). ad PPARD gene delivery to high-fat fed Ldlr Ϫ / Ϫ mice increased both hepatic lipogenic gene expression and hepatic TG accumulation ( 25 ). None of these studies actually measured hepatic FA oxidation or synthesis.
The claim that PPAR ␦ activation increases hepatic steatosis is counterintuitive because PPAR ␦ agonists are known to improve whole-body insulin sensitivity and lipid homeostasis and stimulate Cpt1a -mediated FA oxidation in a variety of cell types and tissues (19)(20)(21). Furthermore, in muscle, GW1516 stimulated FA oxidation due, in part, to increased AMPK activity ( 26,27 ). In a model of hepatic steatosis, GW1516 prevented diet-induced inactivation of hepatic AMPK ( 24 ). This suggests that AMPK activation by PPAR ␦ agonists has the ability to regulate hepatic ␤oxidation and/or FA synthesis.
The objective of this study was to determine whether PPAR ␦ activation can reverse preestablished hepatic steatosis and insulin resistance. We demonstrate that intervention with GW1516 in Ldlr Ϫ / Ϫ mice fed a high-fat, cholesterol-containing (HFHC) diet decreased hepatic lipid deposition, a result of attenuated lipogenesis, and increased FA oxidation. Decreased FA synthesis was due to GW1516-mediated correction of selective hepatic insulin resistance. In addition, we provide evidence that AMPK activation was required for the PPAR ␦ -mediated attenuation of hepatic de novo lipogenesis but was not required for PPAR ␦ -mediated induction of FA oxidation. Furthermore, reduced liver TG content was coupled to attenuated hepatic infl ammation.

Male Ldlr
Ϫ / Ϫ mice on the C57BL/6 background (Jackson Laboratory, Bar Harbor, MA) were housed in pairs in standard cages at 23°C. The animals were cared for in accordance with the Canadian Guide for the Care and Use of Laboratory Animals. All experimental procedures were approved by the Animal Care Committee at the University of Western Ontario. Mice 10-12 weeks of age (n = 16) were fed ad libitum a purifi ed rodent chow diet (14% of calories from fat; Harlan Teklad TD8604, Madison, WI) for 12 weeks. Another group of mice (10-12 weeks of age, n = 48) were fed an HFHC Western diet (42% of calories from fat, 0.2% cholesterol; Harlan Teklad TD09268) for 4 weeks. For the subsequent 8 weeks, half of these mice (n = 24) remained on the HFHC diet; the other half (n = 24) were fed the HFHC diet supplemented with 3 mg/kg/day GW1516 (Enzo Life Sciences, Ann Arbor, MI). A similar dose of GW1516, administered by oral gavage, has been used previously in mice ( 24,28,29 ). In the present study, at the end of the dark cycle and light cycle, nonfasting hepatic steatosis can result in nonalcoholic steatohepatitis, cirrhosis, and eventually liver failure ( 3 ). However, few therapeutic strategies exist that effectively correct hepatic steatosis in the setting of insulin resistance. At a molecular level, insulin binding to its cognate receptor leads to receptor-mediated tyrosine phosphorylation of insulin receptor substrates (IRS-1 and/or IRS-2), which in turn activate phosphoinositide 3-kinase (PI3-K) to simulate the phosphorylation and activation of protein kinase B (Akt) ( 5 ). Normally, insulin-stimulated Akt activation results in the suppression of hepatic gluconeogenesis due to phosphorylation and inactivation of forkhead box (Fox) O1, and the promotion of de novo lipogenesis due to phosphorylation and activation of the mammalian target of rapamycin complex (mTORC) 1 ( 6 ). However, in the insulin-resistant liver, Akt loses its ability to inactivate FoxO1 but paradoxically maintains its ability to activate mTORC1 ( 6 ). Consequently, mTORC1-driven transcription of the master regulator of lipogenesis, sterol-regulatory element binding protein (SREBP) 1c, remains chronically active ( 6 ). In addition, insulin increases the amount of proteolytically processed active SREBP-1c through mechanisms that remain poorly understood ( 7 ). In this way, hyperinsulinemia and accompanying hepatic insulin resistance leads to persistent FA synthesis and eventually excessive hepatic lipid accumulation.
In addition to unregulated lipogenesis, decreased fat oxidation exacerbates hepatic lipid accumulation during insulin resistance (8)(9)(10). The AMP-activated protein kinase (AMPK) controls cellular and whole-body energy metabolism ( 11,12 ). Specifi cally, hepatic AMPK is a pivotal regulator of fat oxidation and synthesis, primarily via direct phosphorylation and inhibition of acetyl-CoA carboxylase (ACC) (12)(13)(14). Biochemically, this reduces malonyl-CoA levels in the liver, which i ) depletes FAS of substrate in the lipogenic pathway and ii ) results in the derepression of carnitine palmitoyl transferase 1a (CPT1a) in the FA oxidation pathway ( 15 ). Thus, activation of AMPK provides a potential mechanism for the attenuation of hepatic steatosis.
PPAR ␦ belongs to a class of ligand-dependent transcription factors involved in regulation of glucose and lipid homeostasis ( 16 ). A number of animal and human studies have highlighted a potential role for activation of this receptor in the treatment of metabolic disease ( 17,18 ). In mice, genetic manipulations of Ppard as well as prevention experiments involving administration of PPAR ␦ agonists revealed that activation of this receptor attenuates dyslipidemia and hyperglycemia, improves whole-body insulin sensitivity, and prevents diet-induced obesity (19)(20)(21). However, there are seemingly confl icting and controversial reports with respect to PPAR ␦ activation and hepatic lipid metabolism in mice ( 19,(22)(23)(24)(25). The PPAR ␦ agonist GW1516 attenuated diet-induced hepatic steatosis; however, the 2-fold increase in hepatic acetyl-CoA oxidase ( Acox ) expression suggested a PPAR ␣ -dependent effect ( 22 ). Tanaka et al. ( 20 ) reported that increased expression of genes involved in hepatic fat oxidation resulted in reduced hepatic lipid droplets in high-fat fed mice treated Primary mouse hepatocyte isolation, lipogenesis, and FA oxidation Primary mouse hepatocytes were isolated from chow-fed WT or AMPK ␤ 1 Ϫ / Ϫ C57BL/6J mice by the collagenase perfusion method as described ( 35 ). AMPK ␤ 1 subunit null mice on a C57BL/6 background were generated using standard homologous recombination techniques, as described previously ( 35,36 ).  ( 37,38 ). For FA oxidation, cells were incubated for 4 h with serum-free Medium 199 containing [1-14 C]palmitic acid (0.5 Ci/ml) (Amersham Biosciences) and 0.5 mM unlabeled palmitate, with or without GW1516 (100 nM) or A-769662 (100 µM). FA oxidation was determined by measuring labeled CO 2 and acid-soluble metabolites as described ( 39 ).

Plasma, blood, and tissue analyses
Plasma insulin concentrations were determined by ELISA (Alpco Diagnostics, Salem, NH) in EDTA plasma according to the manufacturer's instructions as described previously ( 10 ). Blood glucose was determined using an Ascensia Elite glucometer serum concentrations of GW1516 were 604 ± 72 nM and 369 ± 26 nM, respectively, for a mean concentration of 487 ± 50 nM. This plasma concentration is above the EC 50 for murine PPAR ␦ (20 nM), but below the EC 50 for murine PPAR ␥ (1 M) and well below the EC 50 for murine PPAR ␣ (2.5 M) ( 30 ). Animals were fasted for 4 h prior to analyses or being euthanized. For fasting/ refeeding studies, animals were either fasted for 16 h prior to being euthanized or fasted for 16 h followed by a 2 h acute refeeding period of the experimental diets prior to being euthanized ( 31 ). Blood samples were obtained as previously described ( 9,32 ). Body weights and caloric consumption were determined as described previously ( 8 ).

Activation of AMPK in vivo
Activation of AMPK in vivo was assessed in chow-fed Ldlr Ϫ / Ϫ mice following intraperitoneal injection of GW1516 or A-769662 (Selleck Bio, Houston, TX). Animals were fasted overnight (15:00-07:00 h), followed by a period of free access to food (chow) at 07:00 h for 2 h. At 09:00 h, chow was removed and mice were injected intraperitoneally with vehicle (5% dimethyl sulfoxide in PBS), 3 mg/kg GW1516, or 30 mg/kg A-769662, a synthetic activator of AMPK ( 33 ). Analyses of in vivo phosphorylation of AMPK and ACC were performed in liver samples isolated at euthanization by a freezeclamp method, 90 min after injection of the respective treatments, and stored at Ϫ 80°C until analysis as described ( 34 ).

Energy expenditure
In the induction/intervention studies (week 11), analyses of energy expenditure (EE) and respiratory exchange ratio (RER) were performed using the Oxymax Comprehensive Lab Animal Monitoring System (CLAMS, Columbus Instruments, Columbus, OH) ( 8 ). Mice were provided free access to food and water and acclimatized to the system for 24 h prior to a 24 h data collection period, during which data on O 2 consumption and CO 2 production were collected every 10 min. EE and RER were calculated as described previously ( 8 ). (Neomarkers, Fremont, CA), stripped, and reprobed with either the ␤ -actin antibody for the postnuclear fraction or a polyclonal lamin A/C antibody (Santa Cruz Biotechnology, Santa Cruz, CA) for the nuclear fraction ( 41 ). Quantitation of protein on all blots was determined by densitometry as described ( 41,42 ).

Quantitative real-time PCR gene abundance analyses
Total RNA was isolated from liver tissue or primary mouse hepatocytes using TRIzol ® reagent as per the manufacturer's instructions. Specifi c mRNA abundances [PPAR gamma coactivator 1a ( Pgc1a ), Ppara , Acox , Cpt1a , adipocyte differentiation-related protein ( Adfp ), sterol-regulatory element binding factor 1c , heat shock 70kDa protein 5 ( Hspa5) , and Gapdh ] were measured via quantitative real-time PCR using an ABI Prism (7900HT) Sequence Detection System (Applied Biosystems, Foster City, CA) as previously described ( 9,44 ). mRNA abundances were calculated using the standard curve method. Primer and probe sets were obtained from Applied Biosystems (Streetsville, CA) inventoried gene expression arrays with the exception of murine Srebf1c ( Srebp1c ), which was generated as described previously ( 9,44 ).
(Bayer Healthcare, Toronto, Ontario, Canada) ( 10 ). Liver lipids were extracted from 100 mg of tissue using the method of Folch et al. ( 40 ) and quantitated as described previously ( 8 ). FA synthesis was measured following intraperitoneal injection of [1-14 C] acetic acid as described ( 9 ). Hepatic FA oxidation was determined in tissue homogenates of fresh liver by conversion of [

Immunoblotting and densitometry
Total tissue or cell lysates were isolated from mouse liver or from primary mouse hepatocytes as previously described ( 41,42 ). Proteins were separated by SDS-PAGE, transferred to polyvinylidene difl uoride (PVDF) membranes, and immunoblotted ( 42 ). Membranes were probed using antibodies against mouse phosphorylated and total Akt, FoxO1, mTOR, AMPK, and ACC, as well as 78 kDa glucose-regulated protein (GRP78), CCAAT/ enhancer binding protein-homologous protein (CHOP), and ␤actin (Cell Signaling, Danvers, MA). The anti-phosphorylated ACC antibody recognizes both pACC1 Ser76 and pACC2 Ser212, which recently have been shown to have overlapping functions ( 43 ). For SREBP-1, liver lysates were separated into postnuclear and nuclear fractions as described ( 41,42 ). Each fraction was separated by SDS-PAGE (4-15% acrylamide gradient) and transferred to PVDF membranes ( 42 ). Membranes from both fractions were probed with a monoclonal antibody to SREBP-1 test was used. Signifi cance thresholds were P values < 0.05 and marked by different uppercase or lowercase letters as well as asterisks, as indicated in the fi gure legends.

Statistical analyses
Data are expressed as means ± SEM. Student's paired t -test was used to determine signifi cant differences between two groups. One-way ANOVA followed by pair-wise comparisons by the Tukey's test was used to determine differences between three or more groups. Ϫ / Ϫ hepatocytes, and an asterisk (*) indicates statistical signifi cance between WT and ␤ 1 Ϫ / Ϫ within the same treatment group ( P < 0.05); two-way ANOVA with post hoc Tukey's test ( P < 0.05). All data are presented as mean ± SEM (n = 3-4 from at least three independent experiments). mice were fed the HFHC diet supplemented with either vehicle or GW1516 (3 mg/kg/day) for an additional 8 weeks. In mice fed the HFHC diet for 4 weeks, prominent hepatic steatosis developed, as evidenced by signifi cantly increased TG, total cholesterol (TC), and cholesteryl ester (CE) ( Fig. 1A-C ). These lipids continued to increase over the subsequent 8 weeks in HFHC-fed mice. In contrast, the addition of GW1516 to the HFHC diet for 8 weeks decreased hepatic lipids by 30-50% demonstrating a significant slowing of steatosis progression ( Fig. 1A-C ). As we reported previously using a similar protocol ( 45 ), GW1516 intervention reduced the rate of weight gain ( Ϫ 60%), whereas caloric consumption was unaffected ( Fig. 1D, E ). We reasoned that GW1516 attenuates liver TG accumulation via increased FA ␤ -oxidation and/or reduced FA synthesis. With respect to FA oxidation, suppressed Pgc1a expression in livers of mice fed the HFHC diet for 12 weeks ( Ϫ 20% compared with chow-fed mice) was not further affected by GW1516 treatment ( Fig. 2A ). Furthermore, at 12 weeks, the expression of Ppara and the PPAR ␣ -target gene Acox was unaffected by any diet ( Fig. 2B, C ). In contrast, Cpt1a mRNA abundance was signifi cantly enhanced (35%) in livers isolated from GW1516-treated animals, which was associated with a signifi cant 50% increase in FA oxidation as compared with HFHC-fed animals ( Fig. 2D, E ). Expression of the PPAR ␦ -specifi c target gene Adfp ( 46 ) was signifi cantly increased ( ‫ف‬ 60%) in livers from GW1516-treated animals ( Fig. 2F ). Collectively, these results suggest that GW1516 attenuates liver TG accumulation partly due to increased hepatic FA ␤ -oxidation, primarily the consequence of activating the PPAR ␦ -target gene Cpt1a .
To further investigate the GW1516-induced increase in hepatic FA oxidation, we assessed energy balance in a metabolic monitoring system. Total EE was signifi cantly higher (16%) in mice receiving GW1516 compared with mice remaining on the HFHC diet alone ( Fig. 2G ). The RER profi les, which refl ect the relative utilization of carbohydrate (RER ‫ف‬ 1.0) versus fat (RER ‫ف‬ 0.7), were similar between the HFHC-fed and GW1516-intervention groups ( Fig. 2H ). Given the signifi cant increase in total EE in GW1516treated mice, the lack of difference in RER profi les suggests that both carbohydrate and fat utilization are increased by PPAR ␦ activation.

AMPK activation is not required for the GW1516mediated increase in fat oxidation
GW1516 has been shown to activate AMPK in muscle ( 26 ). Given the GW1516-induced stimulation of hepatic FA oxidation in vivo, we hypothesized that AMPK activation may be involved. To evaluate the ability of GW1516 to acutely activate hepatic AMPK in vivo, we used a fasting, feeding, injection, and refasting protocol ( 34 ). In livers isolated 90 min after the injection of GW1516, we observed a signifi cant 2-fold increase in phosphorylation of AMPK and its downstream effector ACC ( Fig. 3A ). Mice were also injected with the potent synthetic AMPK activator A-769662 ( 33 ), which increased AMPK and ACC phosphorylation ‫ف‬ 2-fold ( Fig. 3A ).
The requirement of AMPK for the GW1516-mediated stimulation of FA oxidation was determined in isolated primary mouse hepatocytes from WT or AMPK ␤ 1 Ϫ / Ϫ mice [referred to as AMPK ␤ 1 subunit knockout ( ␤ 1 Ϫ / Ϫ ) mice]. Deletion of the AMPK ␤ 1 subunit results in 90% loss of hepatic AMPK activity ( 35 ). As depicted in Fig. 3B , both GW1516 and A-769662 increased phosphorylation of AMPK and ACC in WT hepatocytes but not in ␤ 1 Ϫ / Ϫ hepatocytes. Furthermore, both GW1516 and A-769662 enhanced FA oxidation in WT hepatocytes by 30% ( Fig. 3C ). The effect of A-769662 was lost in ␤ 1 Ϫ / Ϫ hepatocytes, consistent with an AMPK ␤ 1-dependent effect ( 36 ). However, as in WT cells, GW1516 was able to increase FA oxidation in ␤ 1 Ϫ / Ϫ hepatocytes by ‫ف‬ 30% ( Fig. 3C ). To reconcile this, we examined Cpt1a expression, a PPAR ␦ -target gene, in WT and ␤ 1 Ϫ / Ϫ hepatocytes. GW1516-treatment significantly enhanced Cpt1a expression ( ‫ف‬ 2-fold) in both WT and ␤ 1 Ϫ / Ϫ hepatocytes ( Fig. 3D ). In contrast, A-769662 had no effect on Cpt1a mRNA abundance in hepatocytes of either genotype ( Fig. 3D ). The PPAR ␦ -specifi c target gene Adfp was increased 2-fold in isolated hepatocytes from either genotype, whereas the PPAR ␣ -specifi c target Acox was unaffected by genotype or treatment ( Fig. 3E, F ). Taken together, these results demonstrate that the GW1516-induced increase in Cpt1a -mediated FA oxidation does not require AMPK activation and does not involve PPAR ␣ activation.
In vivo, selective hepatic insulin resistance contributes to FA synthesis due to hyperinsulinemia-driven mTORC1 activation of SREBP-1c ( 4 ). As we reported previously ( 45 ), the HFHC diet resulted in continued progression of is shown. Different letters indicate signifi cant differences; one-way ANOVA with post hoc Tukey's test ( P < 0.05). I: Synthesis of FA in liver obtained 60 min postinjection (intraperitoneal) with [ 14 C]acetic acid (n = 6-8/group). Different letters indicate signifi cant differences; one-way ANOVA with post hoc Tukey's test ( P < 0.05). All data are presented as mean ± SEM. nSREBP-1 was associated with a decrease in Fasn expression ( Fig. 4F, H ) and was coupled to a complete inhibition of the HFHC-induced increase in FA synthesis from 4 to 12 weeks of GW1516 intervention ( Fig. 4I ). Together with results in primary mouse hepatocytes, these data indicate that PPAR ␦ activation inhibits hepatic lipogenesis through activation of AMPK as well as correction of selective hepatic insulin resistance, both of which contribute to the attenuation of liver TG accumulation.

PPAR ␦ activation restores dynamic regulation of hepatic FoxO1, which slows the development of hyperglycemia
Given that PPAR ␦ activation improves hepatic insulin sensitivity in db/db mice ( 19 ), we hypothesized that the GW1516-mediated correction of the HFHC diet-induced bifurcation in hepatic insulin signaling would normalize FoxO1 signaling. At 12 weeks, livers isolated from HFHC-fed mice lost the ability to stimulate FoxO1 phosphorylation and suppress Pck1 expression in the fastingto-feeding transition ( Fig. 5A, B ). This is consistent with our previous hyperinsulinemic euglycemic clamp studies in which the livers of Ldlr Ϫ / Ϫ mice fed a high-fat diet were insulin resistant ( 10 ). In contrast, animals receiving the GW1516 intervention regained the ability to dynamically regulate fasting/refeeding of both FoxO1 phosphorylation and Pck1 expression, similar to that observed in chow-fed mice ( Fig. 5A, B ). As we reported previously ( 45 ), fasting blood glucose levels were increased (1.5-fold) in mice fed HFHC for 12 weeks, which was partially attenuated by GW1516 intervention ( Fig. 5C ).
fasting hyperinsulinemia throughout the study ( Fig. 4C ). Fasting hyperinsulinemia was strongly attenuated by intervention with GW1516 to the HFHC diet ( Fig. 4C ). Selective hepatic insulin resistance was evaluated using a fasting (16 h)/refeeding (2 h) protocol. Compared with chow-fed mice, phosphorylation of hepatic Akt was signifi cantly elevated in the fasted state in HFHC-fed mice, and the response to feeding was enhanced ( Fig. 4D ). A similar pattern for pAkt was observed in GW1516-treated mice. However, in HFHC-fed mice, the phosphorylation of mTORC1 was elevated in the fasted state, and the response to feeding was exaggerated. In contrast, GW1516 intervention completely restored the fasting/refeeding responses of pmTORC1 to levels observed in chow-fed controls ( Fig.  4D ). This is consistent with increased sensitivity in the lipogenic mTORC1 branch of the insulin signaling cascade in HFHC-fed mice ( 4 ) and its normalization following GW1516 treatment.
Compared with chow-fed controls, the hyperinsulinemia and increased hepatic pmTORC1 observed in fasted HFHCfed mice at 12 weeks was associated with increased expression of Srebf1c , increased nSREBP-1, decreased expression of both Insig1 and Insig2a , and increased expression of Fasn ( Fig. 4E-H ), as well as markedly enhanced FA synthesis ( Fig. 4I ). GW1516-intervention attenuated the expression of Srebf1c , inhibited formation of nSREBP-1, and increased the expression of the PPAR ␦ -target gene Insig1 ( Fig. 4E-G ), which is known to block processing of SREBP-1c ( 23 ). The expression of Insig2a , which is dependent on Akt signaling ( 7 ), was unaffected by GW intervention ( Fig. 4G ). Reduced Fig. 5. GW1516 corrects the gluconeogenic branch of insulin signaling during selective hepatic insulin resistance, which improves fasting hyperglycemia. A: Immunoblots of insulin signaling protein pFoxO1 in liver lysates from fasted (F) and refed (RF) mice (n = 6-8/group). Representative immunoblots with quantitations are shown. B: mRNA abundance of Pck1 in liver lysates isolated from F and RF animals (n = 6-8/group). C: Blood glucose levels in F and RF mice (n = 6-8/group). Data are presented as mean ± SEM. Different uppercase letters indicate statistical signifi cance among fasted animals, different lowercase letters indicate statistical signifi cance among refed animals, and asterisk (*) indicates statistical signifi cance between fasted and refed within the same diet group ( P < 0.05); two-way ANOVA with post hoc Tukey's test ( P < 0.05).
These data suggest that the abnormal gluconeogenic branch of insulin signaling, resulting from diet-induced selective hepatic insulin resistance, is corrected by intervention with GW1516, leading to improved fasting blood glucose concentrations.

GW1516 inhibits hepatic infl ammation and induction of endoplasmic reticulum stress
Infl ammation is a prominent feature of hepatic insulin resistance and steatosis ( 3 ). Given that GW1516 intervention attenuated hepatic steatosis and corrected selective hepatic insulin resistance, we postulated that this would be associated with reduced hepatic infl ammation. As shown in Fig. 6A, B , expression of the proinfl ammatory M1 cytokines Tnf , Icam1 , Il1b , Ccl2 , Ccl3 , and iNos was markedly induced (2-to 15-fold) in livers of mice fed the HFHC diet at 12 weeks. In contrast, these cytokines were signifi cantly attenuated ( Ϫ 50 to Ϫ 65%) in livers from the GW1516-intervention group ( Fig. 6A, B ). Furthermore, HFHC feeding strongly suppressed hepatic expression of the M2 antiinfl ammatory marker Arg1 , resulting in a greatly exacerbated iNos/Arg1 ratio compared with chow-fed control mice ( Fig. 6B ). GW1516 intervention reversed or completely prevented this expression pattern ( Fig. 6B ). Together these data suggest that PPAR ␦ activation promotes an antiinfl ammatory M2 cytokine milieu in the liver. To assess the direct effect of GW1516 on the infl ammatory response, we examined the expression of infl ammatory genes in primary hepatocytes from chow-fed mice incubated with GW1516. Of the cytokines examined, the expression of Ccl3 ( Ϫ 25%) and Il1b ( Ϫ 50%) were signifi cantly decreased ( Fig. 6C ). A similar response of these two cytokines to GW1516 was reported in cultured macrophages ( 47 ), suggesting that a direct effect of GW1516 on the expression of Ccl3 and Il1b contributed to their decreased expression in liver following GW intervention ( Fig. 6A ). Furthermore, it is likely that the attenuated hepatic expression of Tnf , Icam1 , and Ccl2 were the consequence of reduced hepatic steatosis. Infl ammation is commonly interwoven with endoplasmic reticulum (ER) stress in the development of hepatic insulin resistance ( 3,48 ). Accordingly, HFHC feeding signifi cantly increased hepatic GRP78 ( Fig. 6D ), a marker of the unfolded protein response (UPR), which is the precursor to the ER-stress response ( 49 ). GW1516 intervention completely normalized the diet-induced increase in Fig. 7. GW1516 intervention inhibits progression of hepatic steatosis in mice. Selective insulin resistance in the liver is induced in mice fed a high-fat diet. Despite hyperinsulinemia, insulin fails to decrease gluconeogenesis due to resistance of the FoxO1 pathway (loss of fasting/refeeding regulation of phospho-serine, pS), whereas the mTORC1/SREBP-1c pathway maintains insulin sensitivity (fasted pS-mTOR is elevated and increased even further with refeeding), leading to increased FA synthesis and TG accumulation. Pathways affected by diet are denoted by black arrows. Red arrows and crosses indicate the effects of GW1516 intervention. PPAR ␦ activation attenuated hepatic TG accumulation as a consequence of increased FA oxidation and reduced FA synthesis. GW1516 stimulated FA oxidation due to increased Cpt1a expression. Increased hepatic AMPK activation by GW1516 was, in part, required for reduction of FA synthesis. The PPAR ␦ agonist suppressed hyperinsulinemia, corrected signaling through the FoxO1 pathway, and normalized signaling through the mTORC1/SREBP-1c pathway. GW1516 decreased Srebf1c expression and inhibited the conversion of precursor SREBP-1c (pSREBP-1c) to the active nuclear form (nSREBP-1c), in part, due to increased expression of Insig1 , which collectively contributed to the decreased FA synthesis. PPRE, PPAR response element; SCAP, SREBP cleavage-activating protein; SRE, sterol response element. GRP78 ( Fig. 6D ). CHOP, the downstream effector of the ER-stress response, was not increased by the HFHC diet, nor was it affected by GW1516 ( Fig. 6D ). This suggests that although the HFHC diet initiated the UPR, GRP78 expression was suffi cient to prevent the full downstream ER-stress response. Nevertheless, PPAR ␦ activation attenuated the HFHC diet-induction of the UPR. The expression of Hspa5 , which codes for GRP78, was unaffected by GW1516 in primary hepatocytes (data not shown), suggesting that the GW-induced reduction in GRP78 in vivo was secondary to decreased hepatic steatosis.

DISCUSSION
We evaluated the ability of the PPAR ␦ agonist GW1516 to attenuate the progression of diet-induced hepatic steatosis.
GW1516 intervention inhibited the progression of liver TG accumulation, the consequence of reduced FA synthesis and increased FA oxidation. GW1516 activated hepatic AMPK in vivo and in vitro. Exposure of isolated hepatocytes defi cient in AMPK activity to GW1516 revealed that inhibition of lipogenesis required AMPK, whereas the induction of FA oxidation was mediated through increased expression of Cpt1a , rather than by AMPK activation. Hepatic lipogenesis was also attenuated through GW1516mediated correction of selective hepatic insulin resistance, which was associated with reduced hepatic infl ammation ( Fig. 7 ).
The role of PPAR ␦ activation in liver TG metabolism has been controversial ( 19,(22)(23)(24)(25). One study showed that GW1516 prevented diet-induced suppression of hepatic AMPK activation, which was associated with increased we demonstrate that PPAR ␦ activation attenuates the progression of the selective hepatic insulin-resistant phenotype, as dynamic regulation of fasting-to-feeding pmTORC1, pFoxO1, and Pck1 expression was restored in the GW1516intervention cohort. These data elaborate on the body of evidence that PPAR ␦ activation not only protects from insulin resistance (19)(20)(21), but can also reverse preestablished hepatic insulin resistance.
Liver infl ammation has been linked to hepatic steatosis and insulin resistance ( 3,51 ). Vascular chronic low-grade infl ammation is, in part, mediated by aortic lipid accumulation and insulin resistance ( 52,53 ). Given the selective insulin-resistant phenotype and TG acquisition in livers of HFHC-fed animals and correction by GW1516 intervention, it is tempting to hypothesize that similar mechanisms govern induction and attenuation of vascular and hepatic infl ammation. Moreover, Kupffer cell-specifi c deletion of Ppard in mice resulted in increased proinfl ammatory cytokine expression and reduced anti-infl ammatory cytokine expression, which was coupled to increased liver TG accumulation and hepatic dysfunction ( 54 ). Therefore, our results are consistent with an anti-infl ammatory role for PPAR ␦ activation in the liver, similar to the effect in macrophages and in the aorta ( 28,45,47 ). The relative impact of reduced infl ammation versus correction of insulin sensitivity to the attenuation of hepatic steatosis cannot be discerned from the present experiments and requires further elucidation.
Previous studies raise the possibility that the reduction in hepatic steatosis with GW1516 intervention was an indirect consequence of attenuated body weight gain or the small decrease in adipose tissue mass ( ‫ف‬ 10%) ( 45 ), as GW1516 is known to also target adipose tissue ( 21 ). Wang et al. ( 21 ) reported that adipose tissue-specifi c overexpression (3-fold) of a PPARD expression vector in HFD-fed C57BL/6 mice decreased adiposity ( ‫ف‬ 50%), primarily due to an increase in adipocyte Cpt1a expression and enhanced FA oxidation. This was associated with a decrease in lipid droplets within liver sections. Using a similar approach, Qin et al. ( 23 ) reported that hepatic overexpression of PPAR ␦ following injection of PPARD adenovirus in db/db mice suppressed the hepatic expression of genes involved in FA synthesis, decreased SREBP-1c processing, and attenuated Oil Red Ostained hepatic lipid. Although neither FA oxidation nor insulin resistance were measured, these latter results are consistent with the concept that direct hepatic activation of PPAR ␦ is able to attenuate hepatic steatosis.
In the present study, we show in primary mouse hepatocytes from chow-fed mice and in liver from HFHC-fed mice that PPAR ␦ activation by GW1516 increases the expression of the PPAR ␦ -specifi c target gene Adfp . Furthermore, the PPAR ␦ -target gene Cpt1a was increased directly or consequent to PPAR ␦ -induced AMPK activation, leading to increased FA oxidation. In addition, GW1516 suppressed lipogenesis in hepatocytes and liver due to activation of AMPK and phosphorylation of ACC, as well as decreased expression of Srebf1c , and suppression of SREBP-1 processing to its active nuclear form. The latter was associated with increased expression of Insig1 , considered to be a direct expression of genes involved in FA oxidation and increased plasma ␤ -hydroxybutyrate ( 24 ). Despite these observations, GW1516 did not affect hepatic TG content ( 24 ). Another study demonstrated that injection of ad PPAR ␦ into Ldlr Ϫ / Ϫ mice signifi cantly increased hepatic AMPK phosphorylation, which was thought to contribute to glucose lowering ( 25 ). However, the impact of increased hepatic AMPK activation on lipid metabolism was not explored ( 25 ). Here we provide direct evidence that PPAR ␦ activation increases hepatic AMPK and ACC phosphorylation in vivo as well as in primary mouse hepatocytes. It is possible that this effect was due to changes in adenylate charge ( 24 ). Nevertheless, the GW1516-mediated increase in pACC was AMPK dependent as this effect was lost in ␤ 1 Ϫ / Ϫ hepatocytes. We demonstrate that PPAR ␦ activation stimulates hepatic FA oxidation in vivo through PPAR ␦ -specifi c activation of Cpt1a . We recapitulated these results in primary mouse hepatocytes and showed that AMPK activation is not a requirement for GW1516-induced FA oxidation, as enhanced Cpt1a expression and FA oxidation persisted in GW1516-treated ␤ 1 Ϫ / Ϫ hepatocytes.
Studies that have examined the role of PPAR ␦ activation in hepatic de novo lipogenesis have yielded both positive and negative results ( 19,23,25 ). On one hand, both ad PPAR ␦ injection and GW1516 treatment have been shown to increase hepatic expression of genes involved in lipogenesis, resulting in increased liver TG accumulation ( 19,25 ). On the other hand, delivery of ad PPAR ␦ or the synthetic PPAR ␦ agonist GW0742 have demonstrated reduced SREBP-1c processing, decreased lipogenic gene expression, and prevention of hepatic steatosis ( 23 ). The data presented here are consistent with and extend this latter concept. We provide evidence that intervention to an HFHC diet with GW1516 in mice halts progression of hepatic steatosis and corrects selective hepatic insulin resistance by normalizing signaling through mTORC1, resulting in suppression of lipogenic gene expression. Not only was Srebf1c expression decreased, but processing of SREBP-1 to its active nuclear form was inhibited. The GW1516-induced expression of Insig1 likely contributes to this inhibition ( 23 ). Other insulin-regulated factors known to infl uence SREBP-1 processing were either unchanged ( Insig2a ) or were not examined (Gsk3B and Lipin1) ( 7,23,50 ). Collectively, the GW1516-induced suppression of the SREBP-1c pathway contributed to the prevention of any further increase in FA synthesis. Furthermore, GW1516 reduced de novo lipogenesis in WT primary mouse hepatocytes, but not in ␤ 1 Ϫ / Ϫ hepatocytes, demonstrating that inhibition of FA synthesis was AMPK dependent. Although two different mechanisms contribute to the observed reduction in lipogenesis by GW1516 treatment, the relative contributions of these pathways require further study. Coupled to hyperinsulinemia, the bifurcation in insulin signaling, in which FoxO1 becomes insulin resistant and the mTORC1/SREBP-1c pathway maintains insulin sensitivity, contributes to hepatic steatosis, dyslipidemia, and hyperglycemia ( 4 ). In the present study, we provide evidence that hepatic insulin signaling does in fact bifurcate in a model of diet-induced insulin resistance. Importantly, target of PPAR ␦ ( 23 ). These results suggest that the attenuation of hepatic steatosis and improved hepatic insulin signaling are the consequence of a direct effect of GW1516 within the liver.
In summary, the data reported here provide physiological and molecular evidence that intervention with PPAR ␦specifi c activation in the liver alleviates diet-induced hepatic steatosis, insulin resistance, and infl ammation. We conclude that PPAR ␦ agonists may serve as therapeutic options for the treatment of patients with hepatic steatosis.