Divergent Role of Estrogen-Related Receptor α in Lipid- and Fasting-Induced Hepatic Steatosis in Mice

Abstract

Given the increasing prevalence of obesity and the metabolic syndrome, identification of intrinsic molecular programs responsible for ensuring fuel homeostasis and preventing metabolic disease is needed. We investigated whether the orphan nuclear receptor estrogen-related receptor α (ERRα), a major regulator of energy metabolism, plays a role in lipid homeostasis and the development of nonalcoholic fatty liver disease (NAFLD) in response to chronic high-fat diet (HFD) consumption and long-term fasting. Systemic ablation of ERRα in mice demonstrated clear beneficial effects for loss of ERRα function in protection from HFD-provoked body weight gain manifested not only from a reduction in white adipose tissue stores but also from an impediment in intrahepatic lipid accumulation. The prevention of HFD-induced NAFLD in ERRα-null mice was underscored by transcriptional repression of de novo lipogenesis, which was upregulated in wild-type mice, a known contributing factor to lipid-stimulated hepatic steatosis. Surprisingly, given these findings, ERRα deficiency had no significant impact on the degree of fasting-induced NAFLD, involving the mobilization of adipocyte triglyceride (TG) stores into the liver. However, the presence of ERRα was essential for acute refeeding–mediated reversal of fasting-induced hepatic TG accretion, underpinned by impaired downregulation of adipose TG lipolysis and reduced hepatic mitochondrial oxidative activity. Taken together, the regulation of lipid handling by ERRα depended on the nutritional state, suggesting that negative modulation of ERRα activity could be envisaged to prevent lipid-induced NAFLD, whereas inducing its activity would be useful to treat and reverse the instilled disease.

Nonalcoholic fatty liver disease (NAFLD), marked by an excess of hepatic triglyceride (TG) levels, is a growing worldwide health concern with common risk factors including obesity and type 2 diabetes (1–3). Maintenance of TG homeostasis is essential to prevent the development of hepatic steatosis. In response to nutritional stress such as increased intake of a lipid-rich diet or cycles of food deprivation and intake, organisms adapt by modulating metabolic gene and metabolite levels. Chronic consumption of a high-fat diet (HFD) leads to obesity and the metabolic syndrome, characterized by a surplus of TG storage in white adipose tissue (WAT) and the liver. The concomitant development of insulin resistance and induction of hepatic de novo lipogenesis further contribute to the pathology of NAFLD (4). In response to prolonged fasting, the liver and WAT cooperate to mobilize WAT TG stores in the liver for storage and ketone body production and secretion to meet the basic energy needs of the body (5). This is a normal, biological adaptive response to maintain systemic energy homeostasis, and refeeding reverses the hepatic steatosis in parallel with restoration of WAT lipid stores (6). Adipose TG catabolism involves the transcriptional and posttranslational control of lipolytic lipases, including adipose TG lipase and hormone-sensitive lipase (HSL), whereby dysregulation of this process may aggravate fasting-induced NAFLD or impair its postprandial reversal (7). Additionally, a switch from fasting to refeeding stimulates hepatic glucose uptake and use and promotes de novo lipogenesis (8), thus rendering postprandial NAFLD clearance crucial to avoid hepatocyte injury due to lipotoxicity.

Several nuclear receptor (NR) family members play a pivotal role in the transcriptional control of lipid metabolism, including liver X receptor α/β (NR1H3/2), glucocorticoid receptor (NR3C1), hepatocyte nuclear factor 4α (NR2A1), retinoid X receptor α (NR2B1), REV-ERBα (NR1D1), farnesoid X receptor (NR1H4), and peroxisome proliferator–activated receptor α (NR1C1) (9–11). The orphan NR estrogen-related receptor α (ERRα; NR3B1), a master regulator of global energy metabolism (12, 13), also plays a fundamental role in lipid handling, in part by regulating intestinal fat absorption and WAT fat storage (14–16). Although genetic or pharmacological inhibition of ERRα has proven beneficial for diminishing circulating lipid profiles and in the treatment and prevention of obesity (17, 18), we have recently shown that ERRα deficiency increases the susceptibility of mice to rapamycin-induced NAFLD (19). Whether ERRα is critical to maintain hepatic lipid homeostasis during HFD consumption and fasting and refeeding nutritional challenges has not yet been explored, to our knowledge.

Herein, we establish that ERRα inhibition is beneficial to prevent HFD-induced obesity and the metabolic syndrome, and particularly protects against NAFLD development. Unexpectedly, although loss of ERRα had no consequential effect in relation to fasting-induced hepatic steatosis in mice, ERRα was be necessary for postprandial hepatic TG clearance. The data indicate that persistent WAT lipolysis and decreased liver mitochondrial oxidative function in the absence of ERRα are causal factors for the observed impaired reversal of NAFLD during acute refeeding. This work highlights the key nutrition-dependent actions of ERRα in hepatocellular lipid metabolism and has clinical implications for ERRα as a therapeutic avenue for the treatment of NAFLD and the metabolic syndrome.

Materials and Methods

Animals

Wild-type (WT) and ERRα^−/− mice (15) in a C57BL/6J genetic background were housed and fed ad libitum with free access to water in an animal facility at McGill University. All animal experiments were conducted in accord with accepted standards of humane animal care and all protocols were approved by the McGill Facility Animal Care Committee and the Canadian Council on Animal Care. For fasting and refeeding challenges, mice aged 2.5 to 3 months were separated randomly into groups of three mice per cage and were subjected to a 24-hour fast followed by a 2-hour refeeding period with free access to standard chow. During the switch from a fed to fasted state, mice were transferred from cages with corn chip to wood chip bedding. For high-fat diet (HFD) experiments, mice were separated randomly into groups of two to three mice per cage and fed either a control chow diet consisting of 10 kcal percent fat (catalog no. TD.08806; Harlan, Indianapolis, IN) or an HFD consisting of 60 kcal percent fat (catalog no. TD.06414; Harlan) during 15 weeks, initiated at 6 weeks of age. Body weight and food intake were measured weekly. For all mouse experiments, littermates were used and mice were euthanized by cervical dislocation at Zeitgeber time (ZT) 4 for serum and tissue isolations.

Glucose and insulin tolerance tests

Glucose tolerance tests and insulin tolerance tests were performed on mice fed either a control diet or an HFD for 13 or 14 weeks, respectively. For glucose tolerance tests, mice were fasted for 6 hours, initiated at ZT 2, with free access to water prior to intraperitoneal injection with a 20% glucose solution at 1 g/kg body weight. d-glucose (catalog no. 15023-021; Gibco, Thermo Fisher Scientific, Waltham, MA) was dissolved in 0.9% NaCl at 200 mg/mL, then filter sterilized and left overnight at room temperature before use. For insulin tolerance tests, mice were subjected to a 4-hour fast starting at ZT 4, followed by intraperitoneal injection with either 0.75 U or 2 U of insulin (catalog no. I0516; Sigma, St. Louis, MO) in phosphate-buffered saline (PBS) per kilogram of body weight for mice fed a control diet or HFD, respectively. Blood glucose levels were measured during a time course postinjection using a OneTouch Ultra2 glucose meter (LifeScan, Burnaby, BC, Canada).

Histology

All histological procedures were performed at the histology facility of the Goodman Cancer Research Centre. Livers and WAT were fixed in 10% buffered formalin and blocked in paraffin. Paraffin-embedded tissues were sectioned (4 μm) and stained with hematoxylin and eosin. Image J software (National Institutes of Health) was used to determine white adipocyte cell diameter. To assess hepatic steatosis, optimal cutting temperature–embedded liver sections were stained with Oil Red O. In brief, sections were washed with distilled water and then stained with 0.5% Oil Red O in propylene glycol for 16 hours. Subsequently, sections were incubated for 1 minute in 85% propylene glycol and washed twice with water. Finally, slides were counterstained with hematoxylin and the accumulation of lipid droplets in the liver sections was quantified using an Aperio digital image analysis algorithm (Aperio Technologies, Vista, CA).

Electron microscopy

Livers were collected, precut into small pieces, and fixed for 2 to 4 hours in 2.5% glutaraldehyde in 0.1 M sodium cacodylate buffer, pH 7.4, at 40°C. Subsequently, tissues were washed three times for 10 minutes each in sodium cacodylate buffer and then were cut into small cubes ∼2-mm thick. Tissues were then placed in 1% osmium tetroxide with 1.5% potassium ferrocyanide in water for 80 minutes at 40°C. After washing three times for 10 minutes each, the liver fragments were dehydrated in ascending concentrations of ethanol and propylene oxide, followed by epon embedding. After curing in the oven, the epon blocks were trimmed and 1-µm–thick semithin sections were cut, attached to glass slides, and contrasted with toluidine blue. After light microscopy observations and selection of the better fields, the epon blocks were further trimmed and ultrathin sections (60 nm) were cut on a Reichert-Jung ultramicrotome (Nussloch, Germany) using a diamond knife and placed onto formvar-coated one-slot grids. Finally, sections were counterstained with uranyl acetate and lead citrate and examined using a Philips/FEI CM120 electron microscope equipped with a digital camera. Image J software (National Institutes of Health) was used to quantify lipid burden.

Biochemistry measurements

Liver TG content was determined using a commercially available quantification kit (catalog no. ab65336; Abcam, Cambridge, United Kingdom), as were hepatic glycogen levels (catalog no. ab65620; Abcam). Blood glucose and lactate measurements were determined using a OneTouch Ultra2 glucose meter (LifeScan) and Lactate Scout (Lactate.com), respectively. Serum fibroblast growth factor 21 (FGF21) levels were measured using an enzyme-linked immunosorbent assay kit (catalog no. MF2100; R&D Systems, Minneapolis, MN) according to the manufacturer’s instructions, with a correction wavelength of 540 nm to account for optical imperfections in the microplate. Sample concentrations were determined using a four-parameter logistic standard curve fit using Prism 7 software (GraphPad Software, La Jolla, CA).

Serum 3-hydrobutyrate levels were determined by gas chromatography–mass spectrometry (GC-MS) analysis at the McGill University Goodman Cancer Research Centre metabolomics core facility. Briefly, 250 μL of cold 80% liquid chromatography–mass spectrometry–grade methanol was added to 10 μL of serum in addition to 20 μL of 1 mM ¹³C2-β-hydroxybutyrate and 1 μL of D27-mysristic acid (840 ng/μL). Samples were centrifuged at 1°C and 15,000 rpm for 10 minutes. The supernatant was transferred to fresh, precooled Eppendorf tubes (Labconco, Kansas City, MO), which were dried by vacuum centrifugation with sample temperature maintained at −4°C. A two-step derivatization process was used. Calibration standards of β-hydroxybutyrate ranging from 0 to 150 nmol per sample were prepared by serial dilution. GC-MS data were acquired using an Agilent 5975C mass selective detector coupled to a 7890A GC (Agilent Technologies, Santa Clara, CA). Electron ionization was used at 70 eV. GC-MS data were analyzed using ChemStation software (Agilent Technologies).

Hepatic very-low–density lipoprotein–TG secretion

To assess hepatic very-low–density lipoprotein (VLDL)–TG secretion rates, mice first were fasted for 4 hours, then injected intraperitoneally with poloxamer-407 (catalog no. 16758; Sigma), an inhibitor of lipoprotein lipase activity. Poloxamer-407 blocks TG lipolysis in circulation, causing an accumulation of blood TG levels and concomitant decrease in peripheral-tissue TG levels. Poloxamer-407 was dissolved in PBS and injected at 1 g/kg body weight. Blood samples were collected at the indicated times and serum TG levels were quantified using a commercially available kit (catalog no. ab65336; Abcam).

Mitochondrial respiration

Respiration rates of isolated mitochondria from livers of mice were determined. After the mice were euthanized by cervical dislocation, fresh livers were isolated and washed several times in sucrose buffer (200 mM sucrose, 5 mM Tris, 1 mM EGTA, pH 74), minced, homogenized at 1000 rpm three times in sucrose buffer, and spun at 600g for 10 minutes; the supernatant was spun again at 9000g for 10 minutes. The pellet was washed with sucrose buffer and spun at 9000g for 10 minutes, after which the mitochondria were used for assessment of mitochondrial function. Respiration measurements using the isolated mitochondria were made using a Digital Model 10 Clark Electrode (Rank Brothers, Cambridge, United Kingdom). Isolated mitochondria were incubated in KHEB respiration buffer [120 mM KCl, 5 mM KH2PO4, 3 mM HEPES, 1 mM EGTA, and 0.3% bovine serum albumin (weight-to-volume ratio), pH 7.2] at 0.6 mg of mitochondrial proteins per milliliter at 37°C. Respiratory control ratios (RCRs) were used to determine the quality of the mitochondrial suspensions. RCR values were obtained by dividing the rate of oxygen consumption in the presence of adenosine diphosphate (state 3) by that in the presence of oligomycin (state 4). Only mitochondrial suspensions with RCR values greater than 3 were used for further studies.

Western blotting

Mouse white adipose whole-cell protein extracts were prepared using Buffer K (sodium phosphate 20 mM, NaCl 150 mM, NP40 1%, EDTA 5 mM) containing protease and phosphatase inhibitor cocktail tablets (Roche, Laval, QC, Canada) and quantified using the Bradford method (protein assay; Bio-Rad Laboratories, Hercules, CA). Proteins were resolved on 8% sodium dodecyl sulfate–polyacrylamide gel electrophoresis gels and then transferred onto polyvinylidene fluoride membranes (Amersham Biosciences, Mississauga, ON, Canada) and blocked for 1 hour at room temperature in phosphate-buffered saline plus 0.1% Tween-20 (PBS-T) containing 5% milk. Membranes were incubated overnight with primary antibodies diluted in PBS-T containing 5% milk. After three washes in PBS-T containing 5% milk, the membranes were incubated for 1 hour with either an anti-rabbit [catalog no. NA934; Research Resource Identifier (RRID): AB_772206; GE Healthcare, Mississauga, ON, Canada], anti-mouse (catalog no. NA931; RRID: AB_772210; GE Healthcare), or an anti-goat secondary antibody (catalog no. sc-2020; RRID: AB_631728; Santa Cruz Biotechnology, Santa Cruz, CA) diluted in PBS-T containing 5% milk. After washing three times with PBS-T, proteins were detected using enhanced chemiluminescence (ECL), ECL Prime, or ECL Select Western blotting detection reagent (Amersham Biosciences). Primary antibodies used were anti-Hsp60 (catalog no. sc-1052; RRID: AB_631683; Santa Cruz Biotechnology), antioxidative phosphorylation antibody cocktail (catalog no. MS604; RRID: AB_2629281; MitoSciences, Eugene, OR), anti-HSL (catalog no. 4107; RRID: AB_2296900; Cell Signaling Technology, Beverly, MA), anti-phospho-HSL (Ser563; catalog no. 4139; RRID: AB_2135495; Cell Signaling Technology), and anti-phospho-HSL (Ser660; catalog no. 4126; RRID: AB_490997; Cell Signaling Technology).

RNA isolation, reverse transcription, and quantitative reverse transcription polymerase chain reaction

Total RNA from mouse liver and WAT was isolated using the RNeasy Mini Kit and RNeasy Lipid Tissue Mini Kit, respectively (Qiagen, Toronto, ON, Canada). Complementary DNA was made from 2 μg of RNA by reverse transcription with Random Primer Mix, deoxyribonucleotide triphosphates, 5X ProtoScript II RT Reaction buffer, dithiothreitol, RNAse inhibitor, and ProtoScript II reverse transcription (all from New England Biolabs). Complementary DNA was amplified by quantitative reverse transcription polymerase chain reaction (qRT-PCR) using SYBR Green Master Mix (Roche) and a LightCycler 480 instrument (Roche). Specific primers are listed in Supplemental Table 1.

Statistical analyses

Statistical analyses were performed with Prism 7 software (GraphPad Software, San Diego, CA). Differences between two groups were determined by unpaired Student t test (two-tailed) and P < 0.05 was considered significant. Data in figures are expressed as mean ± standard error of the mean.

Results

Absence of ERRα protects against lipid-induced obesity and NAFLD

To investigate the potential role of ERRα in physiologically induced hepatic steatosis, we first subjected WT and ERRα-null mice to a chronic HFD regimen to foster obesity-related metabolic dysfunction. Mice lacking ERRα gained significantly less weight than their WT littermates did during the HF challenge, an effect previously observed under a shorter and lower lipid-rich HFD regimen (5 vs 15 weeks; 45% vs 60% fat; Fig. 1A) (15). The protective effect from obesity in the absence of ERRα was not attributed to a difference in food consumption (Supplemental Fig 1A). Although no difference in circulating lactate levels were observed between genotypes (Supplemental Fig. 1B and 1C), ERRα knockout (KO) mice chronically fed an HFD had significantly reduced blood glucose levels, an effect more pronounced after a 24-hour fast (Fig 1B and 1C). In line with the prevention of a diabetic-like state, genetic loss of ERRα ameliorated glucose uptake, insulin sensitivity, and responsiveness under an HFD regimen, as determined by glucose and insulin tolerance tests, respectively (Fig. 1D and 1E).

Figure 1.

Mice lacking ERRα are resistant to HFD-induced obesity. (A) Body-weight gain of WT and ERRα-null mice chronically fed either chow or an HFD (n = 16 to 23). Blood glucose levels of (B) fed (n = 7 to 8) and (C) fasted (n = 4 to 5) mice subjected to a chow or HFD. Blood glucose levels during (D) GTT (n = 5 to 6) and (E) ITT (n = 6 to 8) of HFD-fed WT and ERRα KO mice. Data are reported as mean ± standard error of the mean. *P < 0.05 by unpaired t test. AUC, area under the curve; GTT, glucose tolerance test; ITT, insulin tolerance test.

ERRα promotes adipocyte and liver lipid storage during chronic HFD consumption

Next, the effect of a chronic HFD challenge on WAT and liver fat accumulation was examined. In agreement with reduced weight gain under an HFD, ERRα-null mice had diminished white adiposity (Fig. 2A and 2B). More striking, however, was the dramatic protection of ERRα KO mice from HFD-induced NAFLD, as determined by Oil Red O staining of liver sections for neutral lipids (Fig. 2C and 2D; Supplemental Fig. 2). Livers of HFD-fed WT mice were pale, reflecting the substantial deposit of lipids in comparison with livers of ERRα-deficient mice (Supplemental Fig. 2). The observed hepatic steatosis in WT mice under HFD was underscored by the significant increase in liver mass and higher liver-to-WAT mass ratio (Fig. 2E and 2F).

Figure 2.

Absence of ERRα protects from HFD-stimulated adiposity and NAFLD. (A–F) Mice were fed either a chow or HFD for 15 weeks; a cohort of HFD-fed mice was switched to a chow diet for an additional 2 weeks (HFD/chow). Representative sections of (A) hematoxylin-and-eosin–stained gonadal WAT and (B) measurement of adipocyte cell diameter (n = 3) in WT and ERRα-null mice subjected to the indicated diets. Representative images of (C) Oil Red O staining and (D) quantification (n = 7 to 8) of liver cross-sections of mice under different nutritional regimens. (E) Correlation between liver and WAT mass of mice fed a chow, HFD, or HFD/chow diet (n = 7 to 8). (F) Liver-to-WAT ratio of WT and ERRα KO mice following different diets (n = 7 to 8). Data are reported as mean ± standard error of the mean. *P < 0.05 by unpaired t test.

These data show that intrahepatic lipid deposition is a major contributor to HFD-induced body weight gain together with WAT TG storage. Switching mice chronically fed an HFD to standard chow for an additional 2 weeks was sufficient to substantially reduce but not completely rescue the observed increase in adiposity and NAFLD developed in WT mice (Fig. 2A–2E). On the other hand, the switch from HFD to chow had no significant impact in ERRα-null mice (Fig. 2A–2E).

HFD-induced hepatic de novo lipogenesis is ERRα-dependent

In the context of NAFLD, hepatic de novo lipogenesis is often augmented, thus further contributing to the disease. In line with this notion, liver qRT-PCR analysis of key genes encoding proteins involved in this process showed a significant upregulation of two of the four genes tested in WT mice fed an HFD compared with those fed chow (Fig. 3A). In parallel, a boost in Srebp1 expression, the major transcriptional inducer of de novo lipid biosynthesis, was observed in these mice subjected to an HFD (Fig. 3B). Moreover, chronic intake of a lipid-rich diet led to a strong and similar induction in hepatic gene expression and circulating levels of FGF21 in WT mice, and a switch to chow diet for 2 weeks dramatically reversed this effect (Fig. 3C and 3D). These results are consistent with the observation that serum and liver messenger RNA (mRNA) profiles of FGF21 positively correlate with the development of hepatic steatosis (20). In stark contrast to WT mice, loss of ERRα prevented the upregulation of a liver lipogenic gene program as well as the induction in transcript and serum levels of FGF21 mediated by a chronic lipid-rich diet (Fig. 3A–3D). These results are in line with the protection from HFD-induced NAFLD in the ERRα KO mouse model (Fig. 2C).

Figure 3.

Transcriptional upregulation of hepatic lipogenesis in response to an HFD is ERRα dependent. Liver transcription levels of (A) genes involved in de novo lipogenesis as well as (B) the major regulator of lipid biosynthesis, Srebp1, from WT and ERRα-null mice fed a chow, HFD, or HFD/chow diet (n = 4). (C) Hepatic messenger mRNA (n = 4) and (D) serum (n = 6) levels of FGF21, a marker of fatty liver, are shown for WT and ERRα KO mice under the different diets. Data are reported as mean ± standard error of the mean. *P < 0.05 by unpaired t test.

Intriguingly, an HFD challenge provoked a robust, hepatic, global transcriptional repression of de novo lipogenesis in ERRα KO mice that was accompanied by a trend toward lower FGF21 levels (Fig. 3A, 3C, and 3D). Furthermore, despite the approximately threefold higher FGF21 levels in chow-fed ERRα-deficient mice relative to WT (Fig. 3C and 3D), no difference in hepatic lipid content was found between the genotypes (Fig. 2D). These disconnects between FGF21 levels and hepatic lipid content in the ERRα-null model suggest FGF21 production in these mice reflects other metabolic stresses.

ERRα is required for postprandial alleviation of fasting-induced hepatic steatosis

As loss of ERRα proved beneficial for the prevention of obesity but also, importantly, NAFLD in response to chronic HFD consumption, we next questioned the influence of ERRα during cycles of food deprivation and refeeding, which are known to promote hepatic steatosis and its reversal, respectively. To this end, WT and ERRα-null mice were subjected to a 24-hour fast followed by a 2-hour refeeding period whereby each condition provoked an augmentation in the ERRα-encoding gene, Esrra (Fig. 4A and 4B). Similar effects on liver glycogen, blood glucose, and lactate levels among the genotypes were observed (Fig. 4C–4E). Also, fasting elicited a comparable loss in body weight between the mice (Supplemental Fig. 3A). In WT mice, prolonged fasting decreased WAT lipid stores that were partially rescued after an acute 2-hour refeeding period (Fig. 4F and 4G). In contrast, absence of ERRα impaired the replenishment of postprandial adipose lipid stores (Fig. 4F and 4G).

Figure 4.

ERRα is important for postprandial reversal of fasting-induced hepatic steatosis. (A) Schematic of the fasting-refeeding regimen. (B) Liver mRNA expression of the ERRα-encoding gene, Esrra, during nutritional stress (n = 3). (C) Hepatic glycogen levels of WT and ERRα-null mice under fed, fasting, and refed conditions (n = 6 to 8). (D) Blood glucose and (E) lactate levels of mice under different nutritional states (n = 14 to 19). (F) Representative sections of hematoxylin-and-eosin–stained WAT and (G) measurement of adipocyte cell diameter (n = 6 to 10) in fed, fasted, and refed mice. (H) Representative images of Oil Red O staining of WT and ERRα-null mouse liver cross-sections during fasting and refeeding. Quantification of (I) liver TG content (n = 6 to 10), (J) hepatic transcript (n = 3), and (K) serum (n = 6 to 8) levels of FGF21 of mice subjected to food deprivation and refeeding. Data are reported as mean ± standard error of the mean. *P < 0.05 by unpaired t test.

Next, we explored the impact of ERRα function on liver physiology. Hepatic Oil Red O histological examination, electronic microscopy, and biochemical analyses established that loss of ERRα had no considerable effect on fasting-driven hepatic TG accumulation (Fig. 4H and 4I; Supplemental Fig. 3B–3D). However, although a 2-hour refeeding almost completely reversed the fasting-induced NAFLD in WT mice, this response was remarkably blunted in ERRα-null mice and not coupled to a difference in food intake during this period (Fig. 4H and 4I; Supplemental Fig. 3E and 3F). The hepatic lipid content observed in the mice during fasting and refeeding (Fig. 4I) robustly followed the abundance of liver Fgf21 mRNA (Fig. 4J). Although a similar profile in circulating FGF21 levels was determined, a switch from fasting to refeeding led to a marked reduction in serum but not transcript levels of FGF21 in the ERRα-deficient mice (Fig. 4J and 4K). Nevertheless, serum FGF21 levels remained significantly higher in this state compared with their WT counterparts (Fig. 4K). Notwithstanding, the results demonstrate the utility of using FGF21 hepatic mRNA or circulating levels as an index for NAFLD in this physiological context.

ERRα in liver-adipose crosstalk during the fasting and refeeding metabolic switch

To gain insight into the compromised clearance of fasting-induced hepatic steatosis in ERRα-null mice, we next assessed de novo lipogenesis and lipid catabolic gene transcriptional profiles. Generally, refeeding de-repressed the effect of fasting on the expression of key genes involved in lipid biosynthesis, and only Fasn and Scd1 had increased postprandial transcript levels in ERRα-null livers (Fig. 5A). In WT mice, qRT-PCR analysis of genes involved from fatty acid uptake to ketone body production showed that the expression of many genes was induced by fasting and repressed by refeeding to levels comparable to the fed state (Supplemental Fig. 4). On the other hand, expression of several genes, induced by fasting in ERRα-null livers, including Slc25a20, Acadm, and Acat1, remained elevated during refeeding (Supplemental Fig. 4).

Figure 5.

Loss of ERRα simultaneously impairs hepatic mitochondrial activity and adipose tissue lipolysis during nutritional stress. (A) Effect of fasting and refeeding on liver mRNA levels of genes associated with de novo lipogenesis and the master regulator of this process, Srebp1 (n = 3). (B) Circulating levels of the major ketone body secreted by the liver, 3-hydroxybutyrate (3-HB), in WT and ERRα KO mice during nutritional stress (n = 7 to 8). (C) Isolated hepatic mitochondrial respiration rates of WT and ERRα-null mice subjected to cycles of fasting and refeeding (n = 3 to 5). Basal (state 2), adenosine diphosphate–stimulated (state 3), and mitochondrial respiration in the presence of the uncoupling agent, FCCP, are shown. Loss of ERRα impairs the (D) mRNA expression and (E) protein levels of nuclear-encoded OXPHOS subunits regardless of nutritional stress (n = 3). (F) Lipolytic gene expression profiles in WT and ERRα-null WAT during fasting and refeeding (n = 3 to 6). (G) Western blot examination of WAT HSL activity during food deprivation and refeeding as a marker of lipolysis (n = 3). Data are reported as mean ± standard error of the mean. *P < 0.05 by unpaired t test. FCCP, cyanide p-trifluoromethoxyphenylhydrazone; OXPHOS, oxidative phosphorylation.

Measurement of circulating 3-hydroxybutyrate (3-HB), the major ketone body secreted by the liver during starvation, established that although loss of ERRα had no effect on fasting-induced 3-HB production, the ERRα KO mice exhibited slightly more elevated serum 3-HB levels in the refed state (Fig. 5B), alluding to a possible impairment in the use of fat for mitochondrial energy production. To this end, the respiration rates of isolated liver mitochondria from WT and ERRα-null mice during fasting/refeeding were determined. Overall, there was a noticeable decline in liver mitochondrial oxygen consumption rates in the absence of ERRα (Fig. 5C). Accordingly, we observed decreased mRNA and protein levels of several nuclear-encoded mitochondrial oxidative phosphorylation (OXPHOS) subunits in livers of ERRα-null mice compared with WT (Fig. 5D and 5E), all previously identified as direct ERRα targets in mouse liver (19, 21).

Adaptation to nutrient deprivation and availability involves the concerted action and communication between liver and adipose tissues; therefore, we next investigated whether loss of ERRα affected liver TG export and/or adipocyte lipolysis. No difference in the hepatic expression of the major genes involved in the assembly (namely, ApoB, ApoC1, and ApoE) or export (Mttp) of TGs as VLDL was found between the mice (Supplemental Fig. 5A). To assess liver VLDL-TG export, WT and ERRα-null mice were injected intraperitoneally with poloxamer-407, an inhibitor of lipoprotein lipase, to prevent tissues from converting circulating TGs to nonesterified fatty acid for uptake. Genetic loss of ERRα had no impact on the resulting serum TG accumulation, being a direct reflection of the rate of TG export (Supplemental Fig. 5B). Notably, increased postprandial expression of the key hepatic fatty acid import genes, Cd36 and Fabp1, was found in ERRα KO mice compared with WT (Supplemental Fig. 5C), both directly targeted by ERRα in liver (19, 21). This observation supports the hypothesis of continued lipid flux from WAT to the liver during the switch from fasting to refeeding. In support of this notion, qRT-PCR analysis of key lipolytic genes showed reduced levels of the Perilipin 1–encoding gene, Plin1, in WAT of refed ERRα-null mice (Fig. 5F). Perilipin 1, acting at the surface of adipose lipid droplets, protects against lipid hydrolysis by HSL (7). Although no difference in the expression of Lipe, encoding for HSL (Fig. 5F), was observed among genotypes, loss of ERRα had a strong posttranslational effect on HSL, as demonstrated by the increased detection of activating HSL phosphorylation at serine 563 and, to a lesser extent, serine 660 in ERRα-null adipose tissue lysates (Fig. 5G). Collectively, the data suggest that dysregulation of adipocyte TG lipolysis together with lower hepatic mitochondrial OXPHOS capacity are underlying factors responsible for the blunted rescue of fasting-induced hepatic TG deposition during food intake in ERRα-deficient mice.

Discussion

The NR ERRα is well known to play a central role in the control of energy metabolism, making it an attractive therapeutic target for the treatment of metabolic disorders. In this study, we explored the potential role of ERRα in hepatic lipid homeostasis during the physiological response to nutritional stress. Mice with systemic ablation of ERRα were resistant to obesity during a chronic HFD challenge, as previously observed under a shorter and less lipid-rich HFD regimen (15). Herein, ERRα-null mice chronically fed an HFD regimen exhibited enhanced whole-body energy metabolism underpinned by significantly reduced fasting glucose levels as well as enhanced glucose tolerance and insulin sensitivity. Importantly, WT mice, but not ERRα-deficient mice, developed HFD-induced NAFLD with a notable transcriptional induction of de novo lipogenesis, a known contributing factor to NAFLD. Remarkably, the lipogenic program was found repressed by HFD in the ERRα-null model. However, it is not clear why these mice have higher lipogenesis gene-expression profiles under chow vs HFD, because this had no consequence on hepatic lipid accretion. Although our previous assumption for the observed resistance to HFD-induced obesity in ERRα-deficient mice was decreased adiposity (15), our current work establishes that a reduction in hepatic lipid accumulation by nearly twofold is also a leading cause for the lower body-weight gain of ERRα-deficient mice chronically fed an HFD (Fig. 2E).

Challenging mice with cycles of food deprivation and intake revealed an important role for ERRα in the fasting-to-refeeding metabolic switch. During prolonged starvation, adipose tissue TG lipolysis is upregulated, leading to hepatic steatosis, an adaptive process reversed upon food intake. Although loss of ERRα in mice had no remarkable effects on the degree of fasting-induced TG accumulation, ERRα played an important role in relieving the fasting-acquired NAFLD during acute refeeding. Although both fasting and refeeding induced ERRα gene expression, loss of ERRα had no impact on hepatic lipid deposition and 3-HB production after food deprivation. These observations suggest that ERRα activation during starvation likely occurs to modulate other metabolic programs.

Dyslipidemia can arise from a wide range of factors; therefore, we examined several biological processes that could account for the impaired postprandial reversal of NAFLD in the absence of ERRα, including food consumed in the refed period, hepatic mitochondrial activity, fatty acid β-oxidation, lipogenesis, and hepatic TG export, as well as adipose TG hydrolysis. Despite the increased expression of two de novo lipogenesis genes, Fasn and Scd1, found in livers of refed ERRα-null mice, this effect seemed unlikely to be solely responsible for the sustained NAFLD we observed. Although an impaired ability of ERRα-deficient mice to export hepatic lipids was excluded, our data indicate that decreased mitochondrial oxidative potential is an underlying factor. Although loss of ERRα had no impact on the generation of the major ketone body 3-HB, a byproduct of fat oxidation under nutritional stress, decreased mitochondrial respiration rates together with lower expression of OXPHOS components denote a reduced ability of ERRα-null mice to use lipids for energy production. In addition, analysis of adipose tissue physiology and function indicated a compromised ability of adipocytes to turn “off” TG catabolism during the switch from fasting to refeeding in the absence of ERRα. Decreased expression of Plin1, a guardian of lipid droplet integrity and attack by lipases, along with enhanced posttranslational activation of HSL, were found in adipocytes of refed ERRα-null mice, which paralleled the increased hepatic expression of the ERRα target genes encoding the main fatty acid importers, CD36 and Fabp1. These findings suggest that the maladaptation of ERRα-deficient mice during the fasting-to-refeeding transition involves not only an impaired ability to burn fat in hepatocytes but also sustained hepatic import of lipids from adipocytes likely secondary to a failure to adequately suppress adipose TG lipolysis.

Taken together, our work reveals important and diverging roles of ERRα in hepatic lipid homeostasis in the contexts of chronic HFD consumption and fasting and refeeding, as illustrated in Fig. 6. In light of our previous report demonstrating that loss of ERRα function promotes rapamycin-induced NAFLD (19), it is increasingly clear that ERRα plays a fundamental role in lipid homeostasis. Considering that rapamycin treatment mimics a starvation-like state, depresses ERRα stability and function, and leads to hepatic steatosis on one hand, but that absence of ERRα in this study had no consequence on the level of liver lipid accumulation after food deprivation on the other hand, implies that these physiological differences involve other underlying factors.

Figure 6.

Divergent role of ERRα in lipid handling in response to different nutritional challenges. (A) Model depicting ERRα as a promoter of TG accumulation in adipose tissue and liver under an HFD, highlighting that ERRα inhibition would prevent lipid-induced NAFLD. (B) In contrast, the presence of ERRα helps alleviate fasting-induced hepatic steatosis by regulating intertissue lipid mobilization, suggesting that ERRα activation may help reverse acquired NAFLD.

There is a growing body of evidence supporting the endocrine hormone FGF21 as a key regulator of systemic energy metabolism being primarily produced in the liver (22). Increased hepatic expression and circulating levels of FGF21 have been shown to serve as a marker of hepatic steatosis, and this is supported in part by our study, but we have identified marked limitations for this purpose in the ERRα-null model in a nutritional context–dependent manner. First, in the experimental setting of the chow diet and HFD, ERRα KO mice (∼5.5 months old) had significantly elevated FGF21 levels compared with WT mice under the control diet, despite having a similar hepatic lipid content. Moreover, an HFD substantially depressed FGF21 levels in ERRα-null mice relative to chow-fed mice—again, not coupled to liver lipid accretion. Although FGF21 levels proved to be useful as a diagnostic tool for NAFLD in WT mice in this nutritional context, this was not true for ERRα-deficient mice, in which they likely reflect other metabolic states. In line with this notion, FGF21 production has been linked to a wide range of physiological states including amino acid deprivation and oxidative stress (22). In contrary to the chow and HFD settings, FGF21 levels correlated very well with the degree of hepatic steatosis in WT and ERRα-null mice during cycles of fasting and refeeding. Although the switch from fasting to refeeding did not repress liver Fgf21 mRNA levels in ERRα KO mice, reduced serum FGF21 levels were observed and, more importantly, remained significantly higher than the levels found in WT mice. This is in line with the impaired clearance of fasting-induced NAFLD in the absence of ERRα. We speculate that the noted discrepancy in postprandial mRNA and serum FGF21 levels may be due, in part, to increased adipose tissue FGF21 uptake in the ERRα-deficient mice. FGF21 has been thought to play a role in the stimulation of WAT TG lipolysis via regulation of lipolytic enzyme activity during food deprivation, whereas nutrient uptake signals the repression of liver FGF21 gene expression and secretion, in turn, to halt adipocyte lipolysis, allowing the reuptake of lipids from the liver and replenishment of TG stores (7, 23–30). To date, few factors are known to drive the postprandial transcriptional repression of FGF21 (30). In our study, loss of ERRα prevented the transcriptional repression of the mouse Fgf21 gene during the transition from a fasted to fed state. Whether dysregulated Fgf21 transcription is a causal factor in the observed sustained hepatic steatosis in refed ERRα-null mice remains to be further investigated.

In conclusion, loss of ERRα in mice protects against HFD-induced obesity and NAFLD but impairs the reversal of fasting-induced NAFLD during refeeding. These results identify a previously unrecognized role of ERRα in the control of hepatic lipid homeostasis during nutritional stress and underscore a greater therapeutic advantage for inhibiting ERRα action to prevent, rather than treat, hepatic hyperlipidemia and steatosis.

Abbreviations:

Abbreviations:
 
• 3-HB

  3-hydroxybutyrate

 
• ECL

  enhanced chemiluminescence

 
• ERRα

  estrogen-related receptor α

 
• FGF21

  fibroblast growth factor 21

 
• GC-MS

  gas chromatography–mass spectrometry

 
• HFD

  high-fat diet

 
• HSL

  hormone-sensitive lipase

 
• KO

  knockout

 
• mRNA

  messenger RNA

 
• NAFLD

  nonalcoholic fatty liver disease

 
• NR

  nuclear receptor

 
• OXPHOS

  oxidative phosphorylation

 
• PBS

  phosphate-buffered saline

 
• PBS-T

  phosphate-buffered saline plus 0.1% Tween-20

 
• qRT-PCR

  quantitative reverse transcription polymerase chain reaction

 
• RCR

  respiratory control ratio

 
• RRID

  Research Resource Identifier

 
• TG

  triglyceride

 
• VLDL

  very-low–density lipoprotein

 
• WAT

  white adipose tissue

 
• WT

  wild-type

 
• ZT

  Zeitgeber time

Acknowledgments

We thank McGill University members M. Ghahremani for technical assistance, Drs. D. Avizonis and G. Bridon for metabolomics studies, and J. Ouellette for electronic microscopy.

Financial Support: This study was supported by the Canadian Institutes for Health Research (Grants MOP-64275 and MOP-125885 to V.G.). The metabolomics Core Facility is supported by the Canada Foundation for Innovation (project no. 21875), the Dr. John R. and Clara M. Fraser Memorial Trust, and the Terry Fox Foundation Oncometabolism Team Grant (TFRI-1048). W.B. was supported in part by the McGill Integrated Cancer Research Training Program.

Disclosure Summary:

The authors have nothing to disclose.

Author notes

These authors contributed equally to this work.

Current Affiliation: J. St-Pierre’s current affiliation is the Department of Biochemistry, Microbiology and Immunology, Ottawa Institute of Systems Biology, Faculty of Medicine, University of Ottawa, Ottawa K1H 8M5, Ontario, Canada.