The members of the BCL-2 family are crucial regulators of the mitochondrial pathway of apoptosis in normal physiology and disease. Besides their role in cell death, BCL-2 proteins have been implicated in the regulation of mitochondrial oxidative phosphorylation and cellular metabolism. It remains unclear, however, whether these proteins have a physiological role in glucose homeostasis and metabolism in vivo. In this study, we report that fat accumulation in the liver increases c-Jun N-terminal kinase–dependent BCL-2 interacting mediator of cell death (BIM) expression in hepatocytes. To determine the consequences of hepatic BIM deficiency in diet-induced obesity, we generated liver-specific BIM-knockout (BLKO) mice. BLKO mice had lower hepatic lipid content, increased insulin signaling, and improved global glucose metabolism. Consistent with these findings, lipogenic and lipid uptake genes were downregulated and lipid oxidation enhanced in obese BLKO mice. Mechanistically, BIM deficiency improved mitochondrial function and decreased oxidative stress and oxidation of protein tyrosine phosphatases, and ameliorated activation of peroxisome proliferator–activated receptor γ/sterol regulatory element-binding protein 1/CD36 in hepatocytes from high fat–fed mice. Importantly, short-term knockdown of BIM rescued obese mice from insulin resistance, evidenced by reduced fat accumulation and improved insulin sensitivity. Our data indicate that BIM is an important regulator of liver dysfunction in obesity and a novel therapeutic target for restoring hepatocyte function.
Introduction
Obesity is a major risk factor for the development of severe complications such as nonalcoholic fatty liver disease (NAFLD), cardiovascular disease, and diabetes (1). The liver has an essential role in the regulation of glucose homeostasis, producing glucose during fasting periods to prevent hypoglycemia, maintaining brain function and survival. In obese individuals, excessive nutrients are stored in the liver as fat droplets (hepatosteatosis), which can eventually result in inflammation, insulin resistance, and diabetes (2,3). Obesity triggers activation of the c-Jun N-terminal kinase (JNK) pathway in hepatocytes with subsequent insulin resistance as well as steatosis (4,5). The mechanisms of obesity-induced JNK activation include endoplasmic reticulum and oxidative stress, Toll-like receptors, and inflammatory cytokines, such as tumor necrosis factor-α (6). Hepatocyte-specific deletion of both JNK1 and JNK2 improved glucose and insulin tolerance, increased hepatic insulin action, and lowered fasting blood glucose levels in obese mice (7). However, the molecular pathways and proteins mediating JNK-induced hepatocyte dysfunction remain to be fully elucidated.
The BCL-2 proteins are key regulators of apoptotic pathways (8), but growing evidence indicates that these proteins can also play an important role in glucose homeostasis and metabolism (9,10). We have recently reported that loss of the BCL-2 protein PUMA influences circulating leptin levels and food intake in mice (11). Moreover, BCL-2–associated death promoter phosphorylation activates glucokinase and thus controls hepatic gluconeogenesis and insulin secretion in pancreatic β-cells (12,13). In addition, BCL-2 proteins have been reported to stimulate glucose consumption (14) and regulate glucose metabolism through mitochondrial activity (15) or Ca2+ trafficking (16,17). This dual role in metabolism/apoptosis of BCL-2 proteins is reminiscent of the action of cytochrome c, a protein that functions as a crucial component for oxidative phosphorylation in the mitochondria, but when displaced to the cytosol induces cleavage of caspases and cell death (9).
BCL-2 interacting mediator of cell death (BIM) is a BCL-2 homology 3–only protein widely expressed in tissues (18). Saturated free fatty acids trigger cell death in hepatocytes in culture through BIM upregulation (19–21). Fat accumulation and hepatocellular carcinoma in liver-specific STAT5 knockout mice is associated with downregulation of BIM and PUMA (22). In line with these findings, liver apoptosis in BCL-XL– or MCL-1–knockout mice is ameliorated by simultaneous BIM deletion (23). Although it is well accepted that BIM has a role in apoptosis induction in hepatocytes, its role in the regulation of hepatic lipid and glucose metabolism in vivo is unclear. Moreover, it remains unknown whether dysregulation of BIM may contribute to the development of metabolic diseases such as liver steatosis and insulin resistance. In this study, we show that BIM is induced in hepatocytes during obesity. To clarify the role of BIM, we generated a novel liver-specific BIM-deficient mouse model and observed that BIM deletion decreases fat accumulation and improves insulin signaling in vivo. These novel observations highlight BIM as a key mediator of liver dysfunction in obesity.
Research Design and Methods
Human Samples
We studied 19 biopsy specimens of patients undergoing a liver biopsy for medical reasons in our institution. The clinical characteristics of these patients are shown in Supplementary Fig. 1E. The biopsies were collected after approval of the Hôpital Erasme Ethics Committee. Written informed consent was obtained from each participant.
Mice
Mice were maintained at St. Vincent’s Institute (Melbourne, Victoria, Australia) animal care facility on a 12-h light/dark cycle in a temperature-controlled room and obtained food and water ad libitum. Bimlox/lox mice were generated on a C57BL/6 background as previously described (24). Tissue-specific deletion of BIM was generated by crossing Bimlox/lox mice with Alb-Cre (C57BL/6) mice (The Jackson Laboratory, Bar Harbor, ME). Male mice were kept on regular chow (20% protein, 6% fat, and 3.2% crude fiber) or placed at 8–10 weeks of age on a high-fat diet (SF04-027; Specialty Feeds, Perth, Western Australia) for 14–20 weeks. The nutritional composition of the high-fat diet was 18.4% protein, 23.5% fat, and 4.7% crude fiber. The calculated composition of fatty acids in the high-fat diet is: 14.31% total saturated fats, 7.54% total monounsaturated fats, and 2.07% total polyunsaturated fats. In this diet, 46% of total energy is from lipids, 20% of total calculated energy from protein, and the remainder from carbohydrates. All animal studies were conducted at St. Vincent’s Institute following the guidelines of the Institutional Animal Ethics Committee.
Metabolic Analysis
After 14 to 15 weeks on the diet and 24 h of acclimatization, oxygen consumption, energy expenditure, respiratory exchange ratio, activity, and food intake were assessed using a Comprehensive Lab Animal Monitoring System (Columbus Instruments, Columbus, OH). Data were averaged for two dark and light cycles. Body and tissue composition were determined by MRI scanning (EchoMRI, Houston, TX). After 18–20 weeks of chow/high-fat feeding, mice were sacrificed, and tissues were weighed and collected for further analysis. Serum triacylglycerides (Wako Diagnostic, Richmond, VA), insulin (Merck Millipore, Billerica, MA), and β-hydroxybutyrate (Sigma-Aldrich, Darmstadt, Germany) concentrations were determined using commercial kits and following the manufacturers’ instructions.
Mice were fasted for 4 h before performing an intraperitoneal insulin tolerance test. Insulin (Actrapid; Novo Nordisk, Bagsværd, Denmark), at a dose of 0.65 mU/g, was injected intraperitoneally, and blood glucose was measured after tail bleeding. Mice were fasted for 6 h before an intraperitoneal glucose (2 mg/g; Baxter, Deerfield, IL) or pyruvate (2 mg/g; Sigma-Aldrich) tolerance test.
Cell Culture and Treatments
Mouse hepatocytes were isolated by a two-step collagenase A (0.05% weight for volume; Roche Diagnostics) perfusion method as described previously (25). Where indicated, cells were treated with 0.5 mmol sodium palmitate in the presence of 1% weight for volume fatty acid–free BSA (26). The JNK inhibitor SP600125 (Santa Cruz Biotechnology, CA) was used at 10 μmol (27). JNKs (JNK1 and JNK2) were knocked down transiently in primary hepatocytes using JNK-specific small interfering RNAs (siRNAs; #6232; Cell Signaling Technology) as described (25); scramble siRNAs were used as a control.
Adenovirus Production and Viral Transduction
Adenoviruses carrying green fluorescent protein (GFP) as reporter and BIM short hairpin RNA (shRNA) or scrambled control were developed by Vector Biolabs (Malvern, PA). The 21-bp targeted sequence of BIM was GACGAGTTCAACGAAACTTAC. Virus amplification, purification, titration, and verification were done by Vector Biolabs. Adenoviruses with cytomegalovirus promoter driving the expression of Cre recombinase protein fused to the enhanced GFP were developed by the University of Iowa Viral Vector Core Facility (Iowa City, IA). Adenoviral transduction of hepatocyte cultures was carried out 24 h after plating at a viral dose of 2 × 106 plaque-forming units (pfu)/mL. For in vivo treatment, 19-week-old lox/lox or 20-week high fat–fed (HFF) C57BL/6 mice were intravenously injected with a viral dose of 1 × 109 pfu in 200 μL saline.
Real-time PCR, RNA Sequencing, and Data Analysis
Total RNA was isolated using the RNeasy Mini Kit (Qiagen, Venlo, the Netherlands). RNA quality analysis, library preparation, and sequencing were performed by the Australian Genome Research Facility (Melbourne, Australia). The sequencing was performed on a HiSeq2500 (Illumina, San Diego, CA) using v4 chemistry. An average of 21.6 million single-end reads of 100 nucleotides long was obtained per sample. The raw data generated during the sequencing procedure (Supplementary Table 1) were deposited in the Gene Expression Omnibus database (access number GSE102982). The lists of up-/downregulated genes/transcripts and association with canonical pathways were determined using Ingenuity Pathway Analysis (Supplementary Table 2).
Real-time PCR was performed using the Rotor-Gene RG-3000 machine (Corbett Research; Qiagen) and the TaqMan PCR Master Mix (AmpliTaq Gold with GeneAmp kit; Applied Biosystems). Analyses were performed with the delta-delta threshold cycle method using 18s as internal controls. Probe details are provided in Supplementary Table 3.
Western Blot
Muscle, liver, and white adipose tissue (gonadal) were lysed using RIPA buffer, and total proteins were extracted and resolved by SDS-PAGE, transferred onto a nitrocellulose membrane, and immunoblotted with primary antibodies (Supplementary Table 4). The intensity values for the protein bands were corrected by the values of the housekeeping protein β-actin or α-tubulin used as loading controls. Total (reversible and irreversible) protein tyrosine phosphatase (PTP) oxidation was assessed as previously described (25).
Biochemistry Analysis
Primary mouse hepatocytes were isolated from lox/lox and liver-specific BIM-knockout (BLKO) mice and analyzed on a Seahorse XF24 Flux Analyzer (Seahorse Bioscience, North Billerica, MA) according to a previously described protocol (28). Calculations of parameters of mitochondrial respiratory function (28) included subtraction of nonmitochondrial respiration from all mitochondrial respiration parameters. The data were normalized to total protein content per well measured by bicinchoninic acid assay (Thermo Fisher Scientific, Waltham, MA) according to the manufacturer’s instructions.
Reactive Oxygen Species Determinations
Reactive oxygen species (ROS) was measured in lox/lox and BLKO hepatocytes using MitoSox Red (1.2 μmol; Invitrogen) and 2’,7’-dichlorofluorescin diacetate (DCFDA; 0.2 μmol; Abcam), according to the manufacturer’s instructions.
Immunofluorescence and Histology
Hepatocytes and paraffin sections were fixed with paraformaldehyde (4% cells, 2.5% sections) and permeabilized with Triton X-100 (0.3% cells, 0.1% sections). The fixed cells were incubated overnight at 4°C with the following primary antibodies: rabbit anti-BIM (1/200; Cell Signaling Technology) and mouse anticytochrome c (1/200; BD Biosciences, San Jose, CA). The secondary antibody used for visualization was either fluorescein isothiocyanate (FITC)– or rhodamine-conjugated antibodies developed against rabbit or mouse (Jackson ImmunoResearch Laboratories, West Grove, PA), diluted at 1/200. Nuclei were counterstained with Hoechst 33342 (10 μg/mL; Sigma-Aldrich) before mounting.
For histological analysis, tissue was fixed in formalin, and sections (5 µm) were stained according to standard procedures (25). Guinea pig anti-insulin antibody (DakoCytomation, Glostrup, Denmark) was detected with Alexa Fluor 555–conjugated goat anti-guinea pig IgG antibody (Life Technologies, Carlsbad, CA). Mouse antiglucagon antibody (Sigma-Aldrich) was detected with Alexa Fluor 647–conjugated goat anti-mouse IgG antibody (Life Technologies). Nuclei were stained with DAPI reagent (Life Technologies). Images were analyzed on a Nikon A1R-A1 confocal microscope (Nikon, Tokyo, Japan). The percentage of insulin- and glucagon-positive staining per islet was analyzed using ImageJ software (National Institutes of Health).
Hepatic lipid content was analyzed on frozen sections of lox/lox and BLKO livers by Oil Red O staining following a previously published protocol (29). Slides were imaged using an Aperio digital slide scanner (Leica, Wetzlar, Germany).
TUNEL in paraffin sections was determined using an In Situ Cell Death Detection kit, TMR red (Sigma-Aldrich).
Liver paraffin sections were dewaxed, antigen retrieval was performed, and endogenous peroxidases were neutralized with 0.3% hydrogen peroxide. Liver sections were then blocked in 1% BSA and incubated with antinitrotyrosine antibody (Merck Millipore) overnight at 4°C. Secondary antibody, biotinylated anti-rabbit Ig, 1:500 (DakoCytomation), was then added for 10 min, followed by horseradish peroxidase–conjugated streptavidin, diluted 1:500 (DakoCytomation), and incubated for 30 min in 3,3′-diaminobenzidine tetrahydrochloride (Sigma-Aldrich) with hematoxylin counterstain. Images were visualized under light microscopy (at ×20) and quantitated using Image-Pro Plus (Media Cybernetics, Rockville, MD).
Palmitate Oxidation Assay
Measurement of palmitate oxidation capacity of hepatocytes using [14C] palmitate has been described previously (30). The radioactivity of the culture medium as well as the acid-soluble intermediates after chemical extraction of cellular lipids were measured. Total oxidation was then calculated as a function of both [14C] CO2 produced and incomplete oxidation products.
Statistical Analysis
Statistical significance was determined by a two-tailed paired Student t test or ANOVA with Bonferroni correction. The P values <0.05 were considered statistically significant.
Results
Obesity Induces BIM Expression in Hepatocytes Through JNK Activation
As a first step toward elucidating the role of BCL-2 proteins in the regulation of glucose and lipid metabolism in the liver, we determined whether obesity affects the expression of hepatic BCL-2 proteins in age-matched C57BL/6 mice fed either a high-fat diet (23.5% fat; 46% energy from fat) or standard chow diet for 14 weeks. Consistent with the development of obesity, HFF mice have increased white adipose tissue and liver weights and become glucose intolerant and insulin resistant when compared with chow-fed lean controls (Supplementary Fig. 1A–D). To assess the BCL-2 protein profile, livers from HFF obese and chow-fed lean mice were harvested and homogenized for Western blot analysis with BCL-2 antibodies. We noted a striking increase in the expression of BIM and MCL-1 and decrease of one splicing isoform of BAX in the livers from HFF mice (Fig. 1A). The role of MCL-1 in liver dysfunction in obesity has been previously studied (31), and we thus focused on BIM. Hepatocytes constitute ∼80% of the liver volume. Hence, we next determined whether the observed increased BIM expression could be ascribed to changes specifically in these cells. Primary hepatocytes were freshly isolated from chow and HFF mice, and we confirmed increased BIM protein levels in cells from obese mice (Fig. 1B). Importantly, we observed a trend to increased BIM protein expression in liver biopsies from patients with different degrees of NAFLD when compared with nontumorigenic healthy livers (Fig. 1C and Supplementary Fig. 1E).
Saturated free fatty acids induce BIM expression in hepatocytes through JNK/c-Jun activation. A: Liver extracts from chow-fed versus 16-week HFF C57BL/6 male mice were processed and then analyzed by immunoblotting as indicated (n = 6). B: Primary hepatocytes isolated from chow versus 16-week HFF C57BL/6 male mice were processed and then analyzed by immunoblotting as indicated (n = 2 to 3). Quantification of BIM band intensities is indicated at the bottom as a ratio to the loading control β-actin. C: Box plot of median expression of BIM related to the housekeeping protein β-actin in liver biopsies from patients with NAFLD and control subjects with nontumorigenic healthy livers. D: Primary hepatocytes isolated from 8- to 12-week-old C57BL/6 mice were treated with 0.5 mmol palmitate at different time points and processed for immunoblot analysis as indicated (n = 6). E: Liver extracts from chow versus 16-week HFF C57BL/6 male mice were processed and then analyzed by immunoblotting as indicated (n = 6). Primary hepatocytes isolated from 8- to 12-week-old C57BL/6 mice were treated with 0.5 mmol palmitate at different time points and processed for immunoblot analysis (F) or qPCR (G) as indicated (n = 3 to 4). H: Primary hepatocytes were treated with 0.5 mmol palmitate and the JNK inhibitor SP600125 (10 μmol) at different time points and processed for immunoblot analysis (n = 3). I: Primary hepatocytes were transfected with scramble control or JNK-specific siRNAs prior to palmitate treatment and processed for immunoblot analysis as indicated. The result is representative of two independent experiments. Quantification of band intensities is indicated at the bottom as a ratio to the loading control β-actin. *P < 0.05; **P < 0.01; ***P < 0.001 vs. control. HFD, high-fat diet.
Saturated free fatty acids induce BIM expression in hepatocytes through JNK/c-Jun activation. A: Liver extracts from chow-fed versus 16-week HFF C57BL/6 male mice were processed and then analyzed by immunoblotting as indicated (n = 6). B: Primary hepatocytes isolated from chow versus 16-week HFF C57BL/6 male mice were processed and then analyzed by immunoblotting as indicated (n = 2 to 3). Quantification of BIM band intensities is indicated at the bottom as a ratio to the loading control β-actin. C: Box plot of median expression of BIM related to the housekeeping protein β-actin in liver biopsies from patients with NAFLD and control subjects with nontumorigenic healthy livers. D: Primary hepatocytes isolated from 8- to 12-week-old C57BL/6 mice were treated with 0.5 mmol palmitate at different time points and processed for immunoblot analysis as indicated (n = 6). E: Liver extracts from chow versus 16-week HFF C57BL/6 male mice were processed and then analyzed by immunoblotting as indicated (n = 6). Primary hepatocytes isolated from 8- to 12-week-old C57BL/6 mice were treated with 0.5 mmol palmitate at different time points and processed for immunoblot analysis (F) or qPCR (G) as indicated (n = 3 to 4). H: Primary hepatocytes were treated with 0.5 mmol palmitate and the JNK inhibitor SP600125 (10 μmol) at different time points and processed for immunoblot analysis (n = 3). I: Primary hepatocytes were transfected with scramble control or JNK-specific siRNAs prior to palmitate treatment and processed for immunoblot analysis as indicated. The result is representative of two independent experiments. Quantification of band intensities is indicated at the bottom as a ratio to the loading control β-actin. *P < 0.05; **P < 0.01; ***P < 0.001 vs. control. HFD, high-fat diet.
To determine if the modulation of BCL-2 proteins is because of a direct effect of saturated fats in the liver, we isolated hepatocytes and exposed them to the saturated free fatty acid palmitate. BIM, but not MCL-1/BAX spliced, was differentially upregulated in palmitate-treated cells (Fig. 1D and Supplementary Fig. 2A). The unsaturated free fatty acid oleate, which antagonizes the deleterious metabolic effects of palmitate (32), did not affect BIM expression (Supplementary Fig. 2B). Therefore, we hypothesized that saturated fat accumulation in the liver triggers BIM upregulation in the hepatocytes. It has been described that BIM transcriptional expression in hepatocytes is regulated by FoxO3a and JNK/c-Jun (20,21,33), but we did not observe activation of FoxO3a in the livers of HFF mice or in palmitate-treated hepatocytes (Supplementary Fig. 2C and D). In contrast, JNK and its downstream transcription factor c-Jun were phosphorylated/activated in fatty livers (Fig. 1E). Moreover, palmitate-treated hepatocytes induced JNK/c-Jun phosphorylation and transcriptional upregulation of BIM (Fig. 1F and G). Inhibition of JNK by SP600125 or by specific siRNAs prevented palmitate-induced BIM expression (Fig. 1H and I). These results suggest that saturated fat increases BIM expression in a JNK/c-Jun–dependent fashion and raise the possibility that the deleterious effects of obesity can be mediated, at least in part, via differential modulation of BCL-2 proteins.
A New Mouse Model With Hepatocyte-Specific Deletion of BIM: The BLKO Mice
To ascertain whether the activation of hepatic BIM contributes to liver steatosis and obesity, we deleted BIM specifically in hepatocytes. For this purpose, BIMlox/lox (lox/lox) mice were bred with Alb-Cre mice to generate BLKO mice on a C57BL/6 background (Supplementary Fig. 3A). BLKO mice appeared normal and did not exhibit differences in body weight (Supplementary Fig. 3B). However, the liver-specific BIM-deficient mice have slightly less adiposity, as assessed by monitoring body composition by MRI and measuring epididymal fat pad weights (Fig. 2A and B). Total liver weight was similar when comparing BLKO and lox/lox mice, but hepatic fat accumulation tended to be decreased in the BLKO (Fig. 2C). In keeping with their reduced adiposity, BLKO mice exhibited improved insulin and pyruvate tolerance (Fig. 2D and E). The observed reduction of adiposity was accompanied by a trend for increased oxygen consumption and increased energy expenditure (Supplementary Fig. 3C). No differences were observed regarding respiratory exchange ratio, ambulatory activity, or daily food intake (Supplementary Fig. 3C).
BLKO mice were generated in a C57BL/6 background. A: BIMlox/lox (C57BL/6) were mated with Alb-Cre (C57BL/6) mice for the conditional deletion of BIM in hepatocytes. BLKO and control male mice were maintained on a chow diet for 28 weeks and body composition determined with an EchoMRI scan (n = 10 to 11). B: Gonadal white adipose tissue (WAT), liver, and gastrocnemius muscle relative weights were determined after chow feeding (n = 8–10). C: Fat percentage in livers from 28-week-old BLKO and control mice determined with an EchoMRI scan (n = 5). D: Insulin tolerance tests (0.65 mU/g body weight; i.p.) were conducted after 25 weeks of chow feeding (n = 12–14). E: Pyruvate tolerance tests (2 mg sodium pyruvate/g body weight; i.p.) were conducted after 27 weeks of chow feeding (n = 9). *P < 0.05; **P < 0.01 vs. control. AUC, area under the curve.
BLKO mice were generated in a C57BL/6 background. A: BIMlox/lox (C57BL/6) were mated with Alb-Cre (C57BL/6) mice for the conditional deletion of BIM in hepatocytes. BLKO and control male mice were maintained on a chow diet for 28 weeks and body composition determined with an EchoMRI scan (n = 10 to 11). B: Gonadal white adipose tissue (WAT), liver, and gastrocnemius muscle relative weights were determined after chow feeding (n = 8–10). C: Fat percentage in livers from 28-week-old BLKO and control mice determined with an EchoMRI scan (n = 5). D: Insulin tolerance tests (0.65 mU/g body weight; i.p.) were conducted after 25 weeks of chow feeding (n = 12–14). E: Pyruvate tolerance tests (2 mg sodium pyruvate/g body weight; i.p.) were conducted after 27 weeks of chow feeding (n = 9). *P < 0.05; **P < 0.01 vs. control. AUC, area under the curve.
Obese BLKO Mice Have Improved Insulin Sensitivity and Reduced Liver Steatosis
To assess the impact of hepatic BIM deficiency on the development of obesity, 8- to 10-week-old male lox/lox and BLKO mice were fed a high-fat diet for 14 weeks and their body weight and metabolic status assessed (Fig. 3A). HFF BLKO mice gained similar weight as their lox/lox littermates, and there were no differences in global adiposity or fat pad weights (Fig. 3B and C). Interestingly, liver total weight was lower in HFF BLKO mice (Fig. 3C). The reduced liver weight was associated with decreased steatosis, as assessed by gross morphology, histology (hematoxylin and eosin staining), Oil Red O staining (lipid droplets), and measuring the percentage of hepatic fat by MRI (Fig. 3D–F). HFF BLKO mice also exhibited decreased fasting blood triacylglyceride levels (Fig. 3G).
Hepatic deficiency of BIM reduces liver weight and improves insulin sensitivity in obese mice. A: Eight- to 9-week-old male lox/lox and BLKO mice were HFF for 14 weeks. Body weights were measured on a weekly basis, and the incremental increase in body weight was determined (n = 21–23). B and C: Body composition (n = 8–11) and tissue weights (n = 9–19) in 18-week HFF lox/lox and BLKO mice. D and E: The lox/lox and BLKO male mice were fed a high-fat diet (HFD) for 16–18 weeks and livers extracted, fixed in formalin, paraffin embedded, and processed for histology (hematoxylin and eosin). F: Fat percentage in livers from 20-week HFF BLKO and control mice (n = 8 to 9). G: Nineteen-week-old HFF BLKO and control mice were fasted overnight and triacylglycerides (TAG) quantified in serum (n = 15 to 16). Glucose (2 mg/g; n = 9 to 10; H), insulin (0.65 mU/g; n = 8–11; I), and pyruvate (2 mg/g; n = 10–13; J) tolerance tests were conducted in HFF BLKO and control mice and the areas under the curve (AUC) calculated. K: HFF BLKO and control mice were injected with PBS or insulin (0.65 mU insulin/g body weight, 10 min). Livers were extracted and processed for immunoblotting with the indicated antibodies (n = 3–7). Serum insulin levels (n = 8–11; L) and pancreas insulin area per islet (n = 3; M) in HFF BLKO and control mice. *P < 0.05; **P < 0.01; ***P < 0.001 vs. control. WAT, white adipose tissue.
Hepatic deficiency of BIM reduces liver weight and improves insulin sensitivity in obese mice. A: Eight- to 9-week-old male lox/lox and BLKO mice were HFF for 14 weeks. Body weights were measured on a weekly basis, and the incremental increase in body weight was determined (n = 21–23). B and C: Body composition (n = 8–11) and tissue weights (n = 9–19) in 18-week HFF lox/lox and BLKO mice. D and E: The lox/lox and BLKO male mice were fed a high-fat diet (HFD) for 16–18 weeks and livers extracted, fixed in formalin, paraffin embedded, and processed for histology (hematoxylin and eosin). F: Fat percentage in livers from 20-week HFF BLKO and control mice (n = 8 to 9). G: Nineteen-week-old HFF BLKO and control mice were fasted overnight and triacylglycerides (TAG) quantified in serum (n = 15 to 16). Glucose (2 mg/g; n = 9 to 10; H), insulin (0.65 mU/g; n = 8–11; I), and pyruvate (2 mg/g; n = 10–13; J) tolerance tests were conducted in HFF BLKO and control mice and the areas under the curve (AUC) calculated. K: HFF BLKO and control mice were injected with PBS or insulin (0.65 mU insulin/g body weight, 10 min). Livers were extracted and processed for immunoblotting with the indicated antibodies (n = 3–7). Serum insulin levels (n = 8–11; L) and pancreas insulin area per islet (n = 3; M) in HFF BLKO and control mice. *P < 0.05; **P < 0.01; ***P < 0.001 vs. control. WAT, white adipose tissue.
Steatosis is a key contributor to obesity-induced defects in the insulin receptor (IR) signaling cascade (34). We observed that obese BLKO mice respond better to glucose, insulin, and pyruvate tolerance tests as compared with control lox/lox mice (Fig. 3H–J). Insulin-induced IR phosphorylation in the liver was not different between obese BLKO and lox/lox controls, but the interaction between IR substrate 1 and the p85 regulatory subunit of phosphoinositide 3-kinase and the induction of phosphorylated AKT were significantly enhanced in obese BLKO mice (Fig. 3K and Supplementary Fig. 4A). As expected, insulin-induced AKT phosphorylation was not altered in muscle or white adipose tissue (Supplementary Fig. 4B and C), confirming a liver-specific phenotype. Inactivation of the insulin pathway in the liver induces compensatory pancreatic β-cell mass expansion and increased insulin secretion (34,35), suggesting that the increased insulin sensitivity present in the liver of BLKO mice could induce the opposite phenotype. Consistent with this, we observed reduced levels of circulating insulin and insulin staining in islets from HFF BLKO mice (Fig. 3L and M). The improved glucose homeostasis in HFF BLKO mice was accompanied by enhanced energy expenditure and oxygen consumption (Supplementary Fig. 4D). No differences were observed in respiratory exchange ratio, food intake, and ambulatory capacity (Supplementary Fig. 4D). BIM is associated with apoptosis induction in hepatocytes. Thus, we measured apoptosis by TUNEL analysis (Supplementary Fig. 4E), inflammatory markers by quantitative PCR (qPCR) in the liver (Supplementary Fig. 4F), and serum alanine aminotransferase levels (Supplementary Fig. 4G). There were no differences in all these parameters between obese BLKO and lox/lox mice. Taken together, these results demonstrate that BIM deficiency in the liver reduces steatosis and improves insulin sensitivity in obesity, independently of its effects on apoptosis.
BIM Deficiency Decreases Oxidative Stress and Improves Fatty Acid Metabolism in the Liver
In order to gain mechanistic insight on the direct role of BIM activity in hepatocytes, we run a series of in vitro experiments, starting with RNA-sequencing analysis of three preparations of freshly isolated BLKO and lox/lox control hepatocytes (Supplementary Fig. 5A and Supplementary Table 1). BIM deficiency led to major changes in the metabolic gene network of the cells (Supplementary Fig. 5B and Supplementary Table 2). Interestingly, there was modulation of both mitochondrial-related and oxidative phosphorylation genes (Supplementary Fig. 5C). In line with these findings, the maximal respiratory capacity and ATP production were higher in BIM-deficient hepatocytes as determined by mitochondrial oxygen consumption using Seahorse bioanalysis (Supplementary Fig. 5D). Moreover, confocal microscopy demonstrated that BIM colocalizes with cytochrome c at the mitochondria in liver samples from obese C57BL/6 mice in vivo (Fig. 4A) and in primary hepatocytes after palmitate treatment in vitro (Fig. 4B). BCL-2 proteins have been associated to mitochondrial structure and production of ROS (15,36,37). Indeed, decreased mitochondrial function and enhanced ROS were observed in neuroblastoma cells conditionally expressing BIM (38). We found that saturated free fatty acids induced mitochondrial ROS production in lox/lox hepatocytes but not in BIM-deficient cells, as measured by two different assays, namely mitochondrial-specific MitoSOX Red and DCFDA (Fig. 5A and B). The chronic generation of ROS and consequent oxidative stress observed in fatty livers of obese individuals is at least in part the result of deficient mitochondrial function mediated by enhanced uptake and oxidation of energy substrates (39). PTPs are targets for ROS in the liver through oxidation of their cysteine residue in the catalytic site (25,40). Consistent with our in vitro observations, livers from HFF BLKO mice have decreased oxidative stress as indirectly assessed by nitrotyrosine levels, PTP oxidation, and tyrosine-phosphorylated proteins (Fig. 5C and D). STAT1 and JNK activation, both dependent of ROS production (40,41), were also decreased or tended to decrease in obese BLKO livers, respectively (Fig. 5E).
Saturated free fatty acids induce BIM translocation to the mitochondria. A: Confocal microscopy analysis of BIM distribution in frozen livers from chow-fed versus 16-week HFF C57BL/6 male mice. The profile of fluorescence intensity and colocalization was analyzed using quantification software (Nikon). Images are representative of two independent experiments. B: Confocal microscopy of BIM distribution in 0.5 mmol palmitate-treated or control primary hepatocytes (8-h treatment). The profile of fluorescence intensity and colocalization was analyzed using quantification software (Nikon). Images are representative of three independent experiments.
Saturated free fatty acids induce BIM translocation to the mitochondria. A: Confocal microscopy analysis of BIM distribution in frozen livers from chow-fed versus 16-week HFF C57BL/6 male mice. The profile of fluorescence intensity and colocalization was analyzed using quantification software (Nikon). Images are representative of two independent experiments. B: Confocal microscopy of BIM distribution in 0.5 mmol palmitate-treated or control primary hepatocytes (8-h treatment). The profile of fluorescence intensity and colocalization was analyzed using quantification software (Nikon). Images are representative of three independent experiments.
BIM deficiency decreases oxidative stress and increases fatty acid oxidation. A: Western blot showing upregulation of BIM in palmitate-treated hepatocytes and deficiency of the protein in BLKO cells. MitoSox Red staining (1.2 μmol) of BLKO and control hepatocyte cells treated with 0.5 mmol palmitate for 12 h. Quantification shows the corrected total cell fluorescence in the hepatocytes determined by image analysis. Results are expressed as mean ± SEM (n = 4–7). B: The cell-permeant reagent DCFDA (0.2 μmol), which is later oxidized by ROS into 2’,7’-DCF and detected by FACS, has been used to measure cellular ROS formation induced by 12-h treatment with palmitate in BLKO and control hepatocytes (n = 6). C: Livers were harvested from 4-h–fasted mice and processed for nitrotyrosine staining (n = 5). D: Liver extracts from HFF BLKO and control mice were processed in presence of N-ethylmaleimide and dithiothreitol, applied to a gel-filtration column, treated with pervanadate, and then analyzed by immunoblotting with a PTPox and phosphorylated (p-)Tyr antibodies. E and F: Livers were harvested from 4-h–fasted mice and processed for immunoblot analysis as indicated (n = 6–10). G: Livers were harvested from 4-h–fasted mice and processed for qPCR analysis as indicated (n = 7–10). H: Lipid oxidation determined by indirect calorimetry (n = 8). I: β-Hydroxybutyrate levels in serum from fed BLKO and lox/lox obese mice (20 weeks on high-fat diet [HFD]; n = 9). *P < 0.05; **P < 0.01; ***P < 0.001 vs. control. A.U., arbitrary units.
BIM deficiency decreases oxidative stress and increases fatty acid oxidation. A: Western blot showing upregulation of BIM in palmitate-treated hepatocytes and deficiency of the protein in BLKO cells. MitoSox Red staining (1.2 μmol) of BLKO and control hepatocyte cells treated with 0.5 mmol palmitate for 12 h. Quantification shows the corrected total cell fluorescence in the hepatocytes determined by image analysis. Results are expressed as mean ± SEM (n = 4–7). B: The cell-permeant reagent DCFDA (0.2 μmol), which is later oxidized by ROS into 2’,7’-DCF and detected by FACS, has been used to measure cellular ROS formation induced by 12-h treatment with palmitate in BLKO and control hepatocytes (n = 6). C: Livers were harvested from 4-h–fasted mice and processed for nitrotyrosine staining (n = 5). D: Liver extracts from HFF BLKO and control mice were processed in presence of N-ethylmaleimide and dithiothreitol, applied to a gel-filtration column, treated with pervanadate, and then analyzed by immunoblotting with a PTPox and phosphorylated (p-)Tyr antibodies. E and F: Livers were harvested from 4-h–fasted mice and processed for immunoblot analysis as indicated (n = 6–10). G: Livers were harvested from 4-h–fasted mice and processed for qPCR analysis as indicated (n = 7–10). H: Lipid oxidation determined by indirect calorimetry (n = 8). I: β-Hydroxybutyrate levels in serum from fed BLKO and lox/lox obese mice (20 weeks on high-fat diet [HFD]; n = 9). *P < 0.05; **P < 0.01; ***P < 0.001 vs. control. A.U., arbitrary units.
Steatosis is accompanied by ROS-induced expression of the transcription factors peroxisome proliferator–activated receptor (PPAR)-γ and the downstream sterol regulatory element-binding protein (SREBP) 1, which induce the expression of the lipogenic enzymes stearoyl-CoA desaturase (SCD) 1 and acetyl-CoA carboxylase (ACC) 1 and suppress lipid oxidation (42). PPAR-γ also induces the expression of the fatty acid transporter CD36/FAT, which promotes the uptake of free fatty acids in the liver (43,44). Interestingly, PPAR-γ expression is regulated by c-Jun (and hence JNK activity) during steatosis (45). PPAR-γ and SREBP1 were reduced in the livers of HFF BLKO mice (Fig. 5F and G). Consistent with this result, the expression of liver SCD1 and ACC1 was lower in obese BLKO than in control mice, as was the expression of CD36 (Fig. 5G). Moreover, we observed downregulation of CD36 in our RNA-sequencing analysis of BLKO and lox/lox hepatocytes (Supplementary Tables 1 and 2 and Supplementary Fig. 5C) and confirmed this result by qPCR in control and palmitate-treated cells (Supplementary Fig. 5E). In contrast, lipid oxidation, as assessed by indirect calorimetry and serum levels of β-hydroxybutyrate, was increased in HFF BLKO mice (Fig. 5H and I). Taken together, these results indicate that the deletion of BIM in the liver prevents the deleterious effects of high-fat diet–induced obesity and insulin resistance, as well as hepatic lipogenesis, decreased lipid oxidation, and steatosis.
Partial BIM Inactivation Prevents Diet-Induced Insulin Resistance In Vivo
To test if partial deletion of BIM has also a protective effect against the development of liver steatosis and insulin resistance, we generated male mice with heterozygous hepatic BIM deficiency (∼50% BIM reduction) (Fig. 6A) and challenged them with a high-fat diet. Interestingly, even partial reduction of hepatic BIM was sufficient to improve insulin sensitivity in obesity (Fig. 6B and C). One concern regarding our genetically modified models is that we cannot exclude compensatory effects by developmental recombination of BIM, contributing to the observed phenotype. To address this caveat, 19-week-old lox/lox mice were inoculated with 1 × 109 pfu of either adenovirus AdCRE (expressing the CRE recombinase) or AdControl and insulin tolerance tested 4 weeks later. In line with the data obtained in BLKO mice, lox/lox adult mice with AdCRE-mediated BIM deletion showed an improved insulin tolerance (Fig. 6D and E).
Altered insulin signaling upon hepatic reduction of BIM in obesity. A: Immunoblot analysis of BIM expression in liver samples. Glucose (2 mg/g; B) and insulin tolerance (0.65 mU/g; C) tests were conducted in HFF BLKO heterozygous (HET) and control mice (n = 6). D: Nineteen-week-old lox/lox mice were infected with either AdControl or AdCRE (1 × 109 pfu), and 4 weeks later, the liver was extracted, processed, and analyzed by immunoblotting as indicated (n = 4). Body weights: 33.5 ± 1.7 g; no differences between groups. E: Insulin tolerance tests (0.65 mU/g) of lox/lox mice 4 weeks after adenovirus infection as indicated in F (n = 6). F: Expression of GFP (live microscopy fluorescence) and BIM (Western blot) in mouse hepatocytes 48 h after infection with 2 × 106 pfu/mL of AdshRNABIM or AdshRNAControl (n = 2). G: Total palmitate oxidation measured in hepatocytes isolated from mice 48 h after infection with 2 × 106 pfu/mL of AdshRNABIM or AdshRNAControl (n = 3). H: Liver extracts from 20-week HFF C57BL/6 mice were infected with either AdshRNABIM or AdshRNAControl (1 × 109 pfu) and 7 days later processed and analyzed by immunoblotting as indicated (n = 5). I: Livers from 7-day infected mice with AdshRNABIM and AdshRNAControl were extracted, fixed in formalin, paraffin embedded, and processed for histology (hematoxylin and eosin). J: Fat percentage in livers from 20-week HFF C57BL/6 mice infected with either AdshRNABIM or AdshRNAControl (1 × 109 pfu; n = 6 to 7). K: Insulin tolerance tests (0.65 mU/g) of obese mice before and 5 days after adenovirus infection as indicated (n = 7). Body weights: 43.1 ± 1 g; no differences between groups. *P < 0.05; **P < 0.01; ***P < 0.001 vs. control. AUC, area under the curve; HFD, high-fat diet.
Altered insulin signaling upon hepatic reduction of BIM in obesity. A: Immunoblot analysis of BIM expression in liver samples. Glucose (2 mg/g; B) and insulin tolerance (0.65 mU/g; C) tests were conducted in HFF BLKO heterozygous (HET) and control mice (n = 6). D: Nineteen-week-old lox/lox mice were infected with either AdControl or AdCRE (1 × 109 pfu), and 4 weeks later, the liver was extracted, processed, and analyzed by immunoblotting as indicated (n = 4). Body weights: 33.5 ± 1.7 g; no differences between groups. E: Insulin tolerance tests (0.65 mU/g) of lox/lox mice 4 weeks after adenovirus infection as indicated in F (n = 6). F: Expression of GFP (live microscopy fluorescence) and BIM (Western blot) in mouse hepatocytes 48 h after infection with 2 × 106 pfu/mL of AdshRNABIM or AdshRNAControl (n = 2). G: Total palmitate oxidation measured in hepatocytes isolated from mice 48 h after infection with 2 × 106 pfu/mL of AdshRNABIM or AdshRNAControl (n = 3). H: Liver extracts from 20-week HFF C57BL/6 mice were infected with either AdshRNABIM or AdshRNAControl (1 × 109 pfu) and 7 days later processed and analyzed by immunoblotting as indicated (n = 5). I: Livers from 7-day infected mice with AdshRNABIM and AdshRNAControl were extracted, fixed in formalin, paraffin embedded, and processed for histology (hematoxylin and eosin). J: Fat percentage in livers from 20-week HFF C57BL/6 mice infected with either AdshRNABIM or AdshRNAControl (1 × 109 pfu; n = 6 to 7). K: Insulin tolerance tests (0.65 mU/g) of obese mice before and 5 days after adenovirus infection as indicated (n = 7). Body weights: 43.1 ± 1 g; no differences between groups. *P < 0.05; **P < 0.01; ***P < 0.001 vs. control. AUC, area under the curve; HFD, high-fat diet.
Finally, we developed an adenovirus expressing GFP as reporter with a shRNA to selectively knockdown BIM in adult mice (Fig. 6F) and assessed the potential translational relevance of our findings. The adenovirus infected 100% of hepatocytes in culture, with a multiplicity of infection of 2 × 106 pfu/mL, achieving >90% knockdown of BIM (Fig. 6F). Hepatocytes with BIM knockdown had a greater capacity to oxidize palmitate than control cells (Fig. 6G). Next, mice were placed on a high-fat diet for 20 weeks, and insulin resistance was confirmed by insulin tolerance tests with a chow lean group control. Obese mice were divided into two groups, with no differences in insulin resistance. The mice were then inoculated with 1 × 109 pfu of either AdshRNAControl (group 1) or AdshRNABIM (group 2), this treatment resulted in a >80% reduction in BIM expression in the liver of AdshRNABIM-treated mice (Fig. 6H). Strikingly, after 7 days of adenovirus infection, fat accumulation was reduced in BIM knockdown livers as compared with AdshRNAControl (Fig. 4I and J). Moreover, insulin tolerance was significantly improved in HFF mice treated with AdshRNABIM (Fig. 6K). This result demonstrates that BIM is a master regulator of obesity-driven insulin resistance in the liver and has potential as a therapeutic target for novel antidiabetogenic drugs.
Discussion
NAFLD is a common cause of chronic liver dysfunction in obesity, and previous studies suggested an important role for JNK/c-Jun in hepatic steatosis (4–7). Indeed, a strong correlation has been observed between the grade of steatosis and c-Jun phosphorylation/activity in patients with NAFLD (46). Additionally, activated JNK was demonstrated in liver biopsies from patients with nonalcoholic steatohepatitis (47). We found in the current study that JNK/c-Jun activation is responsible for free fatty acid–mediated induction of BIM and observed a trend to BIM upregulation in a small cohort of patients with different degrees of NAFLD. The key findings of our study are summarized in Fig. 7. Thus, the present in vivo and in vitro data demonstrate a novel metabolic role for BIM in the tight regulation of mitochondrial activity in hepatocytes under lipotoxic stress.
Proposed model of the role of BIM in liver steatosis. In addition to its known function as a potent inducer of apoptosis, BIM is involved in regulation of mitochondrial activity and glucose homeostasis in hepatocytes. The present data clarify a novel mechanism by which obesity triggers activation of the BCL-2 family member BIM, resulting in liver damage and insulin resistance. Further, we are able to rescue obese mice from liver dysfunction by short-term modulation of hepatocyte expressed BIM. FFA, free fatty acid.
Proposed model of the role of BIM in liver steatosis. In addition to its known function as a potent inducer of apoptosis, BIM is involved in regulation of mitochondrial activity and glucose homeostasis in hepatocytes. The present data clarify a novel mechanism by which obesity triggers activation of the BCL-2 family member BIM, resulting in liver damage and insulin resistance. Further, we are able to rescue obese mice from liver dysfunction by short-term modulation of hepatocyte expressed BIM. FFA, free fatty acid.
Compelling evidence links oxidative stress to insulin resistance in humans and rodent models (48–50). In obesity, reduced fatty acid oxidation and excess free fatty acids generate reducing equivalents that exceed the respiratory demand, resulting in high mitochondrial O2•− and H2O2 generation (50,51). Endoplasmic reticulum stress and the unfolded protein response in obesity, together with the enhanced expression and activation of NAD(P)H oxidases, contribute further to ROS generation. Our data postulate BIM as an important contributor of saturated fatty acid–induced mitochondrial dysfunction and oxidative stress in fatty livers (Fig. 7). In agreement with this, neuroblastoma cells conditionally expressing BIM have increased accumulation of mitochondrial ROS and reduced mitochondrial respiration (38).
PTPs are enzymes that dephosphorylate tyrosine-phosphorylated proteins to counteract the actions of protein tyrosine kinases (40). The architecture and low thiol pKa of the cysteine residue in the active site of PTPs renders these proteins susceptible to ROS. We recently demonstrated that PTP oxidation occurs in vivo in the liver under obesity, leading to STAT1 activation and steatosis (25). We observed reduced PTP oxidation/STAT1 phosphorylation in obese BLKO mice and decreased expression of the downstream transcription factor PPAR-γ, which promotes hepatic lipogenesis via SREBP1/ACC1/SCD1 induction, and free fatty acid uptake via CD36 (42,52–54). Importantly, liver biopsies from patients with NAFLD show higher levels of PPAR-γ/CD36 compared with samples from control subjects, highlighting the clinical relevance of this pathway (55,56). Thus, the decreased steatosis observed in BLKO mice may result from the lower STAT1/PPAR-γ activity, occurring as a direct consequence of reduced oxidative stress and PTP oxidation, resulting in decreased lipogenesis via SREBP1/SCD1/ACC1 and free fatty acid uptake via CD36 (Fig. 7). In addition, we observed enhanced β-oxidation and serum β-hydroxybutyrate in obese BLKO mice, probably because of enhanced mitochondrial activity in the hepatocytes, which also contributed to reduced steatosis in these mice. Interestingly, our data indicate that the improved hepatic lipid metabolism in BLKO mice is accompanied by enhanced energy expenditure, a link that has also been shown in alternative mouse models of obesity (25,57). Further studies to clarify this cross talk are clearly warranted.
One caveat of our study is that we found >500 genes with metabolic-associated functions differentially expressed in BIM-deficient hepatocytes. Thus, we cannot exclude additional mechanisms contributing to the observed phenotype in these mice that, based on our gene hits, can be assessed in future studies. Nonetheless, our striking results showing a reversal of the damaged liver phenotype in obese mice targeted with BIM knockdown constructs strongly supports the idea that BIM plays a central role in the regulation of cellular metabolism and liver dysfunction in obesity.
In conclusion, we link obesity with liver steatosis and ROS through the identification of a novel JNK/c-Jun/BIM axis. Our data highlight the potential utility of targeted pharmacological approaches toward BIM in combating the associated hepatic complications of obesity.
E.N.G. is currently affiliated with ULB Center for Diabetes Research, Université Libre de Bruxelles, Brussels, Belgium.
Article Information
Acknowledgments. The authors thank Evan Pappas, Christina Tan, Cameron Kos, and Stacey Fynch (St. Vincent’s Institute) for excellent technical assistance; Philippe Bouillet (Walter and Eliza Hall Institute of Medical Research, Parkville, Victoria, Australia) for the Bimlox/lox mice; Jibran Wali, Thomas Kay, and Shiang Lim (St. Vincent’s Institute) for reagents, protocols, and helpful discussions; and Latifa Bakiri (Centro Nacional de Investigaciones Oncológicas, Madrid, Spain); and Shane Grey (Garvan Institute of Medical Research, Darlinghurst, New South Wales, Australia) for critical reading of the manuscript.
Funding. This work was supported by a National Health and Medical Research Council of Australia project grant (APP1071350). E.N.G. is supported by a JDRF fellowship. St. Vincent’s Institute receives support from the Operational Infrastructure Support Scheme of the Government of Victoria.
Duality of Interest. No potential conflicts of interest relevant to this article were reported.
Author Contributions. S.A.L., L.P., S.G., M.I.-E., W.J.S., J.-V.T., K.L., H.E.T., A.S., D.J.G., and J.B.d.H. researched data and result analysis. E.T. and C.M. provided human liver biopsies, contributed to discussion, and reviewed and edited the manuscript. D.L.E. contributed to data analysis and discussion and reviewed and edited the manuscript. E.N.G. researched data, designed experiments, and reviewed, edited, and wrote the manuscript. E.N.G. is the guarantor of this work and, as such, had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.