Metformin Reduces Potassium Currents and Prolongs Repolarization in Non-Diabetic Heart

Metformin is the first choice drug for the treatment of type 2 diabetes due to positive results in reducing hyperglycaemia and insulin resistance. However, diabetic patients have higher risk of ventricular arrhythmia and sudden cardiac death, and metformin failed to reduce ventricular arrhythmia in clinical trials. In order to explore the mechanisms responsible for the lack of protective effect, we investigated in vivo the effect of metformin on cardiac electrical activity in non-diabetic rats; and in vitro in isolated ventricular myocytes, HEK293 cells expressing the hERG channel and human induced pluripotent stem cells derived cardiomyocytes (hIPS-CMs). Surface electrocardiograms showed that long-term metformin treatment (7 weeks) at therapeutic doses prolonged cardiac repolarization, reflected as QT and QTc interval duration, and increased ventricular arrhythmia during the caffeine/dobutamine challenge. Patch-clamp recordings in ventricular myocytes isolated from treated animals showed that the cellular mechanism is a reduction in the cardiac transient outward potassium current (Ito). In vitro, incubation with metformin for 24 h also reduced Ito, prolonged action potential duration, and increased spontaneous contractions in ventricular myocytes isolated from control rats. Metformin incubation also reduced IhERG in HEK293 cells. Finally, metformin incubation prolonged action potential duration at 30% and 90% of repolarization in hIPS-CMs, which is compatible with the reduction of Ito and IhERG. Our results show that metformin directly modifies the electrical behavior of the normal heart. The mechanism consists in the inhibition of repolarizing currents and the subsequent decrease in repolarization capacity, which prolongs AP and QTc duration.


Introduction
In the last decades, diabetes prevalence in adults has increased in almost every country affecting 422 million adults in 2014 [1,2]. Type 2 diabetes (T2D) accounts for more than 90% of diagnosed cases of diabetes. Since the positive results with overweight diabetic patients reported in the United Kingdom Diabetes Prospective Study in 1998 [3], metformin became the first therapeutic choice for the treatment of T2D. The beneficial effects of metformin include improving glycemic control; reducing insulin resistance; reducing fasting hyperinsulinemia and progression of prediabetes; reducing cardiovascular risk; and may also improve blood lipid profile and reduce proinflammatory state. However, metformin mechanisms of action in the treatment of T2D are complex and not yet fully understood [4].
The incidence of ventricular arrhythmia and sudden cardiac death is higher in diabetic patients [5,6], as diabetes-induced electrical remodeling can contribute to the risk of ventricular arrhythmia [7]. The most common electrical disturbance in diabetic patients is prolonged ventricular repolarization that can be seen on the electrocardiogram as an excessive lengthening of heart rate corrected QT interval (QTc). Prolonged QTc is associated with potentially lethal ventricular arrhythmias and predicts cardiovascular mortality even in healthy individuals [8].
Since many patients with T2D receive medication immediately after diagnosis, data regarding the incidence of QT interval prolongation usually come from studies of patients on glucose-lowering therapy. In addition, large studies do not usually differentiate patients on metformin monotherapy from those on combined therapy to achieve glycemic control. A cross-sectional study of 501 type 2 diabetic patients showed a high prevalence of prolonged QTc (44%). Interestingly, the percentage of patients on metformin was similar in the normal and in the prolonged QTc group [9]. In this sense, we have recently published that metformin treatment controls glycaemia in T2D rats, but does not normalize QTc interval duration and does not reduce the susceptibility to arrhythmia [10]. Among the main families of hypoglycemic drugs, sulfonylureas cause a higher risk of cardiac arrhythmias. Regarding the other families, compared with dipeptidyl peptidase 4 inhibitors, thiazolidinediones or Glucagon-like peptide-1 receptor agonists, metformin was not associated with a difference in the risk of developing ventricular arrhythmia [11]. Furthermore, despite the cardiovascular benefits, metformin failed to reduce ventricular arrhythmia in clinical trials [12][13][14].
The aim of this work was to study a potential mechanism to explain this lack of protective effect of metformin against arrhythmia. We investigated the effect of metformin treatment on cardiac electrical behavior in healthy non-diabetic rats, HEK293 cells and human induced pluripotent stem cells derived cardiomyocytes (hIPSC-CMs). Cardiac transient outward potassium current (I to ) and hERG current (I hERG ), cardiac action potential (AP) and electrocardiograms (ECG) were recorded.

Metformin Treatment Prolongs QTc
To investigate the effects of chronic metformin treatment, we treated healthy Sprague-Dawley rats for 7-8 weeks with metformin 35 mg/kg daily or vehicle (drinking water). Although in our previous work higher metformin doses were required to keep glycemic control in T2D Sprague-Dawley rats [10], we used 35 mg/kg daily because is the maximal approved total daily dose for the treatment of type 2 diabetes [15]. Throughout the treatment, metformin did not cause hypoglycemia or affect body weight gain (Supplementary Figure S1).
The ECG in conscious animals was monitored weekly (Figure 1). We found no differences in the durations of RR, PR, and QRS intervals between animals treated with metformin and control group (not shown). However, metformin lengthened QT and QTc interval durations. At the end of the experimental period, QTc was significantly longer in the metformin group (131 ± 1.9 ms) compared to control (120 ± 2.2 ms; p < 0.01).
Beat-to-beat variability of repolarization duration is an alternative marker used to predict arrhythmia and sudden cardiac death [16,17]. However, no difference was found between control and metformin-treated animals (2.40 ± 0.17 ms vs. 2.45 ± 0.19 ms).

Metformin Treatment Increases In Vivo Arrhythmia Susceptibility
To assess the functional relevance of metformin-induced QTc prolongation, we searched for arrhythmic events. ECG in conscious rats treated with metformin for 7 to 8 weeks, the period when all changes in the electrocardiographic parameters were well established was analyzed. The incidence and type of spontaneous arrhythmia (Figure 2a,b) were different in the two experimental groups. Overall, spontaneous arrhythmic events were infrequent and mild, mainly ventricular premature beats and persistent bigeminy/salvo. However, the percentage of animals who showed no events was smaller in the metformin group (35% vs. 67% in control) and one metformin-treated animal developed non-sustained polymorphic ventricular tachycardia.  In vivo arrhythmia susceptibility assays under cardiac challenge using the caffeine and dobutamine test (Figure 2c,d) were also carried out. Anaesthetized animals did not develop arrhythmia at basal conditions before the challenge. Dobutamine plus caffeine injection differentially induced arrhythmia in control and metformin-treated groups. Again, control animals suffered minor events, mostly persistent bigeminy or salvo (PB/S), whereas metformin-treated rats responded more severely to the challenge and developed ventricular tachycardia and torsades de pointes. An adapted scoring system was used to calculate average arrhythmia score. The mean arrhythmia score was significantly higher in the metformin group (3.5 ± 1.05 metf, 1.33 ± 0.44 contr, p < 0.05). Taken together, these results support the hypothesis that chronic metformin treatment induces cardiac remodeling resulting in prolonged repolarization and enhanced susceptibility to arrhythmias after cardiac challenge.  showing the proportion and severity the proportion of each event after cardiac challenge. Data represent individual data points with mean ± SEM (red marks). * p < 0.05 compared to control. n = 9 control and 6 metformin-treated animals.

Metformin Treatment Reduces Transient Outward Potassium but Not L-Type Calcium Current (I Ca-L )
Prolonged repolarization can result from a reduction in repolarizing currents or an increase in depolarizing currents. The transient outward potassium current (I to ) is the main repolarizing current in the rat heart and the responsible for the phase 1 of the action potential in humans. As shown in Figure 3a-c, in isolated cardiomyocytes chronic metformin treatment caused a 26.7% reduction in the peak I to density without affecting the biophysical properties of the current (Supplementary Figure S2). Regarding the depolarizing I Ca-L , (Figure 3d-f), we found similar current density in both groups at all voltages tested. These results indicate that I to but not I Ca-L contributes to the metformin-induced lengthening in QTc duration.  (n = 17 control and 12 metformin) (d) I Ca-L traces recorded from cardiomyocytes isolated from a control and a metformin-treated rat, (e) current density-voltage relation and (f) peak current density values at +10 mV (n = 7 control and 8 metformin). Diagrams of I to and I Ca-L recording protocols depicted in the insets (a,d). Points represent individual data with mean ± SEM (red marks), except for the voltage-current curves, where mean ± SEM is shown. ** p < 0.01 compared to control group.

In Vitro Metformin Reproduces the Effects Observed In Vivo
To assess whether the effect of metformin on the heart was direct, in vitro experiments in ventricular myocytes isolated from control rats and then incubated with 10 mM metformin for 24 h were performed. First, we tested whether cells incubated with metformin could develop arrhythmic behavior under challenge conditions. Pacing interruption protocol was used as a trigger and the number of spontaneous contractions were analyzed. Cardiomyocytes incubated with metformin had a significantly higher number of spontaneous contractions after interruption of the stimulation trains at 1 and 2 Hz (Figure 4a-c). Moreover, 100% of metformin-treated cells developed spontaneous contractions at the two frequencies tested, whereas in the control group only 50% and 40% of cells presented spontaneous contractions at 1 Hz and 2 Hz, respectively. These results in isolated cells are consistent with the higher susceptibility to develop arrhythmias observed in the whole animal. We then explored the mechanism of arrhythmia by measuring calcium transients in cardiomyocytes isolated from control hearts and incubated for 24 h with or without metformin. The amplitudes of calcium transients recorded at 0.5 and 1 Hz were similar in the two groups (Supplementary Figure S3), indicating that metformin does not cause intracellular calcium overload.    30 ) and 90% of repolarization (APD 90 ) (e,f) of repolarization at stimulation frequencies of 300 and 800 ms (n = 11 control and 11 metformin). Points represent individual data with mean ± SEM (red marks), ** p < 0.01 and * p < 0.05 compared to control at the same frequency.
Next, the effect of metformin on repolarization in vitro was analyzed. Ventricular strips were incubated with or without 10 mM metformin and AP were recorded (Figure 4d-f). Metformin significantly prolonged APD 30 and APD 90 , consistent with the prolongation of QTc interval duration observed in the animals. In cells incubated with metformin for 24 h I to peak density was lower (37%) in the metformin group compared to control (Figure 5a-c). The possibility that metformin could act as an I to channel blocker was ruled out, since acute exposure to the drug (by 30-min perfusion) did not alter current amplitude (Supplementary Figure S4). Therefore, this inhibitory effect is probably due to transcriptional and/or translational causes. Regarding the I Ca-L , no effect was observed after exposure to metformin for 24 h (Figure 5d-f). Taken together, these in vitro results over AP and ionic currents are in agreement with the results observed ex vivo, and confirm the evidence that metformin directly regulates cardiac repolarization.
On the other hand, the rapid delayed rectifier potassium current, I Kr , is the main repolarizing current that drives the human cardiac cells to the resting potential. Since I Kr , which is carried through hERG channels in human cardiac myocytes, is not significantly expressed in the rat heart, we used HEK293 cells that constitutively express the hERG channel (HEK-hERG cells) to study the effect of metformin on this current. As shown in Figure 6a-c, in HEK-hERG cells incubated with metformin for 24 h the I hERG density was 55.7% lower compared with control.

Metformin Prolongs Repolarization in hIPSC-CMs
In a last set of experiments, hIPSC-CMs were used. Cells were incubated with 100 and 300 µM metformin for 24 h and recorded AP using voltage-sensitive dyes (Figure 7). Compared to the control group, metformin slowed down the repolarization of the action potential at phases 1 and 3. Thus, cells incubated with 300 µM metformin showed prolonged APD 30 , as expected from I to inhibition. Besides, APD 90 was longer in the two metformin-treated groups, corroborating the inhibition of I Kr observed in previous experiments. Furthermore, we observed that very high concentrations of metformin lengthened the AP duration to the point of generating early afterdepolarizations (EADs) (Supplementary Figure S5).

Discussion
Here we have studied the effects of metformin exposure on cardiac electrical properties. The major findings are that metformin directly modifies the electrical behavior of the normal heart making it more prone to develop arrhythmia under cardiac challenge. The mechanism involves inhibition of repolarizing currents, which decreases repolarizing capacity and prolongs AP and QTc interval duration.
Compared to the control group we did not observe hypoglycemia or changes in body weight throughout the treatment, suggesting that cardiac alterations were not caused by metabolic defects secondary to metformin-treatment. Next, in vitro recordings of AP and ion currents in cardiomyocytes isolated from control rats and incubated with metformin are consistent with the results observed ex vivo, and confirm that metformin directly regulates cardiac repolarization. These results are in agreement with others that reported direct effects of metformin in neonatal rat cardiomyocytes and HL-1 cells [18].
Metformin-induced reduction of I to or I hERG must involve intracellular signaling pathways, since metformin does not act as a direct channel blocker. Regarding the I Ca-L , Wang et al. reported that metformin inhibited this current and thus shortened APD and QT interval in type 1 diabetic mice [19]. However, we observed no effect on I Ca-L in cardiomyocytes from control animals. Two possible explanations for the discrepancy may be that their type 1 diabetic animals have an increased I Ca-L and that they use doses of metformin from two to ten times higher than ours.
Despite hiPSC-CMs have an immature phenotype characterized by a more depolarized diastolic potential and the absence of notch, due to small I K1 and I to densities [20,21], these cells recreate action potentials similar to those reported for human cardiomyocytes. Therefore, they have become a relevant model for drug safety assays. Metformin lengthened APD 30 and APD 90 in these cardiomyocytes, which is compatible with the reduction in the repolarizing currents I to and I Kr that we observed.
Metformin is known to activate AMP-activated protein kinase or AMPK [22]. Here our results show that metformin significantly reduces I to and I hERG density, in agreement with studies using the classical AMPK activator AICAR that also reported a decrease in these currents in mice, xenopus oocytes and rhabdomyosarcoma cells [23,24]. Despite this evidence, given the high concentrations used in our in vitro experiments, AMPKindependent mechanisms cannot be exclude [15,25].
Electrocardiograms show that metformin increases the probability to develop arrhythmia in treated animals in vivo. Similarly, 100% of the cells incubated in vitro with metformin had spontaneous contractility after pacing interruption. A classical trigger of arrhythmias is the depolarizing current carried by the sodium/calcium exchanger, secondary to cellular Ca 2+ overload that can depolarize the membrane from the resting potential and generate delayed afterdepolarizations [26]. However, we found no differences in cytoplasmic calcium content or the amplitude of Ca 2+ transients in cells incubated with metformin compared to controls. Therefore, calcium overload does not seem to be the mechanism responsible for metformin-induced arrhythmia.
Prolongation of AP repolarization allows calcium and sodium channels to recover from inactivation. The reactivated channels can then reopen, further depolarize the membrane, and generate early afterdepolarizations. Although other mechanisms cannot be ruled out, our results support the hypothesis that the developing of EADs is the trigger for metformin dependent arrhythmias.
It is important to note that metformin use is extending beyond diabetes and diabetes prevention [27]. Because of the potential efficacy to promote weight loss, metformin is used to reduce obesity in non-diabetic patients [28] and in the treatment of antipsychoticinduced weight gain [29]. It has also been used to treat the reproductive and metabolic abnormalities associated with polycystic ovary syndrome [30]. A limitation of the study is that the concentrations used in the in vitro experiments are high and therefore the observed results may be maximal effects. However, this could be relevant to facilitate the development of ectopic activity in especially susceptible individuals or under arrhythmogenic circumstances. Besides ion channel function, other metabolic and structural factors can contribute to diabetic myocardial dysfunction and increase the risk of arrhythmia [5].
Experiments in animal models with type 1 and type 2 diabetes have shown that several cardiac ion currents, mainly I to and I Ks , are significantly reduced. Our results show that metformin (not diabetes) also reduces I to and I hERG . Thus, although metformin corrects hyperglycemia and might improve other metabolic aspects, it cannot revert the diabetesassociated repolarization defects. In this sense, we have recently published that although metformin treatment improves the metabolic and inflammatory profile, it does not reduce the incidence and severity of cardiac arrhythmias in diabetic animals [10]. Our results help to understand why metformin-therapy does not protect against ventricular arrhythmia in diabetic patients [12][13][14].

Ethics Statement and In Vivo Treatment
The investigation fulfills the Spanish (RD 53/2013) and European (D2003/65/CE and R2007/526/CE) rules regulations for the Care of Animals used for experimental and other research purposes, and has been approved by the Ethics Committee for Animal Care of the University of the Basque Country under reference CEBA/273M/2012. Female young Sprague-Dawley rats (200-220 g) from the Universidad del País Vasco UPV/EHU animal facility were housed in a temperature-controlled room (24 • C) under a 12:12 h light:dark cycle and-with free access to water and standard rat chow. Metformin-treated animals received metformin hydrochloride 35 mg/kg daily (Sigma-Aldrich Co., Saint Louis, MO, USA) in drinking water for 7 weeks. Animals were periodically weighted and kept in separated cages. For each animal, metformin was dissolved in 30 mL of water, the average daily drinking volume (n = 35). If needed, additional regular drinking water was also provided. Control animals (n = 20) received regular tap water.

In Vivo ECG Recordings
ECG recordings in conscious animals were performed using a noninvasive method. Stainless steel surgical electrodes, placed under the skin and positioned in the DII lead, were connected to a Biopac MP35 recording system controlled by the Biopac Pro software (Biopac Systems Inc., Goleta, CA, USA). Data were automatically analyzed with the Labchart 7.0 software (AD Instruments) and manually corrected when necessary. Electrocardiographic parameters were calculated as the mean of 50-80 beats. Corrected QT interval was calculated using the Fridericia's formula QTc = QT/(RR/1000) 0. 33 .
At the end of the experimental period, the arrhythmia susceptibility protocol was performed [31]. Animals were anesthetized by 2% isoflurane inhalation and subsequently received an injection of caffeine (intraperitoneal, 120 mg/kg, Sigma-Aldrich Co., St. Louis, MO, USA) and dobutamine (intravenous, 50 µg/kg, Sigma-Aldrich Co., St. Louis, MO, USA). ECG was recorded for 5 min before (basal) and 15 min after caffeine/dobutamine challenge. For analysis of spontaneous and induced arrhythmias, the type of arrhythmia was identified according to the Lambeth Convention guidelines [32]. After the induced arrhythmia assay, anesthetized animals were killed and therefore were not used for cellular electrophysiology experiments.

Patch-Clamp
We used only calcium-tolerant rod-shaped cells to perform these experiments. Transient outward potassium and L-type calcium currents were recorded at room temperature (20-22 • C), using the whole-cell configuration of the Patch-Clamp technique with an Axopatch 200B patch-clamp amplifier (Molecular Devices, San Jose, CA, USA). Recording pipettes were obtained from borosilicate capillary glass (Sutter Instruments, Novato, CA, USA) and had a tip resistance of 1-3 MΩ when filled with the internal solution (in mM): L-aspartic acid (potassium salt) 80; KH 2 PO 4 10; MgSO 4 1; KCl 50; ATP-Na 2 3; EGTA-K + 10; HEPES-K + 5; adjusted to pH 7.2 with KOH. Following the patch rupture, whole-cell membrane capacitances were measured from the integration of capacitive transients elicited by voltage steps from −50 to −60 mV. Series resistances were 80% compensated to minimize errors.
Depolarizing voltage pulses between −30 and +50 mV, starting from a holding potential of −60 mV, were applied at a frequency of 0.1 Hz to allow complete recovery of I to from inactivation. The TEA-resistant, time-independent I ss , was digitally subtracted.
To elicit I Ca-L , the bathing solution was (in mM): NaCl 86; KCl 4; MgCl 2 1; CaCl 2 1.8; TEA-Cl − 50; 4-Aminoprirydine 4; dextrose 12; HEPES-Na + 10; adjusted to pH 7.4 with HCl. Voltage pulses ranged from −30 to +50 mV, starting from a prepulse of 40 ms at −40 mV. All voltage pulses were applied at a frequency of 0.1 Hz. Voltage clamp protocols were controlled with the Clampex program and current recordings analyzed with the Clampfit program of the pClamp 10.2 software (Molecular Devices, San Jose, CA, USA). I to and I Ca-L amplitudes were normalized to cell capacitance and expressed as pA/pF. The propensity to develop arrhythmias was estimated from the number of nonstimulated contractile events, defined as the number of spontaneous contractions developed after interruption of a 3-min stimulation train at 1 or 2 Hz.

hiPSC-CMs Cell Culture
Cryopreserved hiPSC-CMs (iCells ® , CMC-100-110-001, Cellular Dynamics International (CDI), Madison, WI, USA) were plated as previously described [34]. Briefly, the cryovial containing iCell cardiomyocytes was thawed, transferred to a 50 mL conical tube and rinsed with Plating Media (CMM-100-110-001), following addition of 8 mL of this plating media on top. Cells were plated (40,000 cells/well) in 96-well glass bottomed plates (MatTek, Ashland, MA, USA) coated with fibronectin and maintained in a CO 2 incubator. The experiments were performed at day 14, when the cells formed a well-coupled monolayer exhibiting synchronous beating.

Electrical Activity Measurements in hiPSC-CMs Using Voltage Sensitive Dyes
hiPSC-CMs were transiently loaded with 4 µM di-4-ANEPPS (Biotium) in serum-free media (DMEM Gibco, 10 mM Galactose, 1 mM Na-Pyruvate). The media containing the dye was replaced with fresh serum-free media and maintained under this condition throughout the experiment. The cells were kept in the incubator for at least 2 h to reach a steady state before starting an experiment. The multi-well plate was placed on the stage incubator (37 • C, 5% CO 2 , water-saturated air atmosphere) of the CellOPTIQ ® platform (Clyde Biosciences Ltd., Glasgow, Scotland), and spontaneous electrical activity was recorded from the di-4-ANEPPS fluorescence signal from areas (0.2 × 0.2 mm) of hiPSC-CMs in individual wells visualized using a 40× objective (NA0.6). Fluorescence ratiometric signals were digitized at 10 KHz, and the recordings were subsequently analyzed off-line using proprietary software (CellOPTIQ ® ). Action potential duration (optical APD) at 60% and 90% repolarization was obtained from AP complexes that occurred within the 15 s recording period. The incidence of early afterdepolarizations was identified as a positive deflection in the Di-4-ANEPPS signal during the plateau phase of the AP optical signal.

Statistics
Data are presented as mean ± SEM. Comparisons between groups were performed using Student's t test. Values of p < 0.05 were considered statistically significant. For APD in hiPSC-CMs, statistical analysis was performed using Dunnett's test, comparing treatment with control. All analyses were made using GraphPad Prism 5.0 (GraphPad Software, San Diego, CA, USA).