Expression and relevance of the G protein-gated K+ channel in the mouse ventricle

The atrial G protein-gated inwardly rectifying K+ (GIRK) channel is a critical mediator of parasympathetic influence on cardiac physiology. Here, we probed the details and relevance of the GIRK channel in mouse ventricle. mRNAs for the atrial GIRK channel subunits (GIRK1, GIRK4), M2 muscarinic receptor (M2R), and RGS6, a negative regulator of atrial GIRK-dependent signaling, were detected in mouse ventricle at relatively low levels. The cholinergic agonist carbachol (CCh) activated small GIRK currents in adult wild-type ventricular myocytes that exhibited relatively slow kinetics and low CCh sensitivity; these currents were absent in ventricular myocytes from Girk1−/− or Girk4−/− mice. While loss of GIRK channels attenuated the CCh-induced shortening of action potential duration and suppression of ventricular myocyte excitability, selective ablation of GIRK channels in ventricle had no effect on heart rate, heart rate variability, or electrocardiogram parameters at baseline or after CCh injection. Additionally, loss of ventricular GIRK channels did not impact susceptibility to ventricular arrhythmias. These data suggest that the mouse ventricular GIRK channel is a GIRK1/GIRK4 heteromer, and show that while it contributes to the cholinergic suppression of ventricular myocyte excitability, this influence does not substantially impact cardiac physiology or ventricular arrhythmogenesis in the mouse.

Increased output of the parasympathetic branch of the autonomic nervous system leads to the slowing of heart rate (HR) and an increase in heart rate variability (HRV), an indicator of beat-to-beat fluctuations in HR 1,2 . Abnormal parasympathetic regulation of cardiac output has been linked to multiple cardiac disorders, including atrial fibrillation (AF), atrioventricular (AV) block, heart failure, and sudden cardiac death [3][4][5][6][7] . The parasympathetic influence on the heart involves acetylcholine (ACh) acting on sino-atrial nodal (SAN) and atrio-ventricular nodal (AVN) cells, as well as atrial myocytes [8][9][10] . ACh binds to M 2 muscarinic receptors (M 2 R) on these cells, initiating a branched intracellular signaling pathway mediated by inhibitory (G i/o ) G proteins, culminating in the activation of the K + channel I KACh , as well as the suppression of the cAMP/PKA-dependent cation-selective ion channel (HCN/I f ) and voltage-gated Ca 2+ channels [11][12][13][14] .
The atrial I KACh channel is a GIRK channel consisting of GIRK1 and GIRK4 subunits, in 1:1 stoichiometry 15,16 . I KACh is gated by G protein Gβγ subunits liberated following M 2 R activation 17,18 . Genetic ablation of either Girk1 or Girk4 in mice eliminated the I KACh conductance in neonatal atrial myocytes, and yielded a modest increase in HR and decrease in HRV at baseline in adult animals 19,20 . Additionally, HR and HRV responses to ACh and CCh were diminished in isolated hearts from mice lacking GIRK1 or GIRK4 [19][20][21][22] .
While the parasympathetic impact on cardiac function is typically ascribed to its influence on the atria, SA node, and AV node, parasympathetic innervation of ventricular tissue is also evident in mammals 10 . In addition, profound effects of vagal nerve stimulation (VNS) on ventricular physiology have been noted in several species 23 . For example, chronic VNS altered the electrophysiological properties of the heart and reduced susceptibility to ventricular arrhythmias in rats 24 . Additionally, ACh shortened action potential duration (APD) in human ventricular myocytes, in an atropine-sensitive manner 25 . This effect has been attributed to the activation of a GIRK channel 23,[26][27][28] . Consistent with this premise, the ACh-induced decrease in APD and effective refractory period (ERP) in rat papillary muscle was blocked by the non-selective GIRK channel blocker tertiapin, and ACh triggered hyperpolarization and a reduction in APD in right ventricle recordings from wild-type but not Girk4 −/− mice 29 . In addition, a mutation in the human GIRK4/KCNJ5 gene underlies a congenital form of Long QT Syndrome (LQTS13), a ventricular repolarization disorder associated with arrhythmia, syncope, and sudden death 28,30 .
While available evidence supports the contention that a GIRK channel contributes to the cholinergic influence on ventricular physiology, critical details remain unclear. Here, we examined the expression in mouse ventricle of genes implicated in atrial I KACh -dependent signaling, and evaluated the impact of gene ablation on cholinergic signaling in ventricular myocytes. We present functional evidence that the GIRK channel in ventricular myocytes is a GIRK1/GIRK4 heteromer, and that it mediates the impact of cholinergic signaling on APD and excitability of these cells. We also probed the physiological impact of ventricular GIRK-dependent signaling using a novel ventricle-specific Girk1 −/− mouse line. Our findings reveal that, despite the contribution of GIRK channels to the CCh-dependent acceleration of repolarization and decrease in ventricular myocyte excitability, ventricular GIRK-dependent signaling does not exert a significant impact on HR, HRV, or susceptibility to ventricular arrhythmias in mice.

Results
Expression of I KACh -dependent signaling pathway elements in mouse atria and ventricle. Expression of key elements of the atrial M 2 R-I KACh signaling pathway, including GIRK1, GIRK4, M 2 R, and RGS6, has been detected in ventricular tissue from many mammalian species 23,26,28,[31][32][33] . A study in the rat heart revealed higher expression in atria than ventricle for GIRK1, GIRK4, and RGS6 34 . We sought to confirm these findings in the mouse by comparing the expression levels of these targets, as well as M 2 R, in the atria and ventricle of adult mice using quantitative RT-PCR. We found that GIRK1, GIRK4, M 2 R, and RGS6 mRNAs were all present in mouse ventricular tissue, but that the levels of all targets were lower (4-to 10-fold) than those found in mouse atrial tissue (Fig. 1).
Evidence of a GIRK1/GIRK4 heteromer in mouse ventricular myocytes. Previous work suggests that a GIRK4-containing GIRK channel is present in rodent ventricle 29,35 . While GIRK1/GIRK4 heteromeric channels are thought to mediate the atrial I KACh conductance, some evidence suggests that GIRK4 homomeric channels may contribute as well 36,37 . To assess the relative contributions of GIRK1 and GIRK4 to the GIRK channel activity in ventricle, we measured whole-cell currents induced by the non-selective cholinergic agonist carbachol (CCh) in ventricular myocytes from wild-type mice, and from mice lacking GIRK1 or GIRK4. Ventricular myocytes express a prominent constitutive K + current (I K1 ) that masks the ACh/CCh-induced GIRK current under normal whole-cell recording conditions, due to K + accumulation and/or depletion 38 . As such, we included a low concentration of Ba 2+ (5 μM) in our bath solution to selectively suppress I K1 , while preserving any potential GIRK channel activity. Indeed, 5 μM Ba 2+ did not suppress the CCh-induced GIRK current in wild-type SAN cells; indeed, currents were slightly larger in the presence Ba 2+ (Supplementary Fig. S1).
SCieNTifiC REPORTS | (2018) 8:1192 | DOI:10.1038/s41598-018-19719-x smaller in ventricular myocytes, however, as compared to currents reported in SAN cells 43 . Additionally, the CCh-induced current activation and deactivation kinetics were significantly slower for the ventricular GIRK channel as compared to the atrial GIRK channel (Fig. 2c,d) 43 . Furthermore, the ventricular GIRK current was more than an order of magnitude less sensitive to CCh (EC 50 = 4.8 ± 0.6 μM; Fig. 2e) than the SAN cell GIRK current (EC 50 = 0.25 ± 0.03 μM) 43 , reminiscent of the lower ACh sensitivity of the GIRK conductance in human ventricular myocytes as compared to atrial myocytes 44 . Influence of RGS6 on GIRK-dependent signaling in ventricular myocytes. The M 2 R-I KACh signaling pathway in atrial myocytes and SAN cells is negatively regulated by Regulator of G protein Signaling 6 (RGS6) 22,32,43,45 . Indeed, in SAN cells and atrial myocytes from mice lacking RGS6, GIRK currents evoked by CCh exhibited slower deactivation kinetics and an increased sensitivity to CCh 32,43 . To test whether ventricular GIRK-dependent signaling is negatively regulated by RGS6, we compared CCh-induced GIRK currents in ventricular myocytes from adult wild-type and Rgs6 −/− mice. Current densities were slightly, but significantly, larger in ventricular myocytes from Rgs6 −/− mice as compared to wild-type controls (Fig. 2a,b). No effect of Rgs6 ablation, however, was seen on activation ( Fig. 2c) or deactivation (Fig. 2d) kinetics of the CCh-induced, ventricular GIRK current. Moreover, the EC 50 for CCh-induced activation of the GIRK current was comparable in ventricular myocytes from wild-type and Rgs6 −/− mice (Fig. 2e). Thus, the impact of RGS6 on ventricular GIRK-dependent signaling is limited to a modest negative regulation of peak current density.
Influence of GIRK channels on ventricular myocyte APD and excitability. To test whether I KACh modulates ventricular myocyte excitability and repolarization, we evaluated the change in rheobase and APD induced by CCh (100 μM) in ventricular myocytes from wild-type and Girk4 −/− mice. The CCh-induced decrease in ventricular myocyte excitability (increase in rheobase) was larger in ventricular myocytes from wild-type than Girk4 −/− mice (Fig. 3a). And, although there was no significant difference in the CCh-induced shortening of APD 20 , APD 50 , and APD 70 between ventricular myocytes from wild-type and Girk4 −/− mice (Fig. 3b-e), the CCh-induced decrease in APD 90 was blunted in ventricular myocytes from Girk4 −/− mice (Fig. 3f). Thus, GIRK channels contribute to the cholinergic regulation of excitability and repolarization of adult mouse ventricular myocytes.

Generation and characterization of mice lacking GIRK channels in ventricle.
To probe the impact of ventricular GIRK channel on cardiac physiology, we crossed a conditional Girk1 knockout mouse (Girk1 fl/fl ) 46 with a ventricle-specific Cre driver line (MLC2VCre) 47,48 . We first verified the ventricular specificity of Cre-dependent gene ablation conferred by the MLC2VCre driver line using a Cre-dependent fluorescent reporter mouse line (Fig. 4a,b). We next evaluated the functional impact and cell specificity of  (Fig. 4c,d). We observed no difference, however, in CCh-induced current  but this difference did not reach the level of statistical significance (Fig. 5a-c). Similarly, no significant difference in baseline HR was observed in MLC2VCre(+):Girk1 fl/fl mice relative to MLC2VCre(−):Girk1 fl/fl controls (Fig. 5d). As expected, CCh decreased HR in wild-type mice, and this effect was significantly smaller in Girk4 −/− mice (Fig. 5a-c). In contrast, CCh-induced bradycardia was comparable in MLC2VCre(+):Girk1 fl/fl and MLC2VCre(−):Girk1 fl/fl mice (Fig. 5d). Arrhythmic events, defined as instances of AV block or tachycardic episodes, were also quantified before and after CCh injection ( Supplementary Fig. S2). While no arrhythmic events were observed at baseline for any genotype, wild-type mice exhibited more arrhythmic events than    We next evaluated HRV, examining the root mean square of successive differences (RMSSD) in R-R interval before and after CCh injection. While baseline RMSSD in Girk4 −/− mice was slightly lower in comparison to wild-type mice, this difference did not reach the level of statistical significance (Fig. 6a-c). Consistent with published data 22 , the CCh-induced increased in RMSSD was significantly blunted in Girk4 −/− mice relative to wild-type controls (Fig. 6c). MLC2VCre(+):Girk1 fl/fl and MLC2VCre(−):Girk1 fl/fl mice displayed comparable RMSSD values, both prior to and after CCh injection (Fig. 6c).
Frequency domain analysis can offer a more sensitive measure of the relative contributions of the sympathetic and parasympathetic branches of the autonomic nervous system to HRV 49 . While power in the VLF range (<0.4 Hz) at baseline did not differ between wild-type and Girk4 −/− mice, CCh increased VLF power in wild-type but not Girk4 −/− mice (Fig. 7a, left). In the LF range (0.4-1.5 Hz), Girk4 −/− mice exhibited reduced power relative to wild-type controls at baseline and, in contrast to wild-type mice, a lack of increase induced by CCh (Fig. 7b, left). In the HF range (1.5-5.0 Hz), we observed no difference between wild-type and Girk4 −/− mice at baseline. Interestingly, CCh administration decreased power in this frequency range in wild-type mice, but not Girk4 −/− mice (Fig. 7c, left). No difference in power in VLF, LF, and HF domains was observed between MLC2VCre(+):Girk1 fl/fl and MLC2VCre(−):Girk1 fl/fl mice, either at baseline or after CCh administration ( Fig. 7a-c, right).
With respect to key parameters of the ECG waveform (Supplementary Table 1), no significant genotype differences were observed for PR and QT intervals at baseline. CCh significantly prolonged both the PR and QT intervals in wild-type mice, and this effect was blunted in Girk4 −/− mice. PR and QT intervals were also prolonged in MLC2VCre(+):Girk1 fl/fl and MLC2VCre(−):Girk1 fl/fl mice after CCh administration, but there was no difference between genotypes. We also evaluated the corrected QT (QTc) interval, a measure that can account for the influence of HR on the QT interval 50 , using a variation of the human QT correction formula that assumes a typical RR interval for a conscious mouse under normal conditions (100 ms). We observed no difference in QTc interval between wild-type and

Impact of ventricle-specific GIRK channel ablation on pacing-induced ventricular arrhythmia.
To investigate the contribution of the ventricular GIRK channel to ventricular arrhythmogenesis, isolated hearts from MLC2VCre(+):Girk1 fl/fl and MLC2VCre(−):Girk1 fl/fl mice, as well as wild-type and constitutive Girk4 −/− mice, were subjected to burst pacing in the absence and presence of CCh to discern differences in susceptibility to ventricular tachycardia (VT) or fibrillation (VF) (Fig. 8a). There was no significant difference in the incidence of pacing-induced VT/VF at baseline between wild-type and Girk4 −/− hearts (Fig. 8b, left). Similarly, hearts from MLC2VCre(+):Girk1 fl/fl and MLC2VCre(−):Girk1 fl/fl mice did not differ in the incidence of VT/VF in response to pacing stimuli at baseline. After CCh perfusion, there was, again, no significant difference in incidence of pacing-induce VT/VF between wild-type and Girk4 −/− hearts (Fig. 8b, right). Likewise, no difference was observed in the incidence of pacing-induced VT/VF between MLC2VCre(+):Girk1 fl/fl and MLC2VCre(−):Girk1 fl/fl hearts. Collectively, these observations suggest that loss of the ventricular GIRK channel does not impact susceptibility to pacing-induced ventricular arrhythmias.

Discussion
In this study, we present evidence that that GIRK channel found in adult mouse ventricular myocytes is, like the channel described previously in atrial myocytes and SAN cells, a GIRK1/GIRK4 heteromultimeric channel. Nevertheless, the GIRK current in ventricular myocytes exhibited several properties that differed from the GIRK current described previously in atrial myocytes and SAN cells 43,45 . Notably, the ventricular GIRK current was smaller and less sensitive to CCh. The differences in GIRK conductance between atrial and ventricular myocytes could be due to increased expression and/or influence of a negative regulator of GIRK-dependent signaling in ventricle. Our data suggest that increased ventricular expression or influence of RGS6, however, does not explain the functional differences between the mouse atrial and ventricular GIRK conductance. Indeed, RGS6 mRNA levels were substantially lower in ventricle than in atria. Moreover, Rgs6 ablation did not enhance the CCh sensitivity or alter the kinetics of the ventricular GIRK current. It is possible that other RGS proteins account for the differences observed between in atrial and ventricular GIRK-dependent signaling in the mouse heart. Alternatively, elements of the GIRK-dependent signaling pathway may be compartmentalized in distinct fashion in the atria and ventricle, or additional positive and/or negative regulatory elements that impact I KACh -dependent signaling remain to be discovered.
A recent study reported that the ACh-induced shortening of ventricular APD 90 was diminished in isolated hearts from Girk4 −/− mice, suggesting that a GIRK4-containing GIRK channel facilitates ventricular repolarization in the mouse heart 29 . Consistent with this premise, we observed a significant impact of Girk4 ablation on the (within genotype); ##,### P < 0.01 and 0.001, respectively, vs. wild-type (within treatment). (c) Power in the HF (1.5-5.0 Hz) range at baseline and following injection of CCh (1.0 mg/kg i.p.). Two-way ANOVA analysis revealed an interaction between genotype and treatment (F 1,23 = 37.5, P < 0.001) for wild-type and Girk4 −/− mice (left panel). There were no main effects of treatment (F 1,23 = 2.7, P = 0.11) or genotype (F 1, 23  CCh-induced shortening of APD 90 in isolated ventricular myocytes from adult mice. Our efforts in isolated ventricular myocytes also revealed that Girk4 ablation attenuates the CCh-induced decrease in ventricular myocyte excitability. These results support our contention that the GIRK channel in ventricular myocytes contributes to the cholinergic regulation of ventricular myocyte APD and excitability. I KACh is a critical downstream mediator of parasympathetic effects on atrial physiology and HR regulation. Previous studies with ECG telemetry and isolated mouse hearts have shown that a loss of I KACh correlates with altered baseline measures and diminished responses to pharmacologic manipulations that enhance or mimic parasympathetic input to the heart [19][20][21][22] . Here, we replicated several reported observations in Girk4 −/− mice, including the attenuation of CCh-induced bradycardia. While we did not see a significant baseline tachycardia in Girk4 −/− mice in this study, this may reflect an influence of anesthesia. We additionally observed evidence of diminished HRV at baseline (LF), and after CCh administration (RMSSD, VLF, and LF), in Girk4 −/− mice. Our parallel evaluation of mice lacking GIRK channels selectively in ventricle indicate that most, if not all, of the cardiac phenotypes reported in constitutive Girk1 −/− or Girk4 −/− mice are attributable to a loss of atrial I KACh . Surprisingly, we found that CCh decreased power in the HF domain in wild-type mice, but not in Girk4 −/− mice. As fluctuations in HR due to respiration contribute to power in the HF range 51 , our observation could reflect a contribution of a GIRK4-containing GIRK channel to the cholinergic regulation of respiration. The CCh-induced decrease in HF power was not significant in MLC2VCre(−):Girk1 fl/fl mice, however, suggesting that this parameter may not be a reliable indicator of the GIRK channel contribution to HRV in this frequency range.
Despite the fact that GIRK channels contribute to the inhibitory influence of CCh on ventricular myocytes, we did not see an obvious impact of ventricular GIRK channel ablation on QT or QTc intervals, at baseline or following CCh injection. It should be noted that previous attempts to model long-QT syndrome in mice have not reliably yielded mice with prolonged QT intervals 52 . For example, mice lacking KvLQT1, the gene associated with LQT1 in large mammals, do not consistently display prolonged QT intervals 52 . We also cannot exclude the possibility that the influence of ventricular GIRK channel becomes more relevant in pathological conditions such as myocardial infarction or sympathetic stress. We did see a surprising prolongation of the QTc interval after CCh administration in Girk4 −/− mice that was not seen in wild-type controls. While this observation may reflect a failure of the QT correction to adequately compensate for the impact of HR on QT interval or a limitation of this approach in mice under anesthesia 53,54 , the finding is also consistent with the possibility that CCh modulates another effector that exerts an opposing influence from the GIRK channel on the QT interval. Importantly, since MLC2VCre(+):Girk1 fl/fl and MLC2VCre(−):Girk1 fl/fl mice exhibited similar QT and QTc intervals at baseline and after CCh administration, the selective prolongation of QTc interval in Girk4 −/− mice presumably reflects an influence of GIRK channels outside of the ventricle. I KACh -dependent signaling has been linked to a variety of atrial rhythm disorders [55][56][57] . Both genetic ablation of Girk4 and pharmacological inhibition of GIRK channels can restore normal cardiac rhythm in multiple genetic models of supraventricular arrhythmia 55,56 . For example, genetic ablation of Girk4 in mouse models of sick sinus syndrome and AV block reduces the number of ACh-induced arrhythmic episodes in these mice 56 . Additionally, fewer arrhythmic events were observed in Girk4 −/− mice treated with CCh as compared to wild-type controls (this study), and Girk4 −/− mice were resistant to pacing-induced AF 35 . Studies in human AF patients further implicate I KACh in supraventricular arrhythmogenesis. Atrial myocytes from human AF patients displayed both a significant reduction in I KACh expression and an attenuation of CCh-induced shortening of atrial APD 58 . CCh-induced I KACh currents were also smaller in atrial myocytes from AF patients, whereas constitutive GIRK channel activity was elevated [58][59][60] . Remodeling involving I KACh has also been observed in a "tachy-paced" dog model, where atrial tachycardia increased basal I KACh channel activity 61 ; this adaptation correlated with a decrease in ERP and an increase in AF duration. Collectively, these studies highlight the therapeutic potential associated with blocking or suppressing atrial I KACh in multiple distinct atrial arrhythmias. I KACh antagonists have been proposed for use in the treatment of AF and other supraventricular rhythm disorders 56,57,[62][63][64] . Recently, a novel family of direct-acting, small molecule activators and inhibitors of GIRK channels has been identified, members of which act selectively on GIRK1-containing GIRK channels such as I  . While a direct-acting I KACh antagonist could prove useful in the treatment of atrial rhythm disorders, suppressing GIRK channel activity in the ventricle could pose a significant risk. We did not, however, observe a difference in susceptibility to pacing-induced ventricular arrhythmias in either the constitutive or ventricle-specific GIRK channel ablation models, arguing that a loss of ventricular GIRK channel does not predispose to ventricular arrhythmias. However, given the noted differences in the ionic currents underlying ventricular repolarization in mice and humans 53 , further studies are needed to determine whether inhibition of the human ventricular GIRK channel is also without effect. Nonetheless, our data support the contention that targeting I KACh for atrial arrhythmias, via with selective inhibitors or genetic approaches, may constitute a safe alternative to existing anti-arrhythmic therapies that predispose patients to fatal ventricular disorders 64 .

Methods
Animals. All experimental procedures involving mice were approved by the Institutional Animal Care and Use Committee of the University of Minnesota, and were conducted in accordance with guidelines set by the National Institute of Health. The generation of Girk1 −/− , Girk4 −/− , Rgs6 −/− , and Girk1 fl/fl mice was described previously 19,20,45,46 . Girk1 fl/fl mice were crossed with MLC2VCre(+) 47 mice to generate the MLC2VCre:Girk1 fl/fl line. B6.Cg-Gt(ROSA)26Sor tm14(CAG-tdTomato)Hze /J (Ai14-tdTomato) reporter mice were purchased from The Jackson Laboratory (Bar Harbor, ME) and crossed with MLC2VCre(+) mice. Mice were group-housed on a 12-h light/ dark cycle, and given free access to food and water and were used for experiments at age 8-12 wk.
Whole-cell electrophysiology. Coverslips containing adult SAN cells or ventricular myocytes were transferred to a perfusion chamber and electrophysiological recordings were conducted as described 45 . SAN cells were identified as thin, striated cells with capacitance values between 20-40 pF. Ventricular myocytes were identified as quiescent, rod-shaped cells with capacitance values ranging from 100-250 pF. Whole-cell currents and voltages were measured with hardware (Axopatch-700B amplifier, Digidata 1440A) and software (pCLAMP v. 10.4) from Molecular Devices (Sunnyvale, CA). All currents were low-pass filtered at 0.1 kHz, sampled at 5 kHz, and stored on computer hard disk for analysis. The liquid-junction potential, predicted to be −17 mV using JPCalc software (Molecular Devices), was not corrected. CCh-induced currents were measured at a holding potential of −70 mV. Borosilicate patch pipettes were filled with (in mM): 130 K-gluconate, 2 MgCl 2 , 1.1 EGTA, 5 HEPES, 2 Na 2 ATP, 5 phosphocreatine, and 0.3 Na-GTP, pH 7.2 (with KOH). Whole-cell access was achieved in a low-K + bath solution consisting of (in mM): 130 NaCl, 5.4 KCl, 1 CaCl 2 , 1 MgCl 2 , 5.5 glucose, 5 HEPES/NaOH, pH 7.4. A high-K + bath solution (+/−CCh) consisting of (in mM): 120 NaCl, 25 KCl, 1 CaCl 2 , 1 MgCl 2 , 5.5 glucose, 5 HEPES/NaOH, pH 7.4 was applied via ValveLink 8.2 rapid perfusion system (AutoMate Scientific, Berkeley, CA). Activation and deactivation time constants were extracted from appropriate regions of current traces, which were fit with a 1-term Boltzmann equation. Rheobase was evaluated in current-clamp mode in low-K + bath solution by injecting increasing steps (5 pA/500 ms) of depolarizing current, starting at 0 pA, until an action potential spike was elicited. APD 20 , APD 50 , APD 70 , and APD 90 were measured from the average of 10 spikes elicited by injecting a supra-threshold current (1 nA/5 ms) into the cell. The % change in APD and rheobase after CCh perfusion was measured as with the follows: ((baseline − post-CCh/baseline) × 100). Only those experiments in which the access resistances were stable and low (<20 MΩ) were included in the final analysis.
Microscopy. Hearts were fixed with 4% paraformaldehyde in PBS at 4 °C for 16-24 h, transferred to a 10% sucrose/PBS solution at 4 °C for 24 h, and then transferred to a solution of 30% sucrose/PBS solution at 4 °C for at least 48 h. 20 μm thick slices were cut by cryostat at −20 °C and slices were mounted and stained with ProLong Gold Antifade reagent with DAPI (ThermoFisher Scientific; Waltham, MA). Fluorescent images were captured with an Olympus BX51W1 upright microscope with Disk Spinning Unit, using a 2X objective and digital CCD camera (C10600-10B, Hamamatsu Photonic System Corp; Bridgewater, NJ). Images were processed with Metamorph Advanced 7.7.7.0 (Molecular Devices).
In vivo ECG recordings. Mice were anesthetized with 1.5% isoflurane supplemented with an air mixture of 40% O 2 /60% N 2 to sustain stable HR. Platinum needle ECG electrodes connected to an IX-ECG-12 recorder (iWorx; Dover, NH) were placed subcutaneously into each limb. Baseline HR was recorded for 10 min, at which point CCh (1.0 mg/kg i.p.) was administered. HR and HRV analysis was performed on 1-min bins of stable recording from minutes 9-10 (baseline) and 15-16 min (post-CCh), using Kubios HRV 3.02 software 69 . Artifact detection/correction was utilized to reduce the impact of ectopic and skipped beats (either AV block or RR intervals which differ from the neighboring RR intervals by over 25%) on HR and HRV analysis. The frequency bands for HRV analysis were defined as very low frequency (VLF), <0.4; low frequency (LF), 0.4-1.5 Hz; and high frequency (HF), 1.5-5.0 Hz; using a 120 s window with a 50% overlap for Fast Fourier Fransform spectrum analysis using Welch's periodogram method. Arrhythmic events were defined as instances of AV block (P wave with no corresponding QRS complex) and/or instances of tachycardia (at least 3-5 beats). PR and QT intervals were analyzed from the average of 100 cycles from either lead I or lead II during these segments, using LabScribe v3 software, only from recordings where the noise level was sufficiently low to clearly delineate the beginning and end of the indicated waveform deflections. The QT interval was corrected for changes in RR interval using the following formula: QTc = SQRT[QT/(RR/100)] 50 .
SCieNTifiC REPORTS | (2018) 8:1192 | DOI:10.1038/s41598-018-19719-x Statistical analysis. All data are presented as the mean ± SEM. Statistical analyses were performed using Prism 5 (GraphPad Software, Inc.; La Jolla, CA) software. Male and female mice were used in all studies, and groups were balanced for sex. Data were analyzed with Student's two-tailed t-test (unpaired and paired), Fisher's exact test, 1-way ANOVA, and 2-way ANOVA with repeated measures, as appropriate. For studies involving 2-way ANOVA analysis, only the interaction is reported if one was detected. The Bonferroni multiple comparison post hoc test was used following 1-way or 2-way ANOVA analyses, if justified. The level of significance was set at P < 0.05. Data availability. The datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request.