Metabolic Effects of Metformin in the Failing Heart

Accumulating evidence shows that metformin is an insulin-sensitizing antidiabetic drug widely used in the treatment of type 2 diabetes mellitus (T2DM), which can exert favorable effects on cardiovascular risk and may be safely used in patients with heart failure (HF), and even able to reduce the incidence of HF and to reduce HF mortality. In failing hearts, metformin improves myocardial energy metabolic status through the activation of AMP (adenosine monophosphate)-activated protein kinase (AMPK) and the regulation of lipid and glucose metabolism. By increasing nitric oxide (NO) bioavailability, limiting interstitial fibrosis, reducing the deposition of advanced glycation end-products (AGEs), and inhibiting myocardial cell apoptosis metformin reduces cardiac remodeling and hypertrophy, and thereby preserves left ventricular systolic and diastolic functions. While a lot of preclinical and clinical studies showed the cardiovascular safety of metformin therapy in diabetic patients and HF, to confirm observed benefits, the specific large-scale trials configured for HF development in diabetic patients as a primary endpoints are necessary.


Introduction
As first reported in the Framingham study, type 2 diabetes mellitus (T2DM) is not only an independent risk factor for cardiovascular diseases (CVD), but is also associated with a higher incidence of heart failure (HF). The risk of heart failure in patients with diabetes is almost twice higher for men and five times for women when compared with the general population. Moreover, patients with diabetes account for one third of all cases of HF [1]. As the prevalence of diabetes rapidly rises worldwide, the incidence of HF consequently increases in parallel [2]. The concomitance of diabetes and heart failure significantly worsens a prognosis in this group of patients. There is a tight positive correlation between hyperglycemia and heart failure development and progression. Diabetic patients have an 8% increase in the risk of developing HF with every 1% elevation of glycated hemoglobin (HbA1c) [3,4]. While the development of HF is related to poor glycemic control, even patients with good metabolic profiles have an increased risk [5]. Nowadays, there are many classes of anti-diabetic medication; nevertheless their role in the therapy of diabetes with heart failure is still undefined. Some drugs used for diabetes have been associated with increased rate of hospitalization for HF, despite its neutral or positive effect on overall risk of major cardiovascular events. For example, saxagliptin and alogliptin should be avoided in patients with diabetes and HF and pioglitazone is contraindicated in this population [6,7]. On the other hand, it is suggested that some antihyperglycemic drugs can reduce the rate of hospitalization due to heart failure and reduce mortality. These benefits have been proven for empagliflozin, an inhibitor of the sodium-glucose co-transporter 2 (SGLT-2) [8] and metformin, the biguanide derivative. resveratrol, and thiazolidinediones. By contrast, resistin and endocannabinoids inhibit enzyme activity [27].
Metformin-induced energy deficiency in the cell causes AMPK activation and inhibits the processes in which ATP is consumed such as gluconeogenesis, cholesterol, and fatty acids synthesis or glycogen synthesis (Figure 1). Simultaneously activated metabolic pathways producing ATP like free fatty acids (FFA) oxidation in the liver and skeletal muscles or glycolysis [12,23].
In the liver, metformin also affects lipid synthesis and catabolism. By activating AMPK, metformin down-regulates SREBP-1 (sterol regulatory element binding protein 1) gene expression. SREBP-1 regulates transcription of genes encoding lipogenic enzymes, such as fatty acid synthase (FAS). In addition, AMPK activation inhibits acetyl-CoA carboxylase (ACC) and thereby malonyl-CoA synthesis. Malonyl-CoA is the substrate for fatty acid synthesis and an inhibitor of CPT-1 (carnitine palmitoyltransferase 1) [32], the enzyme that transports fatty acids to mitochondria [33]. Therefore, suppression of malonyl-CoA synthesis decreases production of fatty acids and triglycerides in hepatocytes, increasing at the same time FFA oxidation in mitochondria. Moreover, AMPK activation decreases cholesterol synthesis in hepatocytes as a result of HMG-CoA reductase (3-hydroxy-3-methyl-glutaryl-coenzyme A reductase) inhibition [34].
In adipocytes metformin regulates adipogenesis, lipolysis, and FFA oxidation, thus reduces the release of FFA from the adipose tissue. The reduction of plasma FFA level is associated with more efficient glucose uptake by the cells of peripheral tissues [38,39].
In the intestines metformin inhibits absorption of glucose and other simple carbohydrates [40]. In addition, it has been proved that metformin is able to improve the action of intestine-pancreas axis (incretin effect). The drug is considered to directly increase secretion of glucagon-like peptide-1 (GLP-1) from enterocytes. In addition through inhibiting dipeptidyl peptidase-4 (DPP-4), an enzyme responsible for incretin degradation, metformin may prolong the duration and activity of endogenous incretins [37,41]. Metformin may also exert its glucose-lowering and insulin-sensitizing action via modifying microbiota composition [42]. Furthermore, it may contribute to the improvement of pancreatic β-cell function by reducing gluco and lipotoxicity [43,44].
Metformin is slowly absorbed from the proximal small intestine and the rate of absorption depends on the dose. The drug accumulates in the gastrointestinal tissues that may contribute to the gastrointestinal-associated side effects [45]. Intestinal absorption of metformin is probably facilitated by plasma membrane monoamine transporter (PMAT) as well as OCT1 and OCT3 (organic cation transporters) [46]. OCTs play important role in distribution of metformin over body tissues, such as intestine, kidney, and liver [46,47]. Steady-state plasma concentrations of metformin may vary from one to 10 µM [45]. The drug is not metabolized in the liver and is excreted unchanged in urine. It is eliminated by active tubular secretion, with a half-life of approximately five h [46,48]. The uptake of the drug from circulation into renal epithelial cells is mediated through OCT2 and its excretion from the tubular cell into the urine is facilitated by MATE1 and MATE2-K (multidrug and toxin extrusion proteins) [46]. It is suggested that genetic polymorphisms in drug transporters are associated with wide variations in pharmacokinetics profile and a large inter-individual variability in metformin response [47,48].

Diabetic Cardiomyopathy
Diabetes mellitus is an independent risk factor of congestive heart failure [49]. Coronary artery disease, arterial hypertension, and diabetic cardiomyopathy are the main factors that contribute to the development of HF in diabetes. The risk of heart failure in T2DM remains higher compared to non-DM (diabetes mellitus) individuals after adjustment for other risk factors such as serum cholesterol levels, hypertension, or coronary artery disease [50,51]. On the other hand, the presence of DM worsens prognosis in the heart failure population in which the mortality rates are significantly higher than in the non-DM population [52].
Diabetic cardiomyopathy (DMC) is a damage of myocardium that occurs in diabetes independently of concomitant disorders like hypertension and ischemic heart disease [53,54]. Cardiomyopathy is characterized by increased myocardial stiffness and left ventricular mass (cardiac hypertrophy) [55,56]. At the beginning, it is an asymptomatic impairment of left ventricular diastolic function that leads to symptomatic diastolic dysfunction and finally the systolic function becomes impaired too [57][58][59]. Diastolic left ventricular (LV) dysfunction (a restrictive phenotype) has been linked to concentric hypertrophy of LV, and eccentric remodeling of LV has been linked to systolic dysfunction (a dilated phenotype). When becoming symptomatic, diabetic patients with a restrictive phenotype of DMC present heart failure with preserved ejection fraction (HFpEF), which is typically characterized by small LV cavity, normal LV ejection fraction (EF), thick LV walls, elevated LV filling pressure, and large left atrium. Whereas, in patients with a dilated cardiomyopathy, clinical heart failure with reduced ejection fraction (HFrEF) is recognized. These patients have abnormal-sized LV and abnormal LV ejection fraction [60].
On pathological examination in DMC, there was myocardial hypertrophy, fibrosis and microvascular wall thickening. At the tissue level, DMC is characterized by interstitial or replacement fibrosis, increases in extracellular matrix [60,61], and deposition of advanced glycation end-products (AGEs). All these changes contribute to reduced diastolic and systolic compliance and ventricular hypertrophy. At the cellular level in DM-related cardiomyopathy, the hypertrophy as well as fragmentation and degeneration of myocytes are observed [59,60]. The latter is particularly strongly expressed in the dilated type of DMC [60]. At the molecular level, DMC is connected with

Diabetic Cardiomyopathy
Diabetes mellitus is an independent risk factor of congestive heart failure [49]. Coronary artery disease, arterial hypertension, and diabetic cardiomyopathy are the main factors that contribute to the development of HF in diabetes. The risk of heart failure in T2DM remains higher compared to non-DM (diabetes mellitus) individuals after adjustment for other risk factors such as serum cholesterol levels, hypertension, or coronary artery disease [50,51]. On the other hand, the presence of DM worsens prognosis in the heart failure population in which the mortality rates are significantly higher than in the non-DM population [52].
Diabetic cardiomyopathy (DMC) is a damage of myocardium that occurs in diabetes independently of concomitant disorders like hypertension and ischemic heart disease [53,54]. Cardiomyopathy is characterized by increased myocardial stiffness and left ventricular mass (cardiac hypertrophy) [55,56]. At the beginning, it is an asymptomatic impairment of left ventricular diastolic function that leads to symptomatic diastolic dysfunction and finally the systolic function becomes impaired too [57][58][59]. Diastolic left ventricular (LV) dysfunction (a restrictive phenotype) has been linked to concentric hypertrophy of LV, and eccentric remodeling of LV has been linked to systolic dysfunction (a dilated phenotype). When becoming symptomatic, diabetic patients with a restrictive phenotype of DMC present heart failure with preserved ejection fraction (HFpEF), which is typically characterized by small LV cavity, normal LV ejection fraction (EF), thick LV walls, elevated LV filling pressure, and large left atrium. Whereas, in patients with a dilated cardiomyopathy, clinical heart failure with reduced ejection fraction (HFrEF) is recognized. These patients have abnormal-sized LV and abnormal LV ejection fraction [60].
On pathological examination in DMC, there was myocardial hypertrophy, fibrosis and microvascular wall thickening. At the tissue level, DMC is characterized by interstitial or replacement fibrosis, increases in extracellular matrix [60,61], and deposition of advanced glycation end-products (AGEs). All these changes contribute to reduced diastolic and systolic compliance and ventricular hypertrophy. At the cellular level in DM-related cardiomyopathy, the hypertrophy as well as fragmentation and degeneration of myocytes are observed [59,60]. The latter is particularly strongly expressed in the dilated type of DMC [60]. At the molecular level, DMC is connected with the disturbance of calcium transport into cardiomyocytes, disorders of fatty acid metabolism and decrease in Na + , K + -ATPase activity [59,62].
The underlying pathophysiology of DMC is multifactorial but can largely be attributed to metabolic derangements that accompany T2DM like insulin resistance, hyperinsulinemia, hyperglycemia and dyslipidemia. These perturbations disturb myocardial metabolism and impair endothelial function of coronary micro vessels (microangiopathy). The main feature of endothelial dysfunction is reduced bioavailability of NO (nitric oxide), which leads to disrupted nitric oxide (NO)-cyclic guanosine monophosphate (cGMP)-protein kinase G (PKG) signaling pathway and predisposes to concentric LV remodeling and diastolic stiffness of LV (restrictive phenotype). Capillary rarefaction, defined as a decrease in the number of perfused capillaries in an area of tissue, leads to hypoxia and oxidative stress, cardiomyocyte death, and DMC with dilated phenotype [60].
Vascular endothelial dysfunction, manifested with impaired NO production, is thought to be one of key risk factors for diabetic cardiomyopathy. Nitric oxide plays a key role in maintaining cardiac and vascular homeostasis, which is achieved when there is a balance between the generation of NO and superoxide anion (O 2 − ). The cardiac endothelium comprises the endothelial cells of the coronary microvasculature, of the endocardium and the intramyocardial capillaries. The heart cardiomyocytes lies in close proximity to endothelial cells (maximum 3 µm from endothelial cells). This specific location allows adequate blood supply and facilitates the bidirectional communication among those cells [63].
In the heart NO is a powerful anti-hyperthrophic and anti-fibrotic agent. These protective effects of NO are mediated predominantly via activation of intracellular sGC (soluble guanylate cyclase) and subsequent cGMP (cyclic guanosine monophosphate) generation [64]. The reduction of NO-dependent signaling from endothelium to cardiomyocytes might lead to the impairment of NO-sGC signaling pathway and then to cardiac remodeling and fibrosis-consequently to ventricle stiffness, impaired relaxation, and cardiac dysfunction [65,66]. NO exerts its anti-fibrotic effects by the inhibition of the transforming growth factor (TGF-β1)-induced cardiac fibroblast proliferation, transformation and collagen synthesis, and trough activation of cGMP/PKG pathway [67]. Impaired insulin metabolic signaling in the heart results in metabolic stress which manifests in disturbed substrate uptake and utilization, disturbed energy status, mitochondrial dysfunction, increased gene expression of the stiffer protein, and increased fibroblast activity leading to collagen deposition in an extracellular matrix that contributes to myocardial remodeling and diastolic dysfunction [51,[68][69][70]. Insulin resistance-associated hiperinsulinemia can stimulate cardiomyocyte hypertrophy by binding to the IGF-1 (insulin growth factor 1) receptor, as well as stimulating a PI-3K/Akt (phosphatidylinositol-3-kinase/Akt) signaling pathway [51,71].
Most studies discussing diabetic cardiomyopathy have focused only on left ventricle, but the effects on the right ventricle in diabetic cardiomyopathy have not been studied extensively. Nevertheless, from a pathophysiological point of view, all proposed mechanisms leading to LV impairment in type 2 diabetes are systemic changes and therefore might also hamper right ventricular (RV) structure and function. Indeed, impairment of RV function has been reported in Zucker diabetic fatty rats [72] and in human studies [73]. Moreover, impaired parameters of RV diastolic function were shown in patients with type 2 diabetes suggesting that the similar mechanisms responsible for LV stiffness also affect the right ventricle. Alternatively, RV involvement could be the consequence of left ventricle changes, e.g., diffuse fibrotic processes that take place in diabetes could affect the function of both ventricles. Another possibility is that in diabetes, microangiopathy of the lung capillary [74] may cause increased right ventricle afterload leading to its dysfunction.

Metabolic Disorders in Diabetic Cardiomyopathy
Metabolic disorders that are observed in cardiomyocytes in patients with diabetes arise from toxic action of both high levels of FFA (lipotoxicity) and hyperglycemia (glucotoxicity). Physiologically, at normal work loads in aerobic conditions, the energy is acquired by cardiomyocytes mainly from fatty acid β-oxidation (60-90%). Only a small percentage of energy is produced in glycolysis and pyruvate oxidation [68,[75][76][77]. Both β-oxidation and glycolysis leads to the synthesis of acetyl-CoA, which is next oxidized in citric acid cycle [77]. Glycolysis requires less oxygen to synthesis one mole of ATP than FFA oxidation, hence glucose is used as a basic source of energy during ischemia and oxygen deficiency [76,78]. Metabolic disorders that occur in diabetes significantly impair the ability of the heart to adapt to overload conditions, e.g., physical activity [79,80].

Lipotoxicity
The impact of FFA overload on metabolism of the heart shows in Figure 2. In general, the main mechanism leading to insulin resistance is the excessive accumulation of adipose tissue, especially visceral adipose tissue, which is less sensitive to insulin but more susceptible to catecholamines; both these abnormalities result in accelerated lipolysis. Increased release of non-esterified fatty acids from adipocytes raises the level of circulating FFA [81,82] and impairs insulin signaling in the liver, skeletal muscles and adipose tissue [83][84][85][86]. It results in a decreased uptake and consumption of glucose in muscles, abolished the inhibitory effect of insulin on glucose synthesis in hepatocytes [87,88] and facilitated the uptake and accumulation of FFA into cardiomyocytes [49]. Increased activity of lipogenesis-regulating enzymes such as glycerol phosphate acyltransferase (GPAT) causes myocardial triglyceride accumulation, heart steatosis, and eventually cell death [89]. Excess of fatty acids inhibits cardiac glucose utilization. This is because the long-term exposure to high FFA level increases acetyl-CoA concentration in mitochondria, which inhibits the activity of glycolytic enzymes such as phosphofructokinase 1 (PFK-1) and pyruvate dehydrogenase (PDH). As a consequence, glucose oxidation decreases, which results in the accumulation of glycolysis intermediate products such as glucose-6-phosphate, fructose-6phosphate, pyruvate, and lactate [90]. produced in glycolysis and pyruvate oxidation [68,[75][76][77]. Both β-oxidation and glycolysis leads to the synthesis of acetyl-CoA, which is next oxidized in citric acid cycle [77]. Glycolysis requires less oxygen to synthesis one mole of ATP than FFA oxidation, hence glucose is used as a basic source of energy during ischemia and oxygen deficiency [76,78]. Metabolic disorders that occur in diabetes significantly impair the ability of the heart to adapt to overload conditions, e.g., physical activity [79,80].

Lipotoxicity
The impact of FFA overload on metabolism of the heart shows in Figure 2. In general, the main mechanism leading to insulin resistance is the excessive accumulation of adipose tissue, especially visceral adipose tissue, which is less sensitive to insulin but more susceptible to catecholamines; both these abnormalities result in accelerated lipolysis. Increased release of non-esterified fatty acids from adipocytes raises the level of circulating FFA [81,82] and impairs insulin signaling in the liver, skeletal muscles and adipose tissue [83][84][85][86]. It results in a decreased uptake and consumption of glucose in muscles, abolished the inhibitory effect of insulin on glucose synthesis in hepatocytes [87,88] and facilitated the uptake and accumulation of FFA into cardiomyocytes [49]. Increased activity of lipogenesis-regulating enzymes such as glycerol phosphate acyltransferase (GPAT) causes myocardial triglyceride accumulation, heart steatosis, and eventually cell death [89]. Excess of fatty acids inhibits cardiac glucose utilization. This is because the long-term exposure to high FFA level increases acetyl-CoA concentration in mitochondria, which inhibits the activity of glycolytic enzymes such as phosphofructokinase 1 (PFK-1) and pyruvate dehydrogenase (PDH). As a consequence, glucose oxidation decreases, which results in the accumulation of glycolysis intermediate products such as glucose-6-phosphate, fructose-6phosphate, pyruvate, and lactate [90].  When FFA uptake exceeds the ability of their utilization in the heart, FFA accumulates in the cytoplasm. Within the intracellular compartment, fatty acids (long chain fatty acyl-CoA) are converted into toxic ceramides, which induce cell apoptosis, or into diacylglycerol (DAG), which activates protein kinase C (PKC) [59,90]. PKC inactivates pyruvate dehydrogenase in cardiomyocytes and thus impairs glucose oxidation [91,92]. Furthermore, in endothelium PKC reduces the activity of eNOS (endothelial nitric oxide synthase), increases synthesis of endothelin 1 as well as stimulates NADPH oxidase (nicotinamide adenine dinucleotide phosphate oxidase), which generates reactive oxygen species (ROS), further inactivating nitric oxide [92][93][94][95]. High FFA levels may also affect structure and function of cardiomyocyte membranes and raise intracellular calcium concentration leading to disturbances of conduction and myocardial contractility [96][97][98].

Glucotoxicity
Glucose accumulation in cardiomyocytes inhibits FFA oxidation by increasing intracellular malonyl-CoA. Malonyl-CoA is an inhibitor of carnitine palmitoyltransferase I, that is responsible for transport of FFA into mitochondria [90,99]. In addition, hyperglycemia blocks FFA oxidation at the level of gene expression via inhibiting PPARα (peroxisome proliferator-activated receptor alpha) that regulates the expression of enzymes necessary for mitochondrial β-oxidation. In addition, intensified glycation of proteins involved in the insulin signaling pathways such as IRS that reduces cell sensitivity to insulin [90]. Insulin resistance is associated with reduced expression of glucose transporters GLUT-1 and GLUT-4 [83,100].
Moreover, advanced glycation end products generated during hyperglycemia induces the synthesis of reactive oxygen species that impair the function of ion transporters and mitochondria, affect calcium transport between cell compartments, and initiate apoptosis [90]. Moreover, glycation of collagen increases myocardial stiffness and impairs diastolic function [59]. Changes in extracellular matrix disturb conduction of impulses thus results in arrhythmias [61,101]. Oxidative stress caused by hyperglycemia and AGEs contributes to damage of autonomic neurons that regulate contractility of heart and coronary vessels. Parasympathetic neurons are damaged at first, resulting in the predominance of a sympathetic nervous system. Then sympathetic neurons are damaged too. Autonomic neuropathy leads to other disorders such as resting tachycardia and asymptomatic ischemia of myocardium [102].

Glucose and FFA Metabolism
In failing heart, the activation of metabolic pathways controlled by AMPK in cardiomyocytes is a mechanism that allows adapting to conditions under which the energy generation is reduced [103,104]. In this setting, changes in the level of factors regulating phosphofructokinase 1 activity have been observed. For example, the level of AMP, ADP (adenosine diphosphate) and inorganic phosphates, those are PFK-1 allosteric activators, increases and the level of PFK-1 inhibitors, such as ATP, citrate, and H + ions decreases [103]. Moreover, AMPK activates phosphofructokinase 2 (PFK-2), leading to the increased concentration of fructose 2,6-biphosphate, another PFK-1 activator produced in cardiomyocytes ( Figure 3). As PFK-1 is the rate-limiting enzyme of glycolysis, its activation causes intensification of glycolysis [26,78,103]. Furthermore, AMPK activation leads to the translocation of GLUT-4 to cardiomyocyte membrane and accelerates insulin-dependent glucose uptake [90,103]. In vitro studies confirm that increase in PFK-1 activator level and decreased level of citrate, phosphocreatine, and H + ions are accompanied by twofold higher glucose utilization than in control cardiomyocytes, in spite of comparable ATP generation and oxygen utilization [103]. In vitro study showed that metformin added to the incubation medium significantly increased glucose uptake by stimulating the phosphoinositide 3-kinase (PI 3-K)-protein kinase B/Akt pathway and AMPK activation. This positive effect was observed both in insulin resistant cardiomyocytes and in cells with normal insulin sensitivity [105]. Thus, in diabetic hearts due to AMPK activation, the correct glucose metabolism can be restored.
If in diabetic hearts the activation of the AMP-activated kinase cascade increases energy production through intensification of glucose oxidation, paradoxically it could lead to inhibition of β-oxidation and accumulation of FFA in cardiomyocytes and cardiac steatosis. Indeed, it has been proved that obese patients with impaired glucose tolerance or type 2 diabetes can have even twofold higher triglycerides level in the heart than healthy people, even without symptoms of left ventricular dysfunction [50]. Hence a question arises, will metformin treatment not worsen lipotoxicity resulting from accumulation of fatty acids and triglycerides? No unequivocal answer has been found so far. Nevertheless, in vitro studies showed that small doses of metformin did not worsen, but even protected heart muscle from fat-induced apoptosis. It has been observed that AMPK activated acyl-CoA oxidation and inhibited SPT (serine palmitoyltransferase), which is responsible for ceramide synthesis, as well as inhibited caspase 3, is involved in myocytes apoptosis [93].
On the other hand, there are studies showing that metformin at high doses may increase the number of cells undergoing apoptosis as a result of FFA accumulation [106]. As already mentioned, metformin-induced AMPK phosphorylation enhances not only glucose transport and glycolysis, but also fatty acid uptake and β-oxidation ( Figure 4). Furthermore, acetyl-CoA synthetized during β-oxidation blocks the stage of glycolysis in which pyruvate is oxidized by PDH. This inhibition results in a sudden release of lactate dehydrogenase (LDH), converting pyruvate into lactate. Accumulation of lactate causes pH reduction, Ca 2+ overload, and cell death. This effect is not observed when glucose is removed from the incubation medium. Lactate synthesis and pH Figure 3. Regulation of PFK activity in heart failure. ADP, adenosine diphosphate; AMP, adenosine monophosphate; AMPK, AMP activated protein kinase; ATP, adenosine triphosphate; PFK-1, phosphofructokinase 1; PFK-2, phosphofructokinase 2; and Pi, inorganic phosphate. Circles with "+" and "−" represent stimulatory and inhibitory effects, respectively. ↑ and ↓ represent increase and decrease in the concentration of specific compound, respectively.
In vitro study showed that metformin added to the incubation medium significantly increased glucose uptake by stimulating the phosphoinositide 3-kinase (PI 3-K)-protein kinase B/Akt pathway and AMPK activation. This positive effect was observed both in insulin resistant cardiomyocytes and in cells with normal insulin sensitivity [105]. Thus, in diabetic hearts due to AMPK activation, the correct glucose metabolism can be restored.
If in diabetic hearts the activation of the AMP-activated kinase cascade increases energy production through intensification of glucose oxidation, paradoxically it could lead to inhibition of β-oxidation and accumulation of FFA in cardiomyocytes and cardiac steatosis. Indeed, it has been proved that obese patients with impaired glucose tolerance or type 2 diabetes can have even twofold higher triglycerides level in the heart than healthy people, even without symptoms of left ventricular dysfunction [50]. Hence a question arises, will metformin treatment not worsen lipotoxicity resulting from accumulation of fatty acids and triglycerides? No unequivocal answer has been found so far. Nevertheless, in vitro studies showed that small doses of metformin did not worsen, but even protected heart muscle from fat-induced apoptosis. It has been observed that AMPK activated acyl-CoA oxidation and inhibited SPT (serine palmitoyltransferase), which is responsible for ceramide synthesis, as well as inhibited caspase 3, is involved in myocytes apoptosis [93].
On the other hand, there are studies showing that metformin at high doses may increase the number of cells undergoing apoptosis as a result of FFA accumulation [106]. As already mentioned, metformin-induced AMPK phosphorylation enhances not only glucose transport and glycolysis, but also fatty acid uptake and β-oxidation ( Figure 4). Furthermore, acetyl-CoA synthetized during β-oxidation blocks the stage of glycolysis in which pyruvate is oxidized by PDH. This inhibition results in a sudden release of lactate dehydrogenase (LDH), converting pyruvate into lactate. Accumulation of lactate causes pH reduction, Ca 2+ overload, and cell death. This effect is not observed when glucose is removed from the incubation medium. Lactate synthesis and pH reduction are also observed when lower doses of metformin are used, but metformin at low concentrations has no influence on cardiomyocytes survival [106]. reduction are also observed when lower doses of metformin are used, but metformin at low concentrations has no influence on cardiomyocytes survival [106]. In addition, in vivo studies showed that chronic administration of metformin declined FFA level in the plasma and increased their oxidation in non-diabetic cardiomyocytes. At the same time, the inhibition of glucose oxidation was observed, probably as a result of the glycolysis-blocking effect of metformin, directly or indirectly, by acetyl-CoA synthesized during FFA oxidation. Increased β-oxidation and inhibition of glycolysis had no influence on heart failure progression [107].

Protein Synthesis
In vitro studies confirmed that pharmacological activation of AMPK could prevent heart muscle hypertrophy by inhibiting protein synthesis [108,109]. It has been postulated that there are two signaling pathways that play key roles in protein synthesis: eEF-2 (eukaryotic elongation factor 2) pathway and p70S6 (ribosomal protein S6) kinase pathway [109]. eEF-2 regulates the movement of the ribosome along the mRNA during elongation [110], while p70S6 kinase phosphorylases eEF-2 kinase [111] and ribosomal protein S6 [112]. It has been proven that metformin-induced AMPK activation inhibits protein synthesis by increasing the level of phosphorylated, (inactive) eEF-2 protein and decreasing phosphorylation of p70S6 kinase, leading to activation of the eEF-2 kinase, as well as to the inhibition of ribosomal protein S6 activity [108].

Mitochondrial Function
The other mechanism explaining the beneficial effect of metformin on metabolism of failing heart is improvement of mitochondria function in cardiomyocytes. In experimental heart failure induced by myocardial ischemia in mice, four-week administration of small doses of metformin significantly improved left ventricular function and structure and increased animal survival almost by 47%. This effect was related to elevated AMPK phosphorylation and increased the expression of eNOS and PGC-1α (peroxisome proliferator-activated receptor gamma coactivator 1-alpha) [113]. Both eNOS and PGC-1α are important regulators of biogenesis and function of mitochondria [114][115][116], and their activation improves oxidative metabolism in cardiomyocytes, i.e., increases ATP In addition, in vivo studies showed that chronic administration of metformin declined FFA level in the plasma and increased their oxidation in non-diabetic cardiomyocytes. At the same time, the inhibition of glucose oxidation was observed, probably as a result of the glycolysis-blocking effect of metformin, directly or indirectly, by acetyl-CoA synthesized during FFA oxidation. Increased β-oxidation and inhibition of glycolysis had no influence on heart failure progression [107].

Protein Synthesis
In vitro studies confirmed that pharmacological activation of AMPK could prevent heart muscle hypertrophy by inhibiting protein synthesis [108,109]. It has been postulated that there are two signaling pathways that play key roles in protein synthesis: eEF-2 (eukaryotic elongation factor 2) pathway and p70S6 (ribosomal protein S6) kinase pathway [109]. eEF-2 regulates the movement of the ribosome along the mRNA during elongation [110], while p70S6 kinase phosphorylases eEF-2 kinase [111] and ribosomal protein S6 [112]. It has been proven that metformin-induced AMPK activation inhibits protein synthesis by increasing the level of phosphorylated, (inactive) eEF-2 protein and decreasing phosphorylation of p70S6 kinase, leading to activation of the eEF-2 kinase, as well as to the inhibition of ribosomal protein S6 activity [108].

Mitochondrial Function
The other mechanism explaining the beneficial effect of metformin on metabolism of failing heart is improvement of mitochondria function in cardiomyocytes. In experimental heart failure induced by myocardial ischemia in mice, four-week administration of small doses of metformin significantly improved left ventricular function and structure and increased animal survival almost by 47%. This effect was related to elevated AMPK phosphorylation and increased the expression of eNOS and PGC-1α (peroxisome proliferator-activated receptor gamma coactivator 1-alpha) [113].
Both eNOS and PGC-1α are important regulators of biogenesis and function of mitochondria [114][115][116], and their activation improves oxidative metabolism in cardiomyocytes, i.e., increases ATP synthesis and restores normal ratio of ATP synthesis/oxygen use. By contrast, in other studies performed in transgenic animals with inactive AMPKα2 and eNOS genes, this cardioprotective effect of metformin was not observed [113].

iNOS Activity and Calcium Ions Transport
In the heart tissue, there are expressed three isoforms of NO-synthase (NOS)-endothelial NOS (eNOS), neuronal NOS (nNOS), and inducible NOS (iNOS). Endothelial NOS is principally expressed in endothelial cells and cardiomyocytes and is a key source of NO [64]. Neuronal NOS is mainly located in cardiomyocytes. Expression of iNOS has been found in immune cells, cardiomyocytes, and activated fibroblast. In heart muscle, the main physiological role is played by NO generated by eNOS and nNOS [117].
Increased expression of inducible NO synthase (iNOS) is observed during myocardial ischemia in both damaged and non-damaged areas in cardiomyocytes, endothelial cells, and macrophages. It is suggested that the enzyme may be involved in the development of late complications of myocardial infarction such as congestive heart failure [118]. It has been proven that excessive synthesis of NO by iNOS contributes to the impairment of diastolic and systolic LV function. Selective iNOS inhibitors, such as SMTU (S-methylisothiourea) and aminoguanidine improve myocardial contractility. On the other hand, L-arginine, an unselective iNOS substrate, causes the opposite effect, disturbing myocardial function [119]. In contrast to eNOS that synthesizes small nanomolar amounts of NO, iNOS produces NO in micromolar concentration [120]. High cellular NO level, via cGMP, a second messenger, activates PKG and cGMP-activated phosphodiesterase. PKG blocks L-type calcium channels and consequently decreases calcium influx into the cells, while phosphodiesterase breaks down cAMP [119]. Decreases in the intracellular Ca 2+ and cAMP concentrations may contribute to impaired myocardial contractility under conditions of iNOS-related NO overproduction [49,119].
Long term exposition of intra and extracellular structures to toxic NO concentrations can cause tyrosine nitration in specific proteins and induce apoptosis [119]. Metformin in the in vitro studies has been shown to inhibit lipopolysaccharide-induced elevation of iNOS mRNA (messenger ribonucleic acid) expression in macrophages. Therefore, metformin via blocking the excessive NO generation can reduce formation of cardiotoxic peroxynitrite. This effect was partially dependent on AMPK phosphorylation [121]. It should be emphasized that iNOS activity is regulated at the level of gene expression, mainly by proinflammatory cytokines, such as IL-1β (interleukin-1β) [122]. In vitro studies proved that, through AMPK activation, metformin inhibited synthesis of IL-1β in activated macrophages [121]. It is also noteworthy, that metformin, by activating AMPK, facilitates eNOS phosphorylation/activity [123] and reduces the activity of TGF-β1, which is the basic pro-fibrotic growth factor in the cardiovascular system myocardium [124].

Collagen Synthesis and Glycation
In diabetes, the important mechanism leading to impaired heart muscle relaxation is the interstitial accumulation of collagen and its non-enzymatic glycation [125]. Metformin may inhibit both of them.
The preclinical study conducted in normoglycemic mice with left ventricular pressure overload showed that metformin can inhibit collagen synthesis in myocardium, which contributes to left ventricular dimension reduction and significantly decreases diastolic pressure in the LV. This effect resulted from the ability of metformin to inhibit the synthesis of TGF-β1 in myocardium and was independent of its glucose and insulin-lowering effects [126]. The same authors, in another study, showed that metformin in cultured fibroblasts may inhibit phosphorylation of Smad3 factor and its translocation to the nucleus stimulated by TGF-β1 [126]. The TGF-β1/Smad3 signal pathway displays a significant role in the regulation of the expression of extracellular matrix protein genes [127]. Thus, its inhibition by metformin explains the reduction of collagen synthesis. This effect of metformin was not related to AMPK activation [126].
Non-enzymatic protein glycation (NEG) plays an important role in the development of structural and functional abnormalities in diabetic heart. In this process, in the Maillard reaction, carbonyl groups of reducing sugars attach to free amino groups of amino acids (lysine and arginine) of proteins as well as of nucleic acids or phospholipids ( Figure 5) [128][129][130]. In diabetes, in addition to sustained hyperglycemia, inflammation, oxidative stress and polyol pathway are also the important processes that can exacerbate protein glycation ( Figure 6) [131,132]. The Maillard reaction leads to synthesis of reactive dicarbonyls, such as glyoxal, methylglyoxal, and 3-deoxyglucosone [129,130]. Subsequently these compounds may covalently bind to amino groups of proteins forming advanced glycation end products [129]. AGEs may alter protein structure by producing crosslinks within or between proteins. AGEs do not disassociate and accumulate permanently in long-living protein such as collagen in myocardium and vessels, increasing their stiffness and impairing relaxation [59,130]. Non-enzymatic protein glycation (NEG) plays an important role in the development of structural and functional abnormalities in diabetic heart. In this process, in the Maillard reaction, carbonyl groups of reducing sugars attach to free amino groups of amino acids (lysine and arginine) of proteins as well as of nucleic acids or phospholipids ( Figure 5) [128][129][130]. In diabetes, in addition to sustained hyperglycemia, inflammation, oxidative stress and polyol pathway are also the important processes that can exacerbate protein glycation ( Figure 6) [131,132]. The Maillard reaction leads to synthesis of reactive dicarbonyls, such as glyoxal, methylglyoxal, and 3-deoxyglucosone [129,130]. Subsequently these compounds may covalently bind to amino groups of proteins forming advanced glycation end products [129]. AGEs may alter protein structure by producing crosslinks within or between proteins. AGEs do not disassociate and accumulate permanently in long-living protein such as collagen in myocardium and vessels, increasing their stiffness and impairing relaxation [59,130]. Figure 5. Maillard reaction. Carbonyl groups of glucose react with amino groups of proteins (both in dotted frames) to form Schiff bases (yellow) which then are spontaneously converted to Amadori products (yellow). Both of them contribute to formation of low molecular weight reactive dicarbonyls (intermediate products, green) which bind to proteins forming advacned glycation end products (AGEs, red). Continuous and dotted arrows represent one-and multiple-step reactions, respectively. Figure 5. Maillard reaction. Carbonyl groups of glucose react with amino groups of proteins (both in dotted frames) to form Schiff bases (yellow) which then are spontaneously converted to Amadori products (yellow). Both of them contribute to formation of low molecular weight reactive dicarbonyls (intermediate products, green) which bind to proteins forming advacned glycation end products (AGEs, red). Continuous and dotted arrows represent one-and multiple-step reactions, respectively.
Glycation of collagen increases its resistance to the enzymatic breakdown, leading to collagen accumulation and fibrosis [131,133,134]. Metformin is able to inhibit generation of AGEs by the mechanism related to direct neutralization of reactive dicarbonyls, through binding the guanidine group of drug to α-dicarbonyl group of methylglyoxal. Furthermore, metformin activates the glyoxolase, an enzyme that converts methylglyoxal to D-lactate. This beneficial effect was confirmed in numerous preclinical studies [135,136]. For example, it was documented that metformin is able to protect apolipoprotein A-I of HDL (high density lipoprotein) fraction from glycation induced by methylglyoxal [137]. Metformin also inhibits modification of apolipoprotein B in LDL (low density lipoprotein) fraction by glycolaldehyde or methylglyoxal [138], and inhibits AGEs generation in macrophages during incubation with glyoxal [139]. In the rats with streptozotocin-induced diabetes, as a result of administration of metformin, smaller amounts of AGEs in the kidney cortex, lens and sciatic nerve were observed [140]. Moreover, lower levels of circulating AGEs were found in diabetic patients treated with metformin [141]. In addition, a study performed on diabetic dogs revealed that metformin could also prevent glycation of collagen in the heart muscle and thereby could reduce heart stiffness. It has been demonstrated that metformin can restore diastolic myocardial function in these animals, reflected by the normalization of end-diastolic pressure and end-diastolic volume.
In the heart of animals given metformin, there was observed reduced AGEs binding to collagen fibers. It is interesting that although the total amount of collagen in the heart of these animals remained unchanged, metformin significantly improved heart function [125]. Figure 5. Maillard reaction. Carbonyl groups of glucose react with amino groups of proteins (both in dotted frames) to form Schiff bases (yellow) which then are spontaneously converted to Amadori products (yellow). Both of them contribute to formation of low molecular weight reactive dicarbonyls (intermediate products, green) which bind to proteins forming advacned glycation end products (AGEs, red). Continuous and dotted arrows represent one-and multiple-step reactions, respectively.

Figure 6.
Different mechanisms of AGE formation. MGO-methylglyoxal, GO-glyoxal, 3-DG-3-deoxyglucosome. Initial, intermediary and terminal steps are shown in yellow, green and orange, respectively. Dotted arrows and lines represent multiple-step reactions.

Apoptosis
Protective effect of metformin on heart muscle is also mediated by the inhibition of cardiomyocyte apoptosis. It has been demonstrated that metformin, through AMPK activation, prevents apoptosis of cardiomyocytes during their incubation with H 2 O 2 [104]. Similarly, an in vivo study showed that metformin reduced the number of dead cardiomyocytes in the hearts of dogs with experimental heart failure. After four weeks treatment with metformin, improvement of LV function and hemodynamic parameters was observed. These cardioprotective, antiapoptotic properties were partially dependent on AMPK activation and eNOS expression, phosphorylation/activation, resulting in the increase in NO synthesis [104].
The evidence of cardioprotective effect of metformin was also found in a study on rats with heart damage caused by subcutaneous injection of isoproterenol [142]. Isoproterenol induces changes consistent with those that appear in patients during myocardial infarction, like cardiomyocytes necrosis, cardiac arrhythmias, decrease in arterial pressure indices, including LV contractility and relaxation, and increase in the left ventricular end-diastolic pressure [143,144]. These changes can lead to dysfunction of the heart, mainly of the left ventricle. Metformin declined heart weight, increased cardiomyocyte survival, and reduced intercellular matrix accumulation, as well as decelerated heart rate in rats subjected to isoproterenol-induced myocardial toxicity. Additionally, the average blood pressure, left ventricular end-diastolic pressure and left ventricular systolic pressure were normalized [142].

Metformin in Heart Failure
The experimental studies and clinical observations supply a growing number of arguments that confirm safety and benefits of metformin in patients with heart failure. For example, in study involving patients with heart failure and type 2 diabetes, at the first stage of antihyperglycemic treatment, metformin, in monotherapy or in combination with sulfonylureas, reduced the total mortality and risk of death or hospitalization compared to sulfonylureas therapy alone [145]. This result has been confirmed in other trials that involved patients with newly diagnosed type 2 diabetes and heart failure. Metformin, both in monotherapy and polytherapy reduced mortality in comparison with traditional treatment based only on diet and life style changes. The effect of metformin was independent of glycemic control and the BMI (body mass index) value, and has not been observed after thiazolidinediones or insulin administration [146]. In another non-randomized study, during two years observation, treatment with metformin reduced mortality of ambulatory patients with diabetes and heart failure [147]. There is also evidence that metformin monotherapy is associated with a lower risk of heart failure development in patients with recent diabetes than sulfonylurea monotherapy, even at high doses [148]. In addition, it has been proven that metformin use in monotherapy or in combination with other antihyperglycemic drugs such as sulfonylurea derivatives, thiazolidinediones, or insulin have a positive effect on patients with advanced heart failure in III and IV New York Heart Association (NYHA) class. After consideration of differences between groups, the results showed a tendency to reduction of all-cause mortality, reduction of composed endpoint (death or urgent heart transplant) risk and to increase in left ventricular ejection fraction after metformin therapy in comparison to other oral antihyperglycemic drugs or insulin [149].
According to the recommendations of the international guidelines, metformin should not be used in patients with diabetes and heart failure because of a risk of lactic acidosis. However, the connection between metformin blood level and lactate blood level during lactic acidosis is not observed in clinical practice. Lactic acidosis is rather the result of comorbid diseases and the risk of lactic acidosis is similar whether metformin is used or not [150,151]. Metanalysis of nine observational studies demonstrated that metformin can be safely used in heart failure. In none of the trials metformin increased mortality of patients with decreased left ventricular ejection fraction, even in those with III and IV NYHA class of heart failure and patients with chronic kidney failure. The analysis also showed that metformin reduced the risk of all-cause hospitalization and the risk of hospitalization due to heart failure. There was no trial to demonstrate that metformin treatment was associated with a higher risk of lactic acidosis than other antihyperglycemic drugs [151].
In conclusion, based on experimental and clinical data, metformin therapy in diabetes and concomitant heart failure is not associated with higher risk and the benefits of its usage exceed the potential danger. Moreover, it is increasingly suggested that metformin should be a first choice of treatment in this group of patients [152,153]. Therefore, the revising of contraindications of metformin use seems to be reasonable.
As the cardioprotective properties of metformin are not associated with an antihyperglycemic effect, the drug could be also beneficial for patients with heart failure without concomitant diabetes.
Funding: This research received no external funding.

Conflicts of Interest:
The authors declare no conflict of interest.