Nuclear factor erythroid-2 related factor 2 (Nrf2) plays a pivotal role in cytoprotection against both endogenous and exogenous stresses. Here, we establish a novel molecular link between Nrf2, nuclear receptor small heterodimer partner (SHP; NROB2), lipogenic genes, and hepatic lipid homeostasis. Deletion of Nrf2 (Nrf2−/−) in mice resulted in a reduced liver weight, a decrease in fatty acid content of hepatic triacylglycerol, as well as concomitant increases in the levels of serum VLDL-triglyceride (TG), HDL cholesterol, and ketone bodies at 6 mo of age. Liver weight and hepatic TG content were consistently lower in Nrf2−/− mice upon a high-fat challenge. This phenotype was accompanied by downregulation of genes in lipid synthesis and uptake and upregulation of genes in lipid oxidation in older Nrf2−/− mice. Interestingly, SHP expression was induced with age in Nrf2+/+ mice but decreased by Nrf2 deficiency. Forced expression and activation of Nrf2 by Nrf2 activators consistently induced SHP expression, and Nrf2 was identified as a novel activator of the SHP gene transcription. We also identified PPAR-γ, Fas, Scd1, and Srebp-1 as direct targets of Nrf2 activation. These findings provide evidence for a role of Nrf2 in the modulation of hepatic lipid homeostasis through transcriptional activation of SHP and lipogenic gene expression.
- transcription factors
- nuclear receptors
- hepatic lipid homeostasis
nuclear factor erythroid-2 related factor 2 (Nrf2) transcription factor plays a significant role in conferring cytoprotection against endogenous oxidative stress and exogenous xenobiotics (17). Nrf2 exerts its biological regulatory effect by controlling the expression of genes required for free radical scavenging, detoxification of xenobiotics, and maintenance of the redox potential (6). Nrf2 regulated genes, known as phase II enzymes, are involved in glutathione (GSH) and NADPH production, detoxification of reactive oxygen species/reactive nitrogen species and xenobiotics (14, 36). When activated, Nrf2 targets promoters of genes containing antioxidant response element (ARE) through heterodimerization with members of the small Maf family of transcription factors and binds to the ARE region of ARE-regulated genes (37).
Kelch-like ECH-associated protein 1 (Keap1) negatively regulates Nrf2 activity by promoting the proteasomal degradation of Nrf2 (16). Under quiescent conditions, Nrf2 interacts with Keap1 and remains in the cytoplasm, which maintains low expression of Nrf2-regulated genes. With exposure to oxidative and electrophilic molecules, Nrf2 is released from Keap1, translocates to the nucleus, and transactivates the expression of cytoprotective genes that enhance cell survival. Nrf2 activity can be induced by a wide range of natural and synthetic small molecules, including isothiocyanates, 1,2-dithiole-3-thiones, heavy metals, and hydroperoxides (32). Consistently, the Nrf2−/− mice are more sensitive to inflammatory stresses, plus acute and chronic exposures to environmental agents and carcinogens (17). Despite its pivotal role in drug metabolism, the function of Nrf2 in hepatic lipid homeostasis is not fully elucidated.
Nuclear receptors are important lipid metabolic regulators (2). Peroxisome proliferator-activated receptor (PPAR), liver X receptor, and farnesoid X receptor, whose endogenous ligands are fatty acids, oxysterols, and bile acids, play critical roles in the coordinated control of fatty acids and cholesterol metabolism (3), through the regulation of key lipogenic enzymes including sterol regulatory element-binding protein 1, stearoyl-CoA desaturase-1, and fatty acid synthase. The orphan receptor small heterodimer partner (SHP) is also an essential regulator of hepatic lipid accumulation, since the lipid content was significantly decreased in the liver of SHP−/− and OB−/−SHP−/− mice (12, 38, 40) but increased in SHP-overexpressed transgenic mice (4), owing to alterations of critical genes in major hepatic lipid metabolic pathways. Several recent studies that revealed changes in hepatic lipid profile and diet-induced obesity in Nrf2−/− mice (29, 33, 34) raised intriguing questions with regard to a potential function of Nrf2 in metabolic regulation. In this study, we identified a novel molecular link between Nrf2, SHP, and other lipogenic genes in the regulation of hepatic lipid content, elucidating therefore the molecular mechanism by which Nrf2 modulates hepatic lipid homeostasis.
MATERIALS AND METHODS
Nrf2-deficient (Nrf2−/−) mice with a hybrid background C57BL6/SJL were obtained from Dr. Yuet Wai Kan [University of California, San Francisco (UCSF)]. They were backcrossed with the C57BL/6 mice to the fourth generation in our laboratory. Nrf2−/− male mice and their littermate controls Nrf2+/+ (wild-type) used in this study were obtained from Nrf2+/− intercrossed mice. All the experiments used only male mice unless specified in the procedure. The gross liver morphology and Harris hematoxylin and eosin (H&E) staining were done in 6-mo-old mice that were backcrossed with the C57BL/6 mice to the 7th generation. Mice were fed a standard rodent chow (test diet no. 5001), water ad libitum in temperature controlled (23°C) and virus-free facilities. For the high-fat diet (HFD) feeding experiments, male mice with matched age groups were fed with a 42% HFD for 12 wk. The HFD (TD 88137, Harlan Teklad) used contained 21% (wt/wt) milk fat and 49% (wt/wt) carbohydrate from sucrose and corn starch, yielding 42% kcal as fat and a caloric value of 20.1 kJ/g. For antioxidant butylated hydroxyanisole (BHA) feeding, the standard rodent chow containing 0.5% BHA was used to feed 2-mo-old mice for 2 wk. A day earlier than euthanasia, in both the HFD and BHA feeding, mice were fasted overnight and blood was collected. These mice were euthanized by inhaling CO2 and their livers were collected for histological analysis. Livers were fixed, dehydrated, embedded in paraffin, sectioned (8 μm) and stained with H&E (Sigma). For Oil Red O staining (ORO), frozen tissue sections were used as previously described (12). Tissues not used for histology were frozen in liquid nitrogen and stored at −80°C. Protocols for animal use were approved by the Institutional Animal Care and Use Committee at the University of Utah.
Serum and tissue chemistry, RNA analysis.
Briefly, serum or plasma obtained from blood collected from individual Nrf2+/+ and Nrf2−/− male mice (n = 10/genotype) were pooled and stored at −20°C for performing lipid analysis. Hepatic lipid was extracted as described in our previous methodology (12). Triglycerides (Thermo), total cholesterol (Wako), and ketone body levels (Stanbio) were estimated by using the enzymatic assay kits according to the manufacturer's instructions. Plasma lipoprotein profile was analyzed by fast phase liquid chromatography (FPLC) gel filtration as also described previously (12). Total RNA was isolated by use of Trizol reagent (Invitrogen) (12) and the gene expression was analyzed by semiquantitative RT-PCR and real-time quantitative PCR (qPCR). β-Actin housekeeping gene showed stable expression in the liver of C57BL/6 mice and in nonalcoholic fatty liver disease (NAFLD)-susceptible LDL−/− mice (11); thus it was used as an internal control. Abbreviations of genes and the primer sequences are presented in Supplementary Table S1 and Supplementary Tables S2–S3, respectively (the online version of this article contains supplemental data).
LC-GC analysis of fatty acid composition in lipid classes.
Liquid and gas chromatography (LC-GC) was used for chromatographic separation of lipid classes and for quantification of fatty acids and lipid metabolites in the liver and serum (Lipomics Technologies). Blood was collected from wild-type and Nrf2−/− mice fasted overnight and serum was purified. Liver was perfused with PBS to remove blood before being collected. Livers (200 mg) and sera (400 μl) from 10 mice in each group were pooled and used for LC-GC lipid profile analysis. Similar results were obtained in a repeating experiment with regard to changes in fatty acid composition in hepatic lipid cholesterol ester (CE) and triglyceride (TG) classes. The data represent the quantitative measurement of each fatty acid in the free fatty acid lipid class and values are expressed as fatty acid composition in units of nmol per gram of tissue (liver) or liquid material (serum).
Transient transfection, mutagenesis, and ChIP assays.
All the methods have been described elsewhere (12). Mutagenesis was performed using the Excite PCR-based site directed mutagenesis kit provided by Stratagene according to manufacturer's instructions. For luciferase assays, HEK293 cells were transfected by use of Fugene-6 (Roche), and luciferase activity was normalized against β-galactosidase activity (Promega). For chromatin immunoprecipitation (ChIP) assays, HepG2 cells were used: chromatin was cross-linked and immunoprecipitated with rabbit anti-Nrf2 antibodies (sc-13032, Santa Cruz) and rabbit normal IgG (serving as a negative control). Primers used for ChIP assays and mutagenesis were listed in Supplementary Table S4.
Hepatocyte culture, adenoviral transduction, RNA interference.
Established procedures were used to isolate and culture mouse primary hepatocytes (12). Primary hepatocytes from 2-mo-old male mice of each genotype were isolated, infected the next day with green fluorescent protein (GFP) control or GFP-Nrf2 adenovirus (50 multiplicity of infection) (8) for 2 h, or otherwise transfected with a pcDNA3.1 control (10 μg) or pcDNA3.1/Nrf2 expression plasmid (10 μg) or Nrf2 small interfering RNA (siRNA) plasmid (10 μg) (Openbiosystems) by use of Lipofectamine 2000. The cells were harvested 2 days later, RNA was extracted from each group, and their cDNAs were synthesized. Gene expression was analyzed by semiquantitative RT-PCR or real-time qPCR, with β-actin and hypoxanthine phosphoribosyl transferase as internal controls.
In vitro stimulation of Nrf2 expression.
Hepatocytes were treated with 50 μM oxidized docosahexaenoic acid, C22.6 omega-3 (oxDHA) (9) or 0.1% BHA (18) (Sigma) for 5 h. The cells were harvested for total RNA isolation and gene expression analysis.
All the experiments were done in triplicate or repeated at least three times and the error bars represent SE. Statistical analyses were carried out by Student's unpaired t-test; P < 0.01 was considered statistically significant.
Decreased liver weight in old Nrf2−/− mice.
Liver of Nrf2−/− mice was apparently smaller than in wild-type controls at 6 mo of age fed with a standard chow diet (Fig. 1A, left). The average total liver weight and the percentage of liver weight to body weight were significantly reduced in Nrf2−/− mice compared with wild-type mice (Fig. 1A, right). H&E staining revealed similar liver morphology between wild-type and Nrf2−/− mice, although there were more small fat vacuoles in wild-type mice than in Nrf2−/− mice (Fig. 1B). Histological analysis of liver sections by ORO staining did not show marked differences in lipid droplets staining between the wild-type and Nrf2−/− mice (not shown). Microvesicular fatty changes may not be easily detected or quantified by the less quantitative ORO staining method. Nonetheless, the reduced liver weight may be contributed by the moderately decreased hepatic total lipid and TG content in Nrf2−/− mice (Fig. 1C). However, no significant changes in liver weight or morphology were observed in 2-mo-old mice (not shown). On the whole, Nrf2 deficiency in older mice resulted in decreased liver weight.
The gain in body weight of Nrf2−/− mice was higher than in their wild-type counterparts at the same age (Fig. 1D, pink vs. dark blue). When the mice were challenged with a HFD, further weight gain was observed in Nrf2−/− mice compared with wild-type mice (Fig. 1D, light blue vs. green, and Supplementary Fig. S1). Interestingly, the liver weight was slightly but not significantly lower in Nrf2−/− mice compared with the wild-type mice after a HFD (Fig. 1E). The hepatic TG content was significantly reduced (Fig. 1F, left) and the Nrf2−/− liver appeared less pale in color, indicating less neutral lipid accumulation (Fig. 1F, right) than the wild-type controls. An increase in the proportion of fat mass was similar between Nrf2−/− and wild-type mice on HFD (Fig. 1G). The expression of Nrf2 was increased by HFD in the WT mice (Fig. 1H), consistent with its role in hepatic lipid metabolism. An equal weight gain by a HFD in the wild-type and Nrf2−/− mice was also reported by another group (34). A most recent study by Shin et al. (29) using young female Nrf2−/− mice showed less liver weight gain, and decreased hepatic TG and cholesterol levels by a HFD. Taken together, the results suggest that Nrf2 deletion may have differential effects on fat redistribution between adipocytes and hepatocytes, which may also be affected by the sex and age of the mice.
Changes of hepatic CEs and TAGs in aged Nrf2−/− mice.
Total liver lipid content of Nrf2−/− and wild-type mice at 2 mo of age did not differ significantly on a normal-chow diet (not shown). To determine the effect of Nrf2 in regulating hepatic lipid homeostasis, we evaluated lipid profiles in the serum and liver of male wild-type and Nrf2−/− mice using a comprehensive LC-GC analysis. The relative fatty acid composition in each fraction was determined and the relative amount of the fatty acids measured in the CE and triacylglycerol (TAG) fractions was shown in Fig. 2. The relative amount of the major fatty acids (14:0, 16:0, 16:1n7, 18:1n7, 18:1n9, and 18:2n6) in the hepatic CE fraction of Nrf2−/− mice increased by ∼40%, which was accompanied by a concomitant decrease of fatty acids in the triglyceride fraction compared with wild-type controls (Fig. 2A). When the fatty acids were grouped in different classes, including free fatty acids, saturated fatty acids, monounsaturated fatty acids, polyunsaturated fatty acids, n3, n6, n7, and n9A, changes in fatty acid classes reflected changes in individual fatty acid composition (Fig. 2B). In the total lipid fraction, there was ∼30% elevation and ∼40% reduction in the CEs and TAGs, respectively, in the liver of Nrf2−/− mice relative to wild-type mice (Fig. 2C). There were only minor changes in the composition of other fatty acids, such as those measured in lysophosphatidylcholine, phosphatidylcholine, and phosphatidylethanolamine between the wild-type and Nrf2−/− group (Supplementary Figs. S2A–B), suggesting the changes in fatty acid composition in CE and TAG were specific to CE and TAG. Interestingly, the levels of fatty acids in the serum of Nrf2−/− mice did not show marked changes compared with wild-type controls. Taken together, the results suggest the hepatic metabolism of CEs and triglycerides are modulated by Nrf2 deficiency.
Changes of lipid and lipoprotein profiles in Nrf2−/− mice.
Liver TG levels are a reflection of the balance of complex processes of input, output, synthesis, and oxidation of fatty acids. Serum lipoprotein profiles were examined (VLDL secretion) by FPLC gel filtration to determine the hepatic lipid export. FPLC detected a VLDL-TG and a LDL-TG peak in wild-type mice, and similar results were reported in mice and rabbits (23, 25, 26). A significant difference was observed in the distribution of TG in the VLDL and LDL lipoprotein fractions between wild-type and Nrf2−/− mice. The levels of TG in Nrf2−/− mice were ∼2.5-fold higher in the VLDL fraction but were almost absent in the LDL fraction (Fig. 3A, left). In addition, Nrf2−/− mice had ∼1.5-fold increase in cholesterol levels in the HDL fraction relative to wild-type mice (right). Ketones are products of hepatic fatty acid oxidation that are released into the circulation and 3-β-hydroxybutyrate (3HB) is a major ketone body synthesized in the liver (20). 3HB levels were markedly induced by fasting in wild-type mice, but to a much greater extent in Nrf2−/− mice (Fig. 3B). The total cholesterol level in serum (Fig. 3C, left) appeared to be lower in Nrf2−/− mice under the fasting condition, but similar to that in wild-type mice under the fed condition. There is an increase in the serum total cholesterol of both wild-type and Nrf2−/− mice under fasting condition, which may be associated with the increased lipolysis in response to fasting. However, the liver cholesterol did not differ markedly in wild-type and Nrf2−/− mice (right). All these data suggest that both increased VLDL export and fatty acids oxidation may contribute to the reduction of hepatic TG content in Nrf2−/− mice.
CEs in LDL and HDL particles are mainly synthesized by two different enzymes, acyl-CoA:cholesterol acyltransferase (Acat) and lecithin-cholesterol acyltransferase (Lcat), respectively (35). To determine whether Acat and Lcat were involved in modulating the CE profile in Nrf2−/− mice, we measured mRNA expression of Acat1, Acat2, and Lcat by real-time qPCR. The basal mRNA level of Acat1 was notably low (Fig. 3D), consistent with the notion that Acat2 is the major cholesterol-esterifying enzyme in the liver (28). The expression of all three enzymes in wild-type and Nrf2−/− mice was drastically induced by fasting. However, the induction of Lcat was significantly higher in Nrf2−/− mice than in wild-type mice. Decreased Lcat mRNA expression has previously been shown by a parallel decrease in plasma activity (30). Overall, the data suggest that the induction of Lcat by fasting may facilitate the process of cholesterol esterification, which may be associated with increased hepatic CE and serum HDL cholesterol in Nrf2−/− mice.
Expression profile analysis of hepatic genes in Nrf2−/− mice.
To establish the functionality between Nrf2 and hepatic lipid homeostasis, we analyzed the gene expression profile of liver RNA from 6-mo-old mice compared with liver RNA of 2-mo-old mice using semiquantitative PCR (Fig. 4A). Semiquantitative PCR was used because it was less expensive, although it was also less quantitative. To more quantitatively assess the expression, selected genes were further analyzed by qPCR (Fig. 4, B–D). The gene profile revealed several interesting observations. The mRNA for Nrf2 and its target genes Nqo1 and Gclc were significantly upregulated in 6-mo-old wild-type mice compared with 2-mo-old wild-type mice. Genes involved in lipid synthesis and uptake, including Srebp, Fas, Scd-1, PPAR-γ, Cd36, and Ldlr, were generally downregulated in 6-mo-old Nrf2−/− mice compared with the wild-type mice. On the other hand, genes regulating fatty acid oxidation, such as PPAR-α, Aco, and Cpt1α, were generally upregulated in older Nrf2−/− mice relative to wild-type mice. In addition, the expression of some genes was increased by HFD in the wild-type mice, and their expression was markedly lower in Nrf2−/− mice (Supplementary Fig. S3), including PPAR-γ, Cd36, Scd-1, etc.
Semiquantitative PCR analysis did not reveal marked changes in gene expression for selected nuclear receptors that are important for liver function (not shown). To confirm this observation, a more sensitive and quantitative qPCR analysis was performed that revealed no marked differences in liver mRNA levels for the majority of the nuclear receptors when the wild-type and Nrf2−/− mice were compared (Fig. 4B), although a moderate upregulation of liver X receptor was observed, which might be associated to the increased levels of PPAR-α, Aco, and Cpt1α (Fig. 4A). Interestingly, SHP expression showed about a 60% reduction in Nrf2-deficient liver (Fig. 4B, inset). Consistent with the semiquantitative PCR results in Fig. 4A, qPCR also revealed a marked induction of Nrf2 mRNA, as well as SHP, in 6-mo-old wild-type mice compared with 2-mo-old mice (Fig. 4C). Because hepatic lipid levels accumulated progressively with age, the induction of Nrf2 in these mice suggests that Nrf2 may function as a potential lipid metabolic regulator. Considering the fact that Nrf2 is a transcriptional activator, we reasoned that the induction of SHP in aged mice might be associated with the activation of Nrf2. However, SHP mRNA was not markedly lower in 2-mo-old Nrf2−/− mice, suggesting that SHP may be regulated via the induction of Nrf2 under stress conditions.
One important function of SHP is to control bile acid synthesis by repressing the expression of the rate-limiting enzymes Cyp7A and Cyp8B (39). We therefore next assessed the alterations in the expression of genes involved in bile acid metabolism to determine whether Nrf2−/− mice had a similar pattern of changes in gene expression as in SHP−/− mice. Cyp7A was upregulated in SHP−/− mice owing to loss of SHP repression (13). As expected, an increased expression of Cyp7A and Cyp8B was observed in 6-mo-old Nrf2−/− mice in which SHP was downregulated (Fig. 4D). In addition, many other genes exhibited similar changes in SHP−/− (10, 12) and Nrf2−/− mice (Fig. 4E). Some examples included Ntcp (no change), mEH (down), Oatp1 (down), apoA1 (no change), and apoAIV (no change) (10, 12). These results confirmed that the SHP function was diminished in Nrf2−/− mice.
Nrf2 activation of SHP gene expression.
Considering the above results, we next tested the ability of Nrf2 to modulate SHP expression. In agreement with the decreased expression in Nrf2−/− liver, SHP mRNA was also markedly reduced in primary hepatocytes of Nrf2−/− mice (Fig. 5A, lane 2 vs. 1). Reexpression of Nrf2 using an expression plasmid in Nrf2−/− hepatocytes reactivated expression of SHP and the Nrf2 target Nqo1 (lane 4 vs. 3). An induction of SHP was also observed by infection of wild-type hepatocytes with Nrf2 adenovirus (Fig. 5B). A siRNA knockdown approach was used to determine whether SHP expression was dependent on Nrf2. Primary hepatocytes were transiently transfected with either control siRNA or four different siRNA plasmids directed against Nrf2. Because of a low knockdown efficiency, downregulation of Nrf2 by Nrf2 siRNA resulted in a moderate repression of SHP expression (not shown).
A recent study reported that oxDHA were capable of inducing Nrf2-directed gene expression (15). To gain additional insight into the molecular mechanisms underlying the induction of SHP by Nrf2, primary hepatocytes were exposed to oxDHA (50 μM) and both Nrf2 and SHP showed a significant increase in oxDHA=treated cells (Fig. 5C). BHA is a synthetic phenolic antioxidant that can induce Nrf2 expression in rat hepatocytes (16). To determine whether Nrf2-mediated SHP activation represents a common mechanism, we treated the mouse normal hepatocyte cell line NMuli with 0.1% BHA and a significant induction in the levels of Nrf2 and SHP was observed (Fig. 5D). To confirm this result in vivo, wild-type mice were fed with a control diet or a diet supplemented with 0.5% BHA for 2 wk. As expected, the mRNA levels of both Nrf2 and SHP were induced by the BHA containing diet (Fig. 5E). Interestingly, Nqo1 was activated by BHA to a much greater extent than Nrf2, suggesting that a moderate activation of Nrf2 by BHA in vivo would result in a profound induction of Nqo1. In addition, SHP expression was markedly lower in Nrf2−/− hepatocytes treated with BHA compared with Nrf2+/+ hepatocytes (Fig. 5F), confirming a Nrf2-dependent activation of SHP. SHP was moderately decreased in Nrf2−/− than in wild-type mice that were given a dietary BHA (Fig. 5G), suggesting that Nrf2-mediated regulation of SHP expression becomes more dominant under in vitro hepatocytes culture conditions, which may be masked under in vivo conditions likely because of SHP regulation by other factors. This also reflected the fact that gene regulation in vivo was more complicated than in vitro. Moreover, overexpression of SHP in Nrf2−/− hepatocytes induced both TG and cholesterol content (Fig. 5H), which is consistent with the role of SHP in induction of fat accumulation in liver.
The above data strongly suggest that the induction of Nrf2 results in transcriptional activation of SHP. To determine the molecular basis for this observation, we analyzed the mouse SHP promoter and identified a conserved ARE binding site (ggctgggtca, −482 to −474 bp) (Supplementary Fig. S4). To our surprise, Nrf2 was not found to transactivate the SHP promoter when transfected alone into the cells (Fig. 5I, lane 2). We reasoned that the activation of Nrf2 may require the coexpression of its coactivators. Brahma-related gene 1 (Brg1) is a catalytic subunit of SWI2/SNF2-like chromatin remodeling complexes that interacts with Nrf2 and enhances Nrf2 activation of its target gene HO-1 (13, 43). Activating transcription factor 4 (Atf4) has been reported as a Nrf2-interacting protein to activate heme oxygenase-1 gene transcription (10). Coactivator Src3 was shown to stimulate the transactivation of Gal4-Nrf4 (21). Thus we tested the effects of these factors and two other important coactivator family proteins p300 and Src1 in Nrf2-mediated transcriptional activity of the SHP promoter. Transfection of each plasmid expressing Brg1, p300, Src1, and Src3 alone did not activate the SHP promoter (Supplementary Fig. S5), but significantly enhanced the activity of Nrf2 when coexpressed with Nrf2 in the cells (Fig. 5I). Interestingly, Atf4 did not exhibit enhancing effect on the Nrf2 activity, suggesting that Atf4 may function as a Nrf2 coactivator in a promoter-dependent manner. In addition, mutation of the ARE site abolished SHP promoter activity induced by Nrf2 and its coactivators (not shown). Furthermore, ChIP assays confirmed that Nrf2 coimmunoprecipitated with the SHP promoter (Fig. 5J). These data demonstrate that Nrf2 is a transcription factor that is able to induce SHP expression via binding to the ARE in the SHP promoter, which requires the presence of its coactivators. Because the induction of SHP was not completely abolished in Nrf2−/− mice (Fig. 4B) and the fold induction of Nrf2 and SHP in aged mice appeared to be different (Fig. 4C), it is postulated that the in vivo regulation of SHP expression is controlled not only by Nrf2, but also by other transcription factors.
Nrf2 targeting lipid metabolic genes.
Our previous data showed that the expression of the critical lipid metabolic genes PPAR-γ, Fas, Scd1, and Srebp1c were increased in 6-mo-old wild-type mice in which Nrf2 was upregulated, but diminished in Nrf2−/− mice by semiquantitative PCR analysis (Fig. 4A). We confirmed this result using the more quantitative qPCR analysis (Fig. 6A). The results suggest that the transcription of these genes may be activated by Nrf2. We analyzed the promoter regions of these genes and multiple potential ARE sites were identified (Supplementary Figs. S6–S9).
The direct association of Nrf2 with these gene promoters in vivo in liver was assessed by use of liver ChIP assays. Nrf2 was coimmunoprecipitated on the ARE containing the promoter regions of PPAR-γ, Fas, Scd1, and Srebp1c in wild-type but not in Nrf2−/− liver (Fig. 6B). The binding of Nrf2 to the upstream region of PPAR-γ (916) promoter appeared to be much stronger than to the downstream region of the PPAR-γ (334) promoter, which was in agreement with a perfect ARE site contained in the upstream region (Supplementary Fig. S6).
Interestingly, in transient transfection assays, expression of Nrf2 alone did not activate the promoters of PPAR-γ, Fas, Scd1, and Srebp1c (Fig. 6C), whereas its effect was strengthened by the addition of its coactivators. In general, Brg1 appeared to be the strongest coactivator of Nrf2 to activate these lipogenic genes. Forced expression of the coactivators alone did not activate the luciferase expression of these promoters. Overall, the data provide the first evidence that Nrf2 is a transcriptional activator of those lipid metabolizing genes.
Fatty liver is defined as the accumulation of lipid in the liver, primarily in the form of TAGs (24, 42). Fatty liver is often associated with active drug metabolism; however, the direct molecular link between drug and lipid homeostasis remains unclear. Nrf2 plays a crucial role in the defense against endogenous and exogenous oxidative stress by regulating the expression of genes involved in cell stress response, drug metabolism, detoxification, and transport. Animal studies on Nrf2 knockout mice suggest that this transcription factor may be critical for protecting against hepatic and gastrointestinal diseases (1, 27). Recent studies suggest that nuclear receptors may be important therapeutic targets in liver diseases (44). The present study not only determines the importance of Nrf2 as a regulator of hepatic lipid homeostasis, but it also elucidates the molecular mechanisms by which Nrf2 regulates the expression of key genes in lipid metabolism and constitutes the evidence for an existing link between drug and hepatic lipid metabolism.
LC-GC analysis for lipid profiling is rapidly becoming an important research tool for identifying novel mechanisms regulating lipid metabolism (5). To assess the biological potential of Nrf2 in hepatic lipid homeostasis, we performed high-resolution lipidomics analyses to accurately quantify lipid species in wild-type and Nrf2−/− mice. In conjunction with FPLC analysis, we observed that Nrf2 deficiency altered hepatic lipid profiling, as reflected by phenotypically reduced hepatic TG content in older mice, increased VLDL-TG level that represents increased VLDL secretion, increased HDL cholesterol level, increased ketone bodies production that represents increased lipid oxidation, downregulation of genes in lipid uptake (PPAR-γ, Cd36) and synthesis (Fas, Scd1, and Srebp), and upregulation of gene in lipid oxidation (PPAR-α, Aco, and Cpt1α). At the molecular level, we demonstrated that the decreased expression of SHP (repressing lipid secretion), PPAR-γ (increasing lipid uptake), Fas, Scd1, and Srebp1c (increasing lipid synthesis) in Nrf2−/− mice was a result of the loss of Nrf2 transcriptional activation. The diminished function of SHP and the lipogenic genes is expected to have coordinately increased hepatic lipid export, decreased lipid import and synthesis, and eliminated lipid accumulation.
Nrf2 is widely expressed in tissues including liver, heart, brain, spleen, intestine, lung, muscle, kidney, and testis, etc. (7). Thus it is likely that deletion of Nrf2 may affect fatty acid uptake, absorption, and utilization via other peripheral tissues. A recent study showed that Nrf2 deletion in ApoE-null background significantly reduced plaque formation thus attenuated ApoE-mediated atherosclerosis by decreasing the expression of Cd36 and LDL uptake (31). We found a decreased PPAR-γ, Cd36, and Ldlr expression in Nrf2−/− liver, which would presumably decrease lipid uptake to the liver. On the other hand, the increased fatty acid composition in hepatic CE is likely associated with the increased Lcat. It is proposed that extrahepatic CE would be increased as well. The increased extrahepatic CE, along with the upregulated Abca1 and Lcat, may facilitate the formation of HDL cholesterol. It is noted that despite a reduced liver weight in older Nrf2−/− mice, ORO staining did not reveal a marked difference in neutral lipid staining. ORO stains neutral lipids containing both TG and CE. The hepatic TG was elevated but CE was reduced in Nrf2−/− mice relative to the wild-type mice; thus the overall lipid alterations may not be distinguished by ORO.
It has previously been reported (19) that hepatic Nrf2 expression increases in a mouse model of diet-induced obesity and fatty liver (C57BL/6J on a 12 wk high-fat diet). This is in accordance with our findings showing that the expression of Nrf2 is significantly increased with HFD and age: in both conditions more lipids had been accumulated in the liver. We showed that the hepatic TG was increased but CE was decreased in Nrf2-deficient mice. These results are consistent with the observation by Tanaka et al. (34) showing a mild increase of hepatic cholesterol and a decrease of TG in control non-HFD fed Nrf2−/− mice. In fact, Takana et al. showed that hepatic TG was consistently lower in Nrf2−/− mice after a HFD. A similar alteration was also reported in Shin et al.'s study (29) in which a HFD markedly lowered hepatic TG and cholesterol levels in Nrf2−/− mice. Interestingly, Tanaka et al. showed a decline in Nrf2 expression with HFD. Nrf2 is an antioxidant gene and is often induced under stressed conditions. It has been shown that many defense and stress response genes, including GST, are upregulated in wild-type mice fed with a HFD (19). Since GST functions a downstream target of Nrf2, it seems reasonable to see its upregulation under such experimental conditions. Since no molecular mechanisms were elucidated in Tanaka et al. study, it is currently unknown whether the alterations in Nrf2 and other gene expression in their study was due to a direct loss of Nrf2 regulation or due to a secondary effect of the HFD. Another study by Tanaka et al. showed that ANIT induced both Nrf2 and SHP expression in the wild-type mice, suggesting that activation of Nrf2 by ANIT corresponded to the induction of SHP (33). These results are in agreement with our observation that Nrf2 functions as a positive regulator of the SHP gene expression that was demonstrated in our in vitro and in vivo studies.
The hepatic expression of Nrf2 was significantly increased in 6-mo-old mice. This observation indicates that Nrf2 activity is increased under conditions that result in fatty liver and raises the question whether it is activated by specific lipid metabolites. Gao et al. (9) reported that oxDHA, a n-3 fatty acid metabolite, activated Nrf2-dependent ARE-directed gene expression by stabilizing Nrf2. In addition, the fatty acid derivative 15-deoxy-Δ12,14-prostaglandin J2 was shown to serve as an endogenous regulator of Nrf2 signaling (15). It is postulated that when hepatic lipid accumulates, because of age, excess dietary consumption, and the imbalance of endogenous metabolism, more lipid metabolites and oxidation products will be produced. Some metabolites may function as Nrf2 activators to induce Nrf2 and Nrf2-dependent gene expression. Indeed, reduced hepatic lipid content became more evident in older Nrf2−/− mice and was not observed in young Nrf2−/− mice.
Because the lipogenic Nrf2 target genes that we identified can also be directly induced by fatty acids metabolites, we propose that upregulation of these genes by Nrf2 provides an additional pathway to further facilitate the accumulation of hepatic lipid. Indeed, reduced hepatic lipid content became more evident with age in Nrf2−/− mice and was not observed in young Nrf2−/− mice. On the other hand, emerging evidence has shown that TG accumulation in the liver in response to lipid overloading serves as a protective mechanism against lipotoxicity (22, 41). Therefore, activation of lipogenic genes and cytoprotective enzymes by Nrf2 may represent a mechanism to protect cells from lipotoxic and oxidative damages.
Boulias et al. (4) identified SHP as an important regulator of key genes in hepatic cholesterol degradation and lipogenesis and showed that mice constitutively expressing SHP in hepatocytes developed a fatty liver phenotype. It is interesting to note that the hepatic lipid phenotype did not occur in the young Nrf2−/− mice but developed with age in older Nrf2−/− mice. In SHP−/− mice, the decreased lipid accumulation in the liver became more evident when the mice were 6 mo old or challenged with a high-cholesterol (38) or a high-fat diet (40). However, when SHP was deleted in the OB/OB mice, fatty liver of the OB/OB mice was largely diminished even at 2 mo of age (12). Thus the function of SHP in the development of fatty liver appeared to be more significant under severe disease conditions. Our present study identified SHP as a direct target of Nrf2. The hepatic lipid phenotype, such as less hepatic TG content and increased VLDL secretion and HDL cholesterol observed in older Nrf2−/− mice, is in general in agreement with what was seen in the SHP−/− mice. It should be noted that even though SHP is activated by Nrf2, it is also regulated by many other nuclear receptors involved in hepatic lipid homeostasis. Therefore, the phenotype between Nrf2−/− and SHP−/− mice, although similar in many aspects, would not be expected to be identical. Nonetheless, on the basis of the present findings and previously reported data, it may be suggested that Nrf2 regulates lipid metabolism in the liver through two molecular pathways (Fig. 7): either via a direct activation of SHP or via the activation of PPAR-γ, Fas, Scd1, and Srebp1c, which in turn will regulate lipid homeostasis in the liver.
In conclusion, our study elucidates for the first time a molecular mechanism by which Nrf2 may function as a regulator of hepatic lipid genes and highlights the importance of Nrf2 in controlling the overall hepatic lipid homeostasis. These findings may bring new perspectives in the targeting of potential genes involved in NAFLD and the development of novel therapeutics.
This research was supported in part by the American Liver Foundation/American Association for the Study of Liver Diseases (Liver Scholar Award) and National Institutes of Health (DK080440) to L. Wang.
No conflicts of interest, financial or otherwise, are declared by the author(s).
We are grateful to Drs. Curt Hagedorn and Annette Kirchgessner for critically reading the manuscript. We thank Dr. Yuet Wai Kan (UCSF) for the Nrf2 knockout mice, Dr. Jason D. Morrow (Vanderbilt University School of Medicine) for the generous gift of oxDHA, and Dr. Xilin Chen for the Nrf2 adenovirus. We also thank Drs. Tsutomu Ohta (National Cancer Center Research Institute, Tokyo, Japan), Ken Itoh (Hirosaki University Graduate School of Medicine, Hirosaki, Japan), Jawed Alam (Alton Ochsner Medical Foundation, New Orleans, LA), Jongsook K. Kemper (University of Illinois, Urbana, IL), and Jianming Xu (Baylor College of Medicine, Houston, TX) for providing the expression plasmids for Nrf2 coactivators.
- Copyright © 2010 the American Physiological Society