Liver-Specific Inhibition of ChREBP Improves Hepatic Steatosis and Insulin Resistance in ob/ob Mice

Six-week-old male ob/+ and ob/ob mice were purchased from Elevage Janvier (Le Genest Saint Isle, France) and adapted to the environment for 1 week before study. All mice were housed in colony cages with a 12-h light/dark cycle in a temperature-controlled environment. All procedures were carried out according to the French guidelines for the care and use of experimental animals. Mice had free access to water and regular diet (in terms of energy: 65% carbohydrate, 11% fat, and 24% protein), unless specified.

For short-term studies, mice were fasted for 24 h and then fed a high-carbohydrate fat-free diet (in terms of energy: 72.2% carbohydrate, 1% fat, and 26.8% protein; SAFE, Chaumesnil, France) for 18 h. After anesthesia (a mix of ketamine/xylazine), livers were frozen in liquid nitrogen and kept at −80°C until use. For long-term studies, mice were fed on a regular diet for 7 days before they were killed.

A 19nt sequence starting from nucleotide 747 of ChREBP was synthesized as complementary antiparallel oligonucleotides with a loop sequence (ttcaagaga) and BamH1- and HinDIII-compatible ends. The nucleotide sequence for the short hairpin RNA (shRNA) against ChREBP (shChREBP) was as follows: gatccGTGTTGGCAATGCTGACATGttcaagagaCATGTCAGCATTGCCAACAttttttggaa (forward) and gCACAACCGTTACGACTGTACaagttctctGTACAGTCGTAACGGTTGTaaaaaaccttttcga (reverse). The forward and reverse oligonucleotides were annealed and ligated into a pRNAT-H1.1 shuttle vector containing the human H1 promoter and expressing the green fluorescent protein (GFP) (cloral GFP [cGFP]) under the control of the cytomegalovirus promoter (GenScript). The H1-shChREBP-RNA-cGFP cassette was then inserted into the BD-AdenoX expression system (Clontech). Recombinant adenovirus expressing shChREBP (Ad-shChREBP) and GFP adenovirus (Ad-GFP) were produced in HEK293 cells and purified on a cesium chloride gradient before use.

Male mice were anesthetized with isoflurane (Belamont, Paris, France) before the injection through the penis vein with 10⁹ pfu of either Ad-GFP or Ad-shChREBP in a final volume of 300 μl sterile PBS. For insulin signaling experiments, mice were injected with 5 units of regular human insulin (Actrapid Penfill; NovoNordisk) via the portal vein. Three minutes later, tissues were snap frozen in liquid nitrogen. Immunoblot analysis of insulin signaling molecules were performed as previously described (17). Rabbit polyclonal for total Akt, Foxo1, mitogen-activated protein kinase (MAPK) (extracellular signal–related kinase [ERK] 1 and ERK2) and phospho-Akt (S473), phospho-Foxo1 (Ser²⁵⁶), and phospho-MAPK antibodies were purchased from Cell Signaling Technology.

Blood glucose values were determined using an AccuCheck glucometer (Roche). Serum concentrations of triglycerides, free fatty acids, β-hydroxybutyrate, alanine aminotransferease, and aspartate aminotransaminase were determined using an automated Monarch device (Laboratoire de Biochimie, Faculté de Médecine, Bichat, France). G6P, liver (12), and muscle glycogen (18) concentrations were determined as previously described. Liver triglyceride content was measured with a colorimetric diagnostic kit (Triglycerides FS; Diasys). Pyruvate, phosphoenolpyruvate, hepatic acetoacetate, and β-hydroxybutyrate were measured as described (18). Insulin concentrations were determined using a rat insulin enzyme-linked immunosorbent assay kit (Crystal Chem) using a mouse insulin standard. The binding reaction was modified to perform the assay on 10 μl of plasma. Malonyl CoA esters were measured using a modified high-performance liquid chromatography method (19).

Total RNA were extracted using the RNeasy Kit (Qiagen), and 500 ng were reverse transcribed. RTQ-PCR analysis was performed using primers for ChREBP, sterol regulatory element–binding protein (SREBP)-1, glucokinase (GK), liver-pyruvate kinase (l-PK), acetyl-CoA carboxylase (ACC), and fatty acid synthase (FAS) as described (12). Primers used for PEPCK were (sense) 5′-TGGCTACGTC CCTAAGGAA-3′, (antisense) 5′-GGTCCTCCAGATACTTGTCGA-3′; for G6Pase were (sense) 5′-TCTTGTGGTTGGGATACTGG-3′, (antisense) 5′-GCAATGCCTGACAAGACTC-3′; for stearoyl-CoA desaturase (SCD)-1 (sense) 5′-CCGGAGACCCTTAGATCGA-3′, (antisense) 5′-TAGCCTGTAAAAGATT TCTGCA AACC-3′; and for glyceraldehyde 3-phosphate acyltransferase (GPAT) (sense) 5′-CAACACCATCCCCGACATC-3′, (antisense) 5′-GTGACCT TCGATTATGCGATCA-3′. The relative quantification for a given gene was corrected to the cyclophilin mRNA values.

Liver nuclear and cytoplasmic extracts were prepared using the NE-PER extraction reagent kit (Pierce Biotechnology) (12). SREBP-1 was detected with a mouse monoclonal antibody (SREBP-1 Ab-1, NeoMarkers; Interchim) and ChREBP with a rabbit polyclonal antibody (Novus Biologicals). ACC1 and ACC2 protein content was detected in total liver extracts with polyclonal antibodies (ACC1; Alpha Diagnostic International; and ACC2; Cell Signaling) A polyclonal GFP antibody from Clonetech was used. Rabbit polyclonal for total and phospho–glycogen synthase kinase (GSK)3β antibodies were purchased from Cell Signaling Technology. Monoclonal mouse β-actin (clone AC.74; Sigma) and polyclonal rabbit lamin A/C (Cell Signaling) antibodies were used as loading controls. Autoradiograms of Western blots were scanned and quantified using an image processor program (Chemi Genius2 scan; Syngene).

Rates of fatty acid synthesis was measured by intraperitoneally injecting 150 μCi of ³H-labeled water to mice during the early light cycle (20). Two hours later, blood was drawn from the inferior cava vein to determine the plasma ³H-labeled water–specific activity. Rates of fatty acid synthesis were calculated as micromoles of ³H-radioactivity incorporated into fatty acids per hour per gram of tissue.

For histology studies, tissues were fixed in 10% neutral buffered formalin and embedded in paraffin. Then, 7-μmol/l sections were cut and stained with hematoxylin eosin. For the detection of neutral lipids, liver and muscle cryosections were stained with the Oil Red O technique (21), using 0.23% dye dissolved in 65% isopropyl alcohol for 10 min.

Glucose tolerance tests were performed by glucose gavage (1 g d-glucose/kg body wt) after an overnight fast. Insulin tolerance tests were performed by intraperitoneal injection of human regular insulin (1 units insulin/kg body wt, Actrapid Penfill; NovoNordisk) 5 h after food removal. Blood glucose was determined using one-touch AccuCheck glucometer (Roche).

Results are reported as means ± SE. The comparison of different groups was carried out using two-tailed unpaired Student's t test. Differences were considered statistically significant at P < 0.05.

To determine the role of ChREBP in hepatic steatosis, ChREBP expression and protein content were measured in liver of lean (ob/+) and obese (ob/ob) mice. ChREBP mRNA levels were significantly higher in liver of fasted ob/ob mice compared with fasted lean controls (Fig. 1A). Upon feeding with a high-carbohydrate diet, ChREBP mRNA levels were higher in liver of ob/ob compared with ob/+ mice, although a similar fold of induction was observed (Fig. 1A).

While nuclear ChREBP content was undetectable in liver extracts from 24-h–fasted ob/+ mice (Fig. 1B), it was significantly increased in nuclear extracts from 24-h–fasted ob/ob mice and was, in fact, comparable with levels observed in nuclear extracts from high-carbohydrate–fed lean mice (Fig. 1B). A threefold increase in ChREBP protein was observed in nuclear extracts from high-carbohydrate–fed ob/ob mice when compared with fasted ob/ob mice (Fig. 1B). Since the transcription factor SREBP-1c had been previously implicated in the development of fatty livers in ob/ob mice (22), we measured its protein content under our experimental conditions. While no detectable levels in mature SREBP-1c protein was observed in liver of 24-h–fasted ob/+ and ob/ob mice, a significant increase in the amount of nuclear SREBP-1c was observed in liver of high-carbohydrate–fed ob/ob mice compared with ob/+ mice (Fig. 1B).

To assess the effects of ChREBP inhibition in liver, we have developed an adenovirus-mediated RNA interference approach in which shRNAs were used to inhibit ChREBP gene expression in vivo. The delivery of 10⁹ pfu of Ad-shRNA against ChREBP to mice (Ad-shChREBP), which coexpresses the GFP, was liver specific since no GFP protein was detected in tissues other than liver (online appendix Fig. 1A [available at http://diabetes.diabetesjournals.org]). The efficiency of delivery was high since >90% of hepatocytes from Ad-shChREBP–injected mice expressed the GFP protein (online appendix Fig. 1B). A first series of experiments (presented in Figs. 2 and 3 and in Table 1) were performed in which mice received a dose of adenovirus (Ad-shChREBP or GFP) and were either killed 24 h later in the fasted state or 42 h later following an 18-h high-carbohydrate diet refeeding. After the injection of the Ad-shChREBP, an equivalent 60% decrease in ChREBP mRNA levels was observed in liver of both 24-h–fasted and high-carbohydrate–fed ob/ob mice when compared with ob/ob mice receiving an equivalent dose of Ad-GFP (Fig. 2A). A 70–90% decrease in nuclear ChREBP content was also observed in nuclear extracts from Ad-shChREBP–injected ob/ob mice (both in fasted and fed states) (Fig. 2B). Mature SREBP-1c protein content was not affected by ChREBP knockdown (Fig. 2B).

Since ChREBP is required for l-PK gene induction by glucose, we first hypothesized that ChREBP knockdown in liver of ob/ob mice would lead to alterations in glucose metabolism. Both GK and l-PK gene expression was significantly higher in liver of 24-h–fasted Ad-GFP–injected ob/ob compared with ob/+ mice and further increased after high-carbohydrate feeding (Fig. 2C). While GK mRNA levels were not affected by ChREBP knockdown, l-PK mRNA levels were decreased by 60% in liver of both 24-h–fasted and high-carbohydrate–fed Ad-shChREBP–injected ob/ob mice (Fig. 2C). In agreement with an inhibition of glycolysis at the level of l-PK, the pyruvate-to-phosphoenolpyruvate ratio was decreased by 65% in liver of Ad-shChREBP–treated ob/ob mice (Table 1). In contrast, both glucose 6-phosphate and glycogen concentrations were significantly higher in liver of Ad-shChREBP–injected ob/ob mice in the fed state (Table 1), indicating that ChREBP knockdown had led to a redistribution of the glucose 6-phosphate flux from glycolysis to glycogen synthesis. However, despite this redistribution of flux after high-carbohydrate feeding, Ad-shChREBP–injected ob/ob mice remained markedly hyperglycemic (Table 1). Interestingly, after a 24-h fast, Ad-shChREBP–injected ob/ob mice had normal blood glucose levels (Table 1). Both 24-h–fasted and high-carbohydrate–fed Ad-shChREBP–injected ob/ob mice were hyperinsulinemic (Table 1).

The injection of Ad-shChREBP to both 24-h–fasted and high-carbohydrate–fed ob/ob mice caused a 60% reduction in ACC and FAS mRNA levels (Fig. 2C). In fact, ACC and FAS gene expression in liver of Ad-shChREBP–injected ob/ob mice was restored to levels measured in lean mice (Fig. 2C). In addition, mRNA levels for SCD-1, the desaturase responsible for the production of monounsaturated fatty acids and the one of GPAT, which catalyzes the first step of triglyceride synthesis, was significantly higher in liver of Ad-GFP–injected ob/ob mice compared with ob/+ mice in both the fasted and high-carbohydrate–fed states (Fig. 2C). The fact that both SCD-1 and GPAT mRNA levels were significantly decreased after ChREBP knockdown in ob/ob mice (Fig. 2C) suggests that ChREBP knockdown normalized the expression of genes controlling both lipogenesis and triglyceride synthesis. In contrast, SREBP-1c mRNA levels were not affected by ChREBP silencing (Fig. 2C).

We determined whether ChREBP knockdown led to an improvement of hepatic steatosis in ob/ob mice. First, we observed that the liver weight of Ad-shChREBP–injected ob/ob mice (in both the fasted and high-carbohydrate–fed states) was significantly reduced compared with Ad-GFP–injected ob/ob mice (Table 1). Liver sections from Ad-shChREBP ob/ob mice also revealed fewer lipid droplets stained with Oil red O after ChREBP knockdown in both the fasted and fed states (Fig. 3A). This was also associated with a significant reduction in hepatic triglyceride content and in plasma free fatty acid and plasma triglyceride concentrations in these mice (Table 1). Lipogenic rates were also measured in vivo after the incorporation of ³H-labeled water to de novo synthesized lipids (Fig. 3B). As expected, de novo lipid synthesis was low in liver of 24-h–fasted ob/+ mice and increased by ∼20-fold after high-carbohydrate feeding (Fig. 3B). Higher rates of lipogenesis were measured in livers of 24-h–fasted Ad-GFP–injected ob/ob mice. Upon high-carbohydrate feeding, a twofold increase in fatty acid synthesis was further observed in these mice (Fig. 3B). ChREBP knockdown caused a significant decrease in lipogenic rates in both 24-h–fasted and high-carbohydrate–fed ob/ob mice (Fig. 3B). Altogether, our results demonstrate that ChREBP inhibition in liver of ob/ob mice, by causing a significant reduction in lipogenesis, led to the improvement of their hepatic steatosis.

We next investigated whether the improvement in hepatic steatosis observed in Ad-shChREBP–treated ob/ob mice could be linked to an increase in lipid oxidation. A two- to threefold increase in plasma β-hydroxybutyrate concentrations was observed following the injection of Ad-shChREBP to 24-h–fasted and high-carbohydrate–fed ob/ob mice (Table 1), suggesting that β-oxidation was activated in these mice. The rate-limiting step of β-oxidation is the transport of acyl-CoAs into the mitochondria by the liver carnitine palmitoyltransferase 1 (l-CPT 1), which allosteric inhibitor malonyl-CoA is synthesized by ACC (23). ACC1 and ACC2 protein content was significantly lower in liver of Ad-shChREBP–injected ob/ob mice compared with Ad-GFP–injected ob/ob mice (Fig. 3C). The fact that malonyl-CoA concentrations were also found to be decreased in liver of 24-h–fasted Ad-shChREBP–injected ob/ob mice supports the hypothesis of increased β-oxidation rates in these mice during fasting (6.4 ± 0.4 in Ad-GFP ob/+, 7.9 ± 1.0 in Ad-GFP ob/ob, and 5.7 ± 0.4 nmol/g liver in Ad-shChREBP ob/ob mice, n = 4 per group). Another argument in favor of an increased β-oxidation in liver is that both hepatic acetoacetate and β-hydroxybuyrate concentrations were significantly increased in liver of 24-h–fasted Ad-shChREBP–injected ob/ob mice (acetoacetate: 1.3 ± 0.2 in Ad-GFP ob/+, 0.9 ± 0.2 in Ad-GFP ob/ob, and 2.9 ± 0.4 μmol/g liver* in Ad-shChREBP ob/ob mice, n = 4 per group; β-hydroxybuyrate: 1.8 ± 0.5 in Ad-GFP ob/+, 1.0 ± 0.1 in Ad-GFP ob/ob, and 5.9 ± 0.5 μmol/g liver* in Ad-shChREBP ob/ob mice, n = 4 per group; *significantly different from from Ad-GFP–injected ob/ob mice, P < 0.05).

To determine whether the normalization of blood glucose levels in 24-h–fasted Ad-shChREBP–injected ob/ob mice (Table 1) could be due to an improvement of their insulin resistance, the insulin signaling pathway was evaluated after an insulin bolus in liver of 24-h–fasted ob/ob mice injected with either Ad-GFP or Ad-shChREBP (Fig. 3D). In agreement with an alteration of the early steps of insulin signaling in ob/ob mice (24), the stimulation by insulin of Akt, ERK1, and ERK2 phosphorylation was markedly decreased in liver of Ad-GFP–injected ob/ob mice (Fig. 3D). The defect in Akt activation translated downstream to the decreased phosphorylation of Foxo1 (Fig. 3D). Interestingly, ChREBP knockdown resulted in a significant improvement of insulin signaling in ob/ob mice as evidenced by the restoration of Akt, ERK1, ERK2, and Foxo1 phosphorylation by insulin (Fig. 3D).

Insulin signaling was also determined in high-carbohydrate–fed Ad-shChREBP–injected ob/ob mice. Foxo1 phosphorylation was restored to normal levels after ChREBP inhibition in liver of high-carbohydrate–fed ob/ob mice (Fig. 3E). Since the phosphorylation of Foxo1 by Akt inhibits its ability to activate gluconeogenesis (25), we hypothesized that restored Foxo1 phosphorylation may lead to an efficient inhibition of gluconeogenic genes in liver of Ad-shChREBP–treated ob/ob mice (Fig. 3F). Indeed, after Ad-shChREBP treatment, G6Pase and PEPCK mRNA levels were significantly decreased in livers of ob/ob mice (Fig. 3F). Therefore, our results demonstrate that the liver-specific inhibition of ChREBP is associated with a normalization of hepatic insulin signaling in both 24-h–fasted and high-carbohydrate–fed ob/ob mice.

To determine whether ChREBP knockdown under long-term conditions could improve overall glucose tolerance and insulin sensitivity, a 7-day treatment with Ad-shChREBP was performed in ob/ob mice (Fig. 4). No sign of inflammation was observed upon a 7-day adenoviral treatment since both alanine aminotransferease and aspartate aminotransaminase concentrations were similar in Ad-GFP–versus Ad-shChREBP–injected mice (Table 2). ChREBP knockdown was sustained after 7 days of Ad-shChREBP treatment (Fig. 4A), and genes known to be controlled by ChREBP were also downregulated (Fig. 4B). Liver sections from Ad-shChREBP ob/ob mice revealed fewer lipid droplets stained with Oil red O (Fig. 4C, panel C). This was also associated with a significant reduction in hepatic triglyceride content, lipogenic rates (80%), and plasma free fatty acid and triglyceride concentrations (Table 2). Like we observed under short-term conditions, β-oxidation rates were probably increased in liver of long-term–treated Ad-shChREBP–treated mice since both hepatic acetoacetate and β-hydroxybuyrate concentrations were significantly increased after the adenoviral treatment (Table 2). Interestingly, hepatic cholesterol concentrations were significantly decreased after the 7-day treatment with Ad-shChREBP (Table 2). This decrease was correlated with a parallel decrease in both HMGCoA synthase and reductase gene expression (data not shown). In addition, while Ad-GFP–treated ob/ob mice remained markedly hyperglycemic throughout the treatment period, Ad-shChREBP ob/ob mice showed, as early as day 1, a significant drop in their blood glucose concentrations to reach, at day 4, values comparable with the ones measured in Ad-GFP–treated ob/+ mice (Fig. 4D). The improvement in blood glucose concentrations in 7-day–treated Ad-shChREBP ob/ob mice was associated with a 60% decrease in plasma insulin levels (Table 2).

To determine the physiological consequences of the long-term inhibition of ChREBP in liver of ob/ob mice, glucose and insulin tolerance tests were performed (Fig. 4E and F). Fasting blood glucose was reduced and glucose tolerance was significantly improved in Ad-shChREBP–treated ob/ob mice (Fig. 4E). In addition, while blood glucose levels failed to decrease after insulin injection in Ad-GFP–treated ob/ob mice, blood glucose levels were significantly reduced in Ad-shChREBP–treated ob/ob mice after the insulin injection (Fig. 4F). Our results demonstrate that the long-term inhibition of ChREBP in liver of ob/ob mice significantly improves their overall glucose tolerance and insulin sensitivity.

To address the possibility that the improvement in glucose tolerance and insulin sensitivity observed in long-term–treated Ad-shChREBP ob/ob could be due to a restoration of muscle insulin sensitivity, we next evaluated insulin signaling, GSK3β phosphorylation, and glycogen content in muscles of these mice (Fig. 5). Consistent with their state of insulin resistance, insulin-mediated phosphorylation of Akt, ERK1, and ERK2 was low in skeletal muscles from Ad-GFP–treated ob/ob mice. In contrast, a significant improvement in their phosphorylation by insulin was observed after ChREBP knockdown (Fig. 5A). Similarly, while GSK3β phosphorylation and glycogen content were decreased in muscles from Ad-GFP–treated ob/ob mice, both were significantly improved after ChREBP knockdown (Fig. 5B and C). Finally, to address whether improved insulin signaling could be due to a decrease in lipid content, Oil red O staining of intracellular lipid droplets and lipogenic rates were measured in muscles from Ad-shChREBP–treated ob/ob mice (Fig. 5D and Table 2). Indeed, muscle sections from Ad-shChREBP ob/ob mice revealed fewer lipid droplets than muscle sections from Ad-GFP–treated ob/ob mice (Fig. 5D, panels B and C), an observation consistent with the decrease in circulating lipids observed in these mice (Table 2). In addition, a 70% decrease in lipogenic rates was measured in muscles from Ad-shChREBP ob/ob mice (Table 2).

Finally, to address whether insulin sensitivity in white adipose tissue from Ad-ChREBP–treated ob/ob mice was also improved, insulin signaling, ChREBP, and SREBP-1c protein content as well as lipogenic rates were evaluated (Fig. 6 and Table 2). Similar to what we observed in skeletal muscles, insulin-mediated phosphorylation of Akt, ERK1, and ERK2 was improved after a 7-day treatment with the Ad-shChREBP (Fig. 6A). In addition, because circulating glucose and insulin concentrations were improved, lipogenic rates in white adipose tissue were also significantly reduced in these mice (Table 2) and were associated with a parallel decrease in nuclear ChREBP and SREBP-1c content (Fig. 6B). The decrease in ChREBP protein content was not due to a direct effect of the Ad-shChREBP since no GFP protein was detected in white adipose tissue (data shown). As a consequence, the weight of white adipose tissue was reduced by 30% in Ad-shChREBP–treated ob/ob mice (Table 2).

Although NAFLD and hepatic insulin resistance are associated, a causal relationship between hepatic fat accumulation and insulin resistance has not been clearly established. In this report, we provide strong evidence to support the importance of the transcription factor ChREBP in the regulation of de novo lipogenesis and in the development of fatty liver. Our study reveals that liver-specific inhibition of ChREBP, by preventing fat accumulation in liver, significantly improves the hyperlipidemic phenotype and restores normal glucose tolerance and insulin sensitivity in ob/ob mice. Altogether, our studies provide evidence for a molecular mechanism whereby hepatic fat accumulation in ob/ob mice can lead to insulin resistance.

The molecular mechanisms leading to the development of hepatic steatosis is complex. Several studies have shown that genes encoding lipogenic enzymes are elevated in livers of ob/ob mice (22), and, previously, the transcription factor SREBP-1c was shown to contribute to the high rates of lipogenesis in livers of these mice (22,26). Indeed, SREBP-1c content is markedly increased in livers of ob/ob mice (22), and when ob/ob mice are crossed with SREBP-1c–null mice, they show a significant improvement in their hepatic steatosis but not in their overall insulin resistance. The recent emergence of the transcription factor ChREBP in the control of lipogenic gene expression in liver (12,13) prompted us to address its role in the physiopathology of hepatic steatosis. ChREBP directly activates lipogenic gene transcription by binding to the carbohydrate responsive element present in their promoter sequence (14,16). Furthermore, ChREBP silencing in hepatocytes (12) and in mice (13) not only leads to the lack of induction of ACC and FAS genes in response to glucose but also causes a significant reduction in lipid synthesis. The present study reveals that ChREBP expression is markedly increased in liver of ob/ob mice. Therefore, the concomitant increase in nuclear ChREBP and SREBP-1c content we observed in the fed state supports the fact that these two transcription factors contribute to the high rates of lipogenesis that leads to hepatic steatosis in ob/ob mice. It should be noted, however, that only ChREBP content was increased in liver of fasted ob/ob mice, suggesting that ChREBP by itself is responsible for the increased rates of lipogenesis measured after a 24-h fast. While the mechanism by which ChREBP expression is increased in liver of ob/ob mice is still unknown, it could be directly caused by chronic exposure to high glucose concentrations since we have previously demonstrated that glucose metabolism through hepatic GK is required for the induction of ChREBP in liver (12). The fact that both GK expression and glucose metabolism are elevated in liver of ob/ob mice supports this hypothesis.

To address the role of ChREBP in the physiopathology of fatty liver, we have used an shRNA approach to inhibit its expression in vivo. Our study demonstrates that ChREBP knockdown, both under short- and long-term conditions, significantly improves the fatty liver phenotype of ob/ob mice by decreasing rates of lipogenesis, thereby decreasing hepatic fat accumulation. In fact, liver-specific inhibition of ChREBP not only affected the rates of de novo lipid synthesis but also had consequences on β-oxidation. Lipogenesis and β-oxidation are correlated since malonyl-CoA, synthesized by ACC, is the allosteric inhibitor of l-CPT-1, the rate-limiting enzyme of β-oxidation (27). The fact that ACC1 and ACC2 content was lower in liver of fasted Ad-shChREBP–injected ob/ob mice probably led to a constitutive activation of l-CPT-1 activity. The significant decrease in malonyl-CoA concentrations and the increase in both hepatic and plasma β-hydroxybutyrate levels measured in Ad-shChREBP–injected ob/ob mice also support this hypothesis. Therefore, the coordinate modulation in fatty acid synthesis and oxidation led to the overall improvement of lipid homeostasis in Ad-shChREBP–injected ob/ob mice. In agreement with our data, knockouts of lipogenic genes such as ACC2 (28) or SCD-1 (29) are associated with increased rates of fatty oxidation in liver, leading to the improvement of overall lipid homeostasis in these mice.

Interestingly, our study also shows that ChREBP is not only required for the carbohydrate-induced transcriptional activation of enzymes involved in de novo fatty acid synthesis but also in triglyceride synthesis. Indeed, both SCD-1 and GPAT gene expression was increased in liver of ob/ob mice, confirming findings from a previous report (30). The fact that SCD-1 and GPAT gene expression was significantly decreased after ChREBP knockdown supports the hypothesis of a direct transcriptional control of these genes by ChREBP. Indeed, a carbohydrate responsive element has been identified on the promoter sequence of the GPAT gene (31), and SCD-1 gene expression is markedly induced by high-carbohydrate feeding in liver of mice (32). Therefore, ChREBP appears to be a key modulator of hepatic triglyceride concentrations by regulating the expression of genes involved in both lipogenesis and triglyceride synthesis. While recent studies have shown that lipogenesis does indeed significantly contribute to triglyceride synthesis in humans and that this metabolic pathway is increased in individuals with obesity and insulin resistance (9–11), the implication of ChREBP in the development of hepatic steatosis in human disease remains to be determined.

We thought that ob/ob mice would provide a valuable model for the study of the relationship between ChREBP and insulin resistance. If hepatic triglyceride accumulation is truly required for the development of insulin resistance (33,34), then preventing fat accumulation through a decrease in ChREBP expression should prevent insulin resistance. While insulin activation of Akt and MAPK was impaired in livers of Ad-GFP–injected ob/ob mice, Ad-shRNA ChREBP treatment of ob/ob mice restored insulin-stimulated Akt and MAPK activation. Therefore, ChREBP knockdown by “protecting” liver against lipid overload both under short- and long-term conditions blunted the negative effect of intrahepatic triglyceride on liver insulin sensitivity. As a consequence, the restored inhibition of genes from the gluconeogenic pathway (G6Pase and PEPCK) by insulin led to the improvement of blood glucose levels in fasted ob/ob mice. The positive effect of ChREBP knockdown was however more apparent under long-term conditions. Indeed, correction of hepatic steatosis also led to decreased levels of plasma triglycerides and NEFAs, and, as a result, insulin sensitivity was restored in both skeletal muscles and adipose tissue. Skeletal muscle is known to play a determinant role in the physiopathology of insulin resistance, and defects in glycogen synthesis have been particularly implicated in the development of the pathogenesis (35). Consistent with these observations, our results show a decrease in insulin signaling and glycogen content in ob/ob mice as well as a significant accumulation of intramuscular lipids in their skeletal muscles. Interestingly, a 7-day Ad-shChREBP treatment overcame all of these metabolic disorders. First, decreased fat content in skeletal muscles, caused by the concomitant decrease in circulating lipids and in lipogenic rates, led to the improvement in insulin signaling and glycogen synthesis. In turn, blood glucose concentrations were improved. The normalization of glycemia and insulinemia in Ad-shChREBP–treated ob/ob mice also caused a significant decrease in lipogenic rates in adipose tissue contributing to the weight loss of the tissue. Altogether, the overall phenotype of Ad-shChREBP–treated ob/ob mice is a significant improvement in hyperlipidemia, hyperglycemia, and hyperinsulinemia (Fig. 7).

In conclusion, the liver-specific inhibition of ChREBP has helped us define its role in the physiopathology of hepatic steatosis and insulin resistance in ob/ob mice. Since ChREBP knockdown markedly prevents fat accumulation and significantly restores overall insulin sensitivity, ChREBP may represent a potential therapeutic target for the treatment of fatty liver disease and insulin resistance in the future.

This work was supported by a grant from Alfediam/Sanofi-Synthelabo and from the Agence Nationale pour la Recherche (ANR 2005 Cardiovasculaire Obésité et Diabète, ANR-05-PCOD-035-02). R.D. is a recipient of a doctoral fellowship from the Ministère de l’Enseignement Supérieur et de la Recherche and received a financial support from the Société Française d’Endocrinologie.

We thank Evelyne Souil from the Plate-Forme de Morphologie/Histologie, Cochin Institute, for performing staining experiments and Dr. Benoit Viollet for helpful discussion. Mice used in this study were housed in an animal facility equipped with the help of the Région Ile de France.