Rosiglitazone, a Peroxisome Proliferator-Activated Receptor-γ, Inhibits the Jun NH₂-Terminal Kinase/Activating Protein 1 Pathway and Protects the Heart From Ischemia/Reperfusion Injury

The care of animals and all experimental procedures conformed with the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health. Male Wistar rats (300–320 g) were randomly separated into normal and diabetic rats as they received an intravenous injection of vehicle alone or 0.1 mol/l citrate-buffered streptozotocin (STZ; Sigma, St Louis, MO) (pH 4.5) at a dosage of 55 mg/kg body wt. The development of diabetes and its persistence were followed by serial quantitative measurements of glucose in the urine with reagent strips. On the day of the experiment, the animals’ diabetes state was confirmed by measuring glucose concentrations in blood samples collected at the time of heart excision. Only rats with plasma glucose concentration >20 mmol/l were considered to be diabetic. Hearts were excised under thiopental sodium anesthesia (50 mg/kg i.p.), and perfused, as previously described (15).

Briefly, hearts were quickly removed and immersed in ice-cold buffer. The aorta was dissected free and then mounted onto a cannula attached to a perfusion apparatus. A retrograde perfusion of the heart was carried out for 10 min by the Langendorff mode and then switched to perfusion according to the working-heart technique (16). The perfusion fluid was a Krebs-Henseleit buffer (pH 7.4) of the following composition (in mmol/l): NaCl, 118; NaHCO₃, 23; KCl, 4.7; KH₂PO₄, 1.2; MgCl₂, 1.2; CaCl₂, 1.8; glucose, 11; and pyruvate, 2. This buffer was continuously gassed with O₂/CO₂ (95/5%) mixture and maintained at 37°C throughout the experiment. The perfusate was not recirculated. Preload was held at 15 cm H₂O and afterload was maintained at 80 cm H₂O.

To initiate total global ischemia, the perfusion pump was turned off for 30 min after an initial 30-min equilibration period. After ischemia, hearts were reperfused by restoring flow to the control level for an additional 30 min. At the end of the experiments, the hearts were freeze-clamped and stored at −70°C. Both heart rate and systolic pressure were continuously monitored. Aortic, coronary, and cardiac flows and stroke volume were measured by timed collection during the 30 min of control preischemic perfusion and the 30 min of reperfusion.

Initial studies were performed to determine the direct effect of rosiglitazone on cardiac function. In this study, rosiglitazone (0.3, 1, and 3.0 μmol/l) was added to the perfusate after a 15-min equilibration period and perfused continuously for an additional 75 min. This duration of perfusion represented an exposure to rosiglitazone that covered the entire ischemia/reperfusion protocol. For ischemia and reperfusion experiments, rosiglitazone (1 μmol/l) was added to the perfusate 15 min before stopping the flow and was maintained throughout ischemia and reperfusion.

Subsequently, to determine whether PPAR-γ activation was likely to be responsible for the cardiac effects of rosiglitazone, a subset of normal hearts (n = 5) was exposed to 1 μmol/l of another potent and selective PPAR-γ ligand, SB 219994, the (S)-enantiomer of an α-trifluoroethoxy propanoic acid (acyclic) insulin sensitizer; its 1,000-fold-less-potent (R)-enantiomer, SB 219993 (17); or the DMSO (0.1%) vehicle. As with rosiglitazone, these substances were introduced at 15 min into the preischemic equilibration period. Finally, we investigated the effects of chronic treatment with rosiglitazone on isolated perfused normal and diabetic rat hearts ex vivo. For these experiments, rosiglitazone (10 μmol · kg⁻¹ · day⁻¹ orally by gavage) or its vehicle was administered for 4 weeks, starting immediately after STZ or vehicle injection.

To determine lactate release and lactate dehydrogenase (LDH) activity in the perfusate, samples were collected from the coronary effluent before ischemia and after 5 min of reperfusion. LDH and lactate activities were assayed spectrophotometrically using LDH and lactate assay kits (Boehringer Mannheim, Mannheim, Germany).

Rat heart protein concentrations were determined using the bicinchonic acid assay (Pierce Interchim, Montlucon, France). First, 50 μg of cardiac protein were separated by SDS-PAGE and subsequently transferred to a Hybond membrane. Membranes were checked for equal loading by Ponceau red staining. Nonspecific binding sites were then blocked overnight at 4°C with 10% skim milk powder in Tris-buffered saline with Tween (TBST; 20 mmol/l Tris-HCl, 55 mmol/l NaCl, 0.1% Tween 20). Membranes were subsequently incubated for 4 h at room temperature in 5% skim milk-TBST containing antibodies raised against PPAR-γ, GLUT4, JNK (Santa Cruz Biotechnology, Santa Cruz, CA), and active-JNK (Promega, Madison, WI). After incubation with a secondary peroxidase-conjugated antibody, signals were visualized by chemiluminescence (Amersham, Buckinghamshire, U.K.).

An AP-1 double-stranded oligonucleotide (Promega) was end-labeled with [γ-³²P]ATP using T4 polynucleotide kinase according to standard protocols. Nuclear extracts (5 μg) from rat hearts were incubated with 50,000 cpm of labeled probes for 20 min at room temperature in 20 μl buffer containing 10 mmol/l Tris (pH 7.5), 50 mmol/l NaCl, 1 mmol/l dithiothreitol, 1 mmol/l EDTA, 5% glycerol, 0.3 μg BSA, and 2 μg of poly(dI-dC). The reactions were analyzed by electrophoresis in a nondenaturing 5% polyacrylamide gel in 0.5 × Tris/Borate/EDTA buffer. The gels were then dried and exposed at −80°C for autoradiography.

Results are shown as means ± SE. Statistical significance of differences was determined using Student’s t test or ANOVA followed by Student-Newman-Keuls test. Differences with P < 0.05 were considered to be statistically significant.

Preliminary Western blotting experiments using a specific antibody confirmed that PPAR-γ protein is expressed in the nuclear fraction of cardiac tissue (13,14). Furthermore, PPAR-γ protein levels were unchanged by acute treatment with rosiglitazone (1 μmol/l; data not shown). Interestingly, increasing concentrations of rosiglitazone (0.3, 1.0, and 3.0 μmol/l) also did not change cardiac function indexes in isolated working hearts from normal hearts (data not shown).

Diabetic rats exhibited a marked hyperglycemia and reduced body and heart weights 4 weeks after STZ injection. Heart rate was also decreased in diabetic rat hearts (∼24%; P < 0.05), in agreement with results from previous studies (18), although otherwise, at baseline the cardiac performance of diabetic hearts was indistinguishable from that of nondiabetic hearts. When rosiglitazone (1 μmol/l) was included in the perfusate for 75 min, aortic flow, which is considered to be a valid index of ventricular contractility (16), remained unchanged in both normal and diabetic hearts (Table 1). Our results suggested that although the nuclear receptor PPAR-γ is expressed in rat cardiac tissue, as previously demonstrated in rat cardiomyocytes (13,14), acute exposure to rosiglitazone is not associated with any direct functional effect in either normal or diabetic rat hearts.

Activation of PPAR-γ is linked with improved glucose metabolism through enhanced glucose transport (10), and glucose availability is linked with improved postischemic recovery after thrombolysis (19). We hypothesized, therefore, that rosiglitazone would reduce the functional deficit of rat hearts after ischemia and reperfusion.

Changes in aortic flow, cardiac output, heart rate, and stroke volume at baseline, during ischemia, and in reperfusion in both normal and diabetic rat hearts perfused in the absence or presence of rosiglitazone (1 μmol/l) are illustrated in Fig. 1. We observed that 30 min of zero-flow ischemia markedly reduced cardiac function in normal hearts. After 5 min of reperfusion, aortic flow was maximally suppressed (0%), whereas cardiac output, heart rate, and stroke volume (n = 7) represented 3, 12, and 3%, respectively, of baseline values (Fig. 1A). In contrast to the case in normal hearts, restoration of perfusion in diabetic hearts after 30 min of zero-flow ischemia resulted in more rapid recovery of functional indexes (Fig. 1B). After 5 min of reperfusion, aortic flow was 20% of its baseline value, cardiac output was 44%, heart rate was 69%, and stroke volume was 37% (n = 6; P < 0.05 diabetic vs. control). These observations were consistent with previous findings obtained in similar experimental conditions (15).

The pattern of postischemic functional recovery was significantly improved in the presence of rosiglitazone in both normal and diabetic rat hearts (Figs. 1A and B). In normal rat hearts, rosiglitazone increased the recovery of aortic flow (34% of baseline value after 5 min of reperfusion), cardiac output (61%), and stroke volume (61%) (n = 6; P < 0.05 treated hearts vs. controls). Also, in diabetic hearts, the recovery of aortic flow (63%), cardiac output (82%), and stroke volume (64%) were all significantly improved at 5 min of reperfusion (n = 6; P < 0.05 vs. controls). Consistent with the recovery of contractility during reperfusion, heart rate also recovered more rapidly in rosiglitazone-treated hearts and reached baseline values after 5 min of reperfusion in both normal and diabetic rat hearts.

The cardioprotection conferred by rosiglitazone, manifested by improved cardiac function early during reperfusion, was also characterized by evidence of reduced cell damage. First, increased glycolysis at the time of reperfusion, although less in diabetic than in normal rat hearts, as previously demonstrated (18), was inhibited by rosiglitazone (Fig. 2A). Indeed, coronary effluent lactate levels were lower in both diabetic and normal rat hearts treated with rosiglitazone than in control hearts (Fig. 2A). Second, the extent of LDH release in the coronary effluent at the time of reperfusion, which again was less marked in diabetic hearts relative to normal hearts, was further reduced by rosiglitazone, although the effect was only statistically significant in normal rat hearts (Fig. 2B).

Moreover, hearts from animals chronically treated with rosiglitazone exhibited an improved postischemic cardiac function ex vivo. After 10 min of reperfusion, the aortic flow was 25.0 ± 16 and 50.9 ± 3.5%, respectively, of preischemic values in normoglycemic and diabetic rats receiving vehicle, and it was 75.0 ± 11 and 69.5 ± 4.7%, respectively, in normoglycemic and diabetic rats given rosiglitazone (Fig. 3).

These results showed that rosiglitazone, administered either acutely via the perfusate to the hearts or as a chronic oral treatment, induced a marked cardioprotective effect in both normal and diabetic rat hearts. Moreover, as illustrated in Fig. 4, hearts from normal animals exposed acutely to another selective PPAR-γ agonist, SB 219994, also exhibited an improvement in the recovery of aortic flow during postischemic reperfusion compared with controls (67 ± 13% of preischemic values at 10 min of reperfusion vs. 25 ± 15% in the control group). By contrast, the (R)-enantiomer SB 219993, which is ∼1,000-fold less potent as a PPAR-γ ligand than SB 219994 (17), was ineffective in improving postischemic functional recovery (14 ± 7%).

Activation of PPAR-γ in 3T3-L1 adipocytes causes a downregulation of PPAR-γ mRNA and protein expression (20). In mouse adipocytes, the synthesis and translocation of the glucose transporter GLUT4 are increased by rosiglitazone (10). Thus, we next investigated the effect of the acute rosiglitazone perfusion of isolated hearts on PPAR-γ protein level and GLUT4 translocation in the absence and presence of ischemia/reperfusion. The PPAR-γ antibody recognized a single protein band with a molecular mass of ∼55 kDa. Interestingly, in normal or diabetic rat hearts, neither ischemic stress nor rosiglitazone influenced PPAR-γ protein levels (Fig. 5).

By contrast, at baseline in both normal and diabetic rat hearts, rosiglitazone increased the level of membrane-associated GLUT4 protein (Fig. 5). However, after ischemia/reperfusion, which by itself increased membrane-associated GLUT4 protein levels, rosiglitazone had no additional effect, despite enhancing functional recovery of the hearts (Fig. 1). These data cast doubt on the proposition that GLUT4 translocation per se might be entirely responsible for the cardioprotective activity of rosiglitazone.

MAP kinases such as JNK have been reported to be involved in mediating the deleterious effects of ischemia/reperfusion in the heart. Therefore, we investigated whether protection of the ischemic myocardium by rosiglitazone in the present study could be explained by an effect of the drug on the JNK/AP-1 pathway. JNK activity was measured by Western immunoblotting using an antibody specific for the active, phosphorylated form of JNK (anti-P-JNK). As shown in Fig. 6A, staining with the JNK antibodies produced a specific band that migrated with a molecular weight of ∼54 kDa, the reported size of JNK (21). Both normal and diabetic hearts exhibited some JNK activity at baseline, as indicated by the presence of JNK immunoreactivity. Furthermore, consistent with previous observations (5), JNK was activated by ischemia/reperfusion, and this activation was observed without a significant change in total JNK protein concentration. Our results also showed that acute rosiglitazone perfusion of isolated hearts markedly reduced ischemia/reperfusion−induced activation of JNK in normal and diabetic hearts (Fig. 6A). Interestingly, similar results were also observed when other enzymes involved in the apoptosis pathways activated by ischemia/reperfusion in cardiac tissue were investigated, such as p38 MAP kinase (data not shown).

Furthermore, to determine whether rosiglitazone regulates the interaction between AP-1 and its DNA-binding element, gel mobility shift assays of cardiac extracts were performed using a labeled AP-1 consensus oligonucleotide probe (Fig. 6B). Normal and diabetic hearts exhibited some AP-1−binding activity, even at baseline, and this was visibly reduced by rosiglitazone. Rosiglitazone also markedly attenuated the increase in AP-1 binding activity triggered by ischemia/reperfusion insult in these hearts (Fig. 6B). These findings suggest that in the heart, as in endothelial cells (22), rosiglitazone interacts negatively with the AP-1 signaling pathway by preventing the binding of transcription factors of the Fos/Jun family to the AP-1 site. Our overall data, however, suggest that in this case, upstream inhibition of JNK activation was most likely responsible for reduced AP-1 DNA-binding activity, as has been previously shown for other nuclear receptors (23).

PPAR-γ is believed to be an important regulator of several biological processes, with adipocyte differentiation being the most prominent and best understood (9). Furthermore, the pathophysiological importance of PPAR-γ is now well recognized in the context of several diseases, such as type 2 diabetes (24). In this study, we have provided compelling evidence that rosiglitazone, a selective PPAR-γ ligand, protects the heart from acute ischemic/reperfusion injury. This cardioprotective effect, manifested by reduced lactate levels and LDH activity, together with reduced functional impairment during reperfusion, is associated with an inhibition of the stress-activated protein kinase (JNK) pathway and a consequential inhibition of AP-1 DNA-binding activity.

Ischemic heart disease is a complex multifactorial pathology for which several therapeutic strategies are available, such as calcium channel blockers and β-adrenergic receptor antagonists. Unlike results obtained previously with troglitazone (25), the cardioprotective effect of rosiglitazone is not related to a direct hemodynamic effect, as our results showed no increase in aortic or coronary flows, even at concentrations higher than those shown to be effective against postischemic dysfunction. Published data support this distinction, because rosiglitazone is less potent than troglitazone as an inhibitor of rat vascular smooth muscle voltage-dependent calcium channels (26). An inhibitory effect of rosiglitazone on the cardiac β₁-adrenergic receptor cannot be excluded by the present study, but no such activity was detected in receptor-binding assays (R.E.B., unpublished observation).

In the absence of a hemodynamic explanation for the cardioprotective effects of rosiglitazone, another mechanistic explanation for the observed effects may be linked to the drug’s ability to improve glucose metabolism (10). Our present data on cardiac energetics, however, are limited, and it is not clear whether acute rosiglitazone perfusion of isolated hearts reduces glycolysis or rapidly enhances uptake of accumulated lactate early in reperfusion. The latter may be more likely in view of the demonstration that treatment of pigs with troglitazone, another PPAR-γ ligand, enhances uptake of lactate during reperfusion (27).

Based on data showing that rosiglitazone increases GLUT1 and GLUT4 protein levels (28) and GLUT4 translocation to the adipocyte cell membrane (10), we hypothesized that the effects on glucose transporter activity might play a pivotal role in regulating myocardial glucose availability, even under acute exposure conditions. In the present study, we showed that rosiglitazone did, indeed, increase GLUT4 translocation to the myocyte membrane, even in the absence of ischemia, and that this action, if it occurred during the 15 min of drug exposure before ischemia (our measurements were only conducted after 75 min of exposure), may have played a priming role in protecting the myocardium. Nonetheless, the observation that the ischemia/reperfusion itself evoked a massive GLUT4 translocation response, which was not further increased by rosiglitazone, casts doubt on whether this action of rosiglitazone can provide a satisfactory explanation for the improvement in functional performance of the heart.

The cardioprotective action of rosiglitazone was further confirmed by its ability to protect against activation of the JNK/AP-1 cascade. The importance of the JNK/AP-1 pathway, together with the p38 MAPK pathway, which is also stimulated by ischemia/reperfusion (1), is profound in terms of cardiac function and cell survival. The ability of rosiglitazone to improve cardiac oxidative metabolism might be sufficient to prevent activation of the JNK/AP-1 cascade. This suggestion is also supported by the observation that even in the absence of ischemia, the drug simultaneously increases GLUT4 translocation to the cell membrane and inhibits baseline JNK phosphorylation and AP-1 DNA-binding activity. Nonetheless, we cannot, at this stage, exclude the possibility that rosiglitazone directly inhibits the JNK pathway. In this respect, troglitazone antagonized vascular smooth muscle cell proliferation by inhibiting angiotensin II−induced extracellular signal-related kinase 1/2 nuclear translocation (29) and also attenuated interleukin-1−induced JNK phosphorylation in the pancreatic β-cell line RINm5F (30). Similarities in the way these drugs affect signaling pathways in vascular smooth muscle cells are also extended to the heart, as troglitazone (1 μmol/l) also has been shown to enhance the functional recovery of ischemic rat hearts in parallel studies. In keeping with its lower binding affinity for PPAR-γ (17), troglitazone’s cardioprotective action was also less marked than that produced by rosiglitazone at the same concentration (data not shown). At present, the exact mechanisms contributing to the cardioprotective effects of rosiglitazone are unclear and require further study. It is unlikely that changes in nitric oxide activity could explain the protection reported here because we did not observe any change in coronary flow (data not shown).

Our data showed that pretreating normal or diabetic rats with rosiglitazone for 4 weeks also protects against ischemic/reperfusion injury. Under these experimental conditions, rosiglitazone was not present in the perfusate, and, although the presence of residual drug in the tissues cannot be excluded, it is quite possible that the beneficial effects of the drug were conferred by preprogramming of protective pathways via transcriptional events mediated by PPAR-γ. Finally, other PPAR-γ ligands, troglitazone (data not shown) and especially SB 219994, also exhibited acute cardioprotective activity in our study, thereby suggesting a role for PPAR-γ in the observed effects.

The hypothesis that PPAR-γ mediates the cardioprotective action of rosiglitazone may appear questionable in view of the rapidity of events in the acute experiments. Nonetheless, the demonstration that the insulin-sensitizing effects of the PPAR-γ ligand MCC-555 are seen within 30 min in rat cardiomyocytes and are blocked by cycloheximide, thereby confirming the requirement for de novo protein synthesis (13), suggests that effects mediated by PPAR-γ in the heart are quite possible on an acute time scale. Moreover, a recent study by Nakajima et al. (31) demonstrated that activation of PPAR-γ by rosiglitazone provides potent protection against ischemia/reperfusion injury, whereas the absence of PPAR-γ predisposes to intestinal ischemia/reperfusion injury.

Altogether, our findings suggest that rosiglitazone could represent a novel therapeutic approach to reducing or preventing reperfusion injury in the myocardium. The present results, together with the recent findings that rosiglitazone reduces myocardial infarct size by a PPAR-γ−dependent anti-inflammatory effect (32), may suggest that the activation of the PPAR-γ receptor inhibits the cellular damage cascade in which JNK/AP-1 plays a central role. In the context of the present use of rosiglitazone in patients with type 2 diabetes, these results suggest a potential beneficial action of the drug in a patient population in which the risk of cardiovascular events is substantially raised.

P.D. is supported by a grant of the Région Nord-Pas-de-Calais.

We thank J. Albadine and M.C. Le Boulch for technical assistance.