Obesity-induced metabolic imbalance allosterically modulates CtBP2 to inhibit PPAR-alpha transcriptional activity

CtBP2 forms a repressive transcriptional complex with PPARα

Based on our previous finding that CtBP2 adopts a monomeric configuration upon binding to fatty acyl-CoAs and derepresses SREBP1-mediated fatty acid biosynthesis, we hypothesized that CtBP2 may have a broad influence on fatty acid metabolism. As a first step, we examined a potential interaction with PPARα, a master regulator for fatty acid oxidation. Indeed, we were able to observe an interaction between CtBP2 and PPARα in HEK293 cells (Fig. 1A). Based on this finding, we surveyed the primary amino acid sequences of PPARα from several different species to examine whether they have the putative CtBP-binding site(s), Pro-x-Asp-Leu motif (23). The amino acid sequences of PPARα lacked this CtBP-binding motif, indicating an indirect interaction. While CtBP2 has been widely accepted to be a transcriptional corepressor, CtBP2 can also serve as a transcriptional coactivator in some specific cases (24, 25). Therefore, we next examined how CtBP2 modulates the PPARα transcriptional activity through this interaction. The peroxisome proliferator response element (PPRE)-driven reporter was activated by the ectopic expression of PPARα but was reduced by the expression of CtBP2 (53%), suggesting a repressive role of CtBP2 (Fig. 1B). We also examined the effects of CtBP2 on other PPAR isoforms and found that PPARγ may also be repressed by CtBP2. Since PPARδ did not sufficiently activate our luciferase reporter, we were not able to reach a meaningful conclusion regarding the effect on this isoform (Fig. S1A). In line with these findings, overexpression of CtBP2 in HepG2 cells, a human hepatoma cell line, reduced the expression levels of PPARα target genes at baseline compared to the overexpression of a control protein, glucuronidase (GUS), albeit to a moderate extent (Fig. 1C, 30% and 15% for acyl-CoA oxidase 1 [ACOX1] and PPARA, respectively). Pharmacological activation of PPARα with a synthetic agonist pemafibrate increased the expression levels of these genes that were blunted by CtBP2 overexpression (35% and 30% for ACOX1 and PPARA, respectively). We also examined GW7647, another widely employed PPARα agonist, where the repressive activity of CtBP2 on PPARα was reproducible (Fig. S1B). We further examined whether modulation of this transcriptional system exerts a functional influence on fatty acid oxidation by measuring oxygen consumption rate (OCR). As expected, CtBP2 overexpression suppressed palmitate-induced fatty acid oxidation in HepG2 cells (Fig. 1D).

Preferential binding of monomeric CtBP2 to PPARα

We next investigated the metabolite-dependent monomer-dimer equilibrium of CtBP2. Previous studies have shown that CtBP2 adopts a dimeric conformation when bound with NADH/NAD⁺ (17) that can be decomposed into a monomeric conformation upon binding to acyl-CoAs (15). Firstly, we examined the effects of acyl-CoA–mediated monomerization of CtBP2. Since long-chain fatty acyl-CoAs can be incorporated into PPARα and modulate its activity, we tested malonyl-CoA, which has been reported to have negligible effects on PPARα activity (12) and a suppressive effect on fatty acid oxidation through the inhibition of CPT1 (13). In our preceding study, we showed that CtBP2 adopts the monomeric configuration with long-chain fatty acyl-CoAs as well as acetyl-CoA resulting in dissociation from FoxO1 (15). In agreement with this, addition of malonyl-CoA to cell lysates expressing CtBP2 and FoxO1 decreased CtBP2/FoxO1 complex formation, indicating that the conformational equilibrium was shifted toward monomer by malonyl-CoA (Fig. 2A). In contrast, addition of malonyl-CoA to cell lysates expressing CtBP2 and PPARα promoted the interaction, suggesting that monomeric CtBP2 preferentially binds to PPARα (Fig. 2B). The CtBP2 mutant lacking the Rossmann fold pocket (G189,192A) did not respond to malonyl-CoA supplementation (Fig. 2C), consistent with our hypothesis that malonyl-CoA modulates CtBP2 activity through binding to its Rossmann fold pocket. To further support these findings, we performed in silico structural modeling of these interactions (Fig. 2, D and E; Supporting informations 2 and 3). In this analysis, palmitoyl-CoA was found to bind to CtBP2 with its CoA moiety in the Rossmann fold and the acyl-chain moiety at the dimerization interface as reported previously (Fig. 2D and Supporting information 2). Indeed, the CoA moiety of malonyl-CoA was similarly accommodated in the Rossmann fold while the short acyl-chain protruded to the dimerization interface, structurally resembling acetyl-CoA that was investigated in our preceding study (15) (Fig. 2E and Supporting information 3). Based on our previous report (15) that accommodation of acetyl-CoA modestly shifts the conformational equilibrium of CtBP2 to monomer (15), these structural modeling further support our idea that CtBP2 adopts a monomeric conformation with malonyl-CoA. We further explored the effects of different length of acyl chains on the CtBP2 monomer-dimer equilibrium by taking advantage of the CtBP2–FoxO1 complex that was extensively investigated in our preceding study (15). The eight-carbon fatty acyl-CoA, octanoyl-CoA (C8), was as effective as the long-chain fatty acyl-CoA, oleoyl-CoA (C18) to induce the CtBP2 monomeric configuration, while the effects of two-carbon and three-carbon acyl-CoAs, acetyl-CoA (C2) and malonyl-CoA (C3) were modest (Fig. 2F). Despite the difficulty in the fair assessment of the effects of long-chain fatty acyl-CoAs on the CtBP2–PPARα interaction, long-chain acyl-CoAs may also increase the CtBP2–PPARα complex formation to inhibit the activity of PPARα. In addition, we examined the effect of malonyl-CoA on the interaction between CtBP2 and SREBP1. While we reported long-chain fatty acyl-CoAs suppress the CtBP2–SREBP1 interaction (15), the effect of malonyl-CoA was marginal, which may reflect the indirect nature of this interaction (15) and requirement of long acyl chains for complete monomerization of CtBP2 (Fig. S2).

We further took advantage of CtBP2 mutants that favor the monomeric configuration. It has been shown that mutations of the Rossmann fold (G189,192A) shifts the conformational equilibrium to monomer and that mutations at the dimeric interface (R147,169L) abrogate the dimerization (26). Both mutations robustly increased the CtBP2–PPARα interaction (Fig. 3, A and B), further supporting our proposed model. From a therapeutic point of view, mitigation of malonyl-CoA production may liberate PPARα from the repressive complex. In order to address this possibility, we stimulated HepG2 cells with metformin, an antidiabetic drug that activates AMPK to inhibit ACC, the rate-limiting enzyme of malonyl-CoA synthesis. Indeed, metformin activated this pathway, resulting in dissociation of the CtBP2–PPARα complex, suggesting the therapeutic potential of targeting this transcriptional system (Fig. 3C). Since both AMPK-dependent and AMPK-independent mechanisms have been reported to underlie metformin’s metabolic benefits (27), we further validated this finding with small molecules targeting this pathway. 5-aminoimidazole-4-carboxamide-1-β-d-ribofuranoside (AICAR), the most widely used activator of AMPK, decreased the CtBP2–PPARα complex formation in a dose-dependent manner (Fig. 3D). We also directly inactivated ACC with CP640186, a pharmacological ACC inhibitor, to decrease malonyl-CoA production. This resulted in a dose-dependent decrease of CtBP2–PPARα complex formation (Fig. 3E). Furthermore, 2-deoxyglucose, a competitive inhibitor of glycolysis that also activates AMPK, decreased CtBP2–PPARα complex formation (Fig. S3).

Since CtBP2 has been reported to respond to the ratio of NADH/NAD⁺, we also examined the possible involvement of CtBP2’s pyridine dinucleotide-sensing property (19). The effect of NADH supplementation in cell lysates was relatively marginal (Fig. 4A). We further modulated the NADH/NAD⁺ ratio in live cells by changing the extracellular lactate/pyruvate ratio, based on the fact that lactate dehydrogenase is an equilibrium enzyme coupling the conversion of pyruvate to lactate with NADH to NAD⁺ (28). Again, an increase in the NADH/NAD⁺ ratio induced by an increase of the extracellular lactate/pyruvate ratio had a negligible effect on CtBP2–PPARα complex formation (Fig. 4B). We also tested A201H mutant of CtBP2 that favors the dimeric configuration (15) and found that the A201H CtBP2 was comparable to WT CtBP2 (Fig. 4C). Collectively, NADH/NAD⁺ supplementation had little, if any, effect on the CtBP2–PPARα interaction in these experimental settings.

Lastly, we examined the effects of PPARα activation on the CtBP2–PPARα interaction. The activation of PPARα with PPARα agonist fibrates reduced the CtBP2–PPARα complex formation (Fig. 4D).

The CtBP2–PPARα complex is increased in the liver of obese mice

Having observed these in vitro findings, we investigated the in vivo relevance of this transcriptional complex. As a first attempt, we examined gene expression in the liver-specific CtBP2 KO mice (15). Indeed, genetic deletion of CtBP2 in the liver increased the expression of PPARα target genes (1.2 ∼ 1.4-fold increase), reflecting the liberation of PPARα from CtBP2-mediated repression, although the difference did not reach statistical significance for Cpt1a gene (p = 0.13) (Figs. 5A and S4).

We next examined CtBP2–PPARα complex formation in the livers of multiple animal models of obesity. In the liver of high fat diet-induced obese mice, the protein expression of PPARα was decreased, potentially reflecting reduced PPARα activity. In accordance with our hypothesis, the CtBP2–PPARα interaction was increased in the livers of obese mice (2.8-fold increase based on our densitometric quantification, Fig. 5B). Similarly, in the livers of genetically obese mice, the protein expression levels of PPARα were reduced. Even in the presence of this reduced protein expression, CtBP2 binding to PPARα was maintained. In other words, CtBP2 bound to PPARα on a per molecule basis tended to be increased in mice with genetic obesity (1.6-fold increase based on our densitometric quantification, p = 0.10, Fig. 5C). To further clarify the interplay between CtBP2 and PPARα in the promoters, we performed chromatin immunoprecipitation (ChIP) experiments. Despite the decreased protein expression, the recruitment of PPARα to the promoters of its target genes was increased in the liver of both diet-induced obese and ob/ob mice (Fig. 5, D and E). Importantly, CtBP2 recruitment to those promoters was also increased in obesity (Fig. 5, D and E). These data further support our hypothesis that CtBP2 is recruited to those promoters to repress PPARα in obesity. As demonstrated in our previous finding that CtBP2 adopts a monomeric state in obese liver due to acyl-CoA deposition (15), the findings in this study further indicate that CtBP2 represses the transcriptional activity of PPARα particularly in the liver of obesity (Fig. 6).

Plasmids and cells

Human PPARα and CtBP2 complementary DNAs (cDNAs) were amplified by PCR with N-terminal FLAG-tag and N-terminal hemagglutinin (HA) tag, respectively, and cloned into pcDNA3.1 (+) (Thermo Fisher Scientific, V79020). The following CtBP2 mutants were generated using the Q5 Site-Directed Mutagenesis Kit (New England Biolabs, E0554S): Rossmann fold-defective mutant (Gly189Ala and Gly192Ala), dimerization-defective mutant (Arg147Leu and Arg169Leu) (26), and dimerization-prone mutant (Ala201His) (15).

HEK293 human embryonic kidney cells and HepG2 human hepatoma cells were cultured in Dulbecco's modified Eagle's medium (Gibco, 11965) containing 25 mM glucose, 100 U/ml penicillin, and 100 μg/ml streptomycin sulfate supplemented with 10% fetal bovine serum.

HEK293 cells (2 × 10⁵ cells/ml) were plated into each well of a 6-well plate and cultured for 24 h. The cells were then transiently transfected with a control plasmid, FLAG WT PPARα (50) along with either WT CtBP2 or mutated CtBP2 using lipofectamine LTX (Thermo Fisher Scientific, 15338) for 48 h. To examine the effects of extracellular lactate/pyruvate ratios, cells were cultured with the indicated ratios of lactate/pyruvate for 1 h.

HepG2 cells (1 × 10⁵ cells/ml) were plated into each well of a 24-well plate and cultured for 24 h. The cells were then transduced with Ad-beta-GUS or Ad-human CtBP2-HA (1 × 10⁹ VP/ml) for 48 h. Thereafter, cells were treated with either vehicle or 10 μM pemafibrate (Kowa Co Ltd) for 24 h.

HepG2 cells (2 × 10⁵ cells/ml) were plated into each well of a 12-well plate and cultured for 24 h. The cells were then transiently transfected with a control plasmid, FLAG WT PPARα along with WT CtBP2 using lipofectamine 3000 (Thermo Fisher Scientific, L3000008) for 48 h. To examine the effect of AICAR (1 mM and 3 mM, Wako 015-22531), CP640186 hydrochloride (10 μM and 30 μM, Medchemexpress HY-15259A), and metformin hydrochloride (3 mM, TCI M2009), cells were treated with the indicated concentrations of these reagents for 2 h, 8 h, and 24 h, respectively. Cells were treated with 10 mM 2-deoxyglucose-D-glucose (Nacalai, 10722M) to alternatively activate AMPK. Thereafter, the cell lysates were subjected to co-immunoprecipitation.

Animals

The research protocol was approved by the Animal Care Committee, University of Tsukuba, and all experimental procedures involving animals were conducted according to the guidelines. All mice used were male and maintained on a 14-h light and 10-h dark period cycle. Leptin-deficient ob/ob mice (B6. Cg-Lep ob/J, 10 weeks of age upon euthanasia) were purchased from the Jackson Laboratories (Stock #000632). Mice were fed a high-fat diet (D12492, Research Diets) for 12 weeks starting from 4 weeks of age. Liver-specific CtBP2-deficient mice (LCKO, 8 weeks of age) were generated as described previously (15).

Western blot analysis and co-immunoprecipitation experiments

Proteins were extracted from cells or liver samples with buffer A (50 mM Tris–HCl pH 7.4, 150 mM NaCl, 1% Nonidet P-40, 1 mM EDTA, 10 mM NaF, 2 mM Na₃VO₄) with protease inhibitor cocktail (Sigma-Aldrich, P8340) and subjected to SDS-PAGE. Membranes were incubated with the following antibodies: anti-CtBP2 (BD, 612044), anti-PPARα (Santa Cruz, sc-398394), anti-GAPDH (Santa Cruz, sc-32233), anti-SREBP1 (Novus, NB600-582), anti-FLAG (Sigma-Aldrich, F3165), and anti-HA (Cell Signaling, 3724S).

The membranes were incubated with secondary antibody conjugated with horseradish peroxidase (Cell Signaling, 7074S and 7076S) and were visualized using ChemiDoc XRS Plus System (Bio-Rad). To detect endogenous binding of PPARα and CtBP2, PPAR alpha antibody (GeneTex, GTX101098) or Rabbit IgG isotype control (GeneTex, GTX35035) were cross-linked to Dynabeads Protein G (Invitrogen, 10004D) with 50 mM dimethyl pimelimidate (Sigma-Aldrich, D8388). Liver samples were lysed with buffer A, and the protein complexes were immunoprecipitated in buffer A with a reduced concentration of NP40 (0.5%) (50 mM Tris–HCl pH 7.4, 150 mM NaCl, 0.5% Nonidet P-40, 1 mM EDTA, 10 mM NaF, 2 mM Na₃VO₄) for 2 h at 4 °C. The beads were washed four times with buffer A containing 0.5% NP40, eluted with SDS loading buffer, and analyzed by Western blot analysis.

HEK293 cells were transiently transfected with either a control plasmid or FLAG WT PPARα along with either HA WT CtBP2, HA mutant CtBP2 using lipofectamine LTX (Thermo Fisher Scientific, 15338).

Cells expressing the indicated plasmids were lysed with buffer A containing 1% NP40 and immunoprecipitated with FLAG M2 magnetic beads (MBL, M185-11R) or anti-HA magnetic beads (Thermo Fisher Scientific, 88836) in buffer A with 0.5% NP40 for 2 h at 4 °C. The beads were washed four times with buffer A containing 0.5% NP40 and eluted with 0.5 mg/ml of 3x FLAG peptide (Sigma, F4799) or HA peptide (MBL, 3320-205). To evaluate the effects of malonyl-CoA (Sigma, M4263) and NADH (Sigma, N8129), the cell lysates were immunoprecipitated with FLAG M2 magnetic beads with increasing concentrations of malonyl-CoA or NADH for 4 h or 8 h at 4 °C. Thereafter, the PPARα–CtBP2 complex was eluted and analyzed.

Quantitative real-time PCR

Total RNA was isolated using Sepasol-RNA I Super G (Nacalai, 09379), and cDNA was synthesized with PrimeScript RT Master Mix (Takara Bio, RR036A). Quantitative real-time PCR analysis was performed using SYBR Green in a Thermal Cycler Dice Real-Time System (Takara Bio, RR820A). Data were normalized to peptidylprolyl isomerase A (Cyclophilin A) or ribosomal protein, large, P0 (36B4) expression. The primer sequences were as follows.

PPRE luciferase reporter assay

HEK293 cells (5 × 10⁴ cells/ml) were plated into each well of a 48-well plate and cultured for 24 h. The cells were then cotransfected with 50 ng of a PPRE luciferase reporter plasmid and 5 ng of a pRL-SV40 plasmid encoding Renilla (Promega, E2231) using Lipofectamine LTX with Plus Reagent. For overexpression, cells were cotransfected with 100 ng of a control plasmid or WT CtBP2 along with 100 ng of a control plasmid, WT PPARα, PPARγ, or PPARδ. Cells were incubated for 48 h after the transfection, and luciferase activities were measured in a Synergy HTX Multi-Mode Reader (BioTek), using the Dual-Luciferase Reporter Assay System (Promega, E1960). The PPRE luciferase activities were normalized to Renilla activities.

Structural prediction of CtBP2 with acyl-CoAs by docking simulation

The X-ray structure of CtBP2 dimer (PDB ID: 4LCJ) was downloaded from the Protein Data Bank (PDB). Assignment of bond orders and hydrogenation, hydrogen bond optimization, and energy minimization were performed by Protein Preparation Wizard in Maestro (Schrödinger, LLC) as described previously (15). CtBP2–malonyl-CoA complex structure was created by docking simulation using Glide (51). Prepared 4LCJ A-chains were used for docking simulations. The grid box center coordinates with each side of 20 Å ware set to 17.28, −3.85, 7.8. Positional constraints were set on the adenine ring and phosphorus atom of NADH bound to the A chain to output a docking pose where the adenine ring and phosphate of malonyl-CoA overlap. The CtBP2/malonyl-CoA and CtBP2/palmitoyl-CoA monomeric models were aligned on the A and B chains of the CtBP2 X-ray structure (PDB code: 4LCJ). Energy minimization calculation was performed on the two aligned structures, and energy minimized structures were used as the CtBP2/malonyl-CoA and CtBP2/palmitoyl-CoA dimeric forms.

Chromatin immunoprecipitation

Liver chromatin was obtained from liver tissues as reported previously (15). ChIP assay was carried out using Magna ChIP HiSens Chromatin Immunoprecipitation system (EMD Millipore) with minor modifications (15). Chromatin was immunoprecipitated either with control IgG (Cell Signaling), anti-CtBP2 (Active Motif, 61261), and anti-PPARα (Abcam, ab227074). Immunoprecipitated DNA and input DNA were quantified by real-time PCR with primers specific for Cpt1a or Ppara gene promoters (primer sequences are as follows: Cpt1a forward: 5′- gggtccctgcagtatagcct -3′, Cpt1a reverse: 5′- acccacctgcccttgaac -3′, Ppara forward: 5′- tgcgatctagaccagctcaac 3′, Ppara reverse: 5′- ggccaggactgaagttcaag -3′).

Measurement of oxygen consumption

HepG2 cells (5 × 10⁵ cells/ml) were plated into each well of a 96-well black bottom plate and cultured for 24 h. The cells were then transduced with Ad-GUS or Ad-human CtBP2-HA (1 × 10⁹ VP/ml) for 48 h. OCR was measured according to the instruction manual of the Extracellular OCR Plate Assay Kit (Dojindo E297). The cells were treated with either control bovine serum albumin or bovine serum albumin-conjugated palmitate (200 μM) for 30 min, and the fluorescent signals were measured at 10 min intervals. The calculation of OCR was derived from an analysis of the kinetic profiles obtained from measurements.

Statistical analysis

Statistical differences between two groups were analyzed using Student’s t test. Statistical differences between more than three groups were analyzed using one-way ANOVA followed by Tukey’s multiple comparisons test. The bar graphs with error bars represent means ± SEM. Significance is indicated by asterisks: ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001 and sharp: #p < 0.05, ##p < 0.01, ###p <0.001.