Inhibition of G-protein-coupled Receptor Kinase 2 Prevents the Dysfunctional Cardiac Substrate Metabolism in Fatty Acid Synthase Transgenic Mice^,

Abstract

Impairment of myocardial fatty acid substrate metabolism is characteristic of late-stage heart failure and has limited treatment options. Here, we investigated whether inhibition of G-protein-coupled receptor kinase 2 (GRK2) could counteract the disturbed substrate metabolism of late-stage heart failure. The heart failure-like substrate metabolism was reproduced in a novel transgenic model of myocardium-specific expression of fatty acid synthase (FASN), the major palmitate-synthesizing enzyme. The increased fatty acid utilization of FASN transgenic neonatal cardiomyocytes rapidly switched to a heart failure phenotype in an adult-like lipogenic milieu. Similarly, adult FASN transgenic mice developed signs of heart failure. The development of disturbed substrate utilization of FASN transgenic cardiomyocytes and signs of heart failure were retarded by the transgenic expression of GRKInh, a peptide inhibitor of GRK2. Cardioprotective GRK2 inhibition required an intact ERK axis, which blunted the induction of cardiotoxic transcripts, in part by enhanced serine 273 phosphorylation of Pparg (peroxisome proliferator-activated receptor γ). Conversely, the dual-specific GRK2 and ERK cascade inhibitor, RKIP (Raf kinase inhibitor protein), triggered dysfunctional cardiomyocyte energetics and the expression of heart failure-promoting Pparg-regulated genes. Thus, GRK2 inhibition is a novel approach that targets the dysfunctional substrate metabolism of the failing heart.

Introduction

Heart failure is a debilitating syndrome that involves insufficient cardiac performance. Multiple pathomechanisms have been elucidated, but treatment options remain insufficient, and hence the mortality of heart failure is high (1). The causes of heart failure are complex with ischemic heart disease being the most frequently associated condition (2). Co-existing disorders such as diabetes, hypertension, and obesity further deteriorate symptoms (3). Despite having a different etiology, late-stage heart failure is commonly characterized by severe changes in myocardial substrate metabolism, with a switch from fatty acid oxidation toward predominant glycolysis (4–6). Conflicting evidence exists as to whether this substrate switch is beneficial or detrimental (7), but several previous studies have indicated that an increased availability of lipid substrates that counteract the substrate switch could improve cardiac function (7, 8). Moreover, treatment options, which improve substrate availability, are attractive because the failing heart is often considered to be “an engine running out of fuel” (9).

Following this concept, we aimed to investigate the impact of improved cardiac substrate availability by generating transgenic mice with myocardium-specific expression of fatty acid synthase (FASN), the major palmitate-synthesizing enzyme. Such an approach is also supported by data obtained for myocardium-specific Fasn deficiency, which have revealed the cardioprotective potential of Fasn (10). Moreover, hearts from patients with heart failure showed an increased expression and protein level of FASN² (10, 11). By generation of transgenic mice, we found that FASN transgenic mice developed a heart failure-like phenotype with impaired cardiomyocyte substrate use. In search for a treatment approach for the disturbed cardiac substrate metabolism, we focused on the role of GRK2 inhibition because GRK2 inhibition could counteract cardiolipotoxicity by promotion of a cardiomyocyte survival program (12, 13) and resensitization of the cardioprotective adiponectin receptor 1 (14–16). For GRK2 inhibition in vivo, we used GRKInh, a small peptide inhibitor derived from the first intracellular loop of the β2-adrenergic receptor (12, 17). Our data with GRKInh show that GRK2 inhibition counteracts the heart failure-related cardiac metabolic dysfunction and signs of heart failure of FASN transgenic mice.

Experimental Procedures

Generation of Transgenic Mice and Animal Experiments

Transgenic mice were generated as described (12) with minor modifications. Briefly, for the generation of transgenic mice with myocardium-specific expression of FASN, we constructed a transgene that placed the FASN cDNA under the control of the α-myosin heavy chain (α-MHC) promoter (12). For the generation of transgenic mice with myocardium-specific expression of UCP1 (uncoupling protein-1), PPARG, and PPARG-S273A (isoform-1, which lacks amino acids 1–28 of isoform-2; serine 273 refers to the numbering of isoform-2), a similar approach was used. The plasmid sequence was removed by NotI digestion, and the purified linear DNA (2 ng/μl) was injected into fertilized oocytes of superovulated B6 (C57BL/6J) mice. For generation of transgenic Tg-PPARG and Tg-PPARG-S273A mice, fertilized oocytes from non-transgenic and Tg-GRKInh mice (transgenic mice with myocardium-specific expression of GRKInh, a GRK2-specific peptide inhibitor with the peptide sequence MAKFERLQTVTNYFITSE) were used. Transgenic mice with myocardium-specific expression of GRKInh or human RKIP (PEBP1) were generated and characterized previously (12). Mouse lines in the study were deposited into the JAX repository (The Jackson Laboratory) and have the following strain ID numbers: 911818 (C57BL/6Tg(MHCPEBP1)1 Sjaa); 911822 (C57BL/6Tg(MHCGRK-Inh)1 Sjaa); 911826 (C57BL/6Tg(MHCFASN)1 Sjaa); and 911830 (C57BL/6Tg(MHCUCP1)1 Sjaa).

The effect of rosiglitazone-induced Pparg activation was analyzed with 8-month-old male ApoE^−/− mice, which had received 30 mg/kg/day rosiglitazone for 2 months. Untreated, age-matched ApoE^−/−, and non-transgenic B6 mice served as control groups. Abdominal aortic constriction (AAC) was performed in 4-month-old male B6 mice to trigger pressure overload-induced cardiac hypertrophy and signs of heart failure (11). Age-matched control mice underwent the identical surgical procedure except for ligation of the aorta (sham-operated mice). All of the mice were kept on a 12-h light/12-h dark regime and had free access to food and water. The ApoE^−/− mice were fed a rodent chow that contained 7% fat and 0.15% cholesterol (AIN-93-based diet), whereas B6 mice were fed a standard rodent chow containing 4.5% fat.

Transthoracic echocardiography was performed with a Vivid 7 echocardiograph (GE Healthcare) with a 12 MHz linear array transducer similarly as described previously (11). The left ventricular ejection fraction was calculated in the M-mode of the parasternal long axis view using the formula of Teichholz. Recordings were interpreted off line using EchoPac Pc 3.0 software (GE Healthcare).

Animal experiments were performed in accordance with National Institutes of Health guidelines, and they were reviewed and approved by the local committee on animal care and use (University of Zurich).

Whole Genome Microarray Gene Expression Analysis

Whole genome microarray gene expression analysis of cardiac tissue was performed using Affymetrix GeneChip Mouse genome MG430 2.0 arrays essentially as described previously (18). Gene ontology analyses of microarray data were performed with GCOS and/or RMA-processed data using GeneSpring GX software (Agilent). Probe sets, which were significantly up-regulated in failing hearts (fold change ≥2 relative to the respective control group and p ≤ 0.01) were used for gene ontology classification. Microarray gene expression data are available at the NCBI GEO database accession numbers GSE25765-8 (GSE25765, GSE25766, GSE25767, and GSE25768), GSE28031, and GSE49351.

Gene expression of selected genes was also analyzed by real time quantitative (q) RT-PCR with a LightCycler 480 (Roche Diagnostics). Sequences of the forward and reverse primers were as follows: Acaca forward 5′-GCCTCCGTCAGCTCAGATAC-3′ and Acaca reverse 5′-GACCACCGACGGATAGATCG-3′; Adipoq forward 5′-ACTGCAACATTCCGGGACTC-3′ and Adipoq reverse 5′-GAGGCCTGGTCCACATTCTT-3′; Fasn forward 5′-GGCCCCTCTGTTAATTGGCT-3′ and Fasn reverse 5′-CGCTTGTTGGTGGACACTTG-3′; FASN forward 5′-TCGTGTTGACTTCTCGCTCC-3′ and FASN reverse 5′-AAGCCGTAGTTGCTCTGTCC-3′; PPARG forward 5′-GCTCCGTGGATCTCTCCGTA-3′ and PPARG reverse 5′-AGCTTTATCTCCACAGACACGA-3′; Retn forward 5′-GTCCTGCTAAGTCCTCTGCCAC-3′ and Retn reverse 5′-GGCTGCTGTCCAGTCTATCCTTG-3′; and Ucp1 forward 5′-CACTGCCAAAGTCCGCCTTCAGA-3′ and Ucp1 reverse 5′-GCAGGCAGACCGCTGTACAGTT-3′.

Lentivirus-mediated Down-regulation of Fasn and Ucp1 by RNAi in Vivo

For the down-regulation of Fasn expression in vivo, ApoE^−/− mice were transduced by intraperitoneal administration of a replication-incompetent lentivirus (1 × 10⁸ copies/mouse in PBS), which down-regulates Fasn by polymerase II-dependent expression of a pre-miRNA targeting the Fasn RNA by RNAi. Endogenously expressed Ucp1 was down-regulated by the transduction of B6 mice with a lentivirus that expressed a pre-miRNA targeting Ucp1 by RNAi. The lentiviral expression plasmids were generated by inserting the indicated double-stranded oligonucleotides that encoded an engineered pre-miRNA sequence into the pLenti6/V5-Dest Gateway® Vector (Invitrogen): miFasn top strand 5′-TGCTGATAACTTGGAGTTCGGGTCTTGTTTTGGCCACTGACTGACAAGACCCGCTCCAAGTTAT-3′ and miFasn bottom strand 5′-CCTGATAACTTGGAGCGGGTCTTGTCAGTCAGTGGCCAAAACAAGACCCGAACTCCAAGTTATC-3′; miUcp1top strand 5′-TGCTGTTTGATCCCATGCAGATGGCTGTTTTGGCCACTGACTGACAGCCATCTATGGGATCAAA-3 and miUcp1 bottom strand 5′-CCTGTTTGATCCCATAGATGGCTGTCAGTCAGTGGCCAAAACAGCCATCTGCATGGGATCAAAC-3. A pseudotyped lentivirus was produced by co-transfection of 293FT cells with the lentiviral plasmid and a mixture of packaging plasmids pLP1, pLP2, and pLP/VSVG (Invitrogen). To quantify the lentivirus integration in vivo, we used primers that comprised sequences that were derived from the cytomegalovirus immediate early promoter and the pre-miRNA sequence. Down-regulation of Fasn (or Ucp1) expression was confirmed by real time quantitative RT-PCR after the transduction of mice or isolated neonatal mouse cardiomyocytes with miFasn-lentivirus (or miUcp1-lentivirus).

Antibodies

The following antibodies were used for immunohistochemistry, immunofluorescence, and immunoblotting: anti-ARRB1 (β-arrestin-1) antibodies, raised in mouse against recombinant ARRB1 (12); anti-Agtr1 (AT1R) antibodies, which were raised in rat against the carboxyl-terminal region of Agtr1 (19); anti-FASN antibodies, which were raised in rabbit against an antigen encompassing amino acids 2205–2504 of FASN (11); anti-GRK2 (ADRBK1) antibodies, raised in rabbit against recombinant GRK2 protein (12); anti-GRK5 antibodies, raised in rabbit against recombinant GRK5 protein (12); anti-GRKInh antibodies, raised in rabbit against GRKInh (12); anti-Pparg antibodies, raised in rabbit against an antigen encompassing amino acids 8–106 of PPARG (Santa Cruz Biotechnology Inc.) or synthetic phosphopeptides derived from PPARG around the phosphorylation site of Ser-273 or Ser-112 (BIOSS antibodies; Abcam); and Ucp1/UCP1 antibodies raised in rabbit against an antigen encompassing amino acids 288–302 of mouse/human Ucp1/UCP1 (11). For the immunohistological and immunoblot detection of activated phospho-ERK1/2, phospho-ERK1/2-specific antibodies were used detecting activated ERK1/2 phosphorylated at Thr-202 + Tyr-204 of ERK1 and Thr-185 + Tyr-187 of ERK2 (E10 mouse mAb, Cell Signaling). For immunoblot detection of ERK1/2, the ERK1/2-specific antibodies raised in rabbits (Cell Signaling) were used, and immunofluorescence detection of p38 MAPK on cardiac sections was performed with anti-p38 antibodies (Cell Signaling). The immunoblot detection of activated AMPKα (Prkaa1/2; protein kinase, AMP-activated α1/2 catalytic subunit) phosphorylated on Thr-183/172 was detected with antibodies raised in rabbit against a synthetic peptide corresponding to residues that surrounded Thr-172 (40H9, Cell Signaling). Immunoblotting and immunohistochemistry were routinely used to determine and confirm the cross-reactivity of the antibodies with the respective mouse and human proteins.

Immunohistology Analyses and Immunofluorescence

For immunohistology, we used paraffin sections or cryosections of mouse heart specimens. Immunohistological detection of Fasn (FASN) was performed with affinity-purified polyclonal antibodies as described (11). Methods describing Oil Red O staining, immunofluorescence detection of proteins, and immunohistology for activated phospho-ERK1/2 in paraffin sections or cryosections have been described previously (11, 12). Immunohistology sections were imaged with a Leica DMI6000 microscope equipped with a DFC420 camera, and immunofluorescence imaging was performed with a Leica (TCS) confocal laser microscope.

Immunoblot Detection of Proteins

For immunoblot detection of proteins, cardiac tissue was pulverized in liquid nitrogen and extracted with RIPA buffer supplemented with protease/phosphatase inhibitor mixture, as described previously (20). Detection of proteins was performed with affinity-purified antibodies or F(ab)₂ fragments of the respective antibodies (11, 12) after separation of proteins by SDS-PAGE and subsequent electrophoretic protein transfer to PVDF membranes. Bound antibody was visualized with F(ab)₂ fragments of enzyme-coupled secondary antibodies (Dianova) or by enzyme-coupled protein A (Merck Millipore) as applicable and was followed by enhanced chemiluminescent detection (ECL Prime, Amersham Biosciences).

Functional Assays

Mouse or rat neonatal cardiomyocytes were isolated and transfected as described (12, 21). Fibroblasts were removed by preplating for 1 h at 37 °C. Cardiomyocytes were collected and cultivated in minimum essential medium supplemented with 5% FCS and 25 mg/liter BrdU (5-bromo-2′-deoxyuridine). For knockdown of Fasn and Ucp1, neonatal cardiomyocytes were transfected with stealth RNAi targeting the coding sequence of rat or mouse Fasn (nucleotides 428–452 and 1990–2014; Invitrogen) and Ucp1 (nucleotides 289–313 and 401–425; Invitrogen). For cardiomyocyte expression of PPARG and PPARG-S273A, the human cDNAs encoding PPARG and PPARG-S273A were inserted into the KpnI/XbaI sites of pcDNA3 (Invitrogen). All of the mutants and constructs that were generated by PCR were sequenced entirely. DNA strand breaks were determined in situ by the terminal deoxynucleotidyltransferase-mediated dUTP nick end labeling (TUNEL) technique (Roche Applied Science) (12). Measurement of the [ATP] of cardiac tissue extracts was performed as described (11), and Pparg transcription factor DNA binding activity was determined with a Pparg transcription factor assay kit (Abcam). Cellular cAMP, total cardiac free fatty acids (FFA), and triacylglycerol (TAG) contents were analyzed as detailed previously (11, 21). Cardiac contents of diacylglycerol (DAG) and ceramides were determined with the DAG kinase assay method as described (22).

Measurement of Cardiomyocyte Substrate Metabolism

We used a Seahorse XF24 extracellular flux analyzer (Seahorse Bioscience) to determine the cardiomyocyte substrate metabolism. The oxygen consumption rates (OCR) (pmol/min) and extracellular acidification rates (ECAR) (the H⁺ production rate, mpH/min), of neonatal cardiomyocytes (10 000 cells/well) plated on Cell-Tak-coated plates (Discovery Labware Inc., Bedford, MA) were measured in assay medium (i.e. unbuffered DMEM supplemented with 5.5 mm glucose and 0.5 mm carnitine) according to the Installation and Operation Manual from Seahorse Bioscience. The oxidation of endogenous fatty acids (without exogenously added palmitate, to detect the function of transgenic FASN expression) was determined by measurement of the absolute and relative OCR that was inhibited by the CPT-1 (carnitine palmitoyl transferase 1) inhibitor, Etomoxir (50 μm). The extent of glycolysis was determined by measurement of the absolute and relative ECAR, which was inhibited by 50 mm 2-deoxyglucose. As indicated, we also determined the effect of an adult-like lipogenic milieu by the cultivation of cardiomyocytes for 10 days with a 3F protocol that consisted of insulin (5 μg/ml), 3-isobutylmethylxanthine (0.25 mm), and dexamethasone (0.5 μm) (23), which were added as supplements to the standard medium. As a control for the 3F protocol, we used cardiomyocytes that were cultivated for 10 days under standard conditions. Long term cultivation of neonatal cardiomyocytes is a model of in vitro senescence characterized by metabolic deficiencies (24), which could account for the overall low β-oxidation rate of 5–25% in our experiments. Metabolic flux experiments were performed on 6 wells of a 24-well plate (technical replicates) and were reproduced at least three times (biological replicates). The oligomycin-insensitive OCR (a measure of mitochondrial uncoupling) was determined after the addition of 2.5 μm oligomycin. The non-mitochondrial OCR that remained after the addition of rotenone/antimycin A (2 μm) was subtracted.

Statistical Analyses

The results are presented as the means ± S.D. unless otherwise specified. The p values were calculated with Student's t test. Analysis of variance was performed for comparisons between more than two groups followed by a post test (Tukey's multiple comparison test unless otherwise specified), and statistical significance was set at a p value of <0.05 unless otherwise stated.

Results

FASN Transgenic Cardiomyocytes Developed a Dysfunctional Cardiac Substrate Metabolism

A dysfunctional cardiac substrate metabolism is a common feature of late-stage heart failure with limited treatment options. To reproduce the energy substrate use of heart failure patients and experimental models, which commonly show up-regulation of the major palmitate-synthesizing enzyme, FASN (10, 11), we generated a transgenic model with myocardium-specific FASN expression under the control of the α-MHC promoter (Fig. 1, A and B). Immunoblot detection of the FASN protein confirmed transgenic protein expression in hearts from mice with stable genomic integration of the FASN transgene, whereas the Fasn protein was barely detectable in non-transgenic B6 hearts (Fig. 1C). Two different transgenic lines (derived from founders numbers 3 and 9) were established, which showed comparable FASN protein levels (Fig. 1C). All of the experiments were independently performed with these two transgenic lines.

FIGURE 1.

FASN transgenic cardiomyocytes developed a dysfunctional cardiac substrate metabolism. A, scheme of the α-MHC-FASN vector used for the generation of FASN transgenic mice. B, identification of the α-MHC-FASN transgene in the genomic DNA of different founder mice by PCR. The asterisks denote the amount of the FASN transgene-specific PCR product (***, **, *, >, =, < amount of PCR product obtained with 2 ng of plasmid template (P)). C, immunoblot (IB) detection of the FASN protein in hearts from different founder mice with FASN-specific antibodies (IB: FASN). The asterisks indicate the amount of transgenic FASN protein relative to the non-transgenic control after normalization to Gnb (***, 2.5-fold; **, 2.3-fold; *, 1.8-fold (lane 5) and 1.4-fold (lane 10) increase over control). The lower panel shows a control immunoblot detecting Gnb. D and E, real time measurement of the OCR and ECAR of neonatal cardiomyocytes isolated from FASN transgenic (Tg-FASN) and B6 control mice under basal conditions (left panel) and after the creation of an adult-like lipogenic milieu by the 3F protocol for 10 days (right panel). The addition of the CPT-1 inhibitor Etomoxir (Etom., 50 μm) and 2-deoxyglucose (2-DG, 50 mm) is indicated by the arrows. The relative values normalized to the baseline (i.e. the second measurement point set to 100%) (D) and the absolute values of OCR and ECAR (E) are presented. F and G, etomoxir-blocked fraction of the OCR represents fatty acid β-oxidation; the 2-deoxyglucose-blocked fraction of the ECAR represents glycolysis, and the ratio of glycolysis/β-oxidation was calculated. The data are shown as the means ± S.D.; n = 6 technical replicates (D and E), and n = 3 biological replicates (F and G); ***, p = 0.0002 (F, left panel) and 0.0004 (G, right panel); **, p = 0.0032 (F, middle panel), 0.0079 (F, right panel) and 0.0071 (G, left panel); p = 0.7324 (G, middle panel).

After the generation of the transgenic mouse lines, the metabolic energetics of isolated neonatal FASN transgenic (Tg-FASN) cardiomyocytes was determined with a Seahorse Bioscience XF24 Extracellular Flux Analyzer. We measured the β-oxidation of endogenous fatty acids by the Etomoxir-sensitive fraction of the OCR, and glycolysis by the 2-deoxyglucose-sensitive ECAR (Fig. 1, D and E). Compared with non-transgenic controls, FASN transgenic cardiomyocytes showed significantly more β-oxidation under basal conditions (22.7 ± 1.5% versus 5.3 ± 0.9%; Fig. 1, F and G). This finding indicated that FASN increases the substrate availability of cardiomyocytes for β-oxidation.

We next determined the effect of an adult-like lipogenic milieu on neonatal cardiomyocytes by treatment with a 3F protocol consisting of insulin, 3-isobutylmethylxanthine, and dexamethasone for 10 days (23). In control cardiomyocytes, the 3F protocol switched the embryo-like metabolism dominated by glycolysis (25) to an adult-like metabolism, which was characterized by an increased baseline OCR (Fig. 1E), and more fatty acid β-oxidation (Fig. 1G). In contrast, FASN transgenic cardiomyocytes developed a heart failure-like phenotype with an overall depressed substrate metabolism (Fig. 1E), and predominant glycolysis (Fig. 1F). Thus, the increased availability of the lipid substrate, palmitate, did not protect neonatal cardiomyocytes from a substrate switch to a heart failure-like phenotype that had depressed bioenergetics dominated by glycolysis.

FASN Transgenic Mice Developed Signs of Heart Failure and Cardiac Lipid Load

The extracellular flux analysis detected a heart failure-like metabolic substrate use in isolated cardiomyocytes from Tg-FASN mice. But the method employs unloaded cardiomyocytes and does not represent the condition in the loaded myocardium, which functions with real heart rates. Therefore, we analyzed the cardiac phenotype in vivo. Our data show that adult FASN transgenic mice developed signs of heart failure as early as 6 months of age, as evidenced by a significantly decreased left ventricular ejection fraction, cardiac hypertrophy with dilatation, and increased cardiomyocyte apoptosis (Fig. 2, A–E). As a control, the body weight of 6 month-old Tg-FASN mice was not different from B6 control mice (i.e. 34.28 ± 1.02 g and 33.45 ± 0.98 g, respectively; ±S.D.; n = 4; p = 0.5909).

FIGURE 2.

FASN transgenic mice developed signs of heart failure and cardiac lipid load. A, decreased left ventricular ejection fraction of 6-month-old Tg-FASN mice. B, histological analysis of a Tg-FASN heart compared with a B6 heart (bar, 2 mm). C, increased heart weight-to-body weight ratio of Tg-FASN mice. D and E, increased number of TUNEL-positive cardiomyocytes of Tg-FASN hearts. E shows representative immunohistological sections (bar, 20 μm). The data shown are the mean ± S.D., n = 5; ***, p < 0.0001 (A, C, and D). F–H, detection of the cardiac FASN protein levels by immunoblotting (IB) (F and G) and immunohistology (H) relative to the B6 control (bar, 40 μm). F shows quantitative data that were obtained by densitometric scanning of immunoblots (± S.D., n = 4; ***, p = 0.0005). I–L, cardiac FFA (I), triacylglycerol (TAG) (J), DAG (K), and total ceramide (L) contents of Tg-FASN mice relative to non-transgenic B6 controls (±S.D., n = 5; ***, p = 0.0006 (I), 0.0004 (J), and 0.0001 (K); **, p = 0.0028). Histology experiments are representative of four different mice/group (B and H).

The cardiac FASN protein in FASN-expressing mice was detected by immunoblotting and immunohistology (Fig. 2, F–H). The FASN protein level in Tg-FASN hearts was increased ∼2.4-fold (Fig. 2, F and G), which is comparable with the up-regulated FASN level of failing human hearts (10, 11).

Concomitantly to the increased FASN protein, cardiac FFA and triacylglycerol contents of Tg-FASN mice with signs of heart failure were elevated 2.2- and 2.1-fold, respectively (Fig. 2, I and J). The cardiac contents of DAG and ceramides, which can be induced by palmitate, i.e. the major lipid synthesized by FASN, were also significantly increased (Fig. 2, K and L). These lipids could be involved in the heart failure phenotype of Tg-FASN mice because they trigger a wide spectrum of cardiotoxic mechanisms, which involves, for example, the excessive formation of reactive oxygen species, an increased endoplasmic reticulum stress, enhanced apoptosis, and mitochondrial dysfunction (22, 26). Additionally, increased cardiac contents of DAG and ceramide could mediate the activation of protein kinase C (PKC), which further decreases heart function (22, 27, 28). Thus, FASN transgenic hearts developed signs of heart failure with cardiotoxic lipid load in addition to the dysfunctional energy substrate metabolism, which was detected in isolated cardiomyocytes.

Up-regulation of the Heart Failure-related Cardiac Lipid Metabolic Process in FASN Transgenic Mice

Whole genome microarray gene expression profiling further confirmed the heart failure phenotype of Tg-FASN hearts by demonstrating the significant up-regulation of the heart failure-related cardiac lipid metabolic process (Fig. 3A). A similar induction of those adipogenic genes was also observed when signs of heart failure were triggered by 6 months of pressure overload (Fig. 3B) (11). In contrast, cardiac hypertrophy, without signs of heart failure (11) and induced by 4 weeks of pressure overload, did not up-regulate the heart failure-related adipogenic gene expression signature (Fig. 3C). Moreover, cardiac hypertrophy in the absence of heart failure signs promoted a significantly decreased expression of two probe sets that detect two enzymes of fatty acid biosynthesis, i.e. Scd1 (stearoyl-CoA desaturase-1) and Acly (ATP citrate lyase) (Fig. 3C). Thus, a heart failure-related adipogenic gene expression signature accompanied the onset of heart failure signs in FASN transgenic mice.

FIGURE 3.

Up-regulation of the heart failure-related cardiac lipid metabolic process in FASN transgenic mice. A, microarray gene expression profiling identified significantly up-regulated heart failure-related genes of the lipid metabolic process in hearts of 6-month-old Tg-FASN mice compared with B6 mice. B, heart failure-related adipogenic gene signature was similarly detected in 10-month-old B6 mice with heart failure induced by 6 months of AAC compared with age-matched sham controls (Sham-6mo). C, adipogenic genes were not significantly up-regulated in 5-month-old B6 mice with cardiac hypertrophy induced by 1 month of ACC (cardiac hypertrophy) in the absence of heart failure signs compared with the sham control (Sham-1mo). D, significantly up-regulated genes of the lipid metabolic process in hearts of 8-month-old ApoE^−/− mice with 2 months of Pparg activation with rosiglitazone compared with untreated ApoE^−/− mice. The probe sets were categorized into genes involved in lipid synthesis, storage, oxidation (Oxid.), and adipocyte (Adipoc.) differentiation. The statistical significance of the microarray data was evaluated using signal intensity values of probe sets (*, p < 0.05; **, p ≤ 0.01; ***, p ≤ 0.001 versus the respective control; ±S.D., n = 2 gene chips per group, cRNA pooled from four mice/gene chip).

Heart Failure-related Adipogenic Genes Triggered by FASN Are Pparg Targets

In search of FASN-induced pathomechanisms, we focused on the adipogenic and heart failure-promoting transcription factor, Pparg (29, 30), because (i) palmitate, the major lipid synthesized by FASN, enhances the activity of Pparg (31), and (ii) adipogenic genes induced by FASN are Pparg targets (32), which are similarly up-regulated by Pparg overexpression in the mouse heart (29). Similarly, the treatment of ApoE^−/− mice for 2 months with the Pparg agonist, rosiglitazone, also significantly up-regulated those heart failure-related adipogenic Pparg target genes, which were induced by FASN (Fig. 3D). As a control, rosiglitazone promoted signs of heart failure in ApoE^−/− mice (cf. Fig. 4, F–I). These findings demonstrate that the heart failure-related adipogenic gene signature induced by Tg-FASN and chronic pressure overload is also triggered by direct Pparg activation with rosiglitazone.

FIGURE 4.

Down-regulation of endogenous Fasn reveals a causal relationship between Fasn and Pparg activation in promoting cardiac dysfunction. A and B, stable integration of lentiviral miFasn-DNA into the genomic DNA of ApoE^−/− mouse hearts (H1, H2), kidneys (K1, K2), and livers (L1, L2) isolated 2 months after lentiviral transduction. Control DNA (cont. DNA) was isolated from an ApoE^−/− mouse. B, genomic integration of lentiviral DNA was quantified by quantitative PCR (±S.D.; n = 6; ***, p < 0.001; *, p < 0.05 versus heart). C, lentivirus-mediated delivery of miFasn decreased the expression of Fasn in hearts of ApoE^−/− mice with 2 months of Pparg activation by rosiglitazone (±S.D.; n = 3; **, p = 0.0029). D and E, immunohistological detection of Fasn (D) with anti-Fasn antibodies (anti-Fasn), and total lipids by Oil Red O staining (E) in cardiac sections of ApoE^−/− mice after 2 months of Pparg activation and transduction of a control lentivirus (miCon) or lentiviral delivery of miFasn (bar, 40 μm). F, lentivirus-mediated delivery of miFasn normalized the enhanced cardiomyocyte apoptosis of ApoE^−/− mice that was triggered by Pparg activation (±S.D.; n = 3; **, p = 0.0031). The right panels show representative immunohistology images of TUNEL staining (bar, 20 μm). G, viral transduction of ApoE^−/− mice with a lentivirus that targets Fasn by RNAi (miFasn) retarded the increase in the heart weight-to-body weight ratio that was induced by 2 months of Pparg activation with rosiglitazone (±S.D.; n = 6; **, p = 0.0011). H, histological analysis of hearts from 8-month-old ApoE^−/− mice isolated after 2 months of Pparg activation and transduction of a control lentivirus (miCon) or miFasn (bar, 2 mm). I, left ventricular ejection fraction of ApoE^−/− mice after 2 months of Pparg activation and transduction of a control lentivirus (miCon) or miFasn (±S.D.; n = 6; ***, p = 0.0002). Histology experiments are representative of four different mice/group (D, E, and H).

Down-regulation of Endogenous Fasn Reveals a Causal Relationship between Fasn and Pparg Activation in Promoting Cardiac Dysfunction

Next, we investigated the impact of Fasn on Pparg-induced cardiac dysfunction, and we knocked down the endogenously expressed Fasn by RNAi in rosiglitazone-treated ApoE^−/− mice. Inhibition of Fasn by lentiviral transduction of a Fasn-specific miRNA (Fig. 4, A–D) retarded the development of rosiglitazone-induced cardiac lipid load (Fig. 4E) and signs of rosiglitazone-induced heart failure (Fig. 4, F–I). Together, these data confirmed the causal relationship between Fasn and enhanced Pparg activation in promoting cardiotoxicity and cardiac dysfunction.

GRK2 Inhibition by GRKInh in FASN Transgenic Mice

In view of the central role of FASN, the inhibition of FASN would be a straightforward treatment approach. However, FASN is an essential enzyme that has indispensable functions in energy homeostasis, membrane biology, and neurogenesis, which preclude long term FASN inhibition in vivo (10, 33, 34). We therefore searched for an alternative strategy to target the dysfunctional cardiac substrate metabolism. We focused on the inhibition of GRK2, which is a well established means of cardioprotection (12, 35). Furthermore, GRK2 inhibition enhances the ERK cascade (12), which promotes (partial) inactivation of Pparg (36). In agreement with heart failure patients (37), the GRK2 protein levels were significantly up-regulated (i.e. 1.81 ± 0.12-fold) in Tg-FASN hearts with signs of heart failure relative to the B6 controls (Fig. 5A).

FIGURE 5.

GRK2 inhibition by GRKInh in FASN transgenic mice. A, cardiac up-regulation of the GRK2 protein level in Tg-FASN relative to B6 hearts was detected by immunoblotting (IB) with GRK2-specific antibodies (n = 4 mice/group). The lower panel is a control immunoblot detecting Gnb. The right panel shows the quantitative immunoblot evaluation (±S.D., n = 4). B, immunoaffinity enrichment of GRKInh (AP, +) with GRKInh-specific antibodies from Tg-GRKInh/FASN hearts, and immunoblot detection (IB) of co-enriched GRK2 protein (left panel) and GRK5 protein (right panel). The control experiment (AP, −) applied an affinity matrix with immobilized control IgG. The lower panels show immunoblot detection of enriched GRKInh. C, quantitative assessment of the GRKInh-GRK interaction. GRK2 and GRK5 protein levels were determined by immunoblotting with GRK2-specific and GRK5-specific antibodies, respectively, of cardiac lysates from Tg-GRKInh/FASN mice before (total) and after (not bound to GRKInh) incubation with an affinity matrix with immobilized anti-GRKInh antibodies. The left and middle panels present data evaluation from three independent experiments (±S.D., n = 3), and the right panels show representative immunoblots. D, immunofluorescence co-localization of Arrb1 with p38 MAPK in cardiac sections from Tg-FASN and Tg-GRKInh/FASN mice (bar, 20 μm). Arrb1 was detected with affinity-purified mouse anti-Arrb1 antibodies followed by F(ab)₂ fragments of Alexa Fluor 546-labeled (red) secondary antibodies, and p38 was detected with affinity-purified rabbit anti-p38 MAPK antibodies followed by F(ab)₂ fragments of Alexa Fluor 488-labeled (green) secondary antibodies. Cell nuclei were stained with DAPI (blue). Immunofluorescence experiments are representative of four different mice/group.

To inhibit GRK2 in vivo, we used a GRKInh derived from the first intracellular loop of the β2 adrenergic receptor (12, 17). We used transgenic mice with myocardium-specific expression of GRKInh, which were established previously (12). The GRKInh peptide interacted specifically with GRK2 in heart tissue extracts from Tg-GRKInh/FASN mice as demonstrated by co-enrichment, whereas the amount of GRK5 co-enriched with GRKInh-specific antibodies was below the limit of detection (Fig. 5B). Quantitative assessment of the GRK2-GRKInh interaction indicated that 83.6 ± 4.2% of the total cardiac GRK2 protein was captured by GRKInh affinity enrichment, whereas the amount of GRK5 protein bound to GRKInh was less than 20% (i.e. 18.9 ± 2.2%) of the total cardiac GRK5 content (Fig. 5C).

The functional effects of GRK2 inhibition in transgenic Tg-GRKInh/FASN hearts were analyzed by the immunofluorescence detection of Arrb1, which translocates to phosphorylated membrane-spanning receptors as a direct consequence of GRK2-mediated phosphorylation (Fig. 5D). In agreement with an increased GRK2 activity, immunofluorescence analysis detected the substantial membrane localization of Arrb1 in a cardiac section from a Tg-FASN mouse with signs of heart failure (Fig. 5D, left panel). In contrast, the double transgenic Tg-GRKInh/FASN heart section showed a largely cytoplasmic localization of Arrb1 as evidenced by co-localization with the cytosolic p38 MAPK (Fig. 5D, right panel). These findings indicate that GRKInh interacts with GRK2 in hearts from double transgenic Tg-GRKInh/FASN mice. As a consequence of the GRKInh-GRK2 interaction, the enhanced GRK2-mediated Arrb1 membrane translocation could be blunted.

GRK2 Inhibition by GRKInh Prevents the Dysfunctional Cardiac Energetics of FASN Transgenic Cardiomyocytes

We characterized the substrate metabolism of isolated neonatal cardiomyocytes from double transgenic Tg-GRKInh/FASN mice compared with single transgenic Tg-FASN mice. Under basal conditions, the presence of GRKInh retarded the premature appearance of an adult-like metabolic phenotype with an increased baseline OCR of Tg-FASN cardiomyocytes (Fig. 6, A and B). Concomitantly, the increased β-oxidation of Tg-FASN cardiomyocytes under basal conditions was also retarded in double transgenic Tg-GRKInh/FASN cardiomyocytes (Fig. 6, A and C). Notably, the extent of β-oxidation of Tg-GRKInh/FASN double transgenic cardiomyocytes was not significantly different from that of the B6 control, i.e. 6.0 ± 1.0% versus 5.4 ± 1.7%, respectively (Fig. 6, A and C). Moreover, under the 3F protocol, GRKInh retarded the shift to a heart failure-like metabolic phenotype of Tg-FASN cardiomyocytes, which is characterized by predominant glycolysis and decreased β-oxidation (Fig. 6, A–E). Similarly to non-transgenic B6 controls, the Tg-GRKInh/FASN cardiomyocytes shifted to an adult-like phenotype, which is characterized by an increased baseline OCR and an elevated β-oxidation (Fig. 6, A–E).

FIGURE 6.

GRK2 inhibition by GRKInh prevents the dysfunctional cardiac energetics of FASN transgenic cardiomyocytes. A and B, development of dysfunctional cardiomyocyte energetics of FASN transgenic cardiomyocytes was retarded in double transgenic mice that co-expressed the GRK2-inhibitor, GRKInh. Real time measurements of OCR and ECAR of neonatal cardiomyocytes isolated from FASN transgenic (Tg-FASN), double transgenic Tg-GRKInh/FASN, and non-transgenic B6 mice were performed under basal conditions (left panels) and after treatment with the 3F protocol (right panels). The OCR and ECAR values normalized to the baseline (A), and the absolute values of OCR and ECAR (B) are presented. C–E, Etomoxir (Etom.)-blocked fraction of OCR, which represents fatty acid β-oxidation (C), the 2-deoxyglucose (2-DG)-blocked fraction of ECAR, which represents glycolysis (D), and the ratio of glycolysis/β-oxidation (E) were determined with cardiomyocytes isolated from Tg-FASN, Tg-GRKInh/FASN, B6, and Tg-GRKInh mice. The data are shown as the means ± S.D.; n = 6 technical replicates (A and B) and n = 3 biological replicates (C–E); *, p < 0.05; **, p < 0.01 and ***, p < 0.001 versus Tg-FASN. F, cardiomyocytes from Tg-GRKInh/FASN (lanes 3 and 4) mice showed significant adiponectin (Adipoq)-stimulated activation of Prkaa relative to Tg-FASN cardiomyocytes (lanes 1 and 2). Cardiomyocytes (3F protocol) were stimulated without (−) and with (+) globular domain adiponectin (2 μg/ml) for 60 min, and the activation of Prkaa was determined by immunoblotting (IB) with phospho-Prkaa (p-Prkaa)-specific antibodies. The left panel presents quantitative data from four independent experiments (±S.D.; n = 4; ***, p < 0.0001), and the middle and right panels show representative immunoblots.

The improved metabolic profile of Tg-GRKInh/FASN cardiomyocytes was accompanied by a resensitization of adiponectin receptor protein 1 (Adipor1)-mediated signaling (Fig. 6F), which is desensitized by GRK2 in the ischemic heart (14). Notably, the inhibition of GRK2 in Tg-GRKInh/FASN cardiomyocytes led to a significantly increased protein level of activated phospho-Prkaa upon adiponectin stimulation, whereas the adiponectin-stimulated signal was largely blunted in Tg-FASN cardiomyocytes (Fig. 6F). This finding could be relevant because Adipor1 and its target AMP-activated protein kinase (Prkaa) could protect against palmitate-induced toxicity (15, 16).

GRK2 Inhibition Retards the Development of Heart Failure Signs in Tg-FASN Mice

The improved substrate metabolism upon GRK2 inhibition was also reflected in vivo, in adult murine hearts. The presence of GRKInh in double transgenic Tg-GRKInh/FASN hearts compared with single transgenic Tg-FASN hearts led to a significantly decreased total FFA and triacylglycerol load compared with single transgenic Tg-FASN hearts (Fig. 7, A and B).

FIGURE 7.

GRKInh retards the development of heart failure signs in Tg-FASN mice and promotes ERK axis-dependent inhibition of Pparg transcriptional activity. A and B, cardiac FFA (A) and triacylglycerol (TAG) (B) contents of Tg-FASN mice relative to double transgenic Tg-GRKInh/FASN, non-transgenic B6, and single transgenic Tg-GRKInh mice (±S.D., n = 5; *, p < 0.05 versus Tg-FASN and Tg-GRKInh; **, p < 0.01 versus Tg-FASN; ***, p < 0.001 versus Tg-FASN). C, cardiac Acaca expression in Tg-FASN, double transgenic Tg-GRKInh/FASN, B6, and Tg-GRKInh mice (±S.D. n = 5; ***, p < 0.001 versus Tg-GRKInh/FASN, B6, and Tg-GRKInh hearts). D, cardiac expression of transgenic FASN (left panel) (± S.D., n = 5; ***, p < 0.001 versus B6), and expression of the endogenously expressed murine Fasn (right panel) (±S.D.; n = 5; ***, p < 0.001 versus Tg-GRKInh/FASN and B6). E, immunoblot (IB) detection of the FASN/Fasn protein in cardiac tissue extracts from Tg-FASN relative to Tg-GRKInh/FASN mice (±S.D.; n = 5 hearts/group; **, p = 0.0016). F–H, left ventricular ejection fraction (F), the heart weight-to-body weight ratio (G), and the cardiac ATP content (H) of Tg-FASN mice relative to double transgenic Tg-GRKInh/FASN mice and age-matched B6 controls (± S.D., n = 5; *, p < 0.05; **, p < 0.01 and ***, p < 0.001 versus Tg-FASN). I, number of TUNEL-positive cardiomyocytes (± S.D., n = 5; ***, p < 0.001 versus Tg-FASN). The right panel shows representative immunohistological sections of TUNEL staining (bar, 20 μm). J, immunohistological detection of activated phospho-ERK1/2 in cardiac sections of a Tg-FASN mouse relative to a Tg-GRKInh/FASN mouse (bar, 40 μm). Histology experiments are representative of four different mice/group. K, immunoblot detection of activated phospho-ERK1/2 (upper panel) and total ERK1/2 (lower panel) in cardiac tissue extracts from Tg-FASN relative to Tg-GRKInh/FASN mice (n = 4 hearts/group). L, immunoblot detection of pS273-Pparg (upper panel) and total Pparg (lower panel) in cardiac tissue extracts from Tg-FASN and Tg-GRKInh/FASN mice (n = 4 hearts/group). M and N, immunoblot detection (M) of pS273-Pparg (upper panel), pS112-Pparg (middle panel), and Pparg (lower panel), respectively, and Pparg activity of nuclear extracts (N) from Tg-FASN relative to Tg-GRKInh/FASN cardiomyocytes (3F protocol) treated without (−) or with (+) the MEK inhibitor PD0325901 (0.5 μm) (±S.D.; n = 4; ***, p < 0.001). O, cardiac expression of heart failure-related Pparg target genes in Tg-FASN, double transgenic Tg-GRKInh/FASN, and B6 control mice (±S.D.; n = 4; ***, p < 0.001 versus Tg-GRKInh/FASN and B6).

The decreased lipid load of Tg-GRKInh/FASN hearts was accompanied by a significantly decreased cardiac expression of the acetyl-CoA carboxylase α (Acaca), which mediates an essential step of fatty acid synthesis by catalyzing the conversion of acetyl-CoA into malonyl-CoA (Fig. 7C). Notably, GRKInh largely prevented the up-regulation of Acaca, i.e. a gene up-regulated by hypoxia (38), which was commonly triggered at the onset of heart failure induced by Tg-FASN, pressure overload, and Pparg (Fig. 7C and cf. Fig. 3, A, B, and D).

Although the expression of the human FASN transgene was not significantly altered between single transgenic Tg-FASN and double transgenic Tg-GRKInh/FASN hearts (Fig. 7D, left panel), GRKInh led to a significantly decreased expression of the endogenous murine Fasn gene, which is also a hypoxia-induced Pparg target (39), and shows up-regulated expression in Tg-FASN hearts with signs of heart failure (Fig. 7D, right panel, and cf. Fig. 3A). Concomitantly, the total cardiac FASN/Fasn protein level of Tg-GRKInh/FASN hearts was significantly decreased relative to that in Tg-FASN hearts (Fig. 7E). Together, these experiments show that GRK2 inhibition retards the FASN-induced dysfunction of the cardiac substrate metabolism and lipid overload. Concomitantly with the decreased lipid load, the development of signs of heart failure, such as cardiac dysfunction, cardiac hypertrophy, cardiac ATP depletion, and FASN/Fasn-mediated cell death, was significantly retarded (Fig. 7, F–I).

GRK2 Inhibition Promotes ERK Axis-dependent Inhibition of Pparg Transcriptional Activity

We investigated the mechanism that accounts for GRK2 inhibition-mediated protection against FASN-induced cardiolipotoxicity and focused on the interrelationship between GRK2 inhibition-mediated ERK axis activation and the inactivation of Pparg. Several lines of evidence support such a relationship. (i) The expression of several heart failure-related Pparg targets such as adiponectin (Adipoq), resistin (Retn), and uncoupling protein 1 (Ucp1) is down-regulated by ERK-dependent inactivation of Pparg, partially by involving serine 273 phosphorylation (36, 40). (ii) GRK2 inhibition enhances the activation of the ERK cascade (12, 41). (iii) Additionally, the reversal of palmitate toxicity can be achieved by ERK activation, e.g. triggered by AMPK signaling (16), i.e. the signaling pathway that was re-sensitized by GRK2 inhibition (cf. Fig. 6F). Conversely, excess palmitate down-regulated the ERK axis (16).

In agreement with palmitate-mediated inhibition of ERK (16), the cardiac content of activated phospho-ERK1/2 was low in Tg-FASN hearts relative to double transgenic Tg-GRKInh/FASN hearts (Fig. 7, J and K). Notably, GRKInh enhanced the activation of ERK2 (Fig. 7K), which is cardioprotective and protects the myocardium against ischemic injury (42). Concomitantly with enhanced ERK1/2 activation, serine 273-phosphorylated Pparg was increased in double transgenic Tg-GRKInh/FASN hearts relative to Tg-FASN hearts (Fig. 7L).

In agreement with the ERK-dependent inactivation of Pparg (36, 40, 43), the enhanced phosphorylation of Pparg on serine 273 and serine 112 of double transgenic Tg-GRKInh/FASN cardiomyocytes was accompanied by a significantly decreased Pparg transcription factor DNA binding activity compared with that of Tg-FASN cardiomyocytes (Fig. 7, M and N). The decreased Pparg activity of Tg-GRKInh/FASN cardiomyocytes was dependent on an activated ERK axis because the MEK inhibitor PD0325901 blunted the phosphorylation of Pparg on serine 273 and serine 112 and led to a significant up-regulation of the Pparg transcriptional activity of Tg-GRKInh/FASN cardiomyocytes (Fig. 7, M and N).

Concomitantly with Pparg inhibition, the expression of ERK-regulated, heart failure-related Pparg targets (i.e. Ucp1 and Adipoq) was significantly lower in double transgenic Tg-GRKInh/FASN hearts compared with Tg-FASN hearts (Fig. 7O). Concordantly with decreased signs of heart failure, GRKInh also led to a significantly decreased expression of the heart failure marker and Pparg target gene, Retn (Fig. 7O). Because Adipoq and Retn are heart failure-related Pparg targets (44, 45) that are induced by serine 273 phosphorylation-deficient PPARG-S273A (40), these data are compatible with the notion that cardioprotective GRK2 inhibition could involve the suppression of Pparg-dependent cardiolipotoxic gene expression by enhanced ERK-mediated serine 273 phosphorylation and the inactivation of Pparg.

Inhibition of Fasn Lowers the Cardiolipotoxicity Induced by Serine 273 Phosphorylation-deficient PPARG-S273A

We analyzed the impact of phosphorylation-deficient PPARG-S273A on cardiomyocyte function. Our experiments showed that both cardiomyocyte FFA content and the Fasn protein were triggered by PPARG activated with the PPARG agonist rosiglitazone and by PPARG-S273A (Fig. 8, A, C, and D). These data are in agreement with those from previous studies, which have shown that PPARG serine 273 dephosphorylation can enhance the expression of Fasn (46).

FIGURE 8.

Inhibition of Fasn decreases cardiolipotoxicity induced by serine 273 phosphorylation-deficient PPARG-S273A. A and B, neonatal rat cardiomyocytes were transfected with PPARG or PPARG-S273A mutant and incubated in the absence (−) or presence (+) of rosiglitazone (Rosiglit., 5 μm), and Fasn was inhibited by RNAi (siFasn) as indicated. The free fatty acid (A) and ATP (B) contents of cardiomyocytes were determined (±S.D.; n = 4; *, p < 0.05; **, p < 0.01; ***, p < 0.001 versus controls transfected with stealth control RNAi (−); Dunnett's multiple comparison test). C, immunoblot detection of Fasn (upper panel) and PPARG (middle panel) in cardiomyocyte lysates. The lower panel is a control immunoblot detecting (IB) Gnb. D, immunofluorescence localization of Fasn (red) and the transmembrane-spanning AT1 receptor (AT1R; green) in cardiomyocytes without (Control) or with transfection of PPARG, PPARG-S273A, siFasn, and treatment with rosiglitazone (Rosiglit.) as indicated. Fasn was detected with affinity-purified rabbit anti-Fasn antibodies followed by F(ab)₂ fragments of Alexa Fluor 546-labeled secondary antibodies, and AT1R was detected with affinity-purified rat anti-AT1R antibodies followed by F(ab)₂ fragments of Alexa Fluor 488-labeled secondary antibodies. Cell nuclei were stained with DAPI (blue; bar, 20 μm). The immunofluorescence data are representative of four independent experiments.

Concomitantly with the FFA load, cardiomyocyte dysfunction developed as evidenced by a significantly decreased cardiomyocyte ATP content induced either by rosiglitazone-activated PPARG or PPARG-S273A, respectively (Fig. 8B). The inhibition of Fasn by RNAi (Fig. 8, C and D) led to significantly decreased cardiomyocyte FFA content and largely prevented the decrease in cardiomyocyte ATP (Fig. 8, A and B). Together, these findings provide evidence that PPARG-S273A deteriorates cardiomyocyte function by regulating Fasn.

GRK2 Inhibition Retards the Up-regulation of the Heart Failure-related Ucp1 and Mitochondrial Uncoupling

In search of additional heart failure-promoting ERK-controlled Pparg target genes, we focused on Ucp1 (36, 47), which exerts mitochondrial uncoupling, a major metabolic feature of the failing heart metabolism (9). GRK2 inhibition by GRKInh led to a decreased cardiac Ucp1 expression and protein level in Tg-GRKInh/FASN mice relative to Tg-FASN mice (Fig. 9A and cf. Fig. 7O). Moreover, cardiomyocytes from Tg-GRKInh/FASN mice showed a significantly decreased oligomycin-insensitive OCR (a measure of mitochondrial uncoupling) compared with Tg-FASN cardiomyocytes (Fig. 9B). Conversely, the inhibition of the ERK axis in Tg-GRKInh/FASN cardiomyocytes significantly increased the Ucp1 protein and mitochondrial uncoupling (Fig. 9, B and C). These findings indicate that GRK2 inhibition decreased Ucp1-dependent mitochondrial uncoupling in Tg-GRKInh/FASN cardiomyocytes by enhanced activation of the ERK axis.

FIGURE 9.

GRK2 inhibition retards up-regulation of heart failure-related Ucp1 and mitochondrial uncoupling. A, immunoblot (IB) detection of Ucp1 protein in heart tissue extracts of Tg-FASN relative to Tg-GRKInh/FASN mice (n = 4 hearts/group). The lower panel shows a control immunoblot detecting Gnb. B, increased oligomycin-insensitive OCR of Tg-FASN cardiomyocytes relative to Tg-GRKInh/FASN cardiomyocytes (3F) treated for 72 h without (−) or with (+) the MEK inhibitor PD0325901 (±S.D.; n = 4; *, p < 0.05 versus TgGRKInh/FASN(−); **, p < 0.01 versus TgFASN). Baseline OCR (3F) of Tg-FASN, Tg-GRKInh/FASN, and Tg-GRKInh/FASN (+PD) was 13.68 ± 0.98, 22.73 ± 2.18, and 18.09 ± 2.62 pmol/min per μg of protein, respectively (±S.D.; n = 4). C, immunoblot detection of Ucp1 protein of Tg-FASN cardiomyocytes relative to Tg-GRKInh/FASN cardiomyocytes treated for 72 h without (−) or with (+) PD0325901 (±S.D.; n = 4; *, p < 0.05 versus Tg-GRKInh/FASN (−); ***, p < 0.001 versus Tg-FASN). The right panels show a representative immunoblot experiment.

Transgenic Tg-UCP1 Mice Developed Signs of Heart Failure and Increased Mitochondrial Uncoupling

To analyze whether an increased cardiac UCP1 level contributed to heart failure pathogenesis in vivo, we generated Tg-UCP1 mice with myocardium-specific UCP1 expression (Fig. 10, A and B). Immunoblot detection confirmed the increased cardiac UCP1 protein in Tg-UCP1 (Tg-6) mice relative to non-transgenic B6 controls (Fig. 10C). In addition to the Tg-6 line, we also used Tg-11 offspring with lower UCP1 protein levels (Fig. 10D).

FIGURE 10.

Tg-UCP1 mice developed signs of heart failure and mitochondrial uncoupling. A, transgenic vector used for the generation of Tg-UCP1 mice with myocardium-specific UCP1 expression. B, PCR identification of founder mice with stable integration of the UCP1 transgene into the genomic DNA. C and D, immunoblots (IB) show the detection of UCP1 protein in 6-month-old Tg-UCP1 hearts (Tg-6) relative to age-matched B6 hearts (C), and Tg-11 hearts (D), respectively (n = 4 hearts/group (C) and n = 3 hearts/group (D); the last lane of D is an additional B6 control). The lower panels show control immunoblots detecting Gnb. E and F, left ventricular ejection fraction (E) and cardiac ATP content (F) of 6-month-old Tg-UCP1 mice (Tg-6) relative to Tg-11 mice and B6 controls (±S.D.; n = 5; **, p < 0.01; ***, p < 0.001 versus B6 control; Dunnett's multiple comparison test). G, heart weight-to-body weight ratio of Tg-UCP1 relative to B6 mice (±S.D.; n = 5). H, representative histological section of a 6-month-old Tg-UCP1 heart relative to an age-matched B6 control (bar, 2 mm). Histology experiments are representative of four different mice/group. I, immunoblot detection of Fasn protein in cardiac tissue extracts from Tg-UCP1 relative to B6 mice (±S.D.; n = 4). The left panel shows a representative immunoblot detection for Fasn. J, increased oligomycin-insensitive OCR of Tg-UCP1 cardiomyocytes relative to B6 cardiomyocytes (±S.D.; n = 4 biological replicates; 3F protocol). The right panel shows baseline OCR values. K, FFA (left panel) and ATP contents (middle panel) of isolated neonatal cardiomyocytes from B6 mice after transfection without (−) or with (+) PPARG, PPARG-S273A, siUcp1, and treatment with rosiglitazone (+rosiglit.) as indicated (±S.D.; n = 4; *, p < 0.05; **, p < 0.01; ***, p < 0.001 versus control transfected with stealth control RNAi (−); Dunnett's multiple comparison test). The right panels show immunoblot detection of Ucp1 and PPARG.

Aged Tg-UCP1 mice from the Tg-6 line developed cardiac dysfunction with a significantly decreased left ventricular ejection fraction and decreased cardiac ATP content compared with non-transgenic B6 mice, whereas cardiac function parameters of the low-expressing Tg-11 line were not significantly different from B6 controls (Fig. 10, E and F). Concomitantly with cardiac dysfunction, Tg-UCP1 mice showed cardiac dilatation and loss of heart muscle, whereas the heart weight-to-body weight ratio was not significantly different from that in the B6 controls (Fig 10, G and H). Signs of heart failure were accompanied by a significant up-regulation of the cardiac Fasn protein in Tg-UCP1 mice compared with non-transgenic B6 controls (Fig. 10I). The up-regulation of the Fasn protein by UCP1 could be a consequence of the impaired cardiac function and insufficient tissue oxygen supply, which could induce Fasn up-regulation because it is a hypoxia-induced gene (39).

The ensuing increase in palmitate may enhance mitochondrial uncoupling by the transgenic UCP1 protein. In support of that notion, Tg-UCP1 cardiomyocytes showed a significantly increased oligomycin-insensitive (uncoupled) respiration compared with B6 cardiomyocytes (Fig. 10J). Together, these experiments indicate that the UCP1 protein could have a major role in the depressed substrate metabolism of a failing heart because up-regulated UCP1 could account for an enhanced palmitate-triggered uncoupled respiration upon FASN induction. Under those conditions, GRK2 inhibition by GRKInh could confer several modes of cardioprotection as follows: (i) by counteracting mitochondrial uncoupling via ERK activation-mediated down-regulation of UCP1 (36); and (ii) by decreasing the FASN-triggered lipid load, which involves, e.g. down-regulation of hypoxia-induced Pparg targets, Acaca and Fasn, and the resensitization of Prkaa.

Inhibition of Ucp1 Counteracts PPARG-S273A-induced Cardiolipotoxicity

The mechanism of ERK-mediated down-regulation of the Pparg target, Ucp1, is not completely understood (36, 47). Ucp1 is a Pparg target that is highly up-regulated in Tg-FASN and pressure overload-induced heart failure models, and by direct Pparg activation with rosiglitazone (cf. Fig. 3). Low ERK axis activity in these models could mediate stabilization of transcriptional cofactors that are necessary for Ucp1 induction, i.e. PGC1a or PRDM16 (36). Moreover, recent data have shown that PPARG-S273 dephosphorylation can also enhance the recruitment of PGC1a as a transcriptional cofactor involved in Ucp1 induction (46). To analyze the interplay between Ucp1 and PPARG-S273A-mediated cardiolipotoxicity, we inhibited Ucp1 by RNAi. The inhibition of Ucp1 led to an increased ATP content of cardiomyocytes with rosiglitazone-activated PPARG and with PPARG-S273A expression, while concomitantly decreasing the cardiomyocyte FFA content (Fig. 10K). These findings obtained with Ucp1 are analogous to the data obtained with Ucp2, which showed the following: (i) the Ucp2 knock-out decreased the cellular lipid content of pancreatic islets as a consequence of higher palmitate oxidation (48); (ii) Ucp1 knockdown attenuated the free fatty acid-induced apoptosis of cardiomyocytes (49); and (iii) conversely, Ucp2 overexpression decreased the cardiomyocyte ATP levels (50). Taken together, the data are compatible with the notion that Ucp1 exerts a detrimental role in cardiolipotoxicity, as triggered, for example, by Pparg activation with rosiglitazone and/or serine 273 dephosphorylation. However, the inhibition of Ucp1 could counteract such cardiotoxic effects.

Cardioprotective GRK2 Inhibition Requires an Intact ERK Axis

We further investigated the impact of ERK activation on cardioprotective GRK2 inhibition and applied the dual-specific GRK2 and ERK cascade inhibitor, RKIP (21). RKIP transgenic mice with myocardium-specific expression of the human RKIP (PEBP1) were previously generated and characterized (12). RKIP and GRKInh transgenic hearts display similar inhibition of GRK2 (12). Similarly, isolated neonatal cardiomyocytes of RKIP transgenic and GRKInh transgenic mice showed a comparable enhancement of the isoproterenol-stimulated cAMP response (Fig. 11A). This observation confirmed that the two different GRK2 inhibitors were expressed at equivalent levels with regard to the sensitization of the β-adrenergic receptor response. However, in contrast to GRKInh, transgenic hearts that expressed the dual-specific GRK2/ERK cascade inhibitor, RKIP, were characterized by a significantly decreased phosphorylation of Pparg on serine 273 (Fig. 11, B and C), which is ERK-dependent (36).

FIGURE 11.

Cardioprotective GRK2 inhibition requires an intact ERK axis. A, isoproterenol-stimulated (100 nm) cAMP levels of neonatal cardiomyocytes isolated from RKIP transgenic, GRKInh transgenic, and B6 mice (±S.D., n = 3; ***, p < 0.001 versus B6). B, immunoblot detection of pS273-Pparg (upper panel) and total Pparg (lower panel) in cardiac tissue extracts of RKIP transgenic mice relative to B6 controls. C, quantitative evaluation of immunoblot data (±S.D., n = 3; ***, p < 0.001 and **, p < 0.01 versus RKIP; and *, p < 0.05 versus B6). D, gene expression data showed up-regulation of the cardiac Pparg-dependent lipid metabolic process of 6-month-old Tg-RKIP mice compared with B6 controls (***, p < 0.001; **, p < 0.01, *, p < 0.05 versus B6). E and F, cardiac FFA (E) and triacylglycerol (TAG) (F) contents of RKIP transgenic relative to B6 control hearts (±S.D., n = 5; ***, p = 0.0006). G, Oil Red O detection of cardiac lipids in a cardiac section from an RKIP transgenic versus a B6 control heart (bar, 40 μm). Histology experiments are representative of four different mice/group. H, decreased left ventricular ejection fraction of RKIP transgenic mice indicates cardiac dysfunction (±S.D., n = 5; ***, p < 0.0001). I and J, real time measurement of OCR (I) and ECAR (J) of neonatal cardiomyocytes from RKIP transgenic mice was performed under basal conditions and after the creation of an adult-like lipogenic milieu by the 3F protocol. Neonatal cardiomyocytes from Tg-GRKInh mice were measured under basal conditions. The relative values normalized to the baseline (upper panels) and the absolute values of OCR and ECAR (lower panels) are presented. K–M, etomoxir (Etom.)-blocked fraction of OCR, which represents fatty acid β-oxidation (K), the 2-deoxyglucose (2-DG)-blocked fraction of ECAR, which represents glycolysis (L), and the ratio of glycolysis/β-oxidation (M) are also given. The data are shown as the means ± S.D.; n = 6 technical replicates (I and J) and n = 3 biological replicates (K–M); ***, p < 0.001 versus RKIP (K and M); **, p < 0.01 versus RKIP-3F (L) and RKIP (M); and *, p < 0.05 versus RKIP-3F (L and M).

We next determined the expression of cardiac Pparg targets of RKIP transgenic hearts because the inhibition of Pparg serine 273 phosphorylation induces Pparg target gene expression (36). Gene expression analysis revealed the significant up-regulation of heart failure-related Pparg targets (Fig 11D). Some of those highly up-regulated heart failure genes, such as adipsin (Adn), adiponectin (Adipoq), fatty acid synthase (Fasn), resistin (Retn), and uncoupling protein-1 (Ucp1), are reportedly induced by the inhibition of ERK and/or serine 273 dephosphorylation of Pparg (36, 40, 46), which was triggered by RKIP (cf. Fig. 11, B and C). In agreement with heart failure-related adipogenic target gene up-regulation, cardiac lipid load developed, and cardiac dysfunction became evident in hearts of aged RKIP transgenic mice (Fig. 11, E–H) (12). In vitro data documented the dysfunctional cardiomyocyte energetics of RKIP transgenic cardiomyocytes compared with the normal metabolism of GRKInh transgenic cardiomyocytes (Fig. 11, I–M). Taken together, our data strongly suggest that an intact ERK axis is required for GRK2 inhibition-dependent protection against dysfunctional metabolic substrate use.

GRK2 Inhibition Retards the Development of Heart Failure Signs, Cardiac Lipid Load, and Pparg Target Gene Induction in a Pressure Overload-induced Heart Failure Model

Thus, we have presented evidence for GRKInh-mediated protection of Tg-FASN hearts. However, up-regulation of the Pparg-dependent lipid metabolic process is also a characteristic feature of heart failure models that imitate major risk factors of patients such as chronic pressure overload (cf. Fig. 3B and Ref. 11). We therefore analyzed the effect of GRKInh on the cardiac lipid metabolism in a chronic pressure overload-induced heart failure model imposed by long term (6 months) AAC.

In agreement with previous data (12), AAC promoted cardiac hypertrophy with dilatation in non-transgenic mice, whereas Tg-GRKInh mice showed a significantly decreased cardiac hypertrophy (Fig. 12, A and B). The development of cardiac dysfunction upon AAC, as assessed by the left ventricular ejection fraction, was also significantly retarded in Tg-GRKInh mice (Fig. 12C). In addition to the improved cardiac function, the AAC-stimulated up-regulation of the cardiac Fasn protein was blunted in Tg-GRKInh mice (Fig. 12D). Concomitantly, Oil Red O staining of cardiac sections indicated that the AAC-triggered lipid load was lower in Tg-GRKInh mice (Fig. 12E). In agreement with the decreased lipid-induced cardiolipotoxicity, Tg-GRKInh mice showed a significantly decreased number of AAC-induced TUNEL-positive cardiomyocytes compared with the number in non-transgenic mice with AAC (Fig. 12, F and G).

FIGURE 12.

GRK2 inhibition retards the development of heart failure signs, cardiac lipid load, and Pparg target gene induction in a pressure overload-induced heart failure model. A, representative histological sections of hearts from a 10-month-old B6 mouse (AAC) relative to an age-matched GRKInh transgenic mouse (AAC+GRKInh) after 6 months of pressure overload imposed by AAC. The lower panels show age-matched sham-operated control hearts, bar, 2 mm. B and C, heart weight-to-body weight ratio (B), and the left ventricular ejection fraction (C) of 10-month-old GRKInh transgenic mice with 6 months of AAC (AAC+GRKInh) relative to age-matched non-transgenic B6 mice with 6 months of AAC (AAC). Age-matched sham-operated non-transgenic B6 mice (Sham) and sham-operated GRKInh transgenic mice (Sham+GRKInh) served as controls. Data are shown as the means ± S.D., n = 4 (*, p < 0.05 versus AAC+GRKInh; **, p < 0.01 versus AAC; ***, p < 0.001 versus AAC). D, immunoblot detection of Fasn in cardiac tissue extracts of 10-month-old B6 mice with 6 months of AAC relative to age-matched GRKInh transgenic mice with 6 months of AAC (n = 4 hearts/group, left blot). Under the experimental conditions, the Fasn protein (lane 1, A, positive control of a 10-month-old B6 heart with 6 months of AAC) was not detectable in cardiac tissue extracts from sham-operated B6 (sham) or GRKInh transgenic mice (n = 2; right blot). The lower panels show control immunoblots detecting Gnb. E, Oil Red O staining of cardiac sections from the different groups of mice (bar, 40 μm). Histology experiments are representative of four different mice/group (A and E). F and G, number of TUNEL-positive cardiomyocytes (±S.D.; n = 4; ***, p < 0.001 versus AAC). F shows representative immunohistological sections of TUNEL staining (bar, 20 μm). H and I, immunoblot (IB) detection of pS273-Pparg (upper panels) and total Pparg (lower panels) in cardiac tissue extracts from the different groups of mice (n = 4 hearts/group (H), and n = 2 hearts/group (I)). J, expression of heart failure-related Pparg targets in hearts from 10-month-old B6 mice with 6 months of AAC (AAC) and age-matched Tg-GRKInh mice with 6 months of AAC (AAC+GRKInh) relative to sham-operated controls (±S.D.; n = 4; ***, p < 0.001 versus AAC+GRKInh, Sham, and Sham+GRKInh).

In view of the decreased AAC-induced cardiolipotoxicity, we analyzed the potential effect of GRK2 inhibition on the Pparg-inhibitory serine 273 phosphorylation. Immunoblot detection indicated an increased cardiac content of serine 273-phosphorylated Pparg of Tg-GRKInh hearts with AAC compared with non-transgenic B6 mice with AAC (Fig. 12, H and I). The increased level of Pparg-inhibitory serine 273 phosphorylation was accompanied by a significantly lower expression of heart failure-related Pparg targets, which are blunted by ERK activation and/or ERK-dependent Pparg serine 273 phosphorylation, i.e. Ucp1, Adipoq, and Retn (Fig. 12J). Taken together, cardioprotective GRK2 inhibition with GRKInh retarded the up-regulation of heart failure-related and ERK-inhibited Pparg targets and enhanced Pparg-inhibitory serine 273 phosphorylation.

Down-regulation of Endogenous Ucp1 Retards the Development of Cardiac Dysfunction in a Pressure Overload-induced Heart Failure Model

Although previous studies have provided evidence for the involvement of Adipoq and Retn in heart failure pathogenesis of patients and animal models (44, 45, 51–53), the role of Ucp1 up-regulation in cardiac dysfunction is less clear. Notably, the onset of heart failure signs in different heart failure models was characterized by a strong cardiac Ucp1 up-regulation (cf. Fig. 3), and transgenic expression of UCP1 in the heart promoted signs of heart failure (cf. Fig. 10). To investigate the effect of Ucp1 up-regulation in the AAC-induced heart failure model, we down-regulated the endogenously expressed Ucp1 by lentiviral transduction of an miRNA that targets Ucp1 by RNA interference (Fig. 13, A and B). The down-regulation of Ucp1 in the AAC-induced heart failure model led to a small but significant retardation of the development of AAC-induced cardiac dysfunction (Fig. 13C). These findings provide further evidence (cf. Fig. 10) for the role of Ucp1 up-regulation in AAC-induced signs of heart failure.

FIGURE 13.

Down-regulation of endogenous Ucp1 retards the development of cardiac dysfunction in a pressure overload-induced heart failure model. A and B, endogenous Ucp1 expression (A) and Ucp1 protein level (B) in hearts of B6 mice with 2 months of AAC and transduction of a control lentivirus (AAC+miCont.) or a lentivirus targeting Ucp1 by RNAi (AAC+miUcp1) relative to sham-operated B6 controls. C, down-regulation of Ucp1 retarded the AAC-triggered decrease in the left ventricular ejection fraction (AAC+miUCP1) relative to AAC-subjected B6 mice transduced with a control lentivirus (AAC+miCont.). The data are shown as the means ± S.D. (n = 4; *, p < 0.05, and ***, p < 0.001 versus Sham B6; Dunnett's multiple comparison test). IB, immunoblot.

Low Efficacy of GRKInh in Retarding the Cardiac Phenotype of PPARG-S273A Transgenic Mice

Our data provided evidence that GRK2 inhibition counteracts the dysfunctional cardiac substrate use of heart failure (at least partially) by ERK-dependent inactivation of Pparg involving serine 273 phosphorylation. To further analyze the role of PPARG and a serine 273 phosphorylation-deficient PPARG mutant (PPARG-S273A) in the heart, we generated transgenic mice with myocardium-specific expression of wild-type PPARG and mutated PPARG-S273A, respectively (Fig. 14A). Histological analysis revealed that transgenic PPARG-S273A mice developed cardiac hypertrophy with dilatation, which was already evident in newborn mice (Fig. 14B). The dilatation of the PPARG-S273A-expressing heart was greater than that of the PPARG-expressing heart (Fig. 14B), although the cardiac PPARG protein level was comparable between the two transgenic groups (Fig. 14C).

FIGURE 14.

Low efficacy of GRKInh in retarding the cardiac phenotype of PPARG-S273A transgenic mice. A, transgenic vector used for the generation of transgenic mice with myocardium-specific expression of PPARG (and PPARG-S273A). B, cardiac sections from newborn transgenic mice that expressed PPARG-S273A or wild-type PPARG relative to a non-transgenic B6 control. Histological sections were stained with hematoxylin-eosin (H&E), and are representative of three mice/groups. C, detection of PPARG/Pparg in cardiac tissue extracts of 4-week-old mice with transgenic PPARG-S273A (S273A) expression or PPARG expression (as indicated) relative to non-transgenic B6 controls (n = 4 hearts/group). The lower panel is a control immunoblot (IB) detecting Gnb. D and E, cardiac expression of PPARG (D), and the heart weight-to-body weight ratio (E) of 4-week-old transgenic mice with expression of PPARG-S273A (S273A), PPARG-S273A+GRKInh, PPARG, and PPARG+GRKInh are shown relative to age-matched non-transgenic B6 mice (±S.D., n = 4; *, p < 0.05; ***, p < 0.001 versus B6 control, Dunnett's multiple comparison test). F, increased postnatal mortality of mice with myocardium-specific expression of PPARG-273A and PPARG-S273+GRKInh compared with mice expressing PPARG and PPARG+GRKInh, respectively. G, gene expression analysis of heart failure-related Pparg targets in hearts from 4-week-old transgenic mice that express PPARG-S273, PPARG-S273+GRKInh, PPARG, and PPARG+GRKInh relative to non-transgenic B6 hearts (±S.D., n = 4, *, p < 0.05, and ***, p < 0.001 versus B6 control, Dunnett's multiple comparison test). H, cardiomyocyte energetics was determined with neonatal cardiomyocytes isolated from transgenic mice with myocardium-specific expression of PPARG-S273A (S273A), PPARG-S273A+GRKInh, PPARG, PPARG+GRKInh, and non-transgenic B6 mice. The etomoxir-blocked fraction of OCR, which represents fatty acid β-oxidation (left panels), the 2-deoxyglucose-blocked fraction of ECAR, which represents glycolysis (middle panels), and the ratio of glycolysis/β-oxidation (right panels) were determined under basal conditions (upper panels) and after the creation of an adult-like lipogenic milieu by the 3F protocol for 10 days (lower panels). The data are shown as the means ± S.D.; n = 3 biological replicates; *, p < 0.05, **, p < 0.01, and ***, p < 0.001 versus B6; Dunnett's multiple comparison test.

Next, we investigated the effect of GRK2 inhibition by GRKInh and compared single transgenic mice (PPARG-S273A and PPARG), with double transgenic PPARG-S273A + GRKInh-expressing and PPARG + GRKInh-expressing mice, respectively. All four groups of mice showed similar cardiac PPARG expression (Fig. 14D). Despite having similar PPARG expression, the GRKInh largely prevented the cardiac hypertrophy of PPARG-expressing mice, whereas the effect of GRKInh on PPARG-S273A-expressing mice was not significant (Fig. 14E). In addition, there was an increased postnatal mortality of PPARG-S273A-expressing mice compared with wild-type PPARG-expressing mice (47.8.% versus 9.5%), which was not rescued by the GRK2 inhibitor (Fig. 14F).

We determined the expression of selected heart failure-related Pparg targets, i.e. Adipoq, Retn, Fasn, and Ucp1. The genes were all significantly up-regulated in the cardiac tissue of 4-week-old PPARG-S273A transgenic mice compared with non-transgenic B6 mice (Fig. 14G), which confirms the heart failure-like phenotype of newborn PPARG-S273A transgenic mice. We found that the effect of GRKInh was not significant in reducing the PPARG-S273A-mediated up-regulation of Adipoq, Retn, Fasn, and Ucp1 (Fig. 14G). In contrast to PPARG-S273A transgenic hearts, the transgenic expression of PPARG caused significantly less up-regulation of selected Pparg targets, and only the expression of cardiac Adipoq and Fasn was significantly increased in 4-week-old mice (Fig. 14G). The up-regulation of Adipoq in PPARG transgenic mice may have contributed to the cardiac hypertrophy (cf. Fig. 14, B and E) because Adipoq is required for pro-hypertrophic signaling during pressure overload (44).

In agreement with the heart failure-like phenotype of PPARG-S273A-expressing mice, the cardiomyocyte energetics of neonatal cardiomyocytes from PPARG-S273A transgenic mice showed an overall depressed substrate metabolism under basal conditions with predominant glycolysis. This heart failure-like substrate use was not rescued by co-expression of GRKInh (Fig. 14H). Conversely, PPARG-expressing cardiomyocytes showed predominant β-oxidation under basal conditions indicative of lipid load, which was retarded by GRKInh (Fig. 14H). Upon induction of an adult-like metabolism by the 3F protocol, PPARG-expressing cardiomyocytes shifted to a heart failure-like metabolic substrate use (Fig. 14H). In contrast to PPARG-S273A, the development of the PPARG-induced heart failure-like phenotype was retarded by GRKInh (Fig. 14H). Together these findings provide evidence for an enhanced cardiac deterioration of PPARG-S273A transgenic mice compared with PPARG-WT mice. Moreover, GRKInh was inefficient in retarding the development of the PPARG-S273A-induced cardiometabolic dysfunction and the up-regulation of PPARG-S273A-regulated targets.

Discussion

In this study, we investigated whether GRK2 inhibition could be a specific approach for targeting of the dysfunctional cardiac substrate metabolism, which is characteristic of late-stage heart failure (4–6). To reproduce the dysfunctional cardiac substrate use, we generated a novel transgenic model with myocardium-specific FASN expression. The model imitated the up-regulation of FASN, which occurs in patients with heart failure (10, 11). In the context of cardiovascular disease and heart failure, the up-regulation of FASN could be a direct consequence of decreased cardiac output and insufficient oxygen supply because FASN is a hypoxia-induced gene (39). Because cardiac ischemia is triggered by major cardiovascular risk factors such as pressure overload and atherosclerosis, up-regulation of FASN could also be an early and causative event in the pathogenesis of heart failure. In agreement with this notion, we found that the sole expression of FASN was sufficient to trigger the signs of heart failure. Additionally, models of heart failure, which imitated cardiovascular risk factors of patients such as chronic pressure overload or advanced atherosclerosis, also showed up-regulation of cardiac Fasn (11).

How could FASN advance the symptoms of heart failure? Initially, up-regulation of FASN might be considered to be beneficial by supplying more energy substrate to the heart muscle. However, in the long term, the uncontrolled accumulation of palmitate as the major lipid synthesized by FASN could activate the heart failure-promoting transcription factor Pparg, as has been documented by up-regulation of the Pparg-dependent lipid metabolic process in Tg-FASN hearts and various other models of heart failure. Because Fasn is also a Pparg target, a vicious cycle of FASN/Fasn-induced Fasn could be triggered, which finally results in cardiotoxic lipid load, dysfunctional substrate use, and mitochondrial uncoupling due to palmitate-triggered activation of Ucp1 (Fig. 15). The accumulation of palmitate further promotes pro-apoptotic signaling and inhibits the pro-survival ERK axis (16). As a result, there is an enhanced expression of heart failure-associated Pparg targets that are triggered by ERK inhibition (36, 40), such as Adipoq (44, 51, 52), Retn (45, 53), and Ucp1. Ensuing cardiomyocyte death and remodeling and impaired cardiac energy generation due to mitochondrial uncoupling could further aggravate the symptoms of heart failure (Fig. 15).

FIGURE 15.

Scheme illustrates cardioprotective effects of GRK2 inhibition, which target the cardiometabolic dysfunction of late-stage heart failure. The histology image presents a cardiac section from an 8-month-old ApoE^−/− mouse with signs of heart failure induced by 2 months of Pparg activation with rosiglitazone.

When we applied the Tg-FASN mice as a model of a dysfunctional cardiac substrate metabolism and an additional pressure overload-induced heart failure model, we found that GRK2 inhibition directly interfered with the cardiac lipid accumulation and mediated a reduction in the cardiac Fasn protein. These activities could be attributed (at least partially) to several mechanisms as follows: (i) the inhibition of the endogenous Fasn up-regulation, a hypoxia-induced Pparg target (39); (ii) an interference with fatty acid synthesis by preventing Acaca up-regulation, which is also a Pparg-regulated gene induced by hypoxia (38); and (iii) the enhancement of the fatty acid metabolism by re-sensitization of Adipor1 and Prkaa-mediated signaling (14). As a consequence, GRK2 inhibition retarded the development of the dysfunctional cardiac substrate use characteristic of late-stage heart failure (Fig. 15).

Cardioprotective GRK2 inhibition required an intact ERK axis to preserve the cardiac energetics because RKIP as a dual-specific GRK2 and ERK cascade inhibitor promoted dysfunction of cardiomyocyte energetics, cardiac lipid load, and signs of heart failure. Concomitantly, inhibition of the ERK cascade by human RKIP was accompanied by decreased ERK-dependent phosphorylation of Pparg. A decreased ERK-dependent phosphorylation of Pparg on serine 273 and serine 112 is known to enhance Pparg activity and/or increase Pparg target gene induction (36, 40, 43). Similarly, heart failure-related Pparg targets are triggered by RKIP, resulting in development of cardiac lipid load and cardiac dysfunction (12).

Conversely, GRK2 inhibition by GRKInh led to an increased ERK activation and enhanced the ERK-mediated phosphorylation of Pparg on serine 273. ERK axis activation could be part of the cardioprotective gene expression program initiated by GRK2 inhibition (12, 13, 41). Concomitantly, the expression of heart failure-promoting Pparg targets was blunted, and the appearance of the dysfunctional cardiac substrate metabolism was retarded. In agreement with a causal role of PPARG-S273 phosphorylation in GRKInh-mediated cardioprotection, the phosphorylation-deficient PPARG-S273A mutant promoted dysfunction of the cardiomyocyte substrate metabolism and caused enhanced postnatal death, which was largely insensitive to GRKInh. In contrast, the phenotype of wild-type PPARG was less severe and could be (at least partially) rescued by GRK2 inhibition. Together, these data indicate that the ERK axis may specifically counteract the heart failure-promoting transcription factor Pparg by preventing heart failure-related Pparg target gene induction (36, 54) and/or could confer protection against palmitate-induced endoplasmic reticulum stress (55).

Several heart failure-related Pparg targets are inhibited by ERK-dependent Pparg inactivation (36, 40) with GRKInh, such as Adipoq (44, 51, 52) and Retn (45, 53). By generating Tg-UCP1 mice with myocardium-specific UCP1 expression, our study identified UCP1 as a heart failure-related ERK-regulated Pparg target (36, 47), which was also down-regulated upon GRK2 inhibition (Fig. 15). Consequently, GRK2 inhibition could decrease excessive mitochondrial uncoupling as a key event that contributes to inefficient cardiac ATP generation and lipid-induced cardiomyocyte death in heart failure (49, 50).

Although the study was performed with experimental mouse models, the data could also be relevant for the human disease because FASN up-regulation is a characteristic feature of patients with heart failure (10, 11). Because PPARG up-regulation occurs in heart failure patients with pressure-overloaded heart and metabolic syndrome (56), GRK2 inhibition is expected to disrupt a vicious FASN/PPARG cycle in patients who suffer from multiple risk factors (Fig. 15). Such a situation was modeled with rosiglitazone-treated ApoE^−/− mice because these mice are prone to the development of atherosclerosis and insulin resistance (57), and thereby mimic the risk profile of patients with enhanced PPARG activation and cardiovascular disease. In this model, Pparg activation triggered the up-regulation of Fasn and signs of heart failure within 2 months. The causal interplay between Fasn and Pparg-induced cardiotoxicity was demonstrated by RNAi-mediated inhibition of Fasn, which prevented the Pparg-induced cardio-lipotoxicity and signs of heart failure. Because GRK2 inhibition also mediated the down-regulation of FASN-dependent cardio-lipotoxicity, patients with high morbidity and multiple risk factors may benefit from GRK2 inhibition. The additional insulin sensitivity-enhancing activity of GRK2 inhibition (13, 58) may further increase the value of such a strategy.

Moreover, heart-specific GRK2 inhibition could become a cardioprotective combination partner for a novel class of insulin sensitivity-enhancing PPARG activators, which rely on the inhibition of ERK-dependent PPARG phosphorylation for antidiabetic activity (36), but they have promoted signs of heart failure in clinical trials (59).

In sum, our study provides strong evidence that cardioprotective GRK2 inhibition specifically targets the dysfunctional cardiac substrate use that is a symptom of late-stage heart failure (Fig. 15). The identified targeting approach could stimulate the development of new therapeutic strategies.

Author Contributions

J. A., M. G., and U. Q. analyzed the data and produced all of the figures. J. A. and X. F. generated the transgenic mice. J. A. and U. Q. wrote the main manuscript text. All of the authors reviewed the results and approved the final version of the manuscript.