Modification of distinct ion channels differentially modulates Ca2+ dynamics in primary cultured rat ventricular cardiomyocytes

Primary cultured cardiomyocytes show spontaneous Ca2+ oscillations (SCOs) which not only govern contractile events, but undergo derangements that promote arrhythmogenesis through Ca2+ -dependent mechanism. We systematically examined influence on SCOs of an array of ion channel modifiers by recording intracellular Ca2+ dynamics in rat ventricular cardiomyocytes using Ca2+ specific fluorescence dye, Fluo-8/AM. Voltage-gated sodium channels (VGSCs) activation elongates SCO duration and reduces SCO frequency while inhibition of VGSCs decreases SCO frequency without affecting amplitude and duration. Inhibition of voltage-gated potassium channel increases SCO duration. Direct activation of L-type Ca2+ channels (LTCCs) induces SCO bursts while suppressing LTCCs decreases SCO amplitude and slightly increases SCO frequency. Activation of ryanodine receptors (RyRs) increases SCO duration and decreases both SCO amplitude and frequency while inhibiting RyRs decreases SCO frequency without affecting amplitude and duration. The potencies of these ion channel modifiers on SCO responses are generally consistent with their affinities in respective targets demonstrating that modification of distinct targets produces different SCO profiles. We further demonstrate that clinically-used drugs that produce Long-QT syndrome including cisapride, dofetilide, sotalol, and quinidine all induce SCO bursts while verapamil has no effect. Therefore, occurrence of SCO bursts may have a translational value to predict cardiotoxicants causing Long-QT syndrome.

The orchestrated mechanical activity of the heart is controlled by electrical pulses initiating from the sino-atrial node and finally conveys to the ventricles leading to rapid depolarization of all ventricular myocytes and coordinated contraction of the heart 1 . The rhythmic cardiac activity can be disrupted under certain circumstances, leading to cardiac arrhythmia. Both abnormally slow (bradycardia) and rapid (tachycardia) heart rates can lead to syncope and sudden death 1,2 . The most dangerous arrhythmias are those that originate from the ventricles, such as torsades de pointes (TdP) ventricular tachycardia and ventricular fibrillation 3,4 . Many studies have demonstrated that gain or loss of function of ion channels could shape cardiac action potentials (APs) and contribute to arrhythmia susceptibility 5 . Voltage-gated sodium channels (VGSCs) are responsible for the AP generation of the cardiomyocytes. Dysfunction of VGSCs by point mutation on the α -subunit leads to several types of arrhythmia, such as Long-QT (LQT) syndrome and Brugada syndromes 6,7 . Voltage-gated potassium channels (VGPCs) participate in the repolarization of the AP. Loss-of-function of Kv conductance results in AP prolongation leading to LQT syndrome while gain-of-function results in shortened AP duration leading to Short QT (SQT) syndrome 8 . Among the Kvs, the hERG channels (Kv11.1, encoded by human Ether-à-go-go Related Gene) are the major contributors to rapid delayed rectifier potassium currents (I Kr ) which are involved in AP repolarization 9 . In many cases, inhibition of hERG channels results in prolonged AP leading to LQT ventricular arrhythmia, and sometimes, sudden cardiac death 10 . Therefore, functional alteration of the sodium and potassium channels tightly associated with the arrhythmia 11 .
Primary cultured cardiomyocytes show spontaneous transient increase in intracellular Ca 2+ concentration (spontaneous Ca 2+ oscillations, SCOs) 12 . These SCOs occur parallel with the AP generation and control ventricular cardiomyocytes contractile events (including systolic and diastolic function) through a process known as excitation-contraction coupling 12 . It is well documented that inappropriate Ca 2+ homeostasis in ventricular cardiomyocytes are associated with the ventricular tachycardia. Re-opening of L-type Ca 2+ channels (LTCCs) or other depolarizing currents before normal repolarization completes contributes to the early afterdepolarization (EAD). Gain-of-function mutations on Cav1.2 (calcium channel subtype 1.2) produces Timothy syndrome which characterized by a heart condition similar to LQT syndrome 13,14 . Aberrant spontaneous, diastolic Ca 2+ leakage from the sarcoplasmic reticulum due to point mutation on type 2 ryanodine receptors (RyR2) contributes to formation of delayed after-depolarization (DAD) which leads to heart failure and catecholaminergic polymorphic ventricular tachycardia (CPVT) 15 . In addition to governing contractile events, dysregulation of intracellular Ca 2+ also undergoes derangements that promote arrhythmogenesis through Ca 2+ -dependent and coupled electrophysiological effects. Aberrant Ca 2+ signals can modulate CaMKII activity which in turn regulates the activity of a variety of ion channels and transporters, for examples Nav1.5 16 , RyR2, and SERCA2a 17,18 .
In this study, we systematically examined the influence of an array of ion channel modulators on SCO patterns by detecting the intracellular Ca 2+ dynamics in primary cultured rat ventricular cardiomyocytes using Fluorescence Imaging Plate Reader (FLIPR) in 96-well format. We demonstrate that modification of distinct ion channels differentially affects SCO patterns. In addition, we demonstrate that clinically-used drugs including cisapride, dofetilide, sotalol, and quinidine which cause LQT syndrome all produce characteristic SCO bursts therefore prolong the SCO/burst duration. Our results demonstrate that occurrence of SCO bursts may have a translational value to predict cardiotoxicants causing LQT syndrome.

Results
Cultured rat cardiomyocytes displayed spontaneous Ca 2+ oscillations. After 24 h, cultured cardiomyocytes displayed elongated and triangular morphology (Fig. 1). The percentage of cardiomyocytes in our culture system was greater than 95% demonstrated by double staining with anti-cTnT, a specific cardiomyocyte marker, and Hoechst 33342 (Fig. 1). The cultured cardiomyocytes displayed spontaneous beating 24 h after plating which can be observed in a phase-contract microscopy (data not shown). It has been well established that the spontaneous beating was correlated with SCO generation 19 in primary cardiomyocyte cultures. We therefore examined SCOs using rapid throughput machine, FLIPR ® TETRA . FLIPR ® TETRA can simultaneous record the fluorescence intensity from the center area of each well in 96-or 384-well formats therefore representing an overall Ca 2+ signals from a population of cells. At a plating density of 1.5 × 10 5 cells/well and after cultured for 60 h, cardiomyocytes displayed rhythmic SCOs, with minimum variability in the frequency, amplitude as well as full width at half maximum (FWHM) (Supplemental Fig. 1). The frequency, amplitude and FWHM were stable during the whole recording time which was 800 s (Supplemental Fig. 2). It should be noted that different batches of ventricular cardiomyocyte cultures showed certain degree of variability in SCO frequency, amplitude as well as FWHM (Supplemental Fig. 2). To diminish the variability, the response on SCO of each compound was normalized to the individual basal parameters (frequency, amplitude and FWHM) from an initial 100 s recording, respectively.

Epinephrine and acetylcholine affected SCO frequency in primary cultured rat cardiomyocytes.
To test whether the source of the SCOs is from extracellular Ca 2+ influx or intracellular Ca 2+ store release in our culture system, the cells were bathed in normal extracellular Ca 2+ (2.  SCO amplitude without effect on SCO duration (Supplemental Fig. 3). These data suggested that extracellular Ca 2+ was required for the generation of SCOs in primary cultured ventricular cardiomyocytes.
LTCCs are directly pertinent to the intracellular Ca 2+ dynamics and are also responsible for the repolarization of the action potentials (APs) 21 . Activation of LTCC by Bay K 8644 produced irregularity on the SCO pattern with the occurrence of SCO bursts accompanied with increased amplitude and longer FWHM (Fig. 3). The EC 50 values for Bay K 8644 stimulating SCO amplitude and mean SCO/bursts duration were 4.63 nM (2.44-8.78 nM, 95% CI) and 45.7 nM (3.50-599 nM, 95% CI), respectively. Nifedipine, an LTCC inhibitor, concentration-dependently suppressed the SCO amplitude (IC 50 = 34.1 nM, 18.6-62.6 nM, 95% CI) while slightly increased the SCO frequency (EC 50 = 53.4 nM, 19.3-149 nM, 95% CI) without significant effect on the SCO duration (Fig. 3).

Modulation of ryanodine receptors affected SCO pattern in primary cultured rat cardiomyocytes.
The ryanodine receptor (RyR) is one of the store operated Ca 2+ channel located in the endoplasmic/sarcoplasmic reticulum and is responsible for Ca 2+ -induced Ca 2+ release (CICR). RyR2 is primarily expressed in cardiac tissues. Upon activation following LTCC mediated Ca 2+ influx, RyR2 releases Ca 2+ from the sarcoplasmic reticulum into the cytosol, enabling cardiac muscle contraction 22 . FLA-365, a RyR inhibitor 23 , concentration-dependently suppressed the SCO frequency in primary cultured cardiomyocytes (IC 50 = 0.93 μ M, 0.81-1.06 μ M, 95% CI) without affecting the SCO duration. Higher concentration of FLA-365 ( ≥3 μ M) eliminated SCOs (Fig. 4). Low concentration of ryanodine, which activated RyRs, also modulated the SCO patterns. Application of ryanodine produced SCO bursts ( Fig. 4) with increased SCO/burst FWHM.
Modulation of voltage-gated sodium channel influenced SCO pattern in primary cultured rat cardiomyocytes. VGSCs are responsible for the AP generation which is required for the rhythmic cardiomyocyte contraction for cardiomyocytes 21 . We therefore examined the influence of both VGSC agonist, veratridine and the blocker, tetrodotoxin (TTX) on the SCOs. Veratridine at 1 μ M significantly increased the SCO FWHM. Clinically-used drugs causing LQT syndrome produced characteristic bursts of SCO in primary cultured rat cardiomyocytes. The clinically-used drugs that were withdrawn from the market due to adverse cardiac effect were primarily attributed to their abilities to induce LQT syndrome or TdP 26 . We therefore tested whether those clinically-used drugs which have been reported to cause LQT syndrome including cisapride, dofetilide, sotalol and quinidine could modulate SCO patterns. As shown in Fig. 7, dofetilide, cisapride and quinidine all produced SCO bursts with long quiescent time between two SCO bursts with higher concentration having greater response. The response of sotalol on inducing SCO bursts was somewhat smaller and the SCO bursts only occurred at 100 μ M. The EC 50 (Fig. 8).

Discussion
In this study, using rapid throughput imaging system, we systematically investigated the influence of an array of ion channel modifiers on intracellular Ca 2+ ([Ca 2+ ] i ) dynamics in primary cultured rat ventricular cardiomyocytes. We demonstrate that altering the activity of Ca 2+ channels which contribute to [Ca 2+ ] i dynamics or ion channels which are involved in AP generation both affect SCO patterns. The potencies of tested reference compounds are generally consistent with their affinities on their respective targets (Table 1) suggesting target-specific Ca 2+ response in cultured rat ventricular cardiomyocytes. The [Ca 2+ ] i not only governs contractile events (including systolic and diastolic function), but also undergoes derangements that promote arrhythmogenesis through Ca 2+ -dependent and coupled electrophysiological effects 27 . Our data therefore imply that modification of SCO in primary cultured ventricular cardiomyocytes detected by FLIPR ® TETRA system may serve as a rapid throughput method capable of identifying potential cardiotoxicants from environment or early stage of drug development. We demonstrate that acetylcholine dramatically decreases the SCO frequency consisting with previous study where acetylcholine decreases cardiomyocytes contraction resulting from activation of subtype 2 muscarinic acetylcholine receptors 28 . Epinephine, a β -adrenergic receptor agonist dramatically increases SCO frequency which is also consistent with clinically observed heart rate increase 29 . Cardiomyocytes contraction is tightly coupled to SCO which is controlled by APs 12 . The APs, SCOs and contractions of ventricular cardiomyocytes occur simultaneously 19 . Therefore, perturbation of SCO frequency may predict cardiotoxicants which are capable of producing tachycardia or bradycardia.
An important finding is that compounds which contribute to [Ca 2+ ] i dynamics or shape APs affect SCO pattern differently. Activation of LTCC by Bay K 8644 30 induces bursts of SCOs and slightly increases SCO/burst amplitude whereas suppression of LTCC concentration-dependently decreases SCO amplitude and duration with concomitant increase of SCO frequency, a phenomenon observed in the clinic 31 . Activation of RyR by low concentrations of ryanodine ( ≤10 μ M) 32 also produces bursts of SCOs. However, in contrast to increased amplitude by Bay K 8644, ryanodine slightly decreases the SCO/burst amplitude. Suppression of RyR by FLA-365 23 decreases SCO frequency without affecting SCO amplitude and duration. Higher concentrations ( ≥3 μ M) of FLA-365 completely abolish SCOs. Activation of VGSCs which is responsible for AP generation also induces busts of SCOs without changing SCO amplitude. TTX and lamotrigine both decrease SCO frequency but not the amplitude and duration. It should be noted that TTX is a VGSC channel blocker while lamotrigine is a use-dependent VGSC inhibitor. Lamotrigine binds to IVS6 segment of sodium channels 33 which is topologically distinct from the TTX binding sites 34 . These data suggest that compounds suppressing VGSCs can produce similar SCO profile regardless of their modes of action. Suppression of Kv channels also produces irregularity with prolonged SCO duration and occurrence of SCO bursts. Altogether, these data demonstrate that distinct targets engagements produce different SCO patterns therefore mapping the SCO profiles produced by cardiotoxicants may be capable of predicting their molecular target(s).
Another important finding is that E-4031, a specific Kv11.1 (hERG) blocker, produces SCO bursts and decreases the SCO/burst frequency. The EC 50 and IC 50 values of E-4031 on SCO/burst duration and frequency are 38.3 nM and 40.6 nM, respectively, which were consistent with its affinity on hERG channels 35 . Blockade of hERG channel is proposed to be the major cause of LQT syndrome of many withdrawn clinically-used drugs 36 . We demonstrate that four clinically-used drugs including cisapride, dofetilide, sotalol and quinidine that were withdrawn due to their ability to produce LQT also induce SCO bursts in ventricular cardiomyocyte cultures. It should be mentioned that cisapride, dofetilide, and sotalol directly inhibit hERG channel activity whereas quinidine decreases the surface expression of hERG channels by suppressing hERG channel translocation to cell membrane surface 37 . These data imply that regardless of mode of action, functional suppression of hERG channel activity by cardiotoxicants produces similar characteristic SCO prolongation/occurrence of SCO bursts. The potencies for dofetilide and sotalol increasing SCO/burst duration are comparable with their affinities on hERG channels 38,39 . However, the EC 50 values for cisapride and quinidine are significantly deviated from their affinities to hERG channels 40,41 . Cisapride has been demonstrated to affect many other channels such as Kv4.1 42 and receptors such as 5-HT4 receptors 43 . In addition to inhibiting hERG, quinidine inhibits both Na + and Ca 2+ currents 44,45 . The integrated response from many ion channels interaction may account for potency difference between hERG blockade and increased SCO duration. Verapamil is an anti-hypertension drug used in the clinic. Although verapamil is a potent hERG channel blocker, no evidence exists that verapamil can produce LQT syndrome 21 . We demonstrate that verapamil has no effect on inducing SCO bursts although it decreases SCO frequency. The later response is likely derived from suppression of LTCC 46 . Therefore, our data demonstrate that not all the hERG channel blockers are capable of inducing SCO busts in primary cultured rat ventricular cardiomyocytes and only those compounds producing LQT syndrome can stimulate SCO.
In addition to clinically-used drugs, we also demonstrate that sodium channel agonist, veratridine and LTCC activator, Bay K 8644 as well as the universal potassium channel blocker, 4-AP produce SCO bursts. It has been demonstrated that veratridine prolongs AP duration in rabbit ventricular myocytes 47 . Although inactive on hERG channels, alfuzosin produces LQT syndrome in the clinic due to its ability to increase the peak current of Nav1.5 48 . Genetic mutations in ion channels also lead to LQT syndrome. Gain of function of VGSCs mutations leads to LQT syndrome and Brugada syndromes 6,7 . Loss-of-function of Kv channel conductance results in AP prolongation leading to LQT syndrome 8 . Gain-of-function mutations on Cav1.2 produce Timothy syndrome which is characterized by a heart condition similar to LQT syndrome 14 . Therefore, these data together demonstrate that induced SCO bursts may predict the compounds with ability to induce LQT syndrome. A large set of LQT positive and negative reference compounds are required to be investigated to fully establish the relationship between prolonged SCO/burst duration and LQT and how accuracy of this model to predict the compounds with ability to induce LQT syndrome.
A variety of compounds-induced SCO patterns have been investigated in induced pluripotent stem cell (iPSC)-derived cardiomyocytes (iPSC-CMs). The hERG channel inhibitors cisapride, terfenadine, astemizole, and pimozide, all characteristically delay SCO recovery 49 . Sodium channel blockers lidocaine and tetrodotoxin slow beating frequency and various other irregularities 49 . Drug with positive chronotropic effect stimulates the SCO frequency 49 . In contract to our observation in cultured rat cardiomyocytes where acetylcholine decreases SCO frequency, no response was observed when iPSC-CMs were challenged with acetylcholine suggesting that iPSC-CMs lacks M2 receptor signaling pathways 49 . Despite the similarity in the expression of specific cardiac biomarkers, electrophysiology, and pharmacology 50 , when comparing hiPSC-CMs to human cardiac tissue, the immature electrophysiological properties and Ca 2+ handling of iPSC-CM were reported 51 . In addition, recent studies have pointed out the limitations of using iPSC-CMs on drugs screening and toxicity test. The cost of iPSC-CMs is currently still high which is an important consideration when using iPSC-CMs for cardiotoxicity screening 21 . Second, the purity of iPSC-CMs is typically less than 75% and the functional status of iPSC-CMs only exists for a short period 52 . Furthermore, iPSC-CM is more sensitive to DMSO exposure with 0.1% DMSO affecting contractile amplitude and beating rates which may also affects the results explanation 53 .
In summary, we demonstrate that modification of distinct targets differentially affects the SCO pattern in primary rat cardiomyocyte cultures. Although a large set of LQT positive and negative compounds were required to be investigated to fully establish the relationship between LQT and SCO bursts, we demonstrate that clinically-used drugs which produce LQT all produce SCO bursts while negative drugs do not. Therefore, this method may serve as a primary screening for detecting the cardiotoxicants in the early stage of drug development. In addition, mapping the profiles induced by cardiotoxicants may predict their molecular target(s). More importantly, this method may have translational value for detecting the compounds which have the potential to cause LQT syndrome.   Table 1. Comparison of potencies of ion channels modifiers tested on SCO responses and that from their respective targets. "-" means no response.

Materials
515-575 nm range was recorded. The basal fluorescence was recorded for approximately 300 s at a sampling rate of 8 Hz. After recording the basal fluorescence, the drugs (25 μ L, 8X) were added to different wells by an automated, programmable pipetting system yielding a final volume of 200 μ L/well. The fluorescence was recorded for additional 500 s at a sampling rate of 8 Hz. The fluorescence data were presented as Δ F/F 0 , where F is the fluorescence signal at different time point minus background fluorescence whereas the F 0 is the basal fluorescence minus the background fluorescence. The background fluorescence was from the sister wells which were not loaded with Fluo-8/AM. To quantify the drug influence on the SCOs, we analyzed SCO frequency, peak amplitude, as well as SCO duration (FWHM). Many drugs induced SCO bursts and in this case, the SCO duration was calculated as SCO burst duration. A burst was defined to be a cluster of SCOs (at least two) where the previous SCO can't recover to below 20% of the previous SCO amplitude when followed SCO occurred. When drugs induced SCO bursts, a burst was count as an SCO.
Immunofluorescence. Cardiomyocytes at 2 DIV were fixed with 4% paraformaldehyde for 15 min and then permeabilized with 0.25% Triton ™ X-100 for 15 min under gentle shaking. After blocking with 5% goat serum in PBS for 30 min, cells were incubated with anti-cTnT (1:400) antibody overnight at 4 °C. Cells were then incubated with Alexa Fluor ® 488-conjugated goat anti-rabbit (1:500) secondary antibody for 2 h at RT. After aspirating secondary antibody, a concentration of 2 μ g/mL of Hoechst 33342 was added to stain the nuclei. Images were taken using a Nikon eclipse fluorescence microscope using FITC filter and DAPI filter.
Data Analysis. Graphing and statistical analysis were performed using GraphPad Prism software (Version 5.0, GraphPad Software Inc., San Diego, CA, USA) and Origin Pro 8 (Origin Lab Corporation, USA). Peaks were fit by Origin Pro 8 (Origin Lab Corporation, USA). The EC 50 values and the 95% confidence intervals were determined by non-linear regression using GraphPad Prism (Version 5.0, GraphPad Software Inc., San Diego, CA, USA). Statistical significance between different groups was calculated using an ANOVA and, where appropriate, a Dunnett's multiple comparison test; p values below 0.05 were considered to be statistically significant.