Introduction

Obesity is an emerging pandemic linked to type 2 diabetes mellitus, hypertension, and cardiovascular disease.¹ Putative mechanisms underlying these associations include insulin resistance, cardiac hypertrophy, and compromised myocardial contractile function and energy metabolism both in rodents and humans.² Besides shifting energy balance, which alters metabolic regulation, models of obesity have also been associated with low-grade inflammation and endoplasmic reticulum (ER) stress in several organs, including the heart.^3,4 Importantly, the ER stress–induced unfolded protein response intersects with inflammatory signals to contribute to insulin resistance in liver, fat, and skeletal muscle.⁵

Clinical Perspective on p 85

Mice on a high-fat diet (HFD) have been used to replicate these and other features of human obesity.⁶ HFD-induced obesity in mice is associated with several adverse cardiac effects including inflammation, hypertrophy, fibrosis, apoptosis, and contractile dysfunction, depending on the dietary lipid composition, duration, and percentage of caloric intake.² In addition, diet-induced obesity may result in insulin resistance and reduced cardiac glucose uptake.⁴ Furthermore, obesity-induced inflammation also contributes to the pathogenesis of diabetes mellitus and obesity-associated heart disease. For example, neutralizing the inflammatory mediator tumor necrosis factor-α (TNF-α) in rats can ameliorate insulin resistance,⁷ and mice with the genetic absence of TNF-α or TNF-α receptor-1 are protected from insulin resistance induced by diet or a genetic model of obesity.⁸ Finally, clinical studies suggest that the inflammation associated with obesity is an independent predictor of heart failure.⁹ More recently, the diet-independent ob/ob mouse model of obesity was shown to manifest myocardial ER stress in concert with contractile dysfunction.¹⁰ Together, these and other epidemiological data linking inflammatory and ER-stress signaling with cardiac dysfunction,¹¹ provide a rationale for therapeutic targeting of these pathways in obesity-induced heart disease.

Glucagon-like peptide-1 (GLP-1) is an incretin hormone that regulates glucose metabolism and may attenuate insulin resistance. This peptide and its analogs also exert cytoprotective actions,¹² with GLP-1 receptor (GLP-1R) activation resulting in reduced inflammation in atherosclerotic lesions,¹³ brain,¹⁴ and pancreatic β-cells.¹⁵ Of particular interest, GLP-1R agonists also induce moderate weight loss through multiple mechanisms, including appetite suppression, delayed gastric emptying, and increased energy expenditure.¹⁶ For example, liraglutide, an approved treatment for type 2 diabetes mellitus that shares considerable amino acid homology with human GLP-1, is being tested for its ability to reduce body weight in obese adults.¹⁷ Nevertheless, few data exist on the effects of liraglutide in preclinical models of obesity.¹⁸ In a recent study examining hepatic steatosis after only 8 weeks of a HFD, Mells et al¹⁸ showed that 4 weeks of treatment with a weight loss–inducing dose of liraglutide (200 µg/kg daily) reversed obesity-induced increases in blood pressure and cardiac hypertrophy, without describing cardiac histology, cardiac function, or any underlying mechanisms. Although the ability of drugs targeting the GLP-1/GLP-1R axis to affect ER stress, unfolded protein response, and autophagy in the liver have also recently been described,¹⁹ the role, if any, of these molecular mechanisms in the pathophysiology of obesity-associated heart disease has not been examined.

Considerable evidence demonstrates that GLP-1 and related peptides protect the isolated mouse heart against ischemia-reperfusion injury²⁰ and protect cardiomyocytes and endothelial cells from oxidative stress and hypoxia-reoxygenation.²¹ Moreover, pretreatment with liraglutide reduces cardiac rupture and adverse ventricular remodeling following myocardial infarction in mice.²² Accordingly, we hypothesized that sustained GLP-1R activation may also prove beneficial in a mouse model of obesity-induced heart disease. Here, we show in both early (16 week) and late (32 week) stages of HFD-induced obesity and heart disease, that treatment for only 7 days with a weight-neutral dose of liraglutide affects cardiac markers of ER homeostasis and reduces markers of inflammation, with an AMP-activated protein kinase (AMPK)–dependent improvement of cardiac function in vivo. These findings were accompanied by rescue of HFD-induced loss of AMPK, extracellular-signal–regulated kinase (ERK)1/2, and glycogen synthase kinase 3β (GSK-3β) signaling, and correction of obesity-induced decreases in endothelial nitric oxide synthase (eNOS), and the cardiac gap junction protein, connexin-43. Liraglutide also reversed HFD-induced increases in cardiac levels of TNF-α, nuclear translocation of nuclear factor kappa B (NFκB), expression of procollagen 1A1, and perivascular fibrosis. To examine if these effects are due to direct cardiac actions of the GLP-1R agonist, we also show that liraglutide protects isolated mouse neonatal cardiomyocytes and human coronary smooth muscle cells from palmitate-induced lipotoxicity and prevents TNF-α–induced endothelial cell–macrophage interactions in human cell lines. Together, these data support the notion that GLP-1R activation may produce salutary effects in obesity-induced heart disease.

Materials and Methods

Animals

Protocols were approved by our institution in accordance with guidelines of the Canadian Council for Animal Care. Male 10- to 12-week-old C57BL/6 mice were obtained from Jackson Labs and housed for ≥2 weeks before experimentation.

Diet and Drug Treatment

Twelve- to 14-week-old mice were fed HFD (45% calories from fat, diet D12451; Research Diets Inc) or RD (regular chow diet) for 16, 20, or 32 weeks with only the final week of each duration including a 7-day treatment with a weight-neutral dose of liraglutide (30 µg/kg SC BID) or placebo (phosphate-buffered saline 100 µL SC BID). To seek evidence for ER stress, a small number of mice were maintained on HFD and RD for 56 weeks. Body weights were recorded before and after each treatment. The experimental schema is shown in Figure 1.

• Download figure
• Download PowerPoint

Online-only Data Supplement

The online-only Data Supplement provides details regarding blood biochemistry, including fasting blood glucose, triglyceride, and cholesterol levels. Histology and morphometry are also described. Online-only Data Supplement Table I lists all antibodies used. Detailed methods for in vivo assessments of cardiac structure, hypertrophy, and function, including echocardiography and invasive hemodynamics, are provided. Assays for cell adhesion, lipotoxicity, X-box binding protein 1 (XBP1) splicing, apoptosis, and metabolic testing, including the intraperitoneal glucose tolerance test and in vivo insulin sensitivity, are also described.

Statistical Analyses

Data are expressed as mean±SE using GraphPad Prism version 5.04 for Windows (GraphPad Software, CA). Significance was defined as 2-sided P<0.05. For analysis of Western blot densitometry, pressure volume loops, echocardiography, blood biochemistry, real-time polymerase chain reaction data, adhesion, and sodium 2,3,-bis(2-methoxy-4-nitro-5-sulfophenyl)-5-[(phenylamino)-carbonyl]-2H-tetrazolium inner salt (XTT) assays, a 1-way analysis of variance with the Tukey post hoc test was used to evaluate differences between groups. Changes in blood glucose concentrations over time during the intraperitoneal glucose tolerance test were analyzed between each treatment group by 2-way repeated-measures analysis of variance with the Bonferroni-Dunn post hoc test. For other comparisons between 2 groups, we used paired or unpaired 2-tailed Student t tests as appropriate. The nonparametric Wilcoxon rank sum test was used for comparison of 2 groups with smaller sample size (ie, n=3–4/group).

Results

A Weight-Neutral Dose of Liraglutide Lowered Fasting Blood Glucose Levels

Mice fed a HFD became markedly obese, with 46±2%, 53±1%, 60±2%, and 114±3% increases in body weight (BW) at the 16-, 20-, 32-, and 56-week time points, respectively (P<0.001 for each comparison; see online-only Data Supplement Table II for absolute values). Treatment with liraglutide (30 µg/kg SC BID) for 1 week did not induce weight loss in RD- or HFD-fed mice at these same time points (online-only Data Supplement Table II). Although chosen to be weight neutral, the dose of liraglutide administered did demonstrate biological activity, with reductions in fasting blood glucose observed in lean and obese groups (online-only Data Supplement Table II).

Obese Mice Manifest Impaired Glucose Handling, Insulin Resistance, Hypercholesterolemia, and Cardiac Lipid Accumulation

The intraperitoneal glucose tolerance test demonstrated glucose intolerance in HFD-fed mice at 16 weeks, which was corrected by liraglutide (Figure 2A). At this stage, insulin-induced phosphorylation of Akt (same as protein kinase B) in both liver and heart were reduced in obese animals (Figure 2B), suggesting insulin resistance. Treatment with liraglutide for 1 week improved all insulin-activated signals examined in the hearts and livers of mice on HFD in comparison with placebo-treated HFD-fed controls, including IRS1, Akt, GSK3β, and ERK1/2 (Figure 2C). Obesity also caused reduced total cardiac levels of the glucose transporter GLUT4 and impaired insulin-induced membrane translocation of GLUT4 (Figure 2D). Liraglutide normalized both of these abnormalities (Figure 2D). Our model also generated significantly increased plasma total cholesterol concentrations, which were partially alleviated by 1-week treatment with liraglutide (Figure 2E). Although no elevations in plasma triglyceride levels were observed at this stage (Figure 2E), qualitative and quantitative immunohistochemistry demonstrated cardiac accumulation of the sphingolipid ceramide (Figure 3E–H), and neutral lipids, as well, as determined by oil red O staining (Figure 3I–J).

• Download figure
• Download PowerPoint

• Download figure
• Download PowerPoint

HFD for 16 Weeks Did Not Affect Cardiac Structure

Despite the above metabolic perturbations and significant increases in body weight (RD, 25.7±0.6 versus HFD, 44.4±1.3 g, P<0.0001, n=18/group), heart weight at 16 weeks did not differ between RD- and HFD-fed mice (122.6±3.1 versus 122.2±6.7 mg, P=NS), with the fall in the heart weight/BW ratio (RD: 4.1 ± 0.1 versus HFD: 2.6±0.1; P<0.0001) in obese mice driven exclusively by their gain in BW. Confirming the absence of cardiac hypertrophy at this time point, no differences in morphometry-defined cardiomyocyte size or histology-defined perivascular and intermyofibrillar fibrosis were observed between RD- and HFD-fed animals (online-only Data Supplement Figure IA and IB).

Liraglutide Corrected HFD-Induced Abnormalities in Cardiac Signaling

Levels of the metabolic sensing/modulator protein–phosphorylated AMPK were reduced in the hearts of mice on a HFD for 16 and 32 weeks. This abnormality was more than reversed by 1 week of treatment with liraglutide (Figure 4A). Curiously, liraglutide did not increase phosphorylation of the survival kinase Akt in the hearts of mice on HFD for 32 weeks (Figure 4B), as it did in the hearts of mice on HFD for 16 weeks (Figure 2C), lean mice (online-only Data Supplement Figure IIA), or in previous studies of mice undergoing myocardial infarction.²² However, liraglutide effectively normalized or increased phosphorylation of other important signaling molecules such as GSK-3β and ERK1/2 in the hearts of obese mice at this later stage (Figure 4C and D). Suggesting selective pathogenesis, neither obesity nor liraglutide had any effect on phosphorylation of the cardiac signaling molecules p38 mitogen-activated protein kinase (Figure 4E) or mammalian target of rapamycin (Figure 4F).

• Download figure
• Download PowerPoint

HFD and Liraglutide Influenced Cardiac ER Homeostasis

To seek definitive evidence of ER stress in the heart, we first examined cardiac mRNA for the presence of spliced XBP1. Consistent with a report showing that HFD did not increase spliced XBP1 levels in islets,²³ we failed to detect spliced XBP1 in the hearts of mice on HFD for 32 weeks (online-only Data Supplement Figure II). In a small group of mice maintained on HFD for 56 weeks (n=3/group) and found to have severe hepatic steatosis (online-only Data Supplement Figure II), we still did not find evidence of XBP1 splicing in heart or liver (Figure 5A). However, protein levels of several ER-stress markers, including the ER chaperones glucose-regulated protein, disulfide isomerase, and phosphorylation of eukaryotic translation initiation factor 2α were increased in cardiac extracts of obese mice fed HFD for 32 weeks (Figure 5B), suggesting HFD-induced accumulation of misfolded proteins in the ER. Unexpectedly, liraglutide enhanced expression levels of these ER markers in obese hearts at both the 16- (online-only Data Supplement Figure II) and 32-week time points (Figure 5B).

• Download figure
• Download PowerPoint

Liraglutide Reversed Obesity-Induced Abnormalities in Cardiac Gene Expression and Fibrosis

To explore the molecular and structural basis of our model, we used gene expression and immunohistochemistry analyses. Both cardiac fibrosis and adverse remodeling of gap junction proteins are known to participate in the pathogenesis of cardiac dysfunction, including models of inflammation, obesity, and diabetes mellitus.^24,25 Indeed, gap junction remodeling precedes contractile dysfunction in an inducible transgene model of inflammatory cardiomyopathy.²⁶ Here, we identify these mechanisms as targets amenable to GLP-1R activation. Liraglutide treatment of obese mice for 1 week restored diminished and disorganized expression of cardiac connexin-43 (Cx-43) (Figure 6A through 6D), reversed the down-regulated levels of cardiac eNOS (Figure 5E), and reduced upregulated levels of procollagen 1A1 and perivascular fibrosis, as determined by Western blot and polarized microscopy (Figure 6F and G).

• Download figure
• Download PowerPoint

Neither HFD Nor Liraglutide Altered Expression Levels of Mitochondrial Markers in the Heart

Obesity is associated with reduced mitochondrial biogenesis and decreased expression of mitochondrial proteins in fat and skeletal muscle of obese rodents.²⁷ As such, we analyzed protein abundance of selected markers of mitochondrial biogenesis (peroxisome proliferator-activated receptor-γ coactivator-1α) and cell respiration (cytochrome c and cytochrome c oxidase IV) in the hearts of our various experimental groups. Levels of peroxisome proliferator-activated receptor-γ coactivator-1α, cytochrome c, and cytochrome c oxidase IV did not change in any group (online-only Data Supplement Figure II).

Liraglutide Reduced Inflammation in the Hearts of HFD-Fed Mice

In contrast to the absence of altered mitochondrial markers, the hearts of HFD-fed mice manifested increased expression of the inflammatory cytokine TNF-α and increased activation (as defined by nuclear translocation) of its downstream signaling mediator NFκB. These abnormalities were reduced by treatment with liraglutide (Figure 7A and B). Because activation of apoptosis has been demonstrated in the hearts of severely obese Zucker rats,²⁸ we examined whether this mechanism might account for the cardiac dysfunction observed in mice fed a HFD for 32 weeks. Despite appreciable increases in intramyocardial lipids, neither terminal deoxynucleotidyl transferase–mediated dUTP nick end-labeling nor Western blot analyses for activated caspase-3 detected any increase in apoptosis in hearts from 16- or 32-week mice fed HFD versus RD (Figure 7C through 7G). Of potential interest, levels of the microtubule-associated protein-1 light chain-3β, a marker of autophagy,²⁹ were also upregulated following treatment with liraglutide (online-only Data Supplement Figure II).

• Download figure
• Download PowerPoint

Liraglutide Corrected Obesity-Induced Cardiac Dysfunction Through an AMPK-Dependent Mechanism

Unlike the relatively silent cardiac phenotype after 16 weeks of HFD, heart weight was significantly increased in mice on HFD for 32weeks (150±5 versus RD, 136±4 mg, P<0.05, n=10/group), as was morphometry-defined cardiomyocyte size (12±0.8% greater surface area versus RD, P<0.01) with the observed fall in heart weight/BW ratio (3.2±0.1 versus 4.2±0.1 mg/g, P<0.0001) attributable to increased BW. Although baseline heart rates did not differ between HFD and RD groups (HFD, 612 ± 34 versus RD, 548 ± 16 beats/min, P=NS, n=10/group), and nor did they differ or change after 1 week of liraglutide (HFD, 573 ± 18, n=10 versus RD, 542±8, n=6, beats/min, P=NS for all comparisons), the load-dependent echocardiographic measure of fractional shortening was reduced in obese mice versus lean controls (30.3±0.7% versus 41.4±1.2%; P<0.05, n=10/group), with subsequent treatment with liraglutide for 1 week correcting this abnormality (preliraglutide, 30.3±0.7; after 1 week of liraglutide, 42.1±1.5%, P<0.05). Additional echocardiography data from the 32-week time point are provided in online-only Data Supplement Table III.

Because analysis of cardiac signaling pathways in obese mice had shown liraglutide-reversible inactivation (ie, dephosphorylation) of the energy sensor AMPK (Figure 4A), we next tested the role of AMPK in mediating the beneficial effects of liraglutide on cardiac function. Left ventricular performance was assessed in mice fed HFD or RD for 20 weeks, with obese animals receiving liraglutide or placebo for 1 week in the presence or absence of the AMPK inhibitor, compound C. Steady-state pressure-volume loops revealed that obesity-induced increases in peak systolic pressure were reduced by liraglutide and that several measures of ventricular performance, including the load-independent isovolumic relaxation constant (τ) were also improved by treatment with liraglutide (Figure 8A through 8G). Importantly, these beneficial effects of the GLP-1 agonist were completely lost in the presence of compound C (Figure 8A through 8G). These data show that hearts of obese mice are functionally responsive to exogenous incretin therapy in an AMPK-dependent manner.

• Download figure
• Download PowerPoint

Liraglutide Normalized Markers of Cardiac Hypertrophy

Quantitative real-time polymerase chain reaction of left ventricular RNA (see online-only Data Supplement Table IV for primer sequences) showed that expression levels of established markers of cardiac hypertrophy, such as atrial natriuretic peptide, brain natriuretic peptide, and β-myosin heavy chain, and the β/α-myosin heavy chain ratio (RD, 0.4±0.1; HFD-P, 2.1±0.2; HFD-L, 0.5±0.1), as well, were increased in mice fed HFD for 20 weeks in comparison with age-matched RD-fed controls (Figure 8H). Importantly, treatment of obese mice with liraglutide returned these markers (and the β/α-myosin heavy chain ratio) to normal levels (Figure 8H; n=5/group, P<0.05).

Evidence for Direct Anti-inflammatory and Cytoprotective Effects of GLP-1R Activation

As treatment with liraglutide in vivo may exert cardiac effects via regulation of insulin, glucagon, or other physiological processes in other organ systems, we assessed the direct actions of liraglutide ex vivo. Both GLP-1 and liraglutide decreased the number of monocytes adhering to TNF-α–activated human umbilical vein endothelial cell cultures. Importantly, this effect could be blocked by the GLP-1R antagonist exendin (9–39) (online-only Data Supplement Figure III).

Obesity is also associated with elevated plasma free fatty acids that participate in the pathogenesis of insulin resistance.³⁰ As one of the most abundant saturated fatty acids in plasma, palmitate is markedly elevated following HFD,³¹ and chronic exposure of rat neonatal cardiomyocytes to bovine serum albumin–bound palmitic acid (PA) produces a model of fatty acid–induced lipotoxicity in vitro.³² Indeed, PA has also been used to induce cytopathology and myofibrillar degeneration in adult rat cardiomyocytes.³³ We investigated whether GLP-1R activation might prevent lipotoxicity in cardiomyocytes. This experiment revealed that the cell swelling and cytoskeletal disintegration caused by PA could be prevented by coincubation with liraglutide, the latter working through a GLP-1R–dependent mechanism (online-only Data Supplement Figure IVA). Moreover, under these harsh in vitro conditions, immunoblotting for activated caspase-3 revealed PA-induced apoptosis was also prevented by liraglutide (online-only Data Supplement Figure IVB).

Because obesity-associated increases in plasma free fatty acids also target other cardiovascular cell types, including vascular cells,³⁴ we conducted similar experiments in human coronary smooth muscle cells. Once again, liraglutide prevented PA-induced cell swelling and loss of cell viability (online-only Data Supplement Figure IVC and IVD).

Discussion

Obesity is a serious health problem. Excess caloric intake from fat can trigger obesity and contribute to dyslipidemia, insulin resistance, and type 2 diabetes mellitus, all of which are known risk factors for cardiovascular disease.³⁵ As observed in our study, HFD can directly affect the heart even before the onset of overt diabetes mellitus, although impaired glucose tolerance and insulin resistance are features of our model. The mechanisms involved include the induction of inflammation, hypertrophy, fibrosis, and contractile dysfunction.² As such, studies identifying therapies that prevent or reverse HFD-induced cardiac pathophysiology are of interest, because their translation to the clinic may impact both the incidence and severity of cardiovascular disease.

As regulators of glucose and lipid homeostasis,¹² potentiating endogenous GLP-1 by dipeptidyl peptidase-4 inhibitors or exogenous GLP-1 analogs show promise for the treatment of type 2 diabetes mellitus. Furthermore, direct benefits on the cardiovascular system may be a particular advantage of incretin-targeted therapies over other classes of antidiabetic drugs.³⁶ Underlying these effects are distinct and complementary mechanisms that include weight loss, improved glycemic control, enhanced insulin signaling, and direct effects on cardiovascular pathophysiology, including protection from ischemia-reperfusion injury and adverse cardiac remodeling following myocardial infarction.^20,22,37

Here, we show that even brief (ie, 1 week) treatment with a weight-neutral dose of the GLP-1R agonist liraglutide can (a) ameliorate HFD-induced disturbances in glucose handling and insulin sensitivity, (b) reverse the expression of inflammatory markers, (c) enhance ER-stress–related adaptive protein levels, (d) normalize key cardiac signaling pathways disrupted by HFD, (e) restore diminished levels of eNOS and Cx-43, and increased levels of procollagen 1A1, (f) normalize expression levels of hypertrophy markers, with (g) morphological evidence of reduced perivascular fibrosis, and (h) functional evidence of improved cardiac performance through an AMPK-dependent mechanism. To our knowledge, this study represents the first demonstration of perturbed cardiac Cx-43 expression in any model of obesity. Also novel are the results showing that liraglutide can correct this abnormality in gap junction protein expression and other obesity-associated decreases in eNOS and cardiac function in vivo without changes in BW. New data supporting direct GLP-1R–dependent effects on endothelial and cardiomyocyte biology include the ability of liraglutide to (i) inhibit adhesion of monocytes to TNF-α-activated endothelial cells, and (j) prevent lipotoxicity in cardiomyocytes and human coronary smooth muscle cells in vitro. By comparing the cardiac consequences of 16 versus 32 weeks of diet-induced obesity in the mouse, our study reveals that HFD-induced markers of inflammation (TNF-α, NFκB) and disturbed ER homeostasis (glucose-regulated protein 78, glucose-regulated protein 94, disulfide isomerase, phospho-eukaryotic translation initiation factor 2α) (at 16 weeks of HFD) are present before the onset of cardiac fibrosis, hypertrophy, and dysfunction (observed at 32 weeks of HFD). Although Ozcan et al³⁸ also showed increased expression of glucose-regulated protein and phospho-eukaryotic translation initiation factor 2α in livers of ob/ob mice and mice fed a HFD for 16 weeks, and also demonstrated increased insulin resistance in mice lacking the ER-stress transcription factor XBP1, no evidence for HFD-induced increases in XBP1 splicing was presented. In our study, we find no evidence for XBP1 splicing in the heart (or liver) after prolonged HFD, suggesting that HFD causes only mild ER stress, which induces an adaptive response (ie, increased chaperone capacity). We posit that liraglutide’s remarkable ability to augment this response may enhance the capacity of the ER to protect the heart (and human coronary smooth muscle cells) against lipotoxicity, as demonstrated in models of hepatocyte¹⁹ and β-cell³⁹ lipotoxicity in vitro.

AMPK, Akt, GSK-3β, ERK1/2, and p38 mitogen-activated protein kinase are signaling molecules involved in the pathophysiology of a variety of cardiac diseases and models. Our studies reveal that HFD-induced obesity resulted in the dephosphorylation of some (ie, AMPK, GSK-3β, and ERK1/2), but not all (ie, p38 mitogen-activated protein kinase) of these molecular mediators in the mouse heart. Treatment with liraglutide effectively reversed abnormalities in each of the pathways affected by HFD-induced obesity. The effect of liraglutide on increased phosphorylation of the energy sensor AMPK is similar to that of the antidiabetic drug metformin, which also induced activation of this molecule in the hearts of obese mice.⁴⁰ Although our invasive hemodynamic studies suggested a key role for AMPK in the cardiovascular actions of liraglutide, further studies are needed to establish if more of the many pleiotropic benefits of liraglutide observed also depend on its ability to activate AMPK. Several lines of evidence support a central role for AMPK in obesity-induced heart disease. AMPKα1-knockout mice are more susceptible to HFD-induced obesity, inflammation, and insulin resistance,⁴¹ and experience worse HFD-induced cardiac hypertrophy and contractile dysfunction.⁴² In addition, upregulation of AMPK signaling improves myocardial perfusion in diabetic mice,⁴³ and activators of AMPK (eg, dexamethasone and metformin) promote survival in cardiomyocytes exposed to TNF-α, an effect abolished by an AMPK inhibitor.⁴⁴ Ko et al⁴ showed that a HFD (55% fat by calories) for only 6 weeks reduced AMPK phosphorylation in cardiac tissue of C57BL/6 mice, a finding they associated with macrophage infiltration and cardiac inflammation. These authors suggested that cytokines interleukin 6 and TNF-α were involved in decreased AMPK activity and pathological modulation of cardiac glucose metabolism.⁴

With regard to the serine/threonine kinase GSK-3β, we demonstrate that it is dephosphorylated (ie, hyperactivated) in the hearts of obese mice. Because increased activity of total GSK-3 (ie, GSK-3α and β-isoforms) in skeletal muscle, liver, and adipose tissues has been associated with reduced insulin action in those tissues,⁴⁵ it is tempting to speculate that HFD-induced activation of GSK-3β may also play a role in obesity-induced heart disease. Again, short-term treatment with liraglutide resulted in an Akt-independent increase in phosphorylation (ie, inactivation) of GSK-3β, approaching levels seen in lean controls. Supporting the importance of this normalizing effect of liraglutide on GSK-3β are data suggesting that inactivating GSK-3β by the antioxidant metallothionein prevented the pathogenetic mechanisms of streptozotocin-induced diabetes mellitus in mice.⁴⁶

Obesity is associated with a low-grade chronic inflammation characterized by infiltration of monocyte/macrophages in skeletal, cardiac, and adipose tissues,^4,47 and by abnormal production of proinflammatory cytokines including TNF-α.⁷ Our results clearly indicate that treatment with a GLP-1 analog can reduce (a) expression of TNF-α and NFκB in the hearts of obese mice, and (b) adhesion of monocytes to TNF-α-activated endothelial cells in a GLP-1R–dependent manner. Importantly, the latter in vitro data were obtained in human cell lines, enhancing the potential translational significance of our findings.

In addition, reduced cardiac eNOS level in obese mice may be an indicator of coronary dysfunction, because eNOS-derived nitric oxide plays an important role in the regulation of coronary blood flow.⁴⁸ Therefore, restored cardiac eNOS in response to 1 week of treatment with liraglutide may have had a significant impact on cardiac function. With regard to Cx-43, it has been shown that cardiac inflammation can downregulate Cx-43 levels in the rat heart in a TNF-α–dependent manner.⁴⁹ Levels of Cx-43 are also reduced in the diabetic rat.²⁵ Because cardiac Cx-43 is required for coordinated electrical activity and is involved in the pathogenesis of cardiac arrhythmias,⁵⁰ repair of Cx-43 deficiency in the hearts of obese mice in our study may be another mechanism involved in the beneficial effects of the GLP-1 analog used.

Finally, exposure of mouse neonatal cardiomyocytes to PA resulted in cytopathological changes, including cellular swelling and cytoskeletal disintegration, which were blocked by cotreatment with liraglutide. That this effect could be abolished by a GLP-1R antagonist clearly supports a direct cardioprotective action of the agent.

In conclusion, the present study has shown, for the first time, how a short course of treatment with a weight-neutral dose of a GLP-1 analog can reverse the complex molecular pathophysiology of obesity-induced heart disease in mice through a variety of putative mechanisms, with a central role for AMPK (online-only Data Supplement Figure V). These data support further clinical exploration of a potential therapeutic role for GLP-1R agonists in obesity-induced cardiac disease.

Sources of Funding

This study was funded by operating grants to Drs Husain and Drucker from the Heart and Stroke Foundation of Ontario (HSFO; T6757 and NA6997 ) and by a grant to Dr Drucker from Novo Nordisk Inc. Dr Husain is recipient of a Career Investigator Award of the HSFO ( CI6824 ). Dr Drucker received support from a Canada Research Chair in Regulatory Peptides and a Banting and Best Diabetes Center-Novo Nordisk Chair in Incretin Biology . Dr Billia is recipient of a Phase I Clinician-Scientist Award from the Canadian Institutes of Health Research (CIHR). Dr Volchuk is supported by an Operating Grant from the CIHR ( MOP-114922 ) and a Canada Research Chair in Diabetes Research .

Disclosures

Drs Husain and Drucker have served as consultants for NovoNordisk, Merck & Co, and Hoffman La Roche with regard to their incretin-targeted therapeutics. They are recipients of investigator-initiated industry funding for distinct research projects from NovoNordisk and Merck & Co. The other authors report no conflicts.

Footnotes