Chronic left ventricular volume overload (VO) causes severe left ventricular dilatation and eccentric hypertrophy (LVH). LVH causes alterations in myocardial energy metabolism. The normal myocardium has a preference for fatty acids as the main substrate for ATP formation. In the hypertrophied heart, an increased reliance on glucose with an overall reduced oxidative metabolism is often observed [1-3]. Impaired myocardial energetics also activate AMP-activated protein kinase (AMPK), leading to increased glucose uptake and glycolysis [4]. AMPK activation can also stimulate fatty acid oxidation and block the mammalian target of rapamycin (mTOR) pro-hypertrophic pathway [5]. Metformin is an AMPK activator widely used for the management of type 2 diabetes but its potential to help maintain normal myocardial metabolism and/or prevent LVH is not well understood [6]. We have previously demonstrated that LV volume overload caused by aortic valve regurgitation causes severe eccentric LVH and also significantly impairs myocardial metabolism [7].

This study was therefore designed to assess the impact of a treatment with an AMPK activator, metformin, on the development of eccentric LVH from volume overload from severe AR in Wistar rats.

Animals

The animal protocol design was a 2 × 2 type study where animals were surgically induced with the aortic valve lesion of only shamoperated and then received or not the metformin treatment. Adult male Wistar rats were purchased from Charles River (Saint-Constant QC, Canada) and divided in 4 groups as follows: 1) Sham-operated animals (SC; n=14); 2) AR (ARC; n=15), 3) Sham treated with metformin (M)(150 mg/kg/d PO in unsweetened fruit gelatin; SF; n=14) and AR on metformin (ARM n=15). The treatment was started one week before surgery in both S and AR groups and continued for 9 weeks until sacrifice. The animals were housed in a 12 hr light/dark cycle. Access to food and water was free. The protocol was approved by the Université Laval’s Animal Protection Committee and followed the recommendations of the Canadian Council on Laboratory Animal Care.

Aortic regurgitation

Severe AR was induced by retrograde puncture of the aortic valve leaflets as previously described [8,9]. A complete echo cardiographic exam was performed two weeks after AR induction and the day before sacrifice 8 weeks later. At the end of the protocol, animals were sacrificed, hearts were quickly dissected and all cardiac chambers were weighed. LV was snap-frozen in liquid nitrogen and kept at -80°C for further analysis. All sacrifices were scheduled at similar times of the day in the fed state to avoid circadian variations in metabolism. Lungs, liver and abdominal fat were rapidly collected and weighed.

Echocardiography

A complete M-mode, 2D, and Doppler echocardiogram was performed on the animals under 1.5% inhaled isoflurane anesthesia using a 12 MHz probe with a Sonos 5500 echograph (Philips Medical Imaging, Andover, Mass). LV dimensions, wall thickness, ejection fraction, cardiac output (ejection volume in the LV outflow tract and heart rate) were evaluated as previously reported [10-12].

Imaging experiments and data analysis were performed essentially as described before [13-16] on a LabPET™ avalanche photodiode-based small animal PET scanner (Gamma Medica, Northridge, CA) at the Sherbrooke Molecular Imaging Centre. [¹⁸F]-fluorodeoxyglucose ([¹⁸F]-FDG), [¹⁸F]-fluorothioheptadecanoic acid ([¹⁸F]-FTHA) or [¹¹C]-acetate(30–40 MBq, in 0.3 ml plus 0.1 ml flush of 0.9% NaCl, respectively) was injected via the caudal vein over 30 s. In one set of experiments, a 10-min dynamic acquisition with [¹¹C]acetate and a 45-min dynamic acquisition with [¹⁸F]FDG were done to determine myocardial oxidative metabolism [O2 uptake (Vo2)] and glucose utilization [myocardial metabolic rate of glucose (MMRG)]. In another experiment, a 45-min dynamic acquisition with [¹⁸F] FTHA was used to determine myocardial non-esterified fatty acid (NEFA) uptake (Km) [14]. Blood samples were taken before and after the scans to determine an average blood glucose level. Image data analysis was performed as described previously [15]. Myocardial NEFA fractional uptake (Ki) was determined by a Patlak graphical analysis of the [¹⁸F] FTHA data [14]. For [¹¹C] acetate, we used a three-compartment kinetic model that estimates the generation of CO2 from the citric acid cycle in the myocardium using the K2 value [16]. The myocardial metabolic rate of glucose (MMRG) was determined by multicompartmental analysis of the [¹⁸F] FDG data [13,14].

Analysis of mRNA accumulation by quantitative RT-PCR

The analysis of LV mRNA levels by quantitative RT-PCR has been described in details elsewhere [17]. Briefly, one μl RNA (500 ng) was converted to cDNA using the QuantiTect Reverse Transcription kit (Qiagen, Valencia, CA). The cDNA obtained was further diluted 11-fold with water prior to amplification. Five μl of diluted cDNA were amplified in duplicate by Q-PCR in a Rotor-Gene thermal cycler (Corbett Life Science. Sydney, Australia), using QuantiTect Primer Assays (pre-optimized specific primer pairs from Qiagen) and QuantiFast^® SYBR Green PCR kits (Qiagen). Cyclophilin A as a housekeeping gene was used as a control. In (Table 1), the list of primers used in this study is illustrated.

Table 1: QuantiTect^® Primer Assays used in Q-PCR analysis of gene expression.

Enzyme activity determinations

Left ventricle samples were kept at -80°C until assayed for maximal (V_max) enzyme activities. Small pieces of LV (20-30mg) were homogenized in a glass-glass homogenizer with 9 or 39 volumes of ice-cold extracting medium pH 7.4 (250mM sucrose, 10mM Tris-HCl, 1mM EGTA) depending on the enzyme activity assayed. Enzymatic activities for hydroxyacyl-Coenzyme A dehydrogenase (HADH), phosphofructokinase (PFK) and citrate synthase (CS) were determined as previously described [7].

Immunoblotting

Crude LV homogenates were separated by SDS-PAGE. Volumes of samples loaded on gel were corrected for the amount of protein. Immunoblotting was performed as described elsewhere [18]. Membranes were hybridized with the indicated primary antibodies. All primary antibodies against the phosphorylated or the total form of the different signaling proteins (Erk 1/2, p38, Jnk, AMP kinase, LKB1 and Akt) were used at a 1:1000 dilution and were purchased from Cell Signaling Technology (Beverly, MA). Bands were visualized and quantified with a ChemiImager system (Alpha Innotech Corporation).

Statistical analysis

Results are presented as mean ± SEM unless specified otherwise. Inter-group comparisons were done using two-way ANOVA and using Bonferroni post-test if necessary. Student t-test was used when two groups were compared head to head. Statistical significance was set at a p<0.05. Data and statistical analysis were performed using Graph Pad Prism version 5.02 for Windows, Graph Pad Software (San Diego CA).

Increased myocardial glucose and decreased fatty acid uptake in AR rat myocardium

We compared using μPET imaging the glucose and fatty acid uptake of left ventricles of AR animals 8 weeks post-induction to sham-operated animals. As illustrate in (Figure 1), glucose uptake was increased in AR rats while fatty acid uptake was decreased. Interestingly, despite these changes, myocardial oxygen consumption remained normal between groups.

Figure 1: Myocardial glucose (top) and fatty acid (middle) uptakes as well as oxidative metabolism (bottom) modulation in the LV myocardium of AR rats as evaluated by μPET. Myocardial rate of glucose (MMRG) and FTHA uptake and myocardial oxidative metabolism using acetate were evaluated as described in the Materials and Methods section in both sham-operated and AR rats 8 weeks post-surgery (n=4/gr.) and are expressed as the mean ± SEM. P values were calculated using the student t-test.

Clinical data and animal characteristics

At the end of the protocol, all animals were alive with the exception of one in the ARC group. Body weight and tibial length was equivalent in all experimental groups (Table 2). Total heart weight was greatly increased in rats with AR. Metformin had no effect on this parameter. The weight of both left and right ventricles were markedly increased in AR animals but again metformin had no effect on the weight of both ventricles respectively.

Table 2: Animal characteristics.

Echocardiographic LV remodeling and hemodynamic data

End-diastolic and end-systolic LV dimensions were increased in AR animals as expected (Table 3). Metformin treatment reduced both systolic and diastolic LV dilatation in AR as shown by the reduced diastolic and systolic diameter in ARM vs. ARC group. Wall thicknesses were increased in the ARM group vs. ARC but similar to the shams treated with metformin. Relative wall thickness (RWT) was therefore normalized in ARM thereby confirming less eccentric remodeling in animals treated with metformin. Interestingly, RWT also increased in sham-operated animals on metformin suggesting that the drug has an impact on LV remodeling even in normal animals. As illustrated in (Figure 2) in Trans axial LV sections, AR was associated with larger LVs and metformin seemed to increase wall thickness. Systolic function evaluated by fractional shortening was clearly improved in ARM animals compared to the ARC. ARM rats had a lower (closer to normal) stroke volume in this group and a closer to normal calculated cardiac output (Figure 3). Pulse pressure increased in both ARM and ARC groups as expected and metformin did not have any effect on this parameter (results not shown).

Table 3: Echocardiography data.

Figure 2: Typical macroscopic examples of left ventricular hypertrophy and remodeling in specific groups. Representative midventricular sections of the left ventricle stained with trichrome-Masson are shown. ShamC: control sham group, ShamM: metformin-treated sham group, ARC: untreated AR group and ARM: metformin-treated AR group. Scale bar at the bottom right of the figure: 1cm.

Figure 3: Metformin treatment effects on stroke volume and cardiac output as evaluated by echocardiography. Results are reported in % of change relative to the untreated control sham group (SC) as mean ± SEM (n = 13–15/group). Two-way ANOVA statistical analysis results are displayed below each graph. D X T: disease and metformin treatment interaction *: p<0.05 vs. untreated corresponding group from Bonferroni post-test if P<0.05 for D x T.

Markers of LV stress and hypertrophy

As expected, ANP and BNP expressions in LV myocardium were increased in AR animals as a result of the important pro-hypertrophic hemodynamic stress caused by AR. Metformin did not decrease the level of expression of ANP or BNP. A similar pattern of expression was also observed for the endothelin-1 mRNA. Fos expression is also usually associated with hypertrophy in the myocardium. As expected, the level of Fos expression was increased by AR groups. Metformin treatment was associated with a strong trend to decrease Fos expression in ARM animals. This decrease in Fos expression by metformin was also present in the sham animals (Figure 4).

Figure 4: Up-regulation of hypertrophy markers in AR animals. LV mRNA levels of atrial natriuretic peptide (ANP), brain natriuretic peptide (BNP), endothelin-1 (ET-1) and c-Fos were evaluated as described in the Material and methods section. Results are reported in arbitrary units as mean ± SEM (n = 13–15/group). Untreated sham (sham operated animals) group mRNA levels were normalized to 1. Two-way ANOVA statistical analysis results are displayed below each graph. D X T: disease and metformin treatment interaction.

Myocardial extracellular matrix remodeling

The expression of various markers involved in myocardial fibrosis was evaluated in LV myocardium. AR increased the expression of collagen III and IV as well as Lox and TGFβ and fibronectin. The expression of collagen I remained normal in ARC. The expression of MMP2 and TIMP1 were also increased by AR. Metformin treatment in AR increased the expression of collagen I and III as well as fibronectin. It also tended to increase TIMP1 expression. However, we found no increase in fibrosis on direct LV tissue Trichrome-Masson staining (results not shown) (Figure 5).

Figure 5: Evaluation by real-time quantitative RT-PCR of the LV mRNA levels of genes related to extracellular matrix remodeling. Results are reported as mean ± SEM (n=10-15/gr). Col I: collagen I; Col III: collagen III; Col IV Fn: fibronectin; LOX1: lysyl oxidase 1; TGFb1: transforming growth factor; MMP2: matrix metallo protease 2 and TIMP-1: tissue inhibitor of metalloprotease 1. Two-way ANOVA statistical analysis results are displayed below each graph. D X T: disease and metformin treatment interaction *: p<0.05 vs. untreated corresponding group from Bonferroni post-test if P<0.05 for D x T.

Myocardial metabolism

Eight weeks after AR induction, changes in energy substrate uptake are observed in the myocardium as illustrated in (Figure 1). We measured the activity levels of various metabolic enzymes in LV tissue of all groups. Results are shown in (Figure 6). The HADH (hydroxyacyl-Coenzyme A dehydrogenase) responsible for fatty acid β-oxidation was less active in the AR group after 8 weeks compared to sham animals. Metformin treatment had no effect on this parameter. The first step of glycolysis is catalyzed by the hexokinase (HK). HK activity levels were significantly increased in all AR animals compared to the shams and metformin further increased this activity. The entry of acetyl-CoA in the citric acid cycle is catalyzed in the mitochondria by the citrate synthase (CS). CS activity levels were however significantly lower in AR animals. Metformin treatment increased CS activity levels in the LV of both sham and AR animals.

Figure 6: Activity levels of enzymes implicated in myocardial energy metabolism. HADH (hydroxyacyl-Coenzyme A dehydrogenase), HK (hexokinase) and CS (citrate synthase) enzymatic activities were measured in LV homogenates from at least 10 animals in each group as described in the Materials and Methods. Results are reported as mean ± SEM (n=10- 15/gr). Two-way ANOVA analyses are displayed below each panel. D X T: disease and metformin treatment interaction.

With the exception of Glut1 mRNA expression, metformin treatment did not modulate other genes related with the control of energy metabolism tested in this protocol (Figure 7). Interestingly, the majority of them were down-regulated in AR animals.

Figure 7: Evaluation by real-time quantitative RT-PCR of the LV mRNA levels of genes related to cardiac metabolism. Glucose transporters 1 and 4 (Glut1 and Glut4), fatty acid transporter (FAT/CD36), carnityl palmitoyl transferases (Cpt 1b and 2), PDK4 (pyruvate dehydrogenase kinase 4) PDH1a (pyruvate dehydrogenase 1alpha), MCAD (medium-chain acyl- CoA dehydrogenase),UCP3 (uncoupling protein 3), DGK ζ (diacylglycerol kinase ζ), Ant1 (adenine nucleotide transferase 1), PPAR α (peroxisome proliferator-activated receptor α) and PGC1 α (peroxisome proliferatoractivated receptor gamma coactivator-1 α). Results are reported in arbitrary units as mean ± SEM (n = 10/group). Untreated sham (sham-operated animals; SC) group mRNA levels were normalized to 1 and are represented by the black lane.*: p<0.05 vs. ARC group.

Metformin treatment impact on signaling in AR rats

We measured the ratio of the phosphorylated form on the total content of a number of signaling molecules implicated in both hypertrophic and energy metabolism pathways. As illustrated in (Figure 8), activated Erk1/2 and Jnk1/2 are more abundant in AR LVs compared to sham-operated ones. Metformin treatment was associated with a decrease in the content of phosphorylated forms of p-38 and Jnk1/2 and an expected increase was observed for the AMP kinase.

Figure 8: MAP kinases, LKB1, AMP kinase and Akt levels of activation in AR rats treated or not with metformin. Evaluation of the phosphorylated and the total protein contents for each signaling molecules were performed by immunoblotting as described in the Materials and methods section. Results are reported in arbitrary units as mean ± SEM (n=10–12/group) of the ratio of the phosphorylated content on the total protein content. Untreated sham (sham-operated animals; SC) group protein contents were normalized to 1 and are represented by the black lane. *: p<0.05 vs. ARC group.

This study shows that an 8-week treatment with metformin reduces eccentric LV remodeling and dilatation in an experimental model of volume overload caused by severe aortic valve regurgitation. Metformin also helped maintain systolic function (ejection fraction). It also increased the activity of metabolic enzymes HK and CS. The treatment was not powerful enough however to prevent LV hypertrophy. This was expected considering the acute severe stress imposed to the LV. The fact that metformin was able to reduce LV dilatation and eccentric remodeling under such acute stress conditions, is in itself remarkable and exciting. In line with the law of Laplace, we can suppose that wall stress was reduced in AR rats treated with metformin (less dilatation and thicker walls). Our results show that metformin has the potential to act as a protector of the LV in volume overload. Long term studies will answer the question whether this hypothesis remains true over time.

The main mechanism of action of metformin is believed to be via the activation of the AMPK [6]. Metformin does not bind directly to AMPK. One hypothesis is that it inhibits the respiratory chain Complex I causing an increase in the AMP/ATP ratio [19]. In our study, metformin treatment led to AMPK activation. After 8 weeks, AMPK levels of activation were similar to controls in the LV of AR rats. On the other hand, LKB1 phosphorylation was reduced in AR rats and metformin restored normal levels of LKB1 activation. LKB1 is a known AMPK kinase together with the calmodulin-dependent protein kinase (CamKK) which is expressed only at low levels in cardiac myocytes but other kinases have also been proposed to be responsible for the metformin-induced phosphorylation of the AMPK such as PKC ζ [20]. It has also been shown the metformin can exert beneficial metabolic effect independently from the AMPK in cardiac myocytes presumably via p38 or PKC-related signaling [21]. Metformin has been reported to decrease hypertrophy and slow the onset of heart failure in several animal models [22-24].

Cardiac hypertrophy is associated with the reappearance of a “fetal profile” of gene expression. We observed such a “fetal profile” in this study and previous ones using our AR model after a longer follow-up (> 6 months) with a modification in the expression of heart myosin heavy chains subtypes [12] and increased expression of Glut 1 and c-Fos. This profile is associated with increased glucose utilization and was at first considered maladaptive [4]. The observation that cardiac over expression of GLUT1 prevented the development of heart failure (and LV dilation) attributable to pressure overload in mice with little effects on overall LVH changed this perception [25]. It is now believed that glucose reliance in the adult heart is not harmful while reduced ability to utilize glucose could be detrimental in cardiac hypertrophy or heart failure [4]. In our study metformin treatment was associated with an increased expression of Glut1 and a raised activity of the hexokinase in AR animals with an already augmented glucose uptake in AR animals. Increased AMPK activity was shown to promote the translocation of the glucose transporters [26].

It is not clear how metformin (via AMPK activation or not) can modulate LV remodeling caused by severe chronic VO. In a recent study in a model of global heart volume overload caused by aortocaval fistulae (ACF) in rats, metformin was shown to have no impact on the heart and LV weight in ACF rats while echo data showed a trend similar to ours (less dilatation). In that study the dose of metformin was twice the one we used [27]. In another study in rats post-myocardial infarction, metformin was shown to decrease heart dilatation while limiting LV wall thinning [28].

We observed that metformin decreased the protein content of the phosphorylated forms of both stress kinases p38 and Jnk. Both kinases have been shown in the past to have pro-hypertrophic action [29]. Inhibition of p38 in H9C2 cells was shown to decrease glycolysis [21]. Metformin has also been shown to inhibit Jnk activation in the past (Jung BBRC 2012). It is not clear whether metformin is acting on LV remodeling in our model via the inhibition of these pathways.

Metformin was shown in the past to slow the development of myocardial interstitial fibrosis [28]. Interstitial fibrosis is usually less present in VO-induced hypertrophy than in pressure overload situations. We have shown in the past that LV collagen content is still normal after 8 weeks in AR rats even though gene expression of many extracellular matrix (ECM) components is increased [10]. Here we observed that metformin further increased collagen type I and III gene expression in AR LVs. A trend for a higher expression of other ECM components was also present in metformin-treated animals. This suggests that metformin treatment could be associated with a higher turnover of ECM proteins. However this may also be an early sign of increased myocardial fibrosis in a more chronic state. Longer studies will be necessary to better characterize this observation.

The results of this study have to be viewed in light of some limitations. This was a relatively short study in which metformin was given as a pre-disease treatment. Eight weeks may be too short to assess the impact of metformin treatment on systolic and diastolic functions and clinical evolution on the long term. Rodent heart metabolism may also differ in some aspects from humans.

In conclusion, metformin treatment is associated with less eccentric remodeling, less LV dilatation and a better systolic function in an experimental model of LV volume overload caused from aortic valve regurgitation. Long term studies evaluating the effects of such treatment on LV function and survival are mandated.