Endogenously produced adiponectin protects cardiomyocytes from hypertrophy by a PPARγ-dependent autocrine mechanism

Abstract

In experimental animal and cell culture models, activation of peroxisome proliferator-activated receptor (PPAR) γ in heart has been shown to have beneficial effects on cardiac function and cardiomyocyte physiology. The goal of this study was to identify the signaling pathway by which PPARγ activation protects cardiomyocytes from the deleterious effects of hypertrophic stimuli. In primary cardiomyocyte cultures, we found that genetic or pharmacological activation of PPARγ protected cells from cardiac hypertrophy induced by α-adrenergic stimulation. Examination of gene expression in these cells revealed a surprising increase in the expression of adiponectin in cardiomyocytes and secretion of the high-molecular-weight form of the hormone into media. Using RNAi to block PPARγ-induced adiponectin production or adiponectin receptor gene expression, we found that the PPARγ-mediated anti-hypertrophic effect required cardiomyocyte-produced adiponectin, as well as an intact adiponectin signaling pathway. Furthermore, mice expressing constitutive-active PPARγ and cardiomyocyte specific adiponectin expression were protected from high-fat diet-induced cardiac hypertrophy and remodeling. These findings demonstrate that functional adiponectin hormone can be produced from the heart and raise the possibility that beneficial effects of PPARγ activation in heart could be due in part to local production of adiponectin that acts on cardiomyocytes in an autocrine manner.

cardiac hypertrophy is an early manifestation in many cardiovascular diseases, including hypertension, myocardial infarction, hypertrophic cardiomyopathy, and diabetic cardiomyopathy (4, 36). The beneficial effects of the thiazolidinedione PPARγ activators (TZDs) on cardiomyocyte function have been observed in multiple experimental models. TZDs have been shown to prevent adrenergic-stimulated cardiac hypertrophy in isolated hearts and neonatal cardiomyocytes (2, 10, 34). In animal models of obesity-related diabetes, they attenuate many of the deleterious effects of diabetes on the heart (18, 33, 37). Although the cardiovascular benefit of TZDs in humans is complicated by their propensity to cause plasma volume expansion (20) and recent reports that rosiglitazone increases the risk of myocardial infarction (21), the overall body of evidence suggests that activation of peroxisome proliferator-activated receptor (PPAR) γ in cardiomyocytes has direct beneficial effects on cardiac function. However, the mechanism by which PPARγ activation protects cardiomyocytes and the heart against adrenergic-stimulated hypertrophy remains to be elucidated.

Adiponectin is an adipocyte-derived cytokine that has also been shown to ameliorate pressure-induced and adrenergic-stimulated cardiac hypertrophy in vivo and in vitro via an AMP-activated protein kinase (AMPK) signaling pathway (27, 28). These observations suggest that adiponectin plays an important role in regulating cardiac metabolism. Using two different experimental systems, TZD-treated primary cardiomyocyte cultures and a transgenic mouse model that expresses a constitutive-active version of PPARγ in heart, we found that PPARγ activation in cardiomyocytes induced the expression and production of functional adiponectin directly from cardiomyocytes. We also show that the full cardioprotective effect of PPARγ activation in isolated cardiomyocytes requires not only the endogenous production of adiponectin but the presence of the adiponectin receptors AdipoR1 and AdipoR2. Together, our findings suggest that the heart contains a PPARγ-dependent autocrine signaling loop generated by the endogenous production of adiponectin and that this previously unrecognized pathway may play a role in mediating the cardiovascular effects of the TZDs and other PPARγ activators.

MATERIALS AND METHODS

Animals

The transgenic mouse line expressing a constitutively active PPARγ (CA-PPARγ) in muscle and heart has been previously described (1). Briefly, the CA-PPARγ cDNA was introduced into an expression vector under the control of the human skeletal actin promoter. This construct was used to generate a transgenic mouse line in C57BL/6JxCBA hybrid mice. Transgenic mice were backcrossed into C57BL/6J mice for eight generations, and subsequently carried as homozygotes for the transgene locus. These transgenic mice expressed CA-PPARγ transgene mRNA exclusively in skeletal muscle and in heart. Male mice were ad libitum fed a high-fat diet containing 40% calories from fat (Research Diets no. D12451) or normal chow (LabDiet Rodent Chow 5001) for 10 wk starting at 8 wk of age. Aged-matched nontransgenic C57BL/6J male mice were used as controls. All animal procedures were reviewed and approved by the Wayne State University Institutional Animal Care and Use Committee.

Cardiomyocyte Cultures

Neonatal ventricular cardiomyocytes were prepared from either Long Evans rat pups or from transgenic and control mouse pups (1–3 days in age) essentially as described (23). This procedure results in cardiomyocyte cultures with minimal contamination with noncardiomyocytes such as fibroblasts. Whole neonatal hearts were dissected, cut into small pieces with a razor blade, and digested with trypsin (50 μg/ml) overnight at 4°C in Hanks' buffered saline solution. The following day trypsin was removed, and cells were further digested with type-2 collagenase (1 mg/ml for 30 min) in Leibovitz L-15 buffer while being gently triturated with a 10-ml serological plastic pipette at 37°C. The disrupted tissue was then filtered through a 60-μm pore size cell strainer. Cells were collected from the filtrate centrifugation at 50 g for 5 min at 4°C, which pellets cardiomyocytes but not red blood cells. Purification of cardiomyocytes from other cell types was carried out by three short serial preplatings in 150-mm dishes of 30 min each in serum-free DMEM (4.5 g/l glucose). This is enough time to allow fibroblasts, macrophages, and other cell types to adhere to the surface of the dish, but not cardiomyocytes, which remain suspended and are transferred to the next plate in the gently collected media. After the third preplating, cardiomyocytes were collected by centrifugation at 800 g for 15 min and plated in growth media (DMEM with 4.5 g/l glucose and 10% FBS) in 35-mm dishes at 2.0 × 10⁶ to 2.5 × 10⁶ cells/plate.

After 24 h of undisturbed cell growth, the plates were washed with serum-free media to remove dead cells, and subsequent cultures were maintained in serum-free DMEM with 4.5 g/l glucose and used in experiments. The purity of the resulting cardiomyocyte cultures was validated by immunocytochemistry for myosin heavy chain (MHC)-α, which is specific for cardiomyocytes. By this criteria, all of the cultures used in experiments were >90% cardiomyocytes. During experiments, primary cardiomyocytes were maintained in serum-free DMEM with 4.5 g/l glucose, and the media was changed every other day to prevent depletion of glucose and the inadvertent activation of AMPK due to reduced cellular energy availability. ITS (insulin/transferrin/selenium) supplement was not used in any of our cell culture media.

In some experiments, cardiomyocyte cultures were treated with rosiglitazone for 48 h and then harvested. A 10 μM concentration of rosiglitazone has been used previously on cultured cardiomyocytes without induction of apoptosis (2, 3, 16), and no cell death was observed in our cultures. For the hypertrophy studies, the cardiomyocytes were first transfected with small-interfering RNA (200 nM adiponectin-specific Smart-pool RNAi mixture from Dharmacon using DharmaFECT-1 in serum-free DMEM) for 24 h or the short-hairpin RNA (shRNA)-based lentivirus medium (containing 2 μg/ml Polybrene), where the titer was ∼4 × 10⁸ infectious units (IFU) per milliliter. Cells were transferred to fresh serum-free media, and rosiglitazone (or vehicle) was added, followed 24 h later by α-adrenergic stimuli (2 μM propranolol) and 30 min later by addition of norepinephrine (1 μM). Cells were harvested 48 h later and examined by Western blot, quantitative RT-PCR, and immunocytochemistry as described. For experiments in which Erk1/2 activity was measured, cells were treated as described above, but with α-adrenergic stimulation for 1 h.

PCR

Quantitative RT-PCR.

RNA was isolated from hearts or cultured cardiomyocytes using Trizol-LS reagent according to the manufacturer's protocol (Invitrogen). Two-step qRT-PCR was carried out using the iScript cDNA synthesis kit (Bio-Rad Laboratories) and Absolute QPCR SYBER Green Mix (ABgene), with 200 ng of total cDNA in a MX3000P real-time PCR detection system (Stratagene). Real-time PCR data were analyzed using the ΔΔC_t method with the peptidyl-prolylisomerase gene serving as reference. Primer pair sequences are provided in the supplemental information.

Immunoblotting

To examine adiponectin secreted by cultured cardiomyocytes, media were collected after the indicated treatments (Fig. 2) and concentrated sevenfold in an Ultraspin column with a 30-kDa molecular weight cutoff (YM-30; Millipore). Concentrated serum was analyzed by Western blot as described below.

Fig. 2.

PPARγ activation in cardiomyocytes induces adiponectin gene expression and secretion of high-molecular-weight (HMW) hormone. A: immunostaining of cardiomyocytes showing that rosiglitazone treatment (10 μm for 48 h) induces the expression of adiponectin, shown as green. Actinomycin-D-stained nuclei are shown as red. B: adiponectin mRNA and protein levels in primary mouse neonatal cardiomyocytes treated with 20 μM rosiglitazone. Data are means ± SE from 3 independent experiments, **P < 0.01. C: concentrated media from rosiglitazone-treated (CM/Rosi) or vehicle-treated (CM/Veh) cardiomyocytes were analyzed by native (nondenaturing and nonreducing) gel electrophoresis and Western analysis. Normal mouse serum (Serum) and media from 3T3-L1 adipocyte cultures (3T3L1) were run as controls. To compare overall amounts of adiponectin secreted in media after rosiglitazone treatment, the same samples as above where analyzed by denaturing gel electrophoresis under standard reducing conditions. Sample normalization was achieved by using equal volumes of media from identically plated cardiomyocyte cultures. LMW, low molecular weight.

Western analyses of adiponectin in whole cell lysates and concentrated media were carried out by standard procedures using an anti-adiponectin antibody (Abcam) and horseradish peroxidase (HRP)-conjugated secondary antibodies. Proteins were denatured using a Lamelli buffer with β-mercaptoethanol heated at 95°C. Denatured and reduced proteins were resolved on 4–16% gradient polyacrylamide gels (Invitrogen) in a SDS running buffer. Proteins were then transferred to nitrocellulose membranes and then blocked for 1 h in 5% milk in TBS with 0.1% Tween 20 (TBST) for 1 h. Membranes were incubated with primary anti-adiponectin antibody overnight (1:1,000 dilution) in 5% milk + TBST and, after washing, were incubated with HRP-conjugated secondary antibody (1:2,000 in 5% milk TBST) for 1 h. Membranes were washed and visualized by chemiluminescence using the Supersignal West Dura Substrate (Pierce) and recorded using a Bio-Rad GelDoc system. Western blot analyses for AMPK (total and phospho), acetyl-CoA carboxylase (ACC) (total and phospho), Erk1/2 (total and phospho), and glyceraldehyde-3-phosphate dehydrogenase were carried out as described above using primary antibodies purchased from Cell Signaling.

Nondenaturing gel electrophoresis was carried out using the concentrated media and media from 3T3-L1 (nonconcentrated) and mouse serum. Samples were mixed with Native Loading Buffer (Invitrogen) and resolved by electrophoresis in Tris-glycine 4–16% gradient acrylamide gels (Invitrogen). Samples were transferred to polyvinylidene diflouride membranes and blotted for adiponectin (Abcam) as indicted above. To determine relative molecular weights, a native molecular weight marker kit was used (GE Health Sciences).

Microscopy

Immunohistochemistry.

Adiponectin was detected in 10-μm frozen sections using a rabbit anti-adiponectin antibody (Abcam) followed by a fluorescein isothiocyanate (FITC)-conjugated goat anti-rabbit IgG (Invitrogen). Nuclei were labeled with the red fluorescent counterstain, 7-actinomycin-D (Molecular Probes). Sections were examined using an Olympus IX2-UCB confocal fluorescent microscope. Neonatal cardiomyocytes were fixed and blocked as described above and then were analyzed for adiponectin using a goat anti-rat specific adiponectin antibody (R&D Systems) followed by a FITC-conjugated donkey anti-goat IgG secondary antibody (R&D Systems). For cell size measurements, cardiomyocytes were fixed in and stained with a phalloidin-conjugated rhodamine solution (Molecular Probes) to visualize sarcomeric F-actin. Quantitative analysis of cell surface area was measured blindly by semiautomatic computer-assisted planimetry (imagePRO) from a minimum of 100 cells selected at random.

Neonatal cardiomyocytes were fixed and permeabilized using 0.01% Triton-100 in PBS for 15 min, washed with PBS, and blocked. Immunostaining proceeded immediately after blocking. Hearts from transgenic mice were flash-frozen in optimum cutting temperature medium in a liquid nitrogen-cooled isopentane bath. Cryosections of 10 μm were taken and fixed in 4% paraformaldehyde. Sections were washed and permeabilized as indicated above, followed by blocking and immunodetection for adiponectin.

Lentivirus.

shRNAs were expressed from the lentivirus-based vector pLL3.7 (24). Anti-adiponectin receptor shRNA sequences were chosen using algorithms from Dharmacon and Invitrogen. To prepare viral stocks, the pLL3.7 clones were cotransfected into 293T cells with viral packaging and helper plasmids (obtained from Addgene). Later (48 h), recombinant virus was collected from media by centrifugation and suspended in PBS. Titers were determined by optical density measurements (260 nm) and used at ∼4–10 × 10⁸ IFU/ml.

Statistical Analysis

The data for the animals and the in vitro studies (mRNA and immunoblot protein densitometric analysis) are presented as means ± SE. Statistical analysis on these data was performed using one-way ANOVA and Student's t-test or with Dunnett's posttest where applicable and was performed using GraphPad InStat version 3.00 for Windows (GraphPad Software, San Diego, CA, www.graphpad.com). A value of P < 0.05 was considered significant.

RESULTS

To determine if activation of PPARγ alters cardiomyocyte responses to hypertrophic stress, we examined the effect of α-adrenergic stimulation of cultured cardiomyocytes in which PPARγ was activated by either pharmacological or genetic means. Treatment of cultured cardiomyocytes with the adrenergic stimuli propranolol and norepinephrine induced a hypertrophic response characterized by increased cell size, atrial natriuretic factor (ANF) expression, and Erk phosphorylation. However, this hypertrophic response was blocked by pharmacological activation of PPARγ with rosiglitazone (Fig. 1, A–C) or by genetic activation using transgenic cardiomyocytes expressing a constitutive active form of PPARγ (Fig. 1, D and E). These findings suggest that PPARγ activation protects cardiomyocytes from hypertrophic stresses and is consistent with previous observations that TZDs have cardioprotective effects in cell culture models of cardiac hypertrophy (25, 29).

Fig. 1.

Activation of peroxisome proliferator-activated receptor (PPAR) γ protects cardiomyocytes from adrenergic-mediated hypertrophy. Pretreatment of neonatal cardiomyocytes with rosiglitazone (Rosi, 10 μM) for 48 h prevented α-adrenergic-mediated hypertrophy. A: the degree of hypertrophy was ascertained by cell size measurement using semiautomated computer-assisted planimetry of phalloidin-stained neonatal cardiomyocytes (A), atrial natriuretic factor (ANF) expression, measured by qRT-PCR (B), and Erk1/2 phosphorylation measured by immunoblot (C). Con, control; Veh, vehicle; NS, not significant; p, phosphorylated; T, total. Data are means ± SE from 4 independent experiments. Surface area measurements were from at least 150 cell measurements from 4 individual experiments. Values for ANF measurements were standardized to peptidyl-prolylisomerase (PPIA) mRNA levels and set as fold change from vehicle control. For Erk phosphorylation, densitometric values were standardized to total Erk levels and set as fold change from vehicle control. For all panels, ***P < 0.001. Cardiomyocytes expressing a constitutive active PPARγ are protected from α-adrenergic-mediated hypertrophy. D: fluorescence micrographs of phalloidin-stained transgenic primary neonatal cardiomyocytes. Cardiomyocytes isolated from transgenic mice and expressing constitutive-active PPARγ (Tg) were resistant to α-adrenergic [norepinephrine (NE)]-mediated hypertrophy (cell size increase) compared with wild-type cardiomyocytes (Wt). E: semiautomatic computer-assisted quantitative measurement of cell surface areas (μm²) from a minimum of 100 random cells/group, stained as shown in A. Data are means ± SE. *** P < 0.001. F: qRT-PCR analysis of ANF mRNA levels in wild-type and transgenic mouse neonatal cardiomyocytes treated with α-adrenergic stimulation. Data are means ± SE from 4 independent experiments and are given as ratios to PPIA mRNA levels. ***P < 0.001.

To explore possible mechanisms for the PPARγ-mediated cardioprotective effect observed above, we measured the expression of known PPARγ target genes in cultured cardiomyocytes after stimulation with the PPARγ agonist rosiglitazone. We observed a striking and unexpected increase in the expression of the adiponectin gene and secretion of all forms of adiponectin, including the high-molecular-weight form, into the media [Fig. 2, A–C, and Supplemental Fig. 1 (Supplemental data for this article may be found on the American Journal of Physiology: Heart and Circulatory Physiology website.)].

Adiponectin has previously been reported to have anti-hypertrophic activity in the heart (17, 27), and our observation that the cardiomyocytes can produce adiponectin hormone raises the possibility of an autocrine loop in which endogenously produced adiponectin acts directly on cardiomyocytes. To examine this possibility, we developed an RNAi-based strategy to block the PPARγ-mediated induction of adiponectin expression in cardiomyocytes (Fig. 3, A and B). Blocking adiponectin expression with this system reduced PPARγ-mediated AMPK activation (Fig. 3C). Because AMPK activation is a known downstream effect of adiponectin signaling (28, 32, 35), these findings indicate that the adiponectin produced by cardiomyocytes is functional and that it is capable of an autocrine signaling in cardiomyocytes.

Fig. 3.

Endogenously produced adiponectin is required for PPARγ-mediated AMP-activated protein kinase (AMPK) activation. Rat neonatal cardiomyocytes were transfected with scrambled (Sc) or adiponectin (Ad) specific RNAi and then treated with rosiglitazone (10 μM). A: relative adiponectin mRNA levels were determined by qRT-PCR. B: relative adiponectin protein levels determined by Western analysis of protein lysate from neonatal cardiomyocytes treated as described. C: PPARγ-mediated stimulation of AMPK phosphorylation does not occur when adiponectin expression was blocked by anti-adiponectin RNAi treatment (Ad) compared with the control RNAi treatment (Sc). Data are ratios of phospho-AMPKα₂ (serine-172) to total AMPKα₂, normalized to vehicle control. Results for mRNA and Western analyses are means ± SE from 4 individual experiments. RNA data are given as ratios to PPIA and normalized to the vehicle control. Protein values for adiponectin were standardized to glyceraldehyde-3-phosphate dehydrogenase (GAPDH) and normalized to the vehicle control. **P < 0.01 and ***P < 0.001.

To determine if the cardioprotective effect of PPARγ activation described above (Fig. 1) is dependent on autocrine signaling by endogenously produced adiponectin, we tested whether PPARγ activation had its usual anti-hypertrophic effect when adiponectin expression was prevented by RNAi. As expected, cardiomyocytes treated with hypertrophic stimuli (propranolol and norepinephrine) showed an increase in three markers of hypertrophy, and this hypertrophic response was effectively prevented by PPARγ activation as evidenced by cell surface area measurements, ANF mRNA, and Erk phosphorylation status (Fig. 4, B–D). When adiponectin expression was prevented by RNAi treatment, the ability of PPARγ activation to protect the cells from hypertrophic stress was significantly reduced (Fig. 4, A–D). These results demonstrate that the protective effects of PPARγ activation in cultured cardiomyocytes are mediated in part by the autocrine action of endogenously produced adiponectin.

Fig. 4.

The anti-hypertrophic effect of PPARγ activation is dependent on endogenous production of adiponectin. A: immunofluorescence of cultured neonatal cardiomyocytes treated with adrenergic stimulation in the absence (NE) or presence (NE + Rosi) of 10 μM rosiglitazone. Cells were pretreated with scrambled (Sc) or adiponectin (Ad) RNAi. B: quantitative analysis of micrographs similar to A. Cell surface area was measured for at least 150 cells/group, as in Fig. 1. Data are means ± SE of surface area measurements in μm² normalized to control for each RNAi group (scrambled and adiponectin). Similar effects of blocking adiponectin expression were observed with other markers of cardiac hypertrophy, ANF gene expression (C), and Erk1/2 phosphorylation (D). RNA values are means ± SE from 4 individual experiments and are presented as ratios to PPIA and normalized to vehicle control. The Erk1/2 phosphorylation data are means ± SE from 4 individual experiments and are standardized to GAPDH and compared with the vehicle control. For all panels, ***P < 0.001, **P < 0.01, and *P < 0.05.

To further confirm the presence and functionality of an adiponectin-based autocrine loop in cardiomyocytes, we examined the role of the adiponectin receptors in mediating the protective effects of PPARγ activation. An effective shRNA-based strategy to block the expression of each of the two adiponectin receptors was developed (Fig. 5A). Blocking expression of either the AdipoR1 or AdipoR2 significantly reduced the ability of PPARγ activation to activate adiponectin signaling as measured by AMPK and ACC phosphorylation (Fig. 5B). We then examined the role of AdipoR1 and AdipoR2 in mediating the PPARγ-dependent cardioprotection using the in vitro cardiac hypertrophy assays as described in Fig. 4. As observed previously, PPARγ activation protected cardiomyocytes from hypertrophic stimuli (Fig. 6, compare the green fluorescent protein controls, with and without rosiglitazone). Consistent with our observation that endogenous adiponectin production was required for PPARγ-mediated protection against hypertrophy (Fig. 4), the presence of the adiponectin receptors AdipoR1 and AdipoR2 was also required for the protective effect of PPARγ activation (Fig. 6).

Fig. 5.

Adiponectin receptors are necessary for PPARγ-mediated AMPK activation. A: lentiviral vectors expressing short-hairpin RNA (shRNA) against the adiponectin receptors AdipoR1 and AdipoR2 selectively reduce expression of each receptor as determined by Western analysis of infected primary cardiomyocytes. B and C: PPARγ-mediated stimulation of AMPK phosphorylation is dependent on adiponectin receptors AdipoR1 and AdipoR2. Blocking expression of either receptor significantly decreased the ability of rosiglitazone to activate AMPK signaling as determined by AMPK phosphorylation (B) and phosphorylation of its downstream target acetyl-CoA carboxylase (ACC) (C). GFP, green fluorescent protein. Data in the bar graphs are ratios of phospho-AMPKα₂ (serine-172) or phospho-ACC to total AMPKα₂ or ACC, respectively, and normalized to vehicle controls. Data are means ± SE from 4 individual experiments. Protein values for adiponectin receptors were standardized to GAPDH. *P < 0.05, **P < 0.01, and ***P < 0.001.

Fig. 6.

Adiponectin receptors are necessary for PPARγ-mediated protection from hypertrophic stress. A: graphical representation of data obtained from surface area measurements of α-adrenergic-stimulated neonatal cardiomyocyte as in Fig. 1. Cells were infected with control lentivirus (GFP) or virus expressing shRNA against AdipoR1 (R1), AdipoR2 (R2), or with both viruses (R1 + R2). After infection, cells were treated with hypertrophic adrenergic stress in the absence or presence of rosiglitazone (NE or NE + Rosi, respectively). Data are means ± SE from a minimum of 100 cells/group, normalized to the GFP vehicle control. B: measurement of ANF mRNA levels in the same neonatal cardiomyocytes treated as described in A. For both panels, ***P < 0.001 relative to vehicle control in the same group; $$P < 0.01 and $$$P < 0.001 relative to the NE bar in the same group.

These findings further support the concept that adiponectin produced in cardiomyocytes in response to PPARγ activation is capable of autocrine signaling through the known adiponectin receptors, AdipoR1 and AdipoR2, and, further, that this autocrine signaling pathway contributes to the anti-hypertrophic effects of PPARγ activation in cultured cardiomyocytes.

Our in vitro findings suggest that PPARγ activation in the heart would have cardioprotective effects. To examine this possibility, we utilized a transgenic mouse line developed in our laboratory that expresses a constitutive-active PPARγ (CA-PPARγ) transgene in heart and skeletal muscle. This mouse line is resistant to high-fat diet-induced diabetes (1). When fed a high-fat diet, the CA-PPARγ transgenic mice showed elevated PPARγ-dependent expression of adiponectin in cardiomyocytes (Fig. 7, A and B). Given the known cardiac hypertrophy effects of a high-fat diet in mice, this transgenic line provides an opportunity to explore, in vivo, the potential anti-hypertrophic effects of cardiomyocyte-specific, PPARγ-mediated adiponectin expression. As expected, feeding a high-fat diet to our nontransgenic control mice induced cardiac hypertrophy as judged by heart weight, ANF expression, and MHCβ expression (Fig. 7, C–E). In contrast, transgenic mice expressing CA-PPARγ in heart were protected from these diet-induced hypertrophic changes. Although this cardioprotection may be due in part to global metabolic differences between wild-type and transgenic animals, these results are consistent with our in vitro findings and with the idea that endogenous expression of adiponectin affords cardiomyocytes a protection against hypertrophic stress. It should be noted that the mechanism of cardiac hypertrophy in this animal model is likely to differ significantly from the in vitro hypertrophy model used above, which depended on catecholamine treatment, rather than exposure to elevated lipids. The fact that endogenously expressed adiponectin is cardioprotective in two distinct models of hypertrophy suggests that it has a general positive effect on the ability of cardiomyocytes to respond to stress.

Fig. 7.

Expression of constitutive-active (CA) PPARγ in heart of transgenic mice prevents high-fat diet-induced cardiac remodeling. Adiponectin is expressed in hearts of CA-PPARγ transgenic mice as shown by qRT-PCR analysis of adiponectin mRNA levels (A) and immunohistochemistry of heart sections (B) where adiponectin is stained green and nuclei red. In A and B, both wild-type and CA-PPARγ (Tg) mice were on a high-fat diet. C: comparison of transgenic and wild-type heart weights in mice fed either a chow or a high-fat diet (40% calories from fat) for 10 wk. mRNA expression levels of ANF (D) and myosin heavy chain β (MHCβ) (E) measured by qRT-PCR. Data are means ± SE from 18 animals/group for A and 6 animals/group for B and C, and are normalized to PPIA mRNA levels. *P < 0.05, **P < 0.01, and ***P < 0.001.

DISCUSSION

The function of PPARγ in the heart is not clearly understood. Its activation has been associated with both beneficial and deleterious effects on the cardiovascular system in both humans and animals. Consistent with previous findings, we observed that, in cultured cardiomyocytes, either pharmacological or genetic activation of PPARγ protected cells from hypertrophy induced by adrenergic stimulation (Fig. 1). In exploring the mechanism for this protective effect of PPARγ activation, we found that activation of PPARγ in cardiomyocytes induced the expression and secretion of functional adiponectin (Figs. 2 and 3). In addition, we demonstrated that the cardioprotective effects of PPARγ activation were dependent on this locally produced adiponectin (Fig. 4) and its signaling through adiponectin receptors present on cardiomyocytes (Figs. 5 and 6).

These results suggest a model in which PPARγ activation in the heart initiates an autocrine loop of endogenously produced adiponectin that acts on the cardiomyocyte to protect it from hypertrophic stress. While this model is based primarily on the in vitro findings described above, it may also explain our in vivo observation that transgenic animals expressing CA-PPARγ in the heart express adiponectin in cardiomyocytes and are protected against high-fat diet-induced cardiac hypertrophy (Fig. 7). Although it appears that, in the transgenic animal system, two conditions must be met for adiponectin to be produced from cardiomyocytes (exposure to a high-fat diet and activation of PPARγ activity), this does not seem to be the case in cultured neonatal cardiomyocytes. In the cultured system, PPARγ activation alone, without exposure to elevated lipids, induces adiponectin expression. The basis for this difference between the in vivo and in vitro systems is not currently understood.

Although this “ectopic” expression and autocrine action of adiponectin is unexpected, it is consistent with our current understanding of adiponectin biology. Protective effects of adiponectin on the cardiovascular system have been observed in multiple animal models and in vitro experimental systems. For example, the treatment of cultured cardiomyocytes with purified adiponectin hormone prevented the adrenergic-mediated hypertrophic response (27, 28). In addition, adiponectin-deficient mice are more susceptible to pressure overload-induced cardiac hypertrophy, while mice engineered to overproduce adiponectin are resistant to pressure overload-induced cardiac hypertrophy (17, 27, 28). Finally, it is known that adiponectin expression in adipocytes is regulated directly by PPARγ (11, 13), and that it can act in an autocrine manner in adipose tissue (9).

Although, to the best of our knowledge, this is the first demonstration of an endogenous adiponectin autocrine loop in cardiomyocytes, the expression of adiponectin in cells of the cardiac system has been previously observed (7, 30). One of these studies reported a reduction in cardiac adiponectin expression in heart biopsies from patients with dilated cardiomyopathy, which led the authors to propose the existence of an endogenous adiponectin signaling system in the heart that, when defective, contributes to the development of cardiomyopathy (30).

Our findings suggest that activation of cardiomyocyte PPARγ would have a direct beneficial effect on the cardiovascular system. This idea is consistent with experimental studies on PPARγ function in the cardiovascular system. In genetically modified mice, heart specific knockout of PPARγ caused cardiac hypertrophy (8, 12), and, in cultured cardiomyocytes, activation of PPARγ ameliorates the deleterious effects of hypertrophic stimuli (2, 34). In rodent models of lipotoxic dilated cardiomyopathy, PPARγ agonists have been shown to improve heart function, prevent lipid accumulation, and reduce the expression of cardiomyopathy markers (33, 37). This is consistent with our previous observations from the CA-PPARγ transgenic animal model that expression of CA-PPARγ in the heart offered protection against high-fat diet-induced lipid accumulation (1).

Although the mechanism by which activation of heart PPARγ affects cardiac physiology is not fully understood, it may be related to observations made in skeletal muscle where PPARγ activation was shown to increase myocellular fatty acid oxidation (5). Our observation that PPARγ activation leads to an increase in AMPK signaling is consistent with this idea, since AMPK activation is known to increase fatty acid oxidation in muscle cells (15, 22).

A notable outcome of the adiponectin receptor knockdown experiment is the apparently predominant role of the AdipoR1 receptor in this system. In addition to abolishing the PPARγ-mediated stimulation in AMPK phosphorylation, the absence of the AdipoR1 receptor actually caused a reduction in the level of AMPK phosphorylation to below the basal (nonstimulated) level. This suggests that signaling through AdipoR1 might be responsible for maintaining a basal level of AMPK activity. This effect seems to be specific for the R₁ receptor, since knock down of AdipoR2 only blocked PPARγ-mediated AMPK activation and did not affect the basal phosphorylation level of AMPK. The strong effect of AdipoR1 on AMPK activity levels may be related to the hypertrophy data shown in Fig. 6, where knock down of the R₁ receptor also had a stronger effect on the PPARγ-mediated anti-hypertrophic effect, compared with the AdipoR2 knock down. Together, these observations suggest that signaling through the AdipoR1 receptor may play a greater role in cardiac metabolism and sensitivity to hypertrophic stress and further that these effects are mediated to a large degree by changes in AMPK activity.

The activation of AMPK via the proposed PPARγ-adiponectin autocrine pathway may also explain the anti-hypertrophic effects of TZDs in this system. A previous study showed that exogenous adiponectin activated AMPK and inhibited agonist-stimulated hypertrophy and Erk activation in cardiomyocytes, and that the anti-hypertrophic effect of adiponectin was dependent on AMPK (28). An important aspect of our study is that the proposed adiponectin autocrine signaling pathway only accounted for part of the anti-hypertrophic effect of PPARγ activation (Fig. 5). This may be related to the fact that cardiac hypertrophy has an inflammatory component (6, 19) and the observation that PPARγ activators also suppress the activity of the proinflammatory NF-κB pathway, which is known to play a role in the development of cardiac hypertrophy (29, 34).

Our results suggest that PPARγ activators, such as the TZDs, may have a beneficial effect on the cardiovascular system in diabetic subjects. Although animal models of cardiovascular PPARγ function support this idea (as discussed above), the effect of PPARγ activators on the cardiovascular system in patients is less clear. Recent reports indicate that rosiglitazone treatment of diabetic patients is associated with increased risk of myocardial infarction (21) and heart failure (14). These deleterious effects of rosiglitazone upon the cardiovascular system may be due to global physiological changes (e.g., altered hemodynamic), rather than the specific effect of PPARγ activation in cardiomyocytes. A recent paper showing that rosiglitazone induces the development of cardiac hypertrophy in mice with a heart-specific knock out of PPARγ (8) supports this possibility.

The complexity of PPARγ function in the cardiovascular system is further highlighted by a recent publication by Son et al. (31), which showed that extreme overexpression of PPARγ in heart causes cardiac hypertrophy. The discrepancy between these findings and our results may be due to significant differences in the levels of transgene expression in the two animal models. In our study, CA-PPARγ was expressed at relatively low levels, probably less than the level of endogenous PPARγ expression in hearts from high-fat-fed mice (1). This is in sharp contrast to the expression level in the Son et al. study where PPARγ transgene levels in heart were much higher than endogenous PPARγ levels. This raises the possibility that at least some of the cardiac phenotype in this study was due to nonphysiological levels of PPARγ. Together with our own results, these findings suggest that the level of PPARγ activity in the heart is finely balanced, and too much or too little activity can lead to deleterious effects on cardiac function. The difficulty in resolving these observations into a coherent model of PPARγ action in the cardiovascular system highlights the incomplete nature of our understanding of this complex transcription factor.

Our model raises several questions that we are unable to answer at present. In particular, our data do not specifically address the degree to which heart-specific expression of adiponectin contributes to the circulating levels of the hormone. It appears that adiponectin expression levels in the heart are much lower than in adipose tissue, so it seems unlikely the heart could make a substantial contribution to the overall level of circulating hormone. However, it is possible that the primary function of hormone produced from the heart is to regulate cardiac physiology in an autocrine fashion and that release of the hormone in the general circulation is minimal.

Multitissue expression analyses generally show very low levels of adiponectin expression in nonadipose tissues (26), and it is likely that expression of the hormone is indeed extremely low under most physiological conditions. In our transgenic animals, the only circumstance where we see significant levels of adiponectin expression in the heart is in high-fat-fed mice in which PPARγ has been activated; chow-fed mice expressing constitutive-active PPARγ did not express higher levels of adiponectin than control animals (data not shown). These observations raise the possibility that adiponectin is expressed in the heart under only rather special physiological circumstances, in this case a combination of insulin resistance and PPARγ activation.

An important issue raised by our findings is the relative contribution of locally produced adiponectin to the response of the heart to TZD therapy. Our data raise the possibility that the PPARγ-mediated autocrine action of locally produced adiponectin plays an important role in mediating the cardiovascular effects of TZD treatment. Additional studies to characterize the role of adiponectin produced by the heart in normal and pathological cardiac function and in the global regulation of metabolism will require the production of a conditional cardiac-specific knockout of the adiponectin gene.

GRANTS

This work was funded by grants from the Michigan Tri-Technology Corridor (MLSC-27) to T. Leff, National Institute of Diabetes and Digestive and Kidney Diseases Grant DK-062102 to T. Leff, a Pfizer Global Research & Development grant to T. Leff, and the American Heart Association Greater Midwest Postdoctoral Fellowship 0620014Z to R. H. Amin.

DISCLOSURES

None.