Transcriptional Regulation of Bim by FoxO3A Mediates Hepatocyte Lipoapoptosis^*

Nonalcoholic fatty liver disease is present in up to 30% of the American population (1). A subset of these individuals, ∼5%, develop hepatic inflammation, a syndrome referred to as nonalcoholic steatohepatitis (NASH)² (2). This hepatic inflammatory disorder can progress to cirrhosis, chronic liver disease with portal hypertension, and hepatocellular carcinoma (2). A major risk factor for NASH is insulin resistance occurring within the context of the metabolic syndrome (3, 4). Insulin resistance results in excessive lipolysis within peripheral adipose tissue (5). This lipolysis liberates free fatty acids from neutral fat. The liberated intracellular free fatty acids (FFA) are released into the serum, where they are taken up by the liver (6). Indeed, the bulk of hepatic neutral fat in NASH is derived from re-esterification of circulating FFA (6). Given this information, it is not surprising that elevated serum FFA are a hallmark of NASH (7, 8), and the magnitude of circulating FFA correlates with liver disease severity (7). Therapies that improve NASH by enhancing insulin sensitivity also decrease plasma FFA (9). Thus, FFA probably play an integral role in the development and progression of NASH.

FFA appear to cause liver injury by inducing hepatocyte apoptosis, a phenomenon termed lipoapoptosis (10, 11). Indeed, hepatocyte apoptosis is a cardinal feature of NASH and correlates with hepatic inflammation and fibrosis (8, 12). Consistent with this concept, elevated serum cytokeratin-18 fragments, markers of hepatocyte apoptosis, distinguish simple hepatic steatosis from NASH (13). FFA also induce hepatocyte apoptosis in vitro; in this model, saturated FFA (e.g. stearic or palmitic acid) are more toxic than the unsaturated FFA (e.g. oleic acid). Toxic, saturated FFA induce hepatocyte apoptosis, in part, by the up-regulation of Bim (Bcl-2-interacting mediator of cell death), a proapoptotic member of the Bcl-2 (B-cell lymphoma 2) protein family (10). In fact, FFA-mediated hepatocyte apoptosis is abrogated by siRNA-targeted knockdown of Bim (10). Bim belongs to a subset of the Bcl-2 protein family which displays sequence conservation exclusively in the short Bcl-2 homology 3 (BH3) domain and are referred to as BH3-only proteins. These proteins are biosensors of cellular stress and induce apoptosis by functioning as intracellular death ligands (14). BH3-only proteins such as Bim bind to multidomain members of the Bcl-2 family, causing mitochondria dysfunction, with release of proapoptotic factors into the cytosol (e.g. cytochrome c, SMAC/Diablo, endonuclease G, and AIF) (15). Cytosolic cytochrome c promotes activation of downstream effector caspase-3 and -7, proteases that dismantle the cell, causing cell death by apoptosis (16). Bim exists at the protein level as three isoforms, including Bim extra long (BimEL), Bim long (BimL), and Bim short (BimS) (17). All three isoforms are regulated by both transcriptional and post-transcriptional mechanisms. FFA markedly enhance cellular Bim mRNA levels and therefore probably regulate Bim expression by a transcriptional mechanism (10).

Bim mRNA expression is regulated by a member of the FoxO (Forkhead box-containing protein, class O) family of transcription factors, FoxO3a (18–20). This transcription factor is deactivated by kinase-directed phosphorylation and activated by phosphatase-mediated dephosphorylation (21, 22). FoxO3a factor resides in the nucleus, where it is a positive regulator of gene expression. However, upon phosphorylation, FoxO3a is transported out of the nucleus and sequestered in the cytosol, where it is bound to the protein 14-3-3 (21). Once sequestered in the cytoplasm, it can be reactivated by dephospohorylation via a serine/threonine phosphatase pathway (22).

Both Akt and serum- and glucocorticoid-induced kinase (SGK) phosphorylate FoxO3a at the same sites, albeit with different affinities, Thr³², Ser²⁵³, and Ser³¹⁵ (23, 24). There is cooperativity between these sites such that identification of Thr³² phosphorylation can be used to indicate phosphorylation of the other two sites (24). Reactivation of phosphorylated FoxO3a probably occurs via the ubiquitous threonine/serine phosphatase, protein phosphatase 2A (22). Thus, both cellular localization, nuclear versus cytosolic, and Thr³² phosphorylation (negative regulator) can be used to assess the FoxO3a activation status.

The major objectives of this study were to determine if FFA regulate Bim expression via FoxO3a. The results indicate that toxic FFA preferentially induce Bim expression and cell death by a FoxO3a-dependent mechanism. This activation (dephosphorylation) of FoxO3a is mediated by FFA-stimulated protein phosphatase 2A activity.

EXPERIMENTAL PROCEDURES

Cells—Huh-7 cells, a human hepatoma cell line, and HepG2 cells, a human hepatoblastoma cell line, were cultured in Dulbecco's modified Eagle's medium containing high glucose (25 mm), 100,000 units/liter penicillin, 100 mg/liter streptomycin, and 10% fetal bovine serum. Mouse hepatocytes were isolated from C57BL/6 (Jackson Laboratories, Bar Harbor, ME) by collagenase perfusion, purified by Percoll gradient centrifugation, and plated as primary cultures (25).

Plasmid and Transfection—The HA-FoxO3a expression plasmid was kindly provided by M. C. Hung (26). Transfection with HA-FoxO3a was performed using Lipofectamine (Invitrogen). Stably transfected clones were selected in medium containing 1200 mg/liter G418 and screened by immunoblot analysis.

Fatty Acid Treatment—Stearic, palmitic, and oleic acid were from Sigma. All three FFA were individually dissolved in isopropyl alcohol at a concentration of 40 mm. Dulbecco's modified Eagle's medium containing 1% bovine serum albumin was used in all experiments. The concentration of fatty acids used was 400 μm. The concentration of the vehicle, isopropyl alcohol, was 1% in final incubations. Control cells were treated with vehicle alone. Cells were treated when ∼60% confluent.

Quantitation of Apoptosis—Nuclei were stained with 5 μg/ml 4′,6-diamidine-2′-phenylindole dihydrochloride (DAPI) for 30 min at room temperature and visualized under fluorescence microscopy (Nikon Eclipse TE200, Nikon Corporation, Japan). Apoptotic cells were quantified by counting 400 random cells per study. Cells with the characteristic nuclear changes of chromatin condensation and nuclear fragmentation were considered apoptotic. Apoptosis was expressed as a percentage of total cells counted.

Caspase-3/7 Activity—Cells were plated in 96-well plates (Corning Glass). A caspase assay was performed using the commercially available Apo-ONE homogeneous caspase-3/7 assay (Promega Corp.) according to the manufacturer's instructions. Briefly, this assay involves cleavage of a profluorescent caspase-3/7 consensus substrate, bis-(N-benzyloxycarbonyl-l-aspartyl-l-glutamyl-l-valyl-aspartic acid amide) conjugated to rhodamine 110 (benzyloxycarbonyl-DEVD-R110) on its C-terminal side. Proteolytic cleavage liberates rhodamine 110, unquenching its fluorescence, and was measured using excitation and emission wavelengths of 498 and 521 nm, respectively. Fluorescence intensity was acquired by a microplate fluorescence reader (FLx800; Bio-Tek Instruments Inc., Winooski, VT). The amount of fluorescent product generated is proportional to the amount of caspase-3/7 cleavage activity present in the sample.

Immunocytochemistry—Cells were cultured on collagen-coated coverslips. The medium was aspirated, and the cells were washed three times with PBS and then fixed with freshly prepared 4% paraformaldehyde for 20 min at room temperature. After another washing step with PBS, cells were permeabilized using 0.2% Triton X-100 in PBS for 20 min at room temperature. Next the cells were incubated in PBS containing 10% fetal bovine serum, 0.1 N NaN₃, and 0.1% Triton X-100 at room temperature for 20 min. Primary antibody incubation was carried out with anti-FoxO3a (1:200 dilution, rabbit, 07-702; Upstate Biotechnology, Inc., Lake Placid, NY) for 1 h at room temperature. After another washing step with PBS, the cells were incubated with the secondary antibody, Cy3-conjugated anti-rabbit antibody (1:1000; Jackson Immunological Research Laboratories, Inc.), for 30 min at room temperature. All antibodies were diluted in PBS plus 5% fetal bovine serum. The slides were stained with 0.2 μg/ml DAPI diluted in PBS for 5 min at room temperature and then washed three times in PBS. Next, the coverslips were removed, and Prolong antifade (Molecular Probes, Inc., Eugene, OR) was used as mounting medium. Images were acquired by confocal microscopy with an inverted Zeiss laser-scanning confocal microscope (Zeiss LSM 510; Carl Zeiss Inc., Thornwood, NJ). Individual nuclei were outlined using DAPI fluorescence, and the nuclear fluorescence of Cy3 was quantified using Zeiss KS400 image analysis software (Carl Zeiss, Inc., Oberkochen, Germany). Nuclear FoxO3a fluorescence was expressed as an increase in total nuclear fluorescence intensity per cell (pixels above threshold × fluorescence intensity).

Immunoblot Analysis—Cells were lysed for 30 min on ice with lysis buffer (50 mm/liter Tris-HCl (pH 7.4), 1% Nonidet P-40, 0.25% sodium deoxycholate, 150 mm NaCl, 1 mm EDTA, 1 mm phenylmethylsulfonyl fluoride, 1 μg/mol aprotinin, 1 μg/mol leupeptin, 1 μg/mol pepstatin, 1 m Na₃VO₄, and 1 mm NaF). After centrifugation at 13,000 × g for 15 min, protein concentration in the supernatant was measured using Bradford's reagent (Bio-Rad). The supernatant protein was denatured by boiling for 10 min. Protein (40 μg) was resolved by SDS-PAGE on a 4–15% gradient gel and then transferred onto nitrocellulose membranes. Blocking was carried out using 5% nonfat dry milk in Tris-buffered saline (20 mm Tris, 150 mm NaCl, pH 7.4) with 0.1% Tween 20 for 2 h at room temperature. Primary antibodies (Ab) were diluted in blocking solution and incubated overnight at 4 °C: polyclonal Ab to FoxO3a (7-702) and polyclonal Ab to phopho-FoxO3a (Thr³²) (7-695) (rabbit, 1:1000 dilution; Upstate Biotechnology); monoclonal Ab to Bim (MAB17001) (rat, 1:1000 dilution; Chemicon, Australia); monoclonal Ab to phospho-Akt (Ser⁴⁷³) (4051) (mouse, 1:1000 dilution; Cell Signaling Technology, Danvers, MA); polyclonal Ab to Akt (9272) (rabbit, 1:1000 dilution; Cell Signaling Technology); polyclonal Ab to IKK-β (2684) (rabbit, 1:1000 dilution; Cell Signaling Technology); polyclonal Ab to phospho-IKK-α/β (2694) (rabbit, 1:1000 dilution; Cell Signaling Technology); polyclonal Ab to IkB-α (sc-371) (goat, 1:500 dilution; Santa Cruz Biotechnology, Inc., Santa Cruz, CA); polyclonal Ab to PP2A catalytic subunit α isoform clone 1D6 (05-421) (mouse, 1:5000 dilution; Upstate Biotechnology); monoclonal phospho-PP2A (Tyr³⁰⁷) (1155-1) (rabbit, 1:5000 dilution; Epitomics, Burlingame, CA); polyclonal Ab to PTEN (9552) (rabbit, 1:1000 dilution; Cell Signaling Technology); polyclonal Ab to phospho-PTEN (Ser³⁸⁰/Thr^382/383) (9554) (rabbit, 1:1000 dilution; Cell Signaling Technology, Danvers, MA); polyclonal Ab to SGK (3272) (rabbit, 1:1000 dilution; Cell Signaling Technology, Danvers, MA); polyclonal Ab to phospho-SGK (Ser²⁵⁵/Thr²⁵⁶) (9554) (rabbit, 1:1000 dilution; Upstate Technology); monoclonal Ab to γ-tubulin (T6557) (mouse, 1:5000 dilution; Sigma); polyclonal Ab to actin (sc-1615) (goat, 1:2000 dilution; Santa Cruz Biotechnology). To detect antigen-antibody complexes, peroxidase-conjugated secondary antibodies (1:3000; BIO-SOURCE, Camarillo, CA) were diluted in blocking solution and incubated for 45 min at room temperature. Immune complexes were visualized using a chemiluminescent substrate (ECL; Amersham Biosciences), and Eastman Kodak Co. X-Omat film.

Immunoprecipitation—Immunoprecipitation for HA-FoxO3a was performed with 2 μg of anti-HA (rat monoclonal Ab; Roche Applied Science) or 2 μg of anti-PP2A-C (Upstate Biotechnology) in 1.0 mg of whole lysate protein. Samples were first precleared with protein G-Sepharose (Zymed Laboratories, Inc.) for 30 min. Precleared lysates were centrifuged to discard the beads and incubated with anti-HA or anti-PP2A-C for 4 h and then incubated overnight with protein G-Sepharose. Samples were washed six times with lysis buffer and then subjected to immunoblot analysis as described above.

Akt and Phospho-Akt Immunoreactivity by Enzyme-linked Immunosorbent Assay—Akt and phospho-Akt immunoreactivity were measured using a commercially available fast activated cell-based enzyme-linked immunosorbent assay (Active Motif, Carlsbad, CA). Briefly, cells were plated in 96-well plates. Cells were fixed with 4% formaldehyde followed by quenching of endogenous peroxidase with 1% H₂O₂ and 0.1% sodium azide in PBS containing 0.1% Triton X-100. Each well was incubated with antibody specific to total Akt or phospho-Akt. Horseradish peroxidase-conjugated secondary antibody was subsequently added and developed with the commercial reagent. Absorption was measured at 450 nm using a spectrophotometer. Cell number was quantitated by crystal violet staining. Akt and phospho-Akt immunoreactivity was normalized to cell number.

Reporter Gene Assay—The luciferase reporter construct 3×IRS-Luc, which contains three copies of the FoxO-responsive element from the IGFBP1 promoter, was obtained from K. L. Guan (27). Twenty-four hours after the transfection, cells were incubated with 400 μm FFA, and control cells were incubated with vehicle alone for 6 h. Cells were washed two times with PBS, and then cell lysates were prepared by adding lysis buffer directly to the cells for 20 min at 4 °C on a rocking shacker platform. Next, both firefly and Renilla luciferase activities were quantitated using the dual luciferase reporter assay system (Promega, Madison, WI) according to the manufacturer's instructions. Luciferase was quantified using a luminometer (TD-20/20, Turner Designs, Sunnyvale, CA).

Real Time Polymerase Chain Reaction—Total RNA was extracted from the cells using the Trizol reagent (Invitrogen) and was reverse-transcribed into complementary DNA with Moloney leukemia virus reverse transcriptase and random primers (both from Invitrogen). Quantification of the complementary DNA template was performed with real time PCR (LightCycler; Roche Applied Science) using SYBR green (Molecular Probes) as a fluorophore. PCR primers were as follows: for human BIM, forward (5′-TATGAGAAGATCCTCCCTGC-3′) and reverse (5′-ATATCTGCAGGTTCAGCCTG-3′); for human PP2A-Cα, forward (5′-GAATCCAACGTGCAAGAGGT-3′) and reverse (5′-CGTTCACGGTAACGAACCTT-3′); for human PP1C, forward (5′-GACAGCGAGAAGCTCAACCT-3′) and reverse (5′-CTCCAGCTCCAGAAGAATGG-3′); for murine bim, forward (5′-CGACAGTCTCAGGAGGAACC-3′) and reverse (5′-CCTTCTCCATACCAGACGGA-3′). As an internal control, primers for 18 S ribosomal RNA were purchased from Ambion (Austin, TX). After electrophoresis in 2% agarose gel, each expected base pair PCR product was cut out and eluted into Tris-HCl using a DNA elution kit (gel extraction kit; Qiagen, Valencia, CA). The concentrations of extracted PCR products (copies/μl) were measured using a spectrophotometer at 260 nm and were used to generate standard curves. The inverse linear relationship between copy and cycle numbers was then determined. Each resulting standard curve was then used to calculate the number of copies/μl in experimental samples. The relative expression level of each product was expressed as a ratio of 18 S ribosomal copies of PCR product for each sample.

siRNA-targeted Knockdown of FoxO3A and PP2A—RNA interference was used to knock down FoxO3a and PP2A protein expression in Huh-7 cells. The FoxO3a siRNA sequence corresponds to the coding region 46–64 relative to the first nucleotide of the start codon. FoxO3a sense (5′-ACUCCGGGUCCAGCUCCAC(dTdT)-3′)/FoxO3a antisense (5′-GUGGAGCUGGACCCGGAGU(dTdT)-3′) targeting human FoxO3a was synthesized using the Silencer siRNA construction kit (Ambion). The protein phosphatase 2A, catalytic subunit, α isoform (siRNA-PP2A-C) was purchased from Dharmacon (Lafayette, CO). As a control, cells were also transfected with scrambled RNA duplex with the sequence 5′-AACGTGATTTATGTCACCAGA-3′. FoxO3a and PP2A-C knockdown were performed by transfection of siRNA into cells using siPORT lipid transfection reagent according to the manufacturer's instructions. Successful targeted knockdown was verified by real time PCR after transient transfection of the cells with siRNA. Briefly, cells were grown in 12-well dishes and were transiently transfected with 50 nm siRNA-FoxO3a or 100 nm siRNA-PP2A-C, using 6 μl/ml siPORT lipid (Ambion) in a total transfection volume of 0.5 ml of Opti-MEM (Invitrogen). After incubation at 37 °C, 5% CO₂ for 4 h, 1 ml of normal growth medium was added. Transfected cells were then analyzed for apoptosis, immunoblot, and real time PCR as described elsewhere under “Experimental Procedures.”

Chromatin Immunoprecipitation Assay—Chromatin immunoprecipitation assay was performed using a commercially available assay kit (Upstate Biotechnology) according to the manufacturer's instructions. Briefly, cells were incubated with 1% formaldehyde in media at 37 °C for 20 min to cross-link DNA and DNA-binding proteins. After washing in PBS, cells were collected and lysed within 200 μl of SDS lysis buffer supplied by the kit. DNA within the cell extracts was sheared by sonication with four sets of 10-s pulses using a sonicator (Sonifier cell disruptor; Heat Systems-Ultrasonics, Plainview, NY) equipped with a 2-mm tip and set to 30% of maximum power. The cell lysates were next subjected to 13,000 × g for 10 min. The DNA concentration in the supernatant was quantified by measuring absorbance at 260 nm. Samples containing 200 μg of DNA were diluted with the chromatin immunoprecipitation dilution buffer to a final volume of 2,050 μl. Fifty microliters from each sample were removed as a nonimmunoprecipitation input control. After preclearing the sample with 75 μl of salmon sperm DNA/protein A-agarose slurry at 4 °C for 30 min, the target protein FoxO3a was immunoprecipitated with 2 μg of anti-FoxO3a polyclonal antibody (Upstate Biotechnology) at 4 °C overnight with rotation using a multipurpose rotator (model 151; Scientific Industries, Bohemia, NY). A mock immunoprecipitation without antibody was also performed. Next, 60 μl of the salmon sperm DNA/protein A-agarose slurry was added and incubated at 4 °C for 60 min. After washing the agarose beads, the protein A-agarose immune complexes were eluted, in two separate 250-μl aliquots, in elution buffer (1% SDS, 0.1 mol/liter NaHCO₃) at room temperature for 30 min. Twenty-five microliters of each eluent was used for immunoblot analysis to determine the immunoprecipitation efficiency. Following the addition of 5 m NaCl (20 μl), protein-DNA cross-linking was reversed by heating at 65 °C for 6 h. After digestion with 2 μl of 10 mg/ml proteinase K at 45 °C for 1 h, DNA fragments were purified via phenol/chloroform extraction and ethanol precipitation, and the DNA pellet was dissolved in 20 μl of water. The following PCR primers were employed to amplify the product spanning the FoxO3a-specific binding site identified in the 5′-flanking region of the human BIM gene: forward, 5′-TCGCGAGGACCAACCCAGTC-3′; reverse, 5′-CCGCTCCTACGCCCAATCAC-3′. The copy number for amplified PCR products was quantitated using real time PCR.

Protein Phosphatase 2A Activity—A commercially available PP2A immunoprecipitation phosphatase assay kit (Upstate Biotechnology) was used to measure phosphate release as an index of phosphatase activity. Briefly, Huh-7 cells and HepG2 cells were treated with 400 μm FFA for 6–8 h, and vehicle was used as control. Then total cellular proteins were extracted in a phosphatase extraction buffer (100 mm Tris-HCl, pH 7.6, 100 mm NaCl, 2 mm MgCl₂, 1 mm CaCl₂, 2 mm EGTA, 1% Triton X-100, 1 mm phenylmethylsulfonyl fluoride, and the protease inhibitor tablet c-complete (Roche Applied Science)), sonicated for 10 s, and spun down. To preclear protein extracts, protein A-agarose slurry was added, followed by rotation at 4 °C for 1 h. Samples were centrifuged to discard the beads, and 1000 μg of total cellular protein was incubated with protein A-agarose slurry with anti-PP2A-C (2 μg/ml, catalog number 05-421; Upstate Biotechnology) or anti-PP1-catalytic subunit (PP1C) (2 μg/ml, sc-7482; Santa Cruz Biotechnology) at 4 °C with constant rocking for overnight. Agarose-bound immune complexes were collected and, following washing with 500 μl of Tris-buffered saline (five times) and 500 μl of optimized Ser/Thr buffer (final wash), were resuspended in 50 μl of Ser/Thr buffer. Then 25 μl/sample were incubated with 250 μm of phosphopeptide (amino acid sequence KRpTIRR (where pT represents phosphothreonine), obtained from the kit) or 250 μm phospho-FoxO3a peptide (amino acid sequence SCpTWPL, FoxO3a amino acids 30–35, synthesized by the Mayo Clinic Proteomics Core), in a total volume of 50 μl. The reaction was started by the addition of the phosphopeptide or phospho-FoxO3a peptide and conducted for 10 min at room temperature in a shaking incubator. Supernatants (25 μl) were transferred in a 96-well plate, and released phosphate was measured by adding 100 μl of malachite green phosphate detection solution. Color was developed for 10 min before reading the plate at 650 nm. The absorbance of the reactions was corrected by subtracting the absorbance in samples treated without Ab. Phosphate concentrations were calculated from a standard curve created using serial dilutions of a standard phosphate solution (0–2,000 pmol). Results were expressed as -fold change of PP2A activity as compared with that of vehicle-treated cells.

Reagents—DAPI, stearic acid (S4751), palmitic acid (P0500), oleic acid (O1008), fumonisin B1 (F1147), 1,2-dioleoyl-sn-glycero-3-phosphate (42491), and actinomycin D (A9415) were from Sigma. Okadaic acid (EI181) was from Biomol Research Laboratories (Plymouth Meeting, PA). The phosphatidylinositol 3-kinase inhibitor LY294002 (440202) and the JNK inhibitor SP600125 (420119) were from Calbiochem.

Statistical Analysis—All data represent at least three independent experiments and are expressed as the means ± S.E. of the mean. Differences between groups were compared using Student's t tests and one-way analysis of variance with post hoc Dunnett test, and significance was accepted at p < 0.05.

RESULTS

The Saturated FFA, Palmitic and Stearic Acid, Preferentially Induce Bim Expression—Given that our prior studies were predominantly in HepG2 cells (10), we first established that FFA induce Bim expression in Huh-7 cells by real time PCR and immunoblot analysis (Fig. 1). The saturated FFA, stearic and palmitic acid, at a concentration of 400 μm increased Bim mRNA between 15- and 20-fold after 8 h of incubation (Fig. 1A). In contrast, under identical conditions, the monounsaturated FFA, oleic acid, only increased Bim mRNA expression 3-fold. In the Huh-7 cells, only Bim_EL was readily identified by immunoblot analysis (Fig. 1B). Consistent with the real time PCR data, both stearic and palmitic acid increased cellular Bim_EL protein expression (Fig. 1B), whereas oleic acid had only a minimal effect on cellular Bim_EL protein levels (data not shown). These data in Huh-7 cells confirm our prior observations in HepG2 cells that saturated FFA increase Bim mRNA and protein expression. This paradigm, enhanced Bim expression by saturated FFA, provided us with a model to dissect the signaling pathways responsible for this observation. We selected Huh-7 cells in this model, because they can be transfected with greater efficiency than HepG2 cells. Finally, although we have previously demonstrated that the JNK inhibitor SP600125 reduces FFA-mediated Bax (Bcl-2-associated x protein) activation, it does not prevent Bim protein induction (Fig. 1B); nor does it reduce Bim mRNA (data not shown). Based on this information, we focused our current studies on FoxO3a regulation of Bim expression.

Saturated FFA Dephosphorylate and Activate FoxO3a—Under basal conditions, FoxO3a was phosphorylated at Thr³² as assessed by phosphoimmunoblot analysis; however, FoxO3a was dephosphorylated (activated) after incubation of Huh-7 cells with stearic or palmitic acid (Fig. 2, A and B). FoxO3a dephosphorylation by palmitic acid and stearic acid was time-dependent and observed at concentrations of either 200 or 400 μm. Oleic acid treatment resulted in less FoxO3a dephosphorylation as compared with the saturated FFA (data not shown). More importantly, as assessed by immunocytochemistry, palmitic and stearic acid, but not oleic acid, promoted the translocation of FoxO3a from the cytoplasm to the nucleus (Fig. 2C). To determine whether the nuclear FoxO3a was transcriptionally active, Huh-7 cells were transfected with a luciferase reporter construct containing three copies of a FoxO response element. Following incubation with 400 μm stearic and palmitic acid for 8 h, luciferase activity increased over 6-fold (Fig. 2D). Taken together, these results demonstrate activation of the transcription factor FoxO3a by the toxic, saturated FFA, stearic and palmitic acid.

Bim Induction by FFA Is FoxO3a-dependent—Having demonstrated that toxic, saturated FFA can activate FoxO3a, we next determined if their induction of Bim is FoxO3a-dependent. A chromatin immunoprecipitation assay was performed to confirm that FoxO3a binds to a putative cognate binding sequence within the bim promoter (Fig. 3A). The amount of precipitated FoxO3a-Bim promoter complex was significantly increased after treatment with either stearic or palmitic acid (Fig. 3A). Although these data demonstrate that FoxO3a binds to the Bim promoter in response to FFA exposure and support an effect on Bim gene transcription, they do not exclude an effect of FFA on Bim mRNA half-life. To more directly address this potential mechanism of Bim regulation, the effect of the saturated FFA (palmitic acid) on Bim mRNA half-life was assessed by quantifying Bim mRNA over time in FFA-treated cells incubated in the presence of actinomycin D, a transcription inhibitor. FFA treatment, however, did not increase the half-life of Bim mRNA (Fig. 3B). To more rigorously ascertain whether other potential pathways and transcription factors could account for induction of Bim by FFA, an siRNA approach was undertaken to knock down FoxO3a. The siRNA targeting FoxO3a employed for these studies markedly reduced FoxO3a cellular protein levels (Fig. 4A) and completely abrogated induction of Bim mRNA and protein by stearic and palmitic acid (Fig. 4, A and B). The siRNA knockdown of FoxO3a was specific, since it had no effect on FoxO1 cellular protein levels (Fig. 4A). Accordingly, FoxO3a knockdown also conferred protection against FFA-mediated apoptosis as assessed both by morphology and caspase-3/7 activity (Fig. 5, A and B). These results appeared to be specific for FoxO3a, since FoxO1 protein levels were unchanged using siRNA to FoxO3a (Fig. 4A). Collectively, these data implicate FoxO3a as a key regulator of Bim expression during FFA-mediated lipotoxicity.

FFA Activation of FoxO3a Requires Protein Phosphatase Activity—The phosphorylation status of a protein represents a balance between direct kinase phosphorylation and phosphatase dephosphorylation. Therefore, we explored this balance in FFA-mediated FoxO3a dephosphorylation. Because FoxO3a is phosphorylated and, therefore, inactivated by the serine/threonine kinases phospho-AKT and phospho-SGK (active forms) (24), we first postulated that saturated FFA would inhibit pathways leading to Akt and SGK activation. Unexpectedly, FFA did not significantly alter cellular levels of active phospho-AKT and phospho-SGK by phosphoimmunoblot analysis (Fig. 6) and enzyme-linked immunosorbent assay (data not shown). In particular, FFA treatment did not reduce the active (phospho) levels of these kinases, which would be necessary for FoxO3a activation. Given these results, we next examined the role of phosphatase activity in FFA-mediated FoxO3a dephosphorylation. To ascertain if a protein phosphatase is responsible for the activation (dephosphorylation), cells were incubated with FFA in the presence and absence of okadaic acid, an inhibitor of multiple serine/threonine protein phosphatases (28). Indeed, okadaic acid prevented dephosphorylation of FoxO3a by the toxic saturated FFA, palmitic or stearic acid (Fig. 7A). It also prevented Bim protein and mRNA induction and apoptosis by these otherwise toxic FFA (Fig. 7, A–C). Although okadaic acid can inhibit protein phosphatase type 1 (PP1) and type 2A (PP2A) activity, it is less effective versus PP1 (K_i of 147 nm) than PP2A (K_i of 0.032 nm). However, to exclude an effect of PP1, we next employed the selective PP1 inhibitor (22, 29), 1,2-dioleoyl-sn-glycero-3-phosphate, and examined its ability to reduce saturated FFA-mediated FoxO3a activation, Bim expression, and apoptosis (Fig. 8). However, this selective PP1 inhibitor neither prevented FoxO3a dephosphorylation, Bim induction, nor apoptosis as measured by the caspase-3/7 assay. These data suggest that FFA stimulate FoxO3a, activating dephosphorylation by a protein phosphatase, probably PP2A.

Saturated FFA Stimulate Protein Phosphatase 2A Activity—Because protein phosphatase 2A (PP2A) has been implicated in FoxO3a regulation (22), we next examined the role of PP2A in FFA-stimulated FoxO3a activation, Bim mRNA expression, and cell death. First, we explored if FFA affect the activity of PP2A. Indeed, PP2A activity was 3-fold greater in stearic and palmitic acid-treated cells as compared with basal values and was 2-fold greater than in oleic acid-treated cells (Fig. 9A). Not only did FFA enhance PP2A activity as assessed by a consensus phosphopeptide substrate, but they also increased its activity as quantified by a FoxO3a peptide substrate containing phospho-Thr³² (Fig. 9A). In contrast, neither palmitic nor stearic acid affected PP1 phosphatase activity in this assay (data not shown). The increase in PP2A activity occurred in the absence of an increase in its cellular protein (Fig. 9B) and mRNA levels (data not shown). Because PP2A is activated by dephosphorylation of Tyr³⁰⁷, we next examined the phosphorylation status of PP2A by phosphoimmunoblot analysis. In fact, FFA-mediated PP2A activity was associated with a reduction in Tyr³⁰⁷ phosphorylation (Fig. 9C). Next, a siRNA approach was employed to selectively knock down this protein phosphatase (Fig. 10, A–E). PP2A is composed of a family of phosphatases (30); the active enzyme consists of a heterotrimer consisting of A (scaffolding) and B (regulatory) subunits and a C (catalytic) subunit. We employed an siRNA targeting the catalytic or Cα subunit for these studies. The PP2A-Cα siRNA reduced the mRNA for its target, but not that for the PP1 catalytic subunit (PPIC) (Fig. 10A). It also and effectively abrogated increases in PP2A activity by FFA (Fig. 10A). Targeted siRNA knockdown of the catalytic (PP2A-Cα) subunit of PP2A also reduced stearic or palmitic acid-mediated Thr³² FoxO3a dephosphorylation, Bim mRNA induction, and apoptosis (Fig. 10, B and C). Taken together, these data suggest that FFA promote FoxO3a activation by stimulating protein phosphatase 2A activity.

To elucidate the mechanism by which PP2A can regulate the activity (phosphorylation) of FoxO3a, we examined if FoxO3a is physically associated with PP2A. To confirm this postulate, we transfected Huh-7 cells with HA-FoxO3a and examined its association with PP2A by immunoprecipitation studies. Indeed, immunoprecipitation studies revealed an FFA-dependent (palmitic acid) association between FoxO3a and PP2A (Fig. 11). These observations suggest that FFA stimulate PP2A activity, which in turn dephosphorylates/activates FoxO3a.

FFA-mediated FoxO3a Dephosphorylation and Bim Expression Is Not Mediated by the de novo Ceramide Synthesis Pathway—Recently, it has been suggested that de novo ceramide synthesis may mediate the effects of saturated FFA on PP2A activity (30). To ascertain if de novo ceramide synthesis was responsible for FoxO3a dephosphorylation and Bim expression in our model, we employed the ceramide synthesis inhibitor fumonisin B1 (30, 31). However, this inhibitor prevented neither FoxO3a dephosphorylation nor the increase in Bim expressed following treatment of Huh-7 cells with palmitic acid (Fig. 12, A and B). It also did not prevent palmitic acid-mediated apoptosis as assessed by caspase-3/7 activity (Fig. 12C). Based on these studies, we cannot implicate de novo ceramide synthesis for saturated FFA-mediated FoxO3a activation, Bim induction, or apoptosis.

FFA Activates FoxO3a by Dephosphorylation in HepG2 Cells and Murine Hepatocytes—Finally, in order to establish that our observations were not unique to Huh-7 cells, we extended our study to include HepG2 cells and primary cultures of murine hepatocytes. HepG2 cells were treated with FFA (400 μm) for 8 h, and Bim mRNA and FoxO3a activation was evaluated by immunoblot analysis. Indeed, FFA increase Bim mRNA and activate (dephosphorylate) FoxO3a (Fig. 13, A and B). Next, to investigate the role of protein phosphatases on FoxO3a activation, okadaic acid was employed. In agreement with our previous results, FFA-stimulated FoxO3a dephosphorylation was inhibited by okadaic acid in HepG2 cells (Fig 13B). Furthermore, saturated FFA stimulate PP2A enzymatic activity (Fig. 13C), and a reduction in PP2A-C Tyr³⁰⁷ phosphorylation (Fig 13D). FFA-mediated lipoapoptosis was also decreased by okadaic acid (Fig. 13E). FFA-induced Bim mRNA expression, dephosphorylation/activation of FoxO3a, and okadaic acid inhibition of apoptosis were also observed in primary murine hepatocytes (Fig. 14, A–C). Thus, a similar pathway of FFA-mediated FoxO3a activation was present in HepG2 and primary murine hepatocytes as observed in HuH-7 cells.

DISCUSSION

The principal findings of this study pertain to the mechanisms of FFA-mediated lipoapoptosis. The results indicate that the toxic, saturated free fatty acids, stearic and palmitic acid, but not the nontoxic monounsaturated FFA oleic acid (i) induce expression of Bim by a FoxO3a-dependent mechanism, ii) promote FoxO3a dephosphorylation (activation) by stimulating protein phosphatase 2A activity, and (iii) mediate apoptosis in part by a protein phosphatase 2A/FoxO3a/Bim pathway. Modulation of this pathway is a potential therapeutic strategy for ameliorating FFA-induced hepatocyte lipoapoptosis.

In this study, we used nonesterified fatty acids or FFA at concentrations of 400 μm incubated in the presence of albumin. The presence of albumin in our medium assures a physiologic balance between bound and free nonesterified fatty acids in the media analogous to the ratio present in plasma (32). Moreover, the fasting plasma concentrations of total FFA in human nonalcoholic steatohepatitis are in the 700 μm range (9). Although the exact species of circulating FFA in humans with the metabolic syndrome has not been well studied, ∼35% of the total serum FFA in humans are saturated (33–35). Thus, our in vitro cell system examining the toxicity of FFA models the circulating concentrations of these lipids in humans during metabolic disturbances.

Bim is a BH3-only protein member of the Bcl-2 family and has been implicated in mediating cell death by a wide variety of toxic stimuli, including chemotherapy, disturbances in the cellular cytoskeleton, death receptors, cytokine withdrawal, and other noxious stress stimuli (14). Although we previously implicated Bim in FFA-mediated lipoapoptosis (10), the mechanism(s) by which FFA activate Bim were not elucidated in that study. Bim is regulated by both transcriptional and post-translational mechanisms. At the post-translational level, Bim can be activated by c-Jun-N-terminal (JNK) kinase phosphorylation (36) or inhibited by extracellular receptor kinase phosphorylation, which targets the protein for proteosomal degradation (37). Bim can also be transcriptionally induced, especially by members of the FoxO family of transcription factors (18, 20, 38). The results herein suggest that toxic, saturated FFA regulate Bim by a transcriptional mechanism. Activation (dephosphorylation) of the transcription factor FoxO3a, its nuclear translocation, and binding to the Bim promoter were all observed following treatment of Huh-7 cells with FFA. Furthermore, siRNA-targeted knockdown of FoxO3a abrogated the increase in Bim and cell death by the saturated FFA, stearic and palmitic acid. Taken together, these data indicate that FFA regulate Bim predominantly via a transcriptional mechanism, although the possibility of concomitant post-translational activation by JNK cannot be excluded, especially given the known activation of JNK by FFA (10).

Enhanced activity of FoxO3a has been implicated in the expression of additional apoptosis effectors, in addition to Bim. For example, FoxO3a can induce the expression of the BH3-only protein Puma (p53-up-regulated modulator of apoptosis) (39), and the death ligands Fas-L (23) and Trail (40). However, in our prior studies, siRNA-targeted knockdown of Bim was sufficient to attenuate FFA-induced hepatocyte apoptosis, whereas inhibition of the death receptor pathway by enforced expression of dominant negative FADD did not prevent cell death (10). Therefore, at least in vitro, expression of these other apoptosis effectors does not appear to contribute to hepatocyte cell death by FFA.

FoxO transcription factors are generally regulated at the post-transcriptional level by phosphorylation and acetylation (41). In regard to FoxO3a, there are three conserved phosphorylation sites, T1 (Thr³²), S1 (Ser²⁵³), and S2 (Ser³¹⁵), that are recognized by Akt and SGK with different affinities (23, 24). However, it is unlikely that a reduction in kinase activity resulted in FoxO3a dephosphorylation and activation. We failed to observe a reduction in the phosphorylated or active form of Akt and SGK in FFA-treated cells (Fig. 6). Therefore, we focused on FoxO3a dephosphorylation by PP2A. Our observations demonstrate stimulation of PP2A activity following treatment of cells by FFA; conversely, siRNA-targeted knockdown or pharmacologic inhibition of PP2A blocked FoxO3a activation. These data suggest that FFA may stimulates PP2A. This observation has significant implications for liver pathophysiology in NASH given the large number of proteins affected by protein phosphatase activity.

It is beyond the scope of the current study to ascertain how FFA effect changes in PP2A activity. However, considerable information is available to help guide such investigations. PP2A is a heterotrimer complex composed of a catalytic subunit (PP2A-C), which is associated with a scaffolding A subunit (PP2A-A) and a regulatory subunit (PP2A-B). Only two isoforms of the A (Aα and Aβ) and C (Cα and Cβ) subunits have been identified (42), although there are four families of B subunits and each subfamily has multiple isoforms (42). The B subunit provides stability to the heterotrimer, influencing substrate specificity and subcellular localization. Protein phosphatase 2A can be regulated by phosphorylation of the catalytic subunit. The catalytic subunit of PP2A can be phosphorylated by the tyrosine kinases pp60^v-src and pp56^lck, the epidermal growth factor, and insulin receptors (43, 44). The phosphorylation occurs on Tyr³⁰⁷, which is located in the C terminus of PP2A-C, and results in inactivation of the enzyme. Tyrosine phosphorylation of PP2A-C is enhanced in the presence of the phosphatase inhibitor okadaic acid, suggesting that, under different conditions, PP2A-C can rapidly reactivate itself in an autodephosphorylation response. Indeed, our results demonstrate that PP2A-C decreases its phosphorylation at Tyr³⁰⁷ under FFA treatment, and thereby increases its catalytic activity.

Protein phosphatase 2A is expressed in all cell types and is involved in a wide range of cellular processes, including cell cycle regulation, cell morphology, development, signal transduction, translation, and apoptosis. Recent studies have suggested that PP2A has a critical role in the regulation of apoptosis (22, 45–51). The activity of PP2A can regulate the responsiveness to death receptor-mediated cell death by Fas (45), tumor necrosis factor-α (46), and TRAIL (47) via regulation of mitogen-activated protein kinases, such as p38, extracellular signal-regulated kinase, and JNK. Furthermore, PP2A has been implicated in regulation of the intrinsic pathway of cell death by dephosphorylation and inhibition of Bcl-2 (49) or by dephosphorylation and activation of Bad (50) and Bax (51). PP2A has also been proposed to induce Bim expression by modulating Akt activation (22); however, in our current studies, neither Akt nor SGK was affected by FFA treatment. Our studies extend these observations by suggesting that PP2A also drives dephosphorylation (activation) of FoxO3a, thereby promoting apoptosis.

One of the findings of our study is that saturated FFA were more cytotoxic that monounsaturated FFA. These data are in accord with our prior observation (10) and a wide variety of different cell types, including pancreatic β-cell (31, 52), astrocytes (53), neurons (54), granulose (55) and Leydig cells (56), breast cancer cells (57), endothelial cells (58, 59), and cardiac myocytes (60). The mechanism by which FFA are more toxic than unsaturated FFA have not been identified. Recently, saturated FFA, but not unsaturated FFA, have been proposed as ligands of the Toll-like receptor 4 (61, 62). Perhaps, like insulin resistance, FFA-mediated cytotoxicity requires engagement of this membrane receptor.

Our cellular model provides additional insight into the potential mechanisms by which FFA induce hepatocyte lipapoptosis. Apoptosis is regulated by members of the Bcl-2 family of proteins. Antiapoptotic, multidomain members of this family (i.e. Bcl-2, Bcl-X_L, Mcl-1 (myeloid cell leukemia 1), and Bcl-w) prevent apoptosis and function as cell guardians. In contrast, proapoptotic members of this family (i.e. Bak (Bcl-2 antagonist/killer) and Bax) serve as executioners in the apoptotic program. BH3-only proteins, such as Bim, are the biosensors of cellular stress and engage the apoptosis program by either depressing antiapoptotic or directly ligating and activating proapoptotic multidomain Bcl-2 proteins (14). Our previously published data suggested that both JNK and Bim were necessary for Bax activation and cell death by toxic FFA (10). Perhaps JNK through a kinase cascade promotes Bax dissociation from cytosolic binding partners and/or its translocation to mitochondria (63). Once Bax has translocated to mitochondria, Bim may in turn directly bind and activate Bax or indirectly activate this proapoptotic protein by derepressing antiapoptotic Bcl-2 proteins (64). Once fully activated, Bax induces mitochondrial dysfunction, release of cytochrome c, activation of effector caspases, and cell death (65). Thus, FFA stimulation of JNK plus FoxoO3a activity appears to work in concert to induce cell death.

Hepatocyte lipoapoptosis is a cardinal feature of NASH, a growing public health problem (3, 12). Our current data provide insight into the mechanisms of FFA-induced lipoapoptosis, namely stimulation of FoxO3a-dependent Bim expression. Modulation of this pathway is a potential therapeutic strategy to ameliorate NASH.