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
• Open in new tab
• Download powerpoint

Figure 2.

Characterization of HFD-induced obesity model. A, HFD impaired glucose tolerance. After a 6-hour fast, serial tail blood glucose measurements were made before and after glucose administration (arrow, 1 g/kg IP). Obese (16HFD-P) mice showed elevated blood glucose levels versus lean controls (16RD), and versus obese mice treated with liraglutide (HFD-L) by 2-way repeated-measures ANOVA with Bonferroni-Dunn post hoc test: *P<0.0001 for HFD-P versus RD; #P<0.001 for HFD-P versus HFD-L, n=4/group. B, Insulin sensitivity in vivo was assessed by measuring insulin-induced phosphorylation of Akt (P-Akt) in cardiac tissues following an overnight fast. Heart (H) and liver (Li) extracts from PBS-treated mice served as negative controls revealing background levels of P-Akt. Insulin resulted in diminished P-Akt levels in HFD- versus RD-fed mice, *P<0.05 by nonparametric Wilcoxon rank sum test (n=3/group). C, Insulin sensitivity in cardiac and hepatic protein extracts. Liraglutide corrected the impaired insulin-induced phosphorylation of downstream targets IRS1, Akt, GSK3β, and ERK1/2 in HFD-fed mice. D, HFD resulted in reduced cardiac GLUT4 levels, 1-way ANOVA: *P<0.05 by Tukey multiple comparison post hoc test, n=3 group). Confocal microscopy revealed improved insulin-induced membrane translocation of GLUT4 in cardiomyocytes of obese (16HFD) mice treated with liraglutide. E, Obese (16HFD) mice have increased plasma total cholesterol (TC), but not triglyceride (TG) concentrations. TC, 1-way ANOVA: *P<0.0001 versus RD, #P<0.05 versus HFD-P by Tukey multiple comparison post hoc test; TG, P=NS; n=4–5/group. ANOVA indicates analysis of variance; HFD, high-fat diet; INS, insulin; NS, not significant; PBS, phosphate-buffered saline; and RD, regular diet.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3.

HFD-induced obesity causes increased cardiac lipid accumulation. Tissue levels of the sphingolipid ceramide were quantified in liver and heart by immunohistochemical staining, confocal microscopy, and digital image analysis. Sections of liver (A) and heart (B) in which the 1^°anticeramide Ab was omitted served as negative controls for nonspecific background staining by the 2^oAb (red). Ab against cardiac troponin-I (TnIc) and Hoechst were used to label cardiac myocytes (green) and nuclei (blue), respectively. Quantitative analysis of signal intensity revealed increased hepatic (C, D) and cardiac ceramide levels after 16 (E, F) and 32 (G, H) weeks of HFD. Oil red O staining also revealed increased cardiac accumulation of neutral lipids after 32 weeks of HFD (I, J); *P<0.05, †P<0.01 by nonparametric Wilcoxon rank sum test, n=4 mice/group. Ab indicates antibody; HFD, high-fat diet; NC, negative control; and RD, regular diet.

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
• Open in new tab
• Download powerpoint

Figure 4.

Treatment with a weight-neutral dose of a GLP-1 analog corrected the abnormalities in cardiac signaling caused by 32 weeks of HFD. Representative Western blots of left ventricular extracts from lean mice fed a RD and obese mice fed a HFD for 32 weeks are shown. Corresponding densitometric quantifications of the fold changes in phosphorylated/total protein relative to placebo-treated RD-fed animals are shown. Liraglutide treatment (L) for 1 week (30 µg/kg SC BID) increased phosphorylation of AMPK (A) but not Akt (B), and normalized phosphorylation of GSK3β (C) and ERK1/2 (D) in obese mice (HFD-L) in comparison with placebo-treated HFD-fed mice (HFD-P). Phosphorylation of p38-MAPK (E) and mTOR (F) did not change in response to HFD or drug treatment. Data shown are mean±SE; n=5 animals/group; 1-way ANOVA: *P<0.05 versus RD-P, #P<0.05 versus HFD-P, by Tukey multiple comparison post hoc test. AMPK indicates AMP-activated protein kinase; ERK, extracellular-signal-regulated kinase; GLP-1, glucagon-like peptide-1; GSK-3β, glycogen synthase kinase 3β; HFD, high-fat diet; mTOR, mammalian target of rapamycin; p38 MAPK, p38 mitogen-activated protein kinase; and RD, regular diet.

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
• Open in new tab
• Download powerpoint

Figure 5.

Treatment with a GLP-1 analog enhanced the increased cardiac expression of ER-related proteins caused by 32 weeks of HFD. A, Total RNA was isolated from ventricular tissue of mice fed either RD or HFD for 56 weeks and analyzed by RT-PCR and ethidium bromide–stained gel electrophoresis to detect unspliced (uXBP1) and spliced (sXBP1) forms of the ER-stress marker XBP1. Mouse islets treated with thapsigargin (Tg) served as a positive control. B, Representative Western blots and corresponding densitometric quantification of fold changes in GAPDH-normalized expression of ER-related markers, GRP, PDI, and peIF2α are shown for mice on RD or HFD for 32 weeks. Despite enhanced signal for ER-related proteins in hearts of HFD-P and HFD-L groups at 32 weeks, no splicing of XBP1 was observed in the hearts or livers of mice after even 56 weeks of HFD. Treatment with liraglutide enhanced cardiac expression levels of these ER-stress proteins in obese mice (HFD-L) in comparison with placebo (P). Data shown are mean±SE; n=5 animals/group; 1-way ANOVA: *P<0.05 versus RD-P, #P<0.05 versus HFD-P by Tukey multiple comparison post hoc test. ANOVA indicates analysis of variance; ER, endoplasmic reticulum; GLP-1, glucagon-like peptide-1; GRP, glucose-regulated protein; HFD, high-fat diet; PDI, disulfide isomerase; peIF2α, eukaryotic translation initiation factor 2α; RD, regular diet; and RT-PCR, reverse transcription polymerase chain reaction.

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
• Open in new tab
• Download powerpoint

Figure 6.

A GLP-1 analog corrected the abnormal cardiac expression levels of connexin 43, eNOS, and collagen caused by 32 weeks of HFD. In obese mice fed HFD for 32 weeks, confocal immunolabeling for the cardiac gap junction protein Cx-43 (pink) in sections costained for nuclei (DAPI, blue) and α-sarcomeric actinin (green) revealed reduced abundance of Cx-43 (B) in comparison with RD-fed controls (A). C, This abnormality was normalized by treating HFD-fed mice with liraglutide (L) for only 1 week. D, Western blot for total ventricular Cx-43 revealed the same result. E, HFD-induced obesity was also associated with decreased levels of cardiac eNOS, which was corrected by the GLP-1 analog. F, Picrosirius red-stained heart sections were imaged with brightfield and polarized light microscopy. Perivascular collagen under brightfield appears red. Thick collagen fibers under polarized light have yellow-orange birefringence, whereas thin collagen fibrils have green birefringence. HFD increased perivascular thick collagen deposition. Lean (RD-fed) mice retain predominantly fibrillar perivascular collagen. Treatment of HFD-fed mice with liraglutide reduced perivascular collagen staining, with reduced thick collagen fiber deposition. G, These histological findings were mirrored by Western blot for procollagen 1A1. Representative Western blots and corresponding densitometric quantification of fold changes in GAPDH-normalized expression of these proteins are shown. Data are mean±SE; n=5–6 animals/group; 1-way ANOVA: *P<0.05 versus RD-P, #P<0.05 versus HFD-P, by Tukey multiple comparison post hoc test. Cx-43 indicates connexin-43; DAPI, 4′,6-diamidino-2-phenylindole; eNOS, endothelial nitric oxide synthase; GLP-1, glucagon-like peptide-1; HFD, high-fat diet; and RD, regular diet.

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
• Open in new tab
• Download powerpoint

Figure 7.

A GLP-1 analog reversed the expression of inflammatory markers in hearts of obese mice, which showed no evidence of apoptosis. Protein levels of activated NFκB-p65 (demonstrated by nuclear translocation) (A) and TNF-α (B) were increased in hearts of placebo (P)-treated mice fed HFD for 32 weeks (HFD-P) versus lean RD-fed controls (RD-P). Liraglutide treatment decreased inflammatory markers in HFD-fed mice (HFD-L). Data shown are mean±SE; n=4–6 animals/group; 1-way ANOVA: *P<0.05 versus RD-P, #P<0.05 versus HFD-P, by Tukey multiple comparison post hoc test. Paraffin sections (6 µm) were stained for TUNEL to identify apoptotic nuclei (red). Negative (C, without terminal transferase) and positive (D, DNase I-treated) controls were included. Almost no TUNEL-positive nuclei are detectable in hearts from 32RD or 32HFD mice (E, F). Absence of apoptosis in these groups was confirmed by Western blot for cleaved caspase 3 (G), with thapsigargin-induced apoptosis in islet cells serving as a positive control (+ctrl). GLP-1 indicates glucagon-like peptide-1; HFD, high-fat diet; NFκB: NFκB, nuclear factor kappa B; RD, regular diet; TNF-α, tumor necrosis factor-α; and TUNEL, terminal deoxynucleotidyl transferase-mediated dUTP nick end-labeling.

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
• Open in new tab
• Download powerpoint

Figure 8.

Liraglutide improved measures of cardiac performance through an AMPK-dependent mechanism. Invasive hemodynamic measurements were obtained in mice fed RD or HFD for 20 weeks. In the final week, some animals were treated with liraglutide (L) or placebo (P), in the presence of either the AMPK inhibitor compound C (CC) or its vehicle control. A, Representative steady-state PV loops from all groups are shown. B, Maximum left ventricular systolic pressure from all groups. C through G, Results of PV loop analyses showing measures of systolic (EF, ESV, dP/dtmax) and diastolic (dP/dtmin) function, including the load-independent isovolumic relaxation constant (Tau). One-way ANOVA: *P<0.05 versus RD-P by Tukey multiple comparison post hoc test. H, Treatment with liraglutide also normalized the increased expression levels of ANP, BNP, and β-MHC observed in HFD-fed mice as determined by qRT-PCR. n=5–6/group; 1-way ANOVA: *P<0.05 versus RD-P by Tukey multiple comparison post hoc test. AMPK indicates AMP-activated protein kinase; ANP, atrial natriuretic peptide; β-MHC, β-myosin heavy chain; BNP, brain natriuretic peptide; EF, ejection fraction; ESV, end-systolic volumes; HFD, high-fat diet; PV, pressure volume; qRT-PCR, quantitative reverse transcription polymerase chain reaction; and RD, regular diet.

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).