Cardiomyocyte expression of PPARγ leads to cardiac dysfunction in mice

Creation of cardiac-specific PPARγ1–transgenic mice.

Mice overexpressing PPARγ in cardiomyocytes via the α-MHC promoter were designated MHC-PPARγ1 mice (Figure 1A). PCR screening of the offspring identified 3 founders harboring the MHC-PPARγ1 transgene. Offspring from 2 of these founders, designated MHC-PPARγ1L (low-expressing) and MHC-PPARγ1H (high-expressing), were used in this report. Male mice were used in the experiments unless indicated otherwise. The PPARγ1 transgene was expressed specifically in the heart; no expression was detected in the liver, skeletal muscle, or fat (Figure 1B). In control mouse hearts, PPARγ1 mRNA levels were low; the ratio of PPARγ to 18S rRNA was much lower in control heart than adipose tissue. In contrast, MHC-PPARγ1L mice had PPARγ to 18S rRNA levels approximately twice those found in adipose tissue (Figure 1C). MHC-PPARγ1H hearts had PPARγ mRNA levels approximately 10-fold higher than those in MHC-PPARγ1L hearts. PPARγ protein levels were 1.9-fold (MHC-PPARγ1L) and 7.2-fold (MHC-PPARγ1H) higher in the cardiac ventricles of transgenic mice than littermate controls (Figure 1, D and E).

Transgenic mice from each of the lines bred normally, and glucose, body weight, and plasma lipid levels were similar to those of nontransgenic mice (Supplemental Table 1; supplemental material available online with this article; doi:10.1172/JCI30335DS1).

PPARγ activity in hearts of MHC-PPARγ1 mice.

We estimated the degree of PPARγ activity in the MHC-PPARγ1 mice by assessing expression of downstream genes. Surprisingly, despite the adipose-like PPARγ1 expression level, hearts from young MHC-PPARγ1L mice (4 months old) had an expression of downstream genes similar to that of controls (Figure 2A). Eight-month-old MHC-PPARγ1L mice, however, had greater expression of acyl-CoA oxidase (AOX) (1.35-fold), carnitine palmitoyl transferase–1 (CPT1) (1.63-fold), FA translocase (CD36) (1.61-fold), FA synthase (FAS) (1.44-fold), and adipose differentiation–related protein (ADRP) (1.81-fold) (Figure 2B).

In contrast, multiple genes were markedly upregulated in the hearts of 4-month-old MHC-PPARγ1H mice (Figure 2C), including genes involved in lipid oxidation, uptake, synthesis, and storage, such as AOX, CPT1, CD36, FAS, ADRP, and SREBP1. Expression of ATP-binding cassette transport A1 (ABCA1), which mediates reverse cholesterol transport to apoA-I and is induced by PPARγ (18), was 1.73-fold higher in 4-month-old MHC-PPARγ1 H mice (Figure 2C).

PPAR gene expression in control and MHC-PPARγ mice with aging.

To determine why the MHC-PPARγL transgene did not alter downstream gene expression in the young mice, we assessed expression of PPAR family members in control and transgenic mice with aging. As noted by others (19), PPARα expression decreased as the mice aged (Supplemental Figure 1A). Moreover, expression of PPARα was reduced by 29% and 60% in the 2- and 9-month-old MHC-PPARγL transgenic mice, respectively, compared with their littermate controls. Neither endogenous and transgenic PPARγ1 nor PPARδ expression differed between 2- and 9-month-old mouse hearts (Supplemental Figure 1, B and C). Thus, changes in PPARα expression might modify the downstream effects of the MHC-PPARγL transgene.

Cardiac lipid content and uptake in MHC-PPARγ1 mice.

Greater CD36 expression in the MHC-PPARγ1 mice could increase uptake of plasma free and lipoprotein-derived FA. Heart TG and FA content were increased in 8-month-old MHC-PPARγ1L mice (Figure 3A). Oil red O staining revealed greater accumulation of intracellular neutral lipids in hearts from MHC-PPARγ1L mice compared with littermate controls (Figure 3, B and C). Electron microscopy showed more lipid droplets within the sarcoplasm of cardiomyocytes, with distortion of the mitochondrial contours in MHC-PPARγ1L mice (Figure 3D). In some areas the cristae were disrupted (Figure 3E).

Cardiac uptake of VLDL-TG in 8-month-old MHC-PPARγ1L hearts was 74% greater than that in littermate control hearts (P < 0.01; Figure 3F), which corresponds with increased expression of lipid uptake genes and cardiac lipid storage. Liver and skeletal muscle lipid uptake were not altered by the PPARγ1 transgene (data not shown).

Glucose transporter 4 expression and glucose uptake in MHC-PPARγ1 mice.

Unlike in MHC-PPARα transgenic mice (8), glucose transporter 4 (GLUT4) and GLUT1 mRNA levels were unchanged in the MHC-PPARγ1L mice (data not shown) and were upregulated in MHC-PPARγ1H mice (Table 1). Pyruvate dehydrogenase kinase 4, the gene regulating glucose oxidation, was also downregulated in the MHC-PPARγ1H mice (Table 1).

The GLUT4 gene undergoes a complex program of gene regulation. The GLUT4–myocyte enhancer factor 2 (GLUT4-MEF2) site is required for metabolic regulation of GLUT4 transcription (20–22). MEF2A, MEF2C, and MEF2D isoforms, but not MEF2B isoforms, are expressed in the heart. PPARγ coactivator 1 (PGC1) also interacts with MEF2C to upregulate endogenous GLUT4 expression and glucose uptake in cultured muscle cells (23). GLUT4 expression in MHC-PPARγ1H mice was associated with increased expression of MEF2C and MEF2D (1.48-fold, P < 0.001, and 1.55-fold, P = 0.009, respectively), but with no change in MEF2A. PGC1β gene expression was also upregulated (1.68-fold; P = 0.01) in the hearts of the MHC-PPARγ1H mice (Table 1).

Myocardial glucose import in vivo was assessed using 2-deoxy-d-[³H]glucose. Heart uptake of glucose was unchanged in the MHC-PPARγ1L hearts (Figure 4A, left panel) and was increased by 37% in MHC-PPARγ1H mice compared with littermate controls (P < 0.05; Figure 4A, right panel). The PAS technique revealed greater cardiac glycogen storage in MHC-PPARγ1L mice compared with littermate controls (Figure 4B). PAS diastase (PAS-D) completely removed the glycogen in cardiac sections (Figure 4C). These data indicate that in the MHC-PPARγ mice, although cardiac lipid uptake was elevated, glucose uptake was not reduced.

Cardiac function in MHC-PPARγ1 mice.

Both MHC-PPARγ1L and MHC-PPARγ1H hearts had dilated left ventricles and impaired systolic function. Heart to body weight ratios were increased in 8-month-old MHC-PPARγ1L and 4-month-old MHC-PPARγ1H mice (Figure 5, A and B). The dilated cardiomyopathy phenotype was grossly visible (Figure 5, C and D) and also seen by echocardiography (Figure 5, E–J). Left ventricular systolic dimension was increased (0.17 ± 0.03 versus 0.14 ± 0.01 cm; P = 0.021), and fractional shortening was reduced from 53% ± 3.6% to 48% ± 5.2% (P = 0.036) by age 8 months in PPARγ1L mice (Figure 5, F and G). The MHC-PPARγ1H mice had more severe cardiac dysfunction. Four-month-old PPARγ1H hearts had greater left ventricular systolic dimension (0.28 ± 0.02 versus 0.17 ± 0.03; P < 0.001) and a greater reduction in fractional shortening (from 47.3% ± 5.3% to 30.2% ± 3.2%; P < 0.001) (Figure 5, I and J).

Expression of heart failure marker genes brain-type natriuretic peptide (BNP) and atrial natriuretic factor (ANF), was increased in MHC-PPARγ1 mice (Table 2). These increases were seen by 4 months in the PPARγ1H line but not until 8 months in the PPARγ1L line.

Effects of rosiglitazone in wild-type and MHC-PPARγ1 mice.

In contrast to other lipotoxic heart models (13, 16, 24), rosiglitazone treatment of 8-month-old MHC-PPARγ1L mice led to further deterioration of cardiac function: increased lipid accumulation, larger hearts, and decreased fractional shortening (Figure 6, A and B). This was associated with increased cardiac gene expression of CD36 and ADRP in rosiglitazone-treated MHC-PPARγ1L mice (Figure 6C). When wild-type mice were treated with rosiglitazone, cardiac expression of usual PPARγ downstream genes was paradoxically reduced in the 8-month-old mice (Figure 6C). Adipose tissue and muscle had increased CD36 expression after rosiglitazone treatment (Supplemental Figure 2).

Pathways leading to cardiac dysfunction in MHC-PPARγ1 mice.

To explore potential mechanisms in the development of lipid-induced cardiomyopathy, we examined heart tissue for evidence of lipid accumulation triggering programmed cell death. Four-month-old MHC-PPARγ1H mice had a 40% increase in levels of cardiac ceramide, a lipid capable of inducing apoptosis (25), compared with littermate controls (Figure 7A). This was associated with increased expression of long chain base 1 (LCB1) and LCB2, the subunits of serine palmitoyl-CoA transferase. Expression of apoptosis-related genes — B cell leukemia/lymphoma 2 (Bcl2), Bcl-2-like protein (Bcl-XL), BCL2-associated X protein (Bax), and caspase-9 — also dramatically increased (Figure 7B). The CCAAT/enhancer-binding protein (C/EBP) homologous protein (CHOP), one of the regulators of ER stress–mediated cell death, was upregulated in the MHC-PPARγ1H mice. iNOS was unchanged. There was evidence of positive TUNEL staining of heart tissue in both MHC-PPARγ1L and MHC-PPARγ1H, but not control, hearts (Figure 7, C and D).

PPARγ1 expression in humans and streptozotocin-induced diabetic mouse hearts.

We then tested whether PPARγ expression could be induced in the hearts of diabetic mice. Hearts of diabetic mice had 2.1-fold higher expression of PPARγ than those of control mice (P = 0.003) (Figure 8A). We also assessed PPARγ expression in hearts from humans. Human hearts had 8.1- to 14.5-fold higher PPARγ mRNA levels than mouse hearts (Figure 8B). These data suggest that PPARγ is more functional in human hearts.

Generation and identification of the MHC-PPARγ mice.

Creation of the mice and all metabolic and genetic studies were reviewed and approved by the Columbia University Institutional Animal Care and Use Committee. A transgenic construct containing a 1.7-kb mouse PPARγ1 cDNA, which differs from PPARγ2 by a lack of 30 additional NH₂-terminal amino acids, was cloned downstream of the α-MHC promoter (5.4 kb). Sac1 and Nae1 digestion of MHC-PPARγ1 produced a linear 7.8-kb fragment that was used for microinjection (Figure 1A). Transgenic mice were produced by microinjection of the MHC-PPARγ1 construct into fertilized 1-cell C57BL/6 × CBA F₁ eggs. Founder mice and transgenic expression of PPARγ1 were identified by analysis of genomic DNA with primer A (5ι-AGCTGTGGTCCACATTCTTC-3ι; a sense primer specific to MHC promoter exon 2) and antisense primer (5ι-AGAATGGCATCTCTGTGTC-3ι) specific to PPARγ1 cDNA nucleotides 173–192. PPARγ1 mRNA expression (both transgene and endogenous) in the heart was confirmed by RT-PCR with PPARγ1-specific primer B, sense: 5ι-GAGTGTGACGACAAGATTTG-3ι, and antisense: 5ι-GGTGGGCCAGAATGGCATCT-3ι. To compare heart PPARγ expression with that of adipose tissue, we measured heart and adipose tissue PPARγ expression simultaneously using primers that were common to PPARγ1 and PPARγ2 (primer C, sense: 5ι-ATCTACACGATGCTGGC-3ι, and antisense: 5ι-GGATGTCCTCGATGGG-3ι) (Figure 1A).

Western blot analysis.

Fresh heart tissues from 4- and 8-month-old MHC-PPARγ1L and 4-month-old MHC-PPARγ1H mice and their littermate controls were prepared and nuclear proteins were isolated using a nuclear extraction kit according to the manufacturer’s instructions (Active Motif). Thirty micrograms of nuclear proteins were subjected to Western blot analysis with PPARγ antibody (sc-7196; 1:100 dilution; Santa Cruz Biotechnology Inc.). For control of protein loading, the blots were stripped and reacted with anti–lamin B1 antibody (gift from Howard Worman, Columbia University). Bands were quantified by densitometry using Molecular Analysis Software (Bio-Rad).

Heart and plasma lipids.

Blood from fasted (6 hours) mice was collected from retro-orbital plexus for the measurement of plasma total cholesterol (TC), TG, and FFAs. To measure tissue lipids, hearts were perfused with PBS and homogenized in 4°C in 1 M NaCl buffer containing lipase inhibitors to prevent TG hydrolysis. Lipids were extracted from heart tissues (50 mg) according to methods modified from that of Folch et al. (50). The dried lipids were solubilized in PBS containing 2% Triton X-100. Heart and plasma TC, TG, and FFA were measured enzymatically using an Infinity kit (Thermo Electron Corp.) and a NEFA C kit (Wako).

Tissue gene expression.

Total RNA was prepared using a Pure Link Micro-to-Midi Total Purification System kit (Invitrogen). One microgram of RNA was initially treated with DNase I (Invitrogen) for 15 minutes. The RNA samples were then reverse transcribed using the ThermoScript RT-PCR Kit (Invitrogen). Real-time quantitative RT-PCR (qRT-PCR) was performed using an ABI 7700 (Applied Biosystems). Amplification was performed using SYBR Green PCR Master Mix (Applied Biosystems). Primers used for PCR amplification are listed in Supplemental Tables 2 and 3. Analysis was performed using Sequence Detection Software (Applied Biosciences). Standard curves were generated using undiluted and diluted (1:10, 1:100, and 1:1,000) cDNA samples from heart tissue. Correlation coefficients were 0.98 or greater. Data were normalized with 18S rRNA.

Histological analysis.

Neutral lipids were assessed in hearts taken from 24-hour-fasted male and female mice perfused with PBS. The hearts were embedded in Tissue-Tek Optimal Cutting Temperature compound (Sakura). Midventricular sections of myocardium (6 μM in thickness) were stained with oil red O and counterstained with hematoxylin.

PAS and PAS-D staining were used to demonstrate heart glycogen. Hearts from 6-hour-fasted male and female mice were placed in 10% neutral buffered formalin. Midventricular sections were fixed in methyl alcohol for 10 minutes and stained with Schiff Reagent (Poly Scientific) and H&E. The specificity of glycogen staining was demonstrated by treating heart sections with diastase of malt (Sigma-Aldrich), which removes the glycogen.

TUNEL staining.

Cardiac ventricular tissues from 4- and 8-month-old MHC-PPARγ1L and 4-month-old MHC-PPARγ1H were fixed in formalin, embedded in paraffin, and sectioned. Tissues were stained for DNA fragmentation by a TUNEL protocol according to the manufacturer’s specifications (R&D Systems). The data were quantified by counting TUNEL-positive myocytes per square millimeter cellular area.

Echocardiographic analysis.

Two-dimensional echocardiography was performed in conscious mice using techniques described previously (Sonos 5500 system; Philips Medical Systems) (10). Two-dimensional echocardiographic images were obtained and recorded in a digital format. Images were than analyzed off-line by a researcher blinded to the murine genotype. Left ventricular end-diastolic dimension (LVDd) and left ventricular end-systolic dimension (LVDs) were measured. Percent fractional shortening was calculated as: % FS = ([LVDd — LVDs] / LVDd) × 100.

Electron microscopy.

Left ventricles from 8-month-old mice were fixed with 2.5% glutaraldehyde in 0.1 M Sorensen’s buffer (0.2 M monobasic phosphate/0.2 M dibasic phosphate 1:4 vol/vol; pH 7.2), postfixed in osmium tetroxide, and embedded in EPON 812 (Electron Microscopy Sciences). Ultrathin sections were stained with uranyl acetate and lead citrate and examined under a JEM-1200ExII electron microscope (JEOL).

Uptake of VLDL and glucose.

Human VLDL was isolated from normal subjects by sequential ultracentrifugation and was labeled with [carboxyl-¹⁴C]triolein (PerkinElmer) via cholesterol ester transfer protein (CETP) as previously described (51). Sixteen-hour-fasted MHC-PPARγ1 mice and littermates were first injected intravenously with 1 × 10⁶ decays per minute (DPM) of 2-deoxy-d-[³H]glucose (PerkinElmer). Fifty-five minutes after 2-deoxy-d-[³H]glucose injection, 1 × 10⁶ DPM of [¹⁴C-TG]VLDL was injected. Five minutes after VLDL injection, blood was collected, and the vasculature was thoroughly perfused with 10 ml of PBS via cardiac puncture. Tissues were then excised, and accumulated radioactivity for [³H]glucose and [¹⁴C-TG]VLDL was measured. Amounts of glucose and VLDL injected were adjusted by plasma radioactivity counts at 30 seconds after each injection and were compared with plasma counts at the end of the experiments. Tissue uptake for transgenic mice was compared with uptake of control littermate controls.

Ceramide content.

Cardiac ceramide levels were determined using the diacylglyceride kinase method as described previously (52).

Human tissues.

Normal human heart tissues were obtained from National Disease Research Interchange. Hearts from 11 orthotopic heart transplant recipients (50.3 ± 8.8 years old) with idiopathic dilated cardiomyopathy were obtained from the Department of Surgery at Columbia University Medical Center (CUMC). The Institutional Review Board at CUMC approved all study protocols. Normal human left ventricle total RNA was purchased from Ambion (no. 5856). Human cardiac PPARγ1 mRNA expression was measured by qRT-PCR and normalized with human 18S rRNA. Primer sequences used for qRT-PCR were shared by both mouse and human genes (Supplemental Table 3). BLAST (http://www.ncbi.nlm.nih.gov/blast/) analysis showed that these primers were 97% and 90% homologous between human and mouse PPARγ1 mRNA and 18S rRNA, respectively.

Rosiglitazone treatment.

Four- and 8-month-old mice were fed rosiglitazone-containing chow for 15 days. The chow was produced by Research Diets and contained 200 mg/kg of drug; this is equivalent to a dose of 10 mg/kg/d in an adult mouse that consumes on average 4.5 g/d of chow. These doses were similar to those used by other investigators (53).

Statistics.

We analyzed data using the Prism software package (GraphPad Software). Comparisons between 2 groups were performed using unpaired 2-tailed Student’s t tests. All values are presented as mean ± SD. Differences between groups were considered statistically significant at P < 0.05.