ION CHANNELS, RECEPTORS AND TRANSPORTERS

Class 3 inhibition of hERG K+ channel by caffeicacid phenethyl ester (CAPE) and curcumin

Seong Woo Choi & Kyung Su Kim & Dong Hoon Shin &

Hae Young Yoo & Han Choe & Tae Hee Ko & Jae Boum Youm &

Woo Kyung Kim & Yin Hua Zhang & Sung Joon Kim

Received: 16 January 2013 /Revised: 3 February 2013 /Accepted: 4 February 2013 /Published online: 26 February 2013# Springer-Verlag Berlin Heidelberg 2013

Abstract Human ether-á-go-go-related gene (hERG) K+

channel current (IhERG) is inhibited by various compoundsand genetic mutations, potentially resulting in cardiac ar-rhythmia. Here, we investigated effects of caffeic acid phe-nethyl ester (CAPE) and curcumin, two natural anti-inflammatory polyphenols, on IhERG in HEK-293 cells over-expressed with hERG. CAPE dose-dependently decreasedrepolarization tail current of hERG (IhERG,tail; IC50, 10.6±0.5 μM). CAPE also shifted half-activation voltage (V1/2) tothe left (from −17.5 to −26.5 mV) and accelerated activationand inactivation kinetics. The CAPE inhibition of IhERG,tailwas not attenuated in the pore-blocker site mutants of hERG(Y652A and F656A). A point mutation of Cys723 (C723S)mimicked the effects of CAPE and caused a left shift of V1/2

and acceleration of IhERG,tail deactivation. However, IhERG,tailinhibition by CAPE was still observed in C723S. Takentogether, CAPE inhibits hERG channel by class 3 mecha-nism, i.e., modification of gating, not by blocking the pore.Curcumin induced changes of IhERG similar to those ofCAPE, while additional interaction with pore-blocking siteswas suggested from attenuated IhERG,tail inhibition in Y652Aand F656A. Interestingly, IhERG induced by human actionpotential voltage clamp was increased by CAPE while de-creased by curcumin. Mathematical simulation of actionpotential derived from the experimental results of CAPEand curcumin supports that CAPE, but not curcumin, wouldinduce shortening of AP duration by facilitation of IhERG.The above results revealed intriguing roles of Cys723 inhERG kinetics and suggested that conventional drug screen-ing by using step pulse protocol for IhERG,tail would overlookthe hERG kinetic modulations that could compensate thedecrease of IhERG,tail.

Keywords hERGK+ channel . Caffeic acid phenethyl ester .

Curcumin . Electrophile

Introduction

Malfunctions of cardiac ion channels are sometimes fatal.One of the undesirable clinical cases is the long QT syn-drome (LQTS) that can be progressed to torsades de pointesand ventricular fibrillation. LQTS can be invoked by manygenetic mutations and various chemical modulators of car-diac ion channels. Although genetic mutations invokingLQTS are scattered among several cardiac ion channels,most small chemical molecules including blockbuster drugsinducing LQTS inhibit the human ether-á-go-go-relatedgene (hERG) channel (Kv11.1) that generates the rapidcomponent of delayed rectifier K+ current (IKr) [25]. The

S. W. Choi :K. S. Kim :D. H. Shin :H. Y. Yoo :Y. H. Zhang :S. J. Kim (*)Department of Physiology, Seoul National University Collegeof Medicine, 103 Daehakro, Chongro-gu,Seoul 110-799, Republic of Koreae-mail: sjoonkim@snu.ac.kr

H. Y. YooIschemic/Hypoxic Disease Institute, Seoul National UniversityCollege of Medicine, 103 Daehakro, Chongro-gu,Seoul 110-799, Republic of Korea

T. H. Ko : J. B. YoumDepartment of Physiology, College of Medicine, Cardiovascularand Metabolic Disease Center, Inje University, 633-165,Gaegeum-Dong, Busanjin-Gu,Busan 614-735, Republic of Korea

H. ChoeDepartment of Physiology, University of Ulsan, Collegeof Medicine, Seoul 138-736, Republic of Korea

W. K. Kim (*)Department of Internal Medicine, Graduate School of Medicine,Dongguk University, Seoul 410-773, Republic of Koreae-mail: wk2kim@naver.com

Pflugers Arch - Eur J Physiol (2013) 465:1121–1134DOI 10.1007/s00424-013-1239-7

inhibition mechanisms of the hERG channel mutations havebeen classified into four classes: disruption of hERG chan-nel synthesis (class 1), protein trafficking (class 2), gating(class 3), or permeation (class 4) [1]. This classification canbe generalized to include small molecule inhibition of hERGchannel. While most mutations in the hERG channel areinhibited via class 2 mechanism [1], so far, most of thehERG channel inhibitors act as class 4 mechanism, i.e.,blocking the pore. The molecular mechanism of class 4hERG inhibition is an open state-dependent trapping of drugmolecules into the inner cavity of the channel [22, 31]. Inthis model, aromatic residues on the S6 helix (Tyr652 andPhe656) and polar residues at the base of the pore helix(Thr623 and Ser624) line the inner cavity of the channel.The aromatic side chains of Tyr652 and Phe656 form π- πand/or hydrophobic interaction with the drug moleculescontaining phenol rings [7, 21].

Caffeic acid phenethyl ester (CAPE) and curcumin aremajor compounds of propolis and curcuminoids from Cur-cuma longa, respectively. These two phenolic compoundshave been reported to possess anti-inflammatory properties[19, 29] and are widely consumed in complementary medicineas well as dietary components [18]. Their anti-inflammatoryand potential anticancer properties in vitro have been attribut-ed to inhibiting a wide variety of signal-transducing proteinslike telomerase, cyclooxygenase, lipoxygenase, etc. [4, 11].They were also proposed as a heme oxygenase-1 inducer thatprotects against inflammatory and neurodegenerative condi-tions [26]. A molecular mechanism of these compounds waswell described, but the effect on ion channel with these reac-tive compounds was not much studied yet.

We previously reported common inhibitory effects ofCAPE and curcumin on ion channels in T cells [23, 27].Both compounds inhibited store-operated Ca2+ entry(SOCE) via Ca2+ release-activated Ca2+ (CRAC) channel.Also, while less potently, voltage-gated K+ (Kv) channeland intermediate conductance Ca2+-activated K+ (SK4)channel were inhibited. Based on known roles of SOCEand K+ channels in T cell activation [9], such inhibition ofmultiple ion channels was suggested as a mechanism forimmunomodulatory effects of both compounds. The molec-ular nature of CRAC has been recently identified as afunctional complex of transmembrane pore-forming unit(Orai) and ER protein (STIM) [8]. Recently, we could pindown the inhibitory mechanism of the CRAC inhibition byCAPE and curcumin. A cystein residue (C195) of Orai1 wasproposed as an electron acceptor (nucleophile) that interactswith CAPE and curcumin, both of which are electrophlilicagents containing an αβ -unsaturated keton group [28].

Previous studies demonstrated that oxidative stress con-ditions (e.g., H2O2) inhibit K

+ current through hERG chan-nels (IhERG) [2]. Recently, a free thiol group of cysteineresidues such as C723 of hERG has been suggested as the

effective target for the oxidizing agents [16]. The cysteine-dependent inhibition of hERG was similar with the oxida-tive inhibition of CRAC, where C195 of Orai1 was a criticalresidue mediating the oxidative stress: a replacement ofC195 with serine (C195S) largely attenuated the inhibitionby H2O2 [3]. Also, we found that the inhibition of CRAC byCAPE was totally abolished in C195S Orai1 [28].

On these backgrounds, we aimed to investigate theeffects of CAPE and curcumin on hERG channel focusingon voltage dependence and unique gating kinetics of hERG,slow activation, instantaneous inactivation, and slow deac-tivation. The slow activation of IhERG is largely overshad-owed by concomitant fast inactivation. Since recovery frominactivation occurs almost instantaneously, the initial peakamplitude of the tail current on repolarization (IhERG,tail) isrelatively large and plays an important role in repolarizationof cardiac action potential. Therefore, conventional pharma-cological hERG screening examines the effects on ampli-tudes of IhERG,tail.

Our pilot study demonstrated decrease in IhERG,tail byCAPE, which was similar with the effect of curcuminreported recently [14]. As for a hypothetical mechanism,an electrophilic interaction with a specific cysteine residue(C723) of hERG might underlie the inhibitory effects in-duced by CAPE and curcumin. Also, direct pore blockadevia interaction with aromatic amino acids (Tyr652 andPhe656) was also a plausible mechanism. Therefore, wealso investigate the effects of of CAPE and curcumin onvarious hERG mutants: C723S, Y652A, and F656A.

Materials and methods

Cell culture and hERG mutant preparation

HEK293 cells stably expressing hERG1a channel weremaintained on minimum essential medium powdered media(MEM), 9.5 g/L (Gibco, Carlsbad, CA, USA) and supple-mented with 10 % fetal bovine serum (Gibco), 2.2 g/Lsodium bicarbonate (Sigma-Aldrich, St. Louis, MO, USA),1 % sodium pyruvate (Gibco), 1 % nonessential amino acid(Gibco), and 100 μg/ml zeocin (Gibco) and kept at 37 °C inan atmosphere of 5 % CO2. For expression of mutant hERG,HEK293 cells were transfected with 1 μg of each DNAvectors (Y652A, F656A, and C723S mutants) usingFuGENE 6 kit (Roche, Penzberg, Germany). Mutant-expressing cells were selected by cotransfection of greenfluorocese protein (GFP)-expressing vector. Current record-ing was performed for 24–48 h after transfection. MutanthERG-expressing HEK293 cells were maintained withDMEM (Gibco) supplemented with 10 % fetal bovine se-rum. C723S mutation was a kind gift from Dr. Heinemann(Friedrich Schiller University, Germany).

1122 Pflugers Arch - Eur J Physiol (2013) 465:1121–1134

Electrophysiology

Conventional whole-cell voltage clamp and action potentialclamp (AP clamp) were performed at room temperature.Microglass patch pipettes (World Precision Instruments,Sarasota, FL, USA) were pulled with PP-830 puller (Nar-ishige, Tokyo, Japan) and pipette resistance was about 2.5MΩ when filled with intracellur solution. The current wasamplified with Axopatch 200B amplifier (Axon Instru-ments, Foster, CA, USA). Currents were digitized at10 kHz and low pass filtered at 1 kHz using Digidata1440B and pCLAMP software 10.1 (Axon Instruments).Analysis was performed with Clampfit 10.1 and Origin 6.0(Microcal, Northampton, MA, USA). Boltzmann function:

I ¼ IMax=1þ e V1=2�Vð Þ=k

was used to analyze the half-maximal activation potential(V1/2) and slope factor k. Time constant (τ) and amplitude Aof slow and fast components were obtained with doubleexponential function:

y ¼ A1e�t=t1 þ A2e

�t=t2

Detailed information of applied voltage protocols wasdecribed in figures and results.

Solutions and drugs

Cells were perfused with normal Tyrode’s solution containing145 mM NaCl, 3.6 mM KCl, 1 mM MgCl2, 10 mM HEPES,5 mM glucose, and 1.3 mM CaCl2, adjusted to pH7.4 withNaOH. Ninety-six millimolars of KCl-modified Tyrode’s so-lution was applied in hERG mutant experiments to overcomepoor expression of hERG channels. Pippette filling solutioncontainted 100 mMK-aspartate, 25 mM KCl, 5 mM NaCl,10 mM HEPES, 1 mM MgCl2, 4 mMMg-ATP, 10 mMBAPTA, adjusted to pH7.25 with KOH. A calculated liquidjunction potential was about 13.4 mV, and therefore, mem-brane voltage was adjusted by −10 mV. CAPE and curcumin(Sigma-Aldrich) were dissolved in DMSO as a stock solutionand diluted into external solution just before bath perfusionbecause of fast degradation of curcumin [32].

Modeling of hERG current

Modeling of hERG current is basically equivalent to that ofIKr [24], which plays a pivotal role in phase III repolariza-tion in native cardiac myocytes. IhERG is described by thefollowing equation :

IhERG ¼ �gfast � xr1;fast þ�gslow � xr1;slow� � � xr2 � V � EKð Þ

where �gfast and �gslow are fast and slow components ofmaximum conductance xr1,fast and xr1,slow are fast and slow

components of activation gating variable, xr2 is the inacti-vation gating variable, V is the membrane voltage, and EK isthe equilibrium potential for K+ (see “Appendix”). Thevoltage dependences of steady-state activation in variousconditions were determined from amplitudes of normalizedIhERG,tail (Fig. 2a for control and CAPE; Fig. 7a for curcu-min) and were used to get equations of best fit. Time con-stants of activation (fast and slow components) wereobtained by fitting double exponential function to timecourses of IhERG on depolarization (Figs. 2a and 7a) andwere used to get equations of best fit for voltage dependenceof activation time constants. Curves of steady-state inacti-vation were assumed to be the same in all conditions (con-trol, CAPE, and curcumin) and were fitted to inactivationdata in Fig. 3b. The voltage dependence of inactivation timeconstants was fitted to decaying parts of IhERG on returnpulses with single exponential function as shown in Fig. 5e.

The�gfast and�gslow were chosen to present similar density ofIhERG in human ventricular myocytes [15] when using thesame pulse protocol.

Model of human ventricular myocytes

Ten Tusscher’s model [30] was employed to simulate APsof human ventricular myocytes with replacement of IKr withour IhERG as described in the previous section. The modelincludes a basic calcium dynamics and most of the majorion currents, and is able to reproduce action potential resti-tution, which plays a critical role in development of reen-trant arrhythmias. The model also reproduces regionaldifferences in the shape of cardiac APs as categorized intoepicardial, endocardial, and M cell types. We chose theepicardial type as a representative platform to simulate theeffect of two polyphenols on APs of human ventricularmyocytes. All the parameter values except for those of IKrare the same as those in Ten Tusscher’s model. Numericalintegration of our model was carried out by a fourth-orderRunge–Kutta method in most cases such as calculation ofion concentration and membrane voltage. In order to calcu-late open probabilities of ion channels at each time step, analternative method was employed to prevent critical errorsarising from their very rapid kinetics, limiting efficiency ofcalculation as previously introduced by Youm et al. [34].The stimulus current in evoking APs in simulated humanventricular myocytes was 7 nA in amplitude and 0.6 ms induration. Stimulus was applied for 10 s at a rate of 1 Hz forthe simulation model to reach steady-state condition.

Data analysis

Data was analyzed with pCLAMP 10.1 and Origin 6.0programs. The data is presented as original recordings andbar graphs of the mean ± SEM (for n trials or cells). Paired

Pflugers Arch - Eur J Physiol (2013) 465:1121–1134 1123

or unpaired Student’s t test was used for the statisticalanalysis where appropriate. Statistical probability (p) isexpressed as *p<0.05, **p<0.01, and ***p<0.001.

Results

CAPE decreased IhERG,tail and shifted activation voltagedependence

From a holding potential of −90mV, hERG channels was fullyactivated with 4 s of depolarization to 30 mV, followed byrepolarization to −50 mV. The repolarization revealed slowlydecaying tail current with large initial peak (IhERG,tail, Fig. 1a).Peak amplitudes of IhERG,tail measured at 20–25 ms afterrepolarization was decreased by CAPE in a concentration-dependent manner (IC50 10.6±0.5 μM, n=5, Fig. 1b).

It was tested whether the inhibition of IhERG,tail was use-dependent requiring an open-state configuration of hERGfor drug binding. After confirming the stable amplitudes ofcontrol IhERG,tail, the amplitudes of the first and secondIhERG.tail after 10 μM CAPE application was compared tocontrol. Between the control and the first pulses, themembrane voltage was held at −90 mV for 3 min.Steady-state inhibition was observed from the first testpulse after the CAPE application, indicating use-independent mechanisms (Fig. 1c, d). The response of

hERG current during initial depolarization (IhERG,depol) at30 mV was also inhibitory (Fig. 1a). Interestingly, how-ever, effects of CAPE on the IhERG,depol at 0 mV were notinhibitory but facilitating and increased amplitude andaccelerated activation (Fig. 1c, arrow).

To investigate the complex voltage-dependent responses toCAPE, current–voltage relations of IhERG,depol and IhERG,tailwere obtained by applying depolarization voltages from −60to 30 mV with repolarization to −50 mV (Fig. 2a). IhERG,depoland IhERG,tail were normalized to the maximum control cur-rents at 0 mVand 30 mV, respectively, in each cell. Current–voltage relation (I/V curve) of the normalized IhERG,depolshowed typical bell shape, and 10 μM CAPE increased thepeak amplitudes of IhERG,depol at voltages below 0 mV whileslightly decreased from above 0 mV (Fig. 2b).

The inhibitory effect of CAPE on IhERG,tail was alsovoltage-dependent; prepulse above −10 mV was required toreveal the inhibition (Fig. 2c). Interestingly, IhERG,tail wasslightly but significantly increased by 10 μM CAPE at pre-pulse voltages below −10 mV (Fig. 2c). To investigate theeffect of CAPE on the voltage dependence of hERG activa-tion, peak amplitudes of IhERG,tail were normalized to theIhERG,tail at 30 mV in each cell with or without CAPE treat-ment (Fig. 2d). Then, the normalized amplitudes were fitted toBoltzmann equation; the half-activation voltage (V1/2,act) wasshifted from −17.5±0.2 to −26.5±0.1 mV (n=4) by applica-tion of 10 μMCAPE (Fig. 2d). The slope factor (k value) was

Fig. 1 Inhibitory effects ofCAPE on IhERG,tail. a After fulldepolarization (30 mV, 4 s) toactivate hERG (IhERG,depol), steprepolarization to −50 mV for 2 srevealed a large outward tailcurrent with slow deactivationkinetics (IhERG,tail). aApplication of CAPE (1, 10,and 30 μM) decreased bothIhERG,tail and IhERG,depol. bConcentration-dependentinhibition of IhERG,tail issummarized and fitted with Hillfunction (n=5). c, d To test use-dependent inhibitory effect ofCAPE, the test pulse (0-mVdepolarization followed by−50-mV repolarization) was notapplied for 180 s (holdingvoltage, −90 mV), while 10 μMof CAPE was perfused. Thefirst (circled digit one) andsecond (circled digit two) testpulses after the incubationperiod showed the same level ofinhibition

1124 Pflugers Arch - Eur J Physiol (2013) 465:1121–1134

not different (6.5±0.16 and 6.4±0.07, respectively; n=4). Theabove results indicate that the shift of voltage dependencecould explain the paradoxical increase of IhERG,depol andIhERG,tail by CAPE at relatively low depolarization pulses.

An alteration of inactivation voltage dependence mightalso contribute to the incresed amplitudes at less depolarizedvoltages. Hence, we investigated the inactivation voltagedependence of hERG channels by using transient hyper-polarizations (10 ms, −120 to 20 mV with 10-mV incre-ment) from depolarized holding voltage (30 mV, Fig. 3a top).Peak amplitudes of outward current upon returning from thetransient hyperpolarization were normalized to the maximumamplitude obtained in each cell and plotted against the pre-pulse voltages, which reflected the voltage dependence ofinactivation (Fig. 3b). The half inactivation voltages werenot changed by 10 μM CAPE (Fig. 3b, −44.7±1.25 and−45.7±2.21 mV, respectively; n=6).

Acceleration of deactivation and activation kineticsby CAPE

In the initial experiment, the decay of IhERG,tail appearedaccelerated (Fig. 1c). To analyze the effects of CAPE ondeactivation kinetics, duration of repolarization pulse periodwas doubled (4 s, Fig. 4a top), and IhERG,tail was fitted withdouble exponential function (Fig. 4a). Time constants ofboth slow and fast components were decreased by CAPE(Fig. 4b; n=4). Interestingly, the initial peak amplitudes (A1

and A2 in the equation) of the slow and fast componentsresponded differently to CAPE; only the slow componentpeak amplitude was reduced (Fig. 4c).

The effect of CAPE on activation kinetics was analyzedwith a stepwise increase of the duration of depolarizationfrom 3 ms to 1,000 ms (Fig. 5a top). The envelope of peakIhERG,tail reflected the rate of activation at 0 mV (Fig. 5a).

Fig. 2 Effects of CAPE on the current–voltage relations of IhERG,depoland IhERG,tail. a Multiple levels of depolarizing pulses (from −70 to30 mV with 10-mV interval, 4 s) were applied followed by a commonrepolarization to −50 mV (2 s). Ten micrograms of CAPE increased themaximum amplitude of IhERG,depol while decreased IhERG,tail with accel-erations in both activation and deactivation. b Amplitudes of IhERG,depolwere normalized to the one obtained at 0 mV in control, and averagedvalues (norm.IhERG,depol) were plotted against the test voltages. c

Amplitudes of IhERG,tail were normalized to the maximum one (usuallyafter 30-mV depolarization) in each control, and averaged values(norm.IhERG,tail) were plotted against the test voltages. d Amplitudesof IhERG,tail were normalized to the maximum ones in control andCAPE-treated condition, respectively, and averaged values (norm.Con-ductance) were plotted against the test voltages and fitted with Boltz-mann function (see “Materials and methods”). b to d filled sqauresreflect the responses to 10 μM CAPE (n=4)

Pflugers Arch - Eur J Physiol (2013) 465:1121–1134 1125

Control and CAPE peak tail currents (Fig. 5a, open andfilled squares) were normalized by their maximum levelsand fitted with single exponential function (Fig. 5b). Therate of activation at 0 mV was accelerated by CAPE (controlτ, 531.9 ms; CAPE τ, 242.7 ms; n=6). Because the apparentacceleration of activation could be partly due to the shift ofV1/2(act), we also compared the activation kinetics at a supra-maximal depolarization level. The acceleration of activationby CAPE was also observed when the depolarizing pulsewas raised to 30 mV, which was less significant than at0 mV (Fig. 5c).

The voltage-dependent inactivation kinetics (time con-stant) was also analyzed by using a protocol with transienthyperpolarization from depolarized holding voltage fol-lowed by variable return voltages (Fig. 5d). The decayingoutward current during the return voltage was fitted with

single exponential function. The obtained time constantswere plotted against return voltages, which was not signif-icantly changed by CAPE (Fig. 5e, n=6).

Effects of CAPE on point mutated hERG channels

It was tested whether pore blocker-resistant mutants of hERG(Y652A and F656A) are less sensitive to CAPE thanwild type(WT) hERG. To minimize the influence of accerated deacti-vation kinetics, effects of CAPE on inward IhERG,tail werecompared at −110 mV with 96 mM [K+] in bath solution(Fig. 6a). The response of F656Awas not different from WT(Fig. 6b). Suprisingly, Y652Awas not resistant to CAPE butmore sensitive than WT (inhibition by 90 %, Fig. 6b).

Next, we investigated effects of CAPE on C723S hERG.Interestingly, C723S already showed faster kinetics and

Fig. 3 No effect of CAPE onvoltage dependence of hERGinactivation. a Ten microsecondsof transient hyperpolarizationspulses (ranging between −120and 20 mV with 10-mV interval)were applied from a full-activating holding voltage(30mV) protocol. bThe outwardtail currents on returning to30 mV were normalized to themaximum amplitudes in control(open squares) and 10 μMCAPE-treated condition (filledsquares), respectively, andplotted against the prepulsevoltages (n=6)

Fig. 4 Acceleration of thehERG deactivation by CAPE. aIhERG,tail on return to −50 mVwas fitted with doubleexponential function (see“Materials and methods”). bBoth slow and fast timeconstants (τ) were decreased by10 μMCAPE (τslow, from 1,126to 578 ms; τfast, from 209 to112 ms, n=4). c Among theinitial amplitudes of each fittedcomponent (A1 and A2 ofdouble exponential equation),the slow component (A1) wasdecreased in the presence ofCAPE (n=4)

1126 Pflugers Arch - Eur J Physiol (2013) 465:1121–1134

negatively shifed activation voltage dependence whencompared with WT hERG (Fig. 7). The V1/2(act) ofC723S was −49.2±0.4 mV (Fig. 7c, n=9). Application of10 μM CAPE did not induce any further shift of V1/2(act) inC723S (Fig. 7c, C723S + CAPE; filled inverted trian-gle, −46.3±0.8 mV; n=5). However, the amplitudes ofC723S IhERG,depol were further decreased by 10 μM CAPE atvoltage more negative than −10 mV (Fig. 7b).

Because voltage-dependent activation was mixed at−50 mV owing to the shift of V1/2(act), the IhERG,tail ofC723S was analyzed at −70 mV (Fig. 7a). When the time-dependent deactivation kinetics at −70 mV (slow and fasttime constants of double exponential fitting) were comparedbetween WT and C723S, the fast τ of C723S was smallerthan WT (Fig. 7d). However, the application of CAPE

significantly decreased both time constants as well as theamplitudes of C723S IhERG,tail (Fig. 7a, d, and e).

Effects of curcumin on hERG

Finally, we tested the effects of curcumin on hERG activity.The application of 10 μM curcumin generally suppressedthe amplitudes of IhERG,depol and IhERG,tail. In addition, sim-ilar to the CAPE effects, V1/2(act) was also negatively shiftedby curcumin, and the deactivation became faster (Fig. 8). Toinvestigate the effects of curcumin on Y652A and F656A,the pore blocker site mutants, inward IhERG,tail analysis wasperformed. Different from the responses to CAPE, the poremutants of hERG (Y652A and F656A) became less sensi-tive to curcumin while not completely resistant (Fig. 9).

Fig. 5 Acceleration ofactivation, but not inactivation,of hERG by CAPE. a As shownin the protocol, the duration ofprepulse depolarization wasincreased on each repeatedpulse (from 3-ms to 1-sinterval). The results of 10pulses are ovelayed in controland 10 μM CAPE-treatedconditions. b The peakamplitudes of IhERG,tail wereplotted against prepulseduration and fitted with singleexponential functions in controland CAPE (time constant; 532and 243 ms, respectively, n=6).c Similar results were obtainedwith 30 mV of prepulse insteadof 0 mV (time constantdecreased from 100 to 63 ms byCAPE, n=6). d A commontransient hyperpolarization(−100 mV, 2 ms) was appliedbefore returning to variablelevels of depolarization (from−20 to 60 mV, 10-mV interval).e The decaying parts on returnpulses were fitted with singleexponential function, and thetime constants (τinact) wereplotted against the returnvoltages (n=6, filled sqauresreflect 10 μM CAPE)

Pflugers Arch - Eur J Physiol (2013) 465:1121–1134 1127

Effects of CAPE on IhERG induced by action potentialvoltage clamp (AP clamp)

Although the decrease of IhERG,tail suggested that CAPEmightinduce AP prolongation, the effects of CAPE on the activationkinetics did not allow such a simple anticipation. Therefore,we carried out AP clamp, in which an AP-like waveform isused as a command potential instead of the conventional steppulse protocol [33]. Initially, we used an AP waveformrecorded from guinea pig ventricular myocyte [20]. The shapeof IhERG was slightly increased during the intial plateau phase(phase II) while decreased during the repolarization period(phase III) by 10 μM CAPE (Fig. 10a). However, with APwaveform from a physiomic model of human ventricular cell(hAP) that has less depolarized plateau with longer duration[10], IhERG was increased by CAPE during both phases II andIII (Fig. 10b). Interestingly, when the amplitude of hAP wasdigitally augmented by 20 % (hAP×1.2, IhERG was decreasedby CAPE during phase III, similar to the results observed withgAP (Fig. 10c). Different from the effects of CAPE, IhERGinduced by the three different AP clamp protocols were com-monly suppressed by 10 μM curucmin (Fig. 10d–f).

Simulation of IhERG and action potential

Validity of the hERG model was examined by reproducing theexperimental voltage clamp recordings in Fig. 2a (control andCAPE) and Fig. 7a (curcumin). As shown in Fig. 11a, simulat-ed currents in control condition closely match experimentally

obtained currents (Fig. 2a; control) in time courses and voltagedependence of amplitudes. In order to reproduce the effects ofCAPE on IhERG, activation (activation gating) time constantsand steady-state activation under perfusion of CAPE wereobtained and formulated into equations as shown inthe “Appendix”. Voltage dependence of steady-state inacti-vation and inactivation time constants were not considered tobe changed by CAPE as evidenced by Figs. 3b and 5e. Asshown in Fig. 11b, modifying equations governing activationprocess faithfully reproduced time course, voltage depen-dence, and relative amplitudes of IhERG recorded under perfu-sion of 10 μM CAPE (Fig. 2a; CAPE). Simulated IhERG,depolwas increased while simulated IhERG,tail was decreased.

The same approach was adopted for curcumin. Modifiedequations for activation process under curcumin perfusionalso faithfully reproduced time courses and voltage depen-dence of currents; however, amplitudes of IhERG,depol werenot much smaller than control but even bigger (data notshown). Although curcumin induces time- and voltage-dependent additional suppression of IhERG,depol above10 mV of depolarization (Fig. 8a; bottom), the adoptedequations of IhERG was not able to fully reflect the time-dependent inhibitory effect of curcumin. However, as sug-gested in Fig. 9, a direct inhibition of hERG by curcuminwas incorporated into the equation of IhERG (see Table 1),which resulted in a similar degree of block by curcumin inexperimentally obtained recordings (Fig. 11c).

Effects of CAPE and curcumin on simulated AP wave-form were examined by incorporating the hERG model into

Fig. 6 Effects of CAPE onpore blocker-resistant mutatedhERG. a Experiments werecarried out in modified bathsolution (96 mMK+) and IhERG,tail was recorded at −110 mV. bIhERG,tail for Y652A was moresensitive to 10 μM CAPE thanWT (61.5±4.5 and 12.4±2.1 %of current remained in WT andY652A, respectively; n=6).The response of F656Awas notdifferent from WT (60.3±5.1 %, n=6)

1128 Pflugers Arch - Eur J Physiol (2013) 465:1121–1134

a preexisting mathematical model of human ventricularmyocytes [30]. The waveform of simulated AP in control(Fig. 11d; control) was very similar to experimentallyobtained AP [6]. Resting potential was about −84.1 mV,overshoot was 36.3 mV, and APD90 was 277 ms. Applyingequations governing activation process under perfusion ofCAPE resulted in a decrease of APD (APD90=254 ms)without affecting resting potential and overshoot (Fig. 11d;CAPE). Applying equations governing activation processunder perfusion of curcumin resulted in an increase ofAPD (APD90=287 ms) without affecting resting potentialand overshoot (Fig. 11d; curcumin). Concomitant timecourses under the condition of control, CAPE, and curcuminduring AP were compared in Fig. 11e. CAPE was found toincrease IhERG during both phases II and III repolarization,while curcumin reduced it from late phase II.

Discussion

In this study, the assay using single level of voltage–pulseprotocol initially suggested that CAPE might be a direct

inhibitor of hERG, especially at higher concentrations(>10 μM, Fig. 1a). In addition, however, the investigationwith various voltage–pulse protocols revealed that CAPEinduced leftward shift of activation voltage dependence andfastened the kinetics of both activation and deactivation(Figs. 2, 4 and 5). Owing to the above complex effects,the late outward current during repolarization period ofhuman AP was facilitated rather than suppressed by micro-molar range of CAPE (Figs. 10b and 11e). As a whole, thenegative shift of V1/2(act) and fastened activation by CAPEwould induce facilitation of IhERG when the plateau phase ofAP is submaximal level for hERG activation, as in thehuman AP. This result implies that a cardiotoxicity screen-ing of chemical compounds simply focusing on the inhibi-tion of IhERG,tail could have missed the actual effects in vivo.We suggest that both IC50 of IhERG,tail and the late outwardIhERG of AP clamp are required for drug screening to preventthe unnecessary dropout of potential novel drugs.

More than a single mechanism seem to underlie theIhERG,tail inhibition by CAPE: acceleration of deactivationand a putative direct inhibition especially at higher concen-trations. However, the latter could not be explained by a

Fig. 7 Electrophysiologicalcharacteristics of C723S hERGand the effects of CAPE. a Theactivation voltage dependence ofC723S was obtained by variouslevels of prepulses followed bycommon repolarization(−70 mV, see inset). b, c Plot ofthe normalized IhERG,depol andIhERG,tail of C723S (openinverted triangles) together withthe WT (open sqaures) and theresposnes of C723S to CAPE(filled inverted triangles). Notethe negatively shifted V1/2(act) ofC723S and no further shift byCAPE (n=5). d Analysis ofC723S deactivation kinetics incomparison with WT. The IhERG,tail on repolarization to −70 mVwas fitted with doubleexponential function; τfast ofC723S was smaller than WT (87vs. 52ms, n=5). Tenmicrogramsof CAPE decreased both τslow(from 297 to 206 ms) and τfast ofC723 (from 52 to 25 ms, n=5). eAmong the inititial amplitudes offitted function, the peak of slowcomponent was more sensitivelyinhibited by CAPE in C723S,similar with WT (see Fig. 4c)

Pflugers Arch - Eur J Physiol (2013) 465:1121–1134 1129

“docking model.” Firstly, there was no use dependence inthe effects of CAPE on IhERG (Fig. 1c, d). Secondly, theinhibition was not prevented by point mutations of specificaromatic residues (Y652A and F656A) that interact with avariety of pore blockers for hERG (Fig. 6). Furthermore, thesensitivity of Y652A to CAPE was higher than WT, themeaning of which is still unclear.

Curcumin had basically similar effects on IhERG kinetics:leftward shift of activation curve and acceleration of deac-tivation (Fig. 8). However, not only IhERG,tail but also IhERG,depol was suppressed by curcumin. In addition, the inhibitionof IhERG,tail by curcumin was more potent than that of CAPE(Figs. 6b and 9b). A direct inhibition of hERG by curcuminvia interaction with the pore blocker binding site was alsosuggested from the reduced inhibition in Y652A and F656Amutants (Fig. 9). Such additive inhibitory effects mightcontribute to the general inhibition of the IhERG induced byAP clamp conditions, which was distinguishabe from CAPEeffects (Figs. 10 and 11e).

Both CAPE and curcumin have electrophilic αβ -unsaturated keton groups that can react with nucleophilic

thiol groups, potentially forming Michael adducts with cys-tein residues. As shown in recent studies, a point replace-ment of cystein residue (C195S) of Orai1 abolished theinhibitory effect of CAPE and H2O2 on ICRAC [3, 28].Interestingly, a previous study demonstrated that H2O2 nega-tively shifted V1/2(act) of hERG and accelerated the decay ofIhERG,tail [2, 16], similar with the effects of CAPE and curcu-min. Among the many cystein residues in hERG, C723 hasbeen recently suggested as a potential target of oxidative stressinhibition of hERG, where inhibition by 2-sulfonatoethylmethanethiosulfonate was partly alleviated in C723S pointmutant of hERG [16]. However, in our present study, theIhERG,tail inhibition and acceleration of deactivation by CAPEwas still observed in C723S (Fig. 7), suggesting that C723could not be the single decisive site interacting with CAPE.

Nontheless, the current mediated by C723S alreadyshowed highly intriguing features similar to the CAPE effects:leftward shift of activation curve and acceleration of deacti-vation. While CAPE further accelerated the deactivation ki-netics of C723S, the half activation voltage (V1/2,act) of C723Swas not shifted by CAPE (Fig. 8c). Such results suggested that

Fig. 8 Inhibitory effects ofcurcumin on IhERG. a–c Thesame type of test pulses asshown in Fig. 2 were applied(a) for the analysis ofnormalized IhERG,depol (b) andnormalized conductance usingIhERG,tail (c). Ten micrograms ofcurcumin also induced leftwardshift of activation curve whiledecreased both IhERG,depol andIhERG,tail (n=5). d, e Effects ofcurcumin on deactivationkinetics of hERG. The pulseand analysis protocol same aswhen in Fig. 4 were applied,and 5 μM of curcumin wasapplied. Curcumin decreasedτfast deactivation kinetics (from291 to 167 ms, n=5, d). Incontrast to CAPE, the initialpeaks of both slow and fastcomponents were decreased bycurcumin (e)

1130 Pflugers Arch - Eur J Physiol (2013) 465:1121–1134

C723 is a critical residue for the unique property of hERG,slow deactivation, and subsequent large tail current. Althoughthe putative electrophilic reactionwith C723might explain thechanges in voltage-dependent gating processes, at least partly,there seems to be definitely more sites interacting with CAPE.The limit of this study is that we have not yet fully elucidatedthe putative mulitple target sites for the inhibitory actions ofCAPE and curcumin.

A molecular basis for the slow kinetics of activationand deactivation has been extensively investigated, andmany critical amino acid residues in hERG have been

identified [31]. However, C723 has not been suggestedas a candidate for such site yet and not directly locatedin any of the kinetically important domains of hERG.Nevertheless, C723 locates relatively close to the proxi-mal region of cyclic nucleotide binding domain (cNBD)in intracellular C-terminal sequence. Because an interac-tion between cNBD in C-terminal and PAS domain in N-terminal is suggested to be critical for slow deactivationkinetics [13], the point mutation or chemical modifica-tion of C723 might somehow alter the interaction be-tween PAS and cNBD.

Fig. 9 Effects of curcumin onY652A and F656A hERG. aWith 96 mM [K+]ext, inwardIhERG,tail was recorded at−110 mV. b Only 14.9 %of WT IhERG,tail remained(Icurcumin/Icontrol) under thetreatment with 10 μMcurcumin, and significantlymore IhERG,tail remained inY652A and F656A (25.3 and41.5 %, respectively, n=5)

Fig. 10 Effects of CAPE onIhERG elicited by actionpotentials of guinea pig (gAP, a,d) and human (hAP, b, e). Theapplied pulses are indicated bydashed lines, and the elicitedcontrol currents (thickthreadlines) and responses to10 μM CAPE or curcumin (thinthreadlines) are overlayed. Theholding voltage was commonly−90 mV, and the horizontaldotted lines indicate 0 mV. c, fIhERG in response to digitallyaugmented hAP by 20 %(hAP×1.2)

Pflugers Arch - Eur J Physiol (2013) 465:1121–1134 1131

The hyperpolarizing shift of activation V1/2 in C723S iseven more perplexing to explain. The voltage-dependentrearrangements of the voltage-sensing S4 domain with re-peated positively charged amino acids are transmitted to theactivation gate via a physical interaction of the S4S5 linkerswith the S6 helices [31]. Although C723 is relatively closeto the cytoplasmic terminals of the S6, it is less likely thatthe S4 domain itself is affected by a point mutation of Cys incytoplasmic C-terminal. The specific sites of interaction

between C723 and voltage-dependent gating domains inhERG remain to be identified.

As mentioned above, CAPE and curcumin are both elec-trophilic agents that can bind with thiol groups of Cys, andthere are multiple Cys residues in hERG [16]. Based on theadditional inhibition of IhERG,tail and fastened deactivation inC723S, it is supposed that the putative interaction of thecompounds with other Cys residues might also affect thedeactivation kinetics and voltage-dependent gatings.

Fig. 11 Simulated hERGcurrent in voltage clamp andAPsunder current clamp. a Simulatedcurrent traces obtained by thesame voltage clamp protocolused in Fig. 2a under the controlcondition. b Simulated currenttraces with incorporation ofmodified equations governingactivation process under theperfusion of 10 μM CAPE. cSimulated current traces withincorporation of modifiedequations for 10μMcurcumin. dSimulated AP waveforms undercontrol, CAPE, and curcumin.Current injection in the modelwas 7 nA in amplitude and0.6 ms in duration. eConcomitant changes in IhERGduring APs in d

Table 1 Model parameters

aFor time-dependent variables,values after running of simula-tion for 10 s with 1-Hz stimula-tion were shown (initialcondition)

Symbols Definition Valuea Unit

xr1,fast Activation gating variable for fast component 1.37·10−5 –�xr1;fast Steady-state value of xr1,fast –

τr1,fast Activation time constant for fast component ms�gfast Maximum conductance for fast component 2.910 1.309 (curcumin) nS nS

xr1,slow Activation gating variable for slow component 7.10·10−2 –�xr1;slow Steady-state value of xr1,slow –

τr1,slow Activation time constant for slow component ms�gslow Maximum conductance for slow component 4.074 1.833 (curcumin) nS nS

xr2 Inactivation gating variable 8.72·10−1 –�xr2 Steady-state value of xr2 –

τr2 Inactivation time constant ms

R Gas constant 8.3143 JK−1mol−1

T Temperature 310 K

F Faraday constant 96.4867 Cmmol−1

V Membrane voltage −84.1 mV

EK Equilibrium (Nernst) potential for K+ −84.8 mV

Ko Extracellular K+ concentration 5.4 mM

Ki Intracellular K+ concentration 129.3 mM

IhERG hERG current 0.17 pA

1132 Pflugers Arch - Eur J Physiol (2013) 465:1121–1134

In terms of IhERG,tail, CAPE inhibits the hERG channel bythe class 3 mechanism, i.e., modulation of gating, morespecifically the activation gate [5]. Most of the small mol-ecule inhibitors block the pore of the hERG channel by class4 mechanism. While there are reports describing the gatingmodification effects of pharmacological agents, the fasten-ing of deactivation by CAPE was most prominent comparedwith those reports [12, 17]. We also suggest that the sulfur-containing C723 residue plays a role for activation voltagesensitivity and deactivation kinetics. Our present study indi-cates that the screening of drug compounds, especially theherbal origins such as CAPE and curcumin, should considerthe potential electrophilic binding with Cys residues inhERG to fully understand the influence to cardiac electricalactivity. We suggest that the application of human AP clampprotocol for each test and comparison with the conventionalsqure pulse protocol would be a reasonable way for drugscreening of cardiotoxicity in terms of hERG inhibition.

Acknowledgments This study was supported by the Basic ScienceResearch Program through the National Research Foundation of Korea(NRF) funded by the Ministry of Education, Science, and Technology(NRF 2011–0017370 and NRF 2012–0000809).

Appendix

Current amplitude

IhERG ¼ �gfast � xr1;fast þ�gslow � xr1;slow� � � xr2 � V � EKð Þ

ActivationFor control:

�xr1;fast ¼ 1

1þ e� Vþ11:37ð Þ=6:5

tr1;fast¼ 1

1:00188 � 10�4 � e�V=14:10186 þ 5:89793 � 10�4 � eV=9:82184

�xr1;slow ¼ �xr1;fast

tr1;slow¼ 1

2:30792 � 10�5 � e�V=17:63893 þ 7:47645 � 10�4 � eV=14:11235

For 10 μM CAPE:

�xr1;fast ¼ 1

1þ e� Vþ21:37ð Þ=6:5

tr1;fast ¼ 1

1:12741 � 10�5 � e�V=7:37214 þ 0:00349 � eV=14:22556

�xr1;slow ¼�xr1;fast

tr1;slow ¼ 1

1:64477 � 10�7 � e�V=5:22954 þ 0:00158 � eV=15:80355

For 10 μM curcumin:

�xr1;fast ¼ 1

1þ e� Vþ16:00ð Þ=6:17

tr1;fast ¼1

2:59718 � 10�5 � e�V=9:76995 þ 0:0055 � eV=20:29395

�xr1;slow ¼�xr1;fast

tr1;slow ¼ 1

1:46618 � 10�4 � e�V=93:70289 þ 0:003 � eV=15:80355

Inactivation

�xr2 ¼ 1

1þ e Vþ42:84ð Þ=21:57

tr2 ¼ 1

4:22157 � 10�4 � e� V�115:79877ð Þ=30:0 þ 0:01681 � e Vþ50:03784ð Þ=32:4696

First-order differential equations

dxr1;fastdt

¼�xr1;fast � xr1;fast

tr1;fast

dxr1;slowdt

¼�xr1;slow � xr1;slow

tr1;slow

dxr2dt

¼ xr2 � xr2tr2

Equilibrium potential of K+

EK ¼ R � TF

� log Ko

Ki

References

1. Anderson CL, Delisle BP, Anson BD, Kilby JA, Will ML, TesterDJ, Gong Q, Zhou Z, Ackerman MJ, January CT (2006) MostLQT2 mutations reduce Kv11.1 (hERG) current by a class 2(trafficking-deficient) mechanism. Circulation 113:365–373

2. Berube J, Caouette D, Daleau P (2001) Hydrogen peroxide modi-fies the kinetics of HERG channel expressed in a mammalian cellline. J Pharmacol Exp Ther 297:96–102

Pflugers Arch - Eur J Physiol (2013) 465:1121–1134 1133

3. Bogeski I, Kummerow C, Al-Ansary D, Schwarz EC, Koehler R,Kozai D, Takahashi N, Peinelt C, Griesemer D, Bozem M, Mori Y,Hoth M, Niemeyer BA (2010) Differential redox regulation ofORAI ion channels: a mechanism to tune cellular calcium signal-ing. Sci Signal 3:ra24

4. D’Archivio M, Filesi C, Di Benedetto R, Gargiulo R, GiovanniniC, Masella R (2007) Polyphenols, dietary sources and bioavail-ability. Ann Ist Super Sanita 43:348–361

5. Delisle BP, Anson BD, Rajamani S, January CT (2004) Biology ofcardiac arrhythmias: ion channel protein trafficking. Circ Res94:1418–1428

6. Drouin E, Charpentier F, Gauthier C, Laurent K, Le Marec H(1995) Electrophysiologic characteristics of cells spanning the leftventricular wall of human heart: evidence for presence of M cells. JAm Coll Cardiol 26:185–192

7. Fernandez D, Ghanta A, Kauffman GW, Sanguinetti MC (2004)Physicochemical features of the HERG channel drug binding site.J Biol Chem 279:10120–10127

8. Feske S, Gwack Y, Prakriya M, Srikanth S, Puppel SH, Tanasa B,Hogan PG, Lewis RS, Daly M, Rao A (2006) A mutation in Orai1causes immune deficiency by abrogating CRAC channel function.Nature 441:179–185

9. Feske S (2007) Calcium signalling in lymphocyte activation anddisease. Nat Rev Immunol 7:690–702

10. Fink M, Noble D, Virag L, Varro A, Giles WR (2008) Contribu-tions of HERG K + current to repolarization of the human ventric-ular action potential. Prog Biophys Mol Biol 96:357–376

11. Fraga CG (2007) Plant polyphenols: how to translate their in vitroantioxidant actions to in vivo conditions. IUBMB Life 59:308–315

12. Frolov RV, Ignatova II, Singh S (2011) Inhibition of HERG potas-sium channels by celecoxib and its mechanism. PLoS One 6:e26344

13. Gustina AS, Trudeau MC (2011) hERG potassium channel gatingis mediated by N- and C-terminal region interactions. J GenPhysiol 137:315–325

14. Hu CW, Sheng Y, Zhang Q, Liu HB, Xie X, Ma WC, Huo R, DongDL (2012) Curcumin inhibits hERG potassium channels in vitro.Toxicol Lett 208:192–196

15. Iost N, Virag L, Opincariu M, Szecsi J, Varro A, Papp JG (1998)Delayed rectifier potassium current in undiseased human ventric-ular myocytes. Cardiovasc Res 40:508–515

16. Kolbe K, Schonherr R, Gessner G, Sahoo N, Hoshi T, HeinemannSH (2010) Cysteine 723 in the C-linker segment confers oxidativeinhibition of hERG1 potassium channels. J Physiol 588:2999–3009

17. Li J, Correa AM (2002) Kinetic modulation of HERG potassiumchannels by the volatile anesthetic halothane. Anesthesiology97:921–930

18. Mainardi T, Kapoor S, Bielory L (2009) Complementary andalternative medicine: herbs, phytochemicals and vitamins and theirimmunologic effects. J Allergy Clin Immunol 123:283–294

19. Michaluart P, Masferrer JL, Carothers AM, Subbaramaiah K,Zweifel BS, Koboldt C, Mestre JR, Grunberger D, Sacks PG,

Tanabe T, Dannenberg AJ (1999) Inhibitory effects of caffeic acidphenethyl ester on the activity and expression of cyclooxygenase-2in human oral epithelial cells and in a rat model of inflammation.Cancer Res 59:2347–2352

20. Milnes JT, Witchel HJ, Leaney JL, Leishman DJ, Hancox JC(2010) Investigating dynamic protocol-dependence of hERG po-tassium channel inhibition at 37 degrees C: cisapride versus dofe-tilide. J Pharmacol Toxicol Methods 61:178–191

21. Mitcheson JS, Chen J, Lin M, Culberson C, Sanguinetti MC(2000) A structural basis for drug-induced long QT syndrome.Proc Natl Acad Sci USA 97:12329–12333

22. Mitcheson JS, Chen J, Sanguinetti MC (2000) Trapping of amethanesulfonanilide by closure of the HERG potassium channelactivation gate. J Gen Physiol 115:229–240

23. Nam JH, Shin DH, Zheng H, Kang JS, Kim WK, Kim SJ (2009)Inhibition of store-operated Ca2+ entry channels and K+channelsby caffeic acid phenethylester in T lymphocytes. Eur J Pharmacol612:153–160

24. Ono K, Ito H (1995) Role of rapidly activating delayed rectifier K+ current in sinoatrial node pacemaker activity. Am J Physiol 269:H453–H462

25. Roden DM, Viswanathan PC (2005) Genetics of acquired long QTsyndrome. J Clin Invest 115:2025–2032

26. Scapagnini G, Foresti R, Calabrese V, Giuffrida Stella AM, GreenCJ, Motterlini R (2002) Caffeic acid phenethyl ester and curcumin:a novel class of heme oxygenase-1 inducers. Mol Pharmacol61:554–561

27. Shin DH, Seo EY, Pang B, Nam JH, Kim HS, Kim WK, Kim SJ(2011) Inhibition of Ca2+ −release-activated Ca2+ channel (CRAC)and K+ channels by curcumin in Jurkat-T cells. J Pharmacol Sci115:144–154

28. Shin DH, Nam JH, Lee ES, Zhang Y, Kim SJ (2012) Inhibition ofCa2+ release-activated Ca2+ channel (CRAC) by curcumin andcaffeic acid phenethyl ester (CAPE) via electrophilic addition toa cysteine residue of Orai1. Biochem Biophys Res Commun

29. Strimpakos AS, Sharma RA (2008) Curcumin: preventive andtherapeutic properties in laboratory studies and clinical trials. Anti-oxid Redox Signal 10:511–545

30. ten Tusscher KH, Noble D, Noble PJ, Panfilov AV (2004) A modelfor human ventricular tissue. Am J Physiol Heart Circ Physiol 286:H1573–H1589

31. Vandenberg JI, Perry MD, Perrin MJ, Mann SA, Ke Y, Hill AP(2012) hERG K(+) channels: structure, function, and clinical sig-nificance. Physiol Rev 92:1393–1478

32. Wang YJ, Pan MH, Cheng AL, Lin LI, Ho YS, Hsieh CY, Lin JK(1997) Stability of curcumin in buffer solutions and characteriza-tion of its degradation products. J Pharm Biomed Anal 15:1867–1876

33. Wilders R (2006) Dynamic clamp: a powerful tool in cardiacelectrophysiology. J Physiol 576:349–359

34. Youm JB, Jo SH, Leem CH, Ho WK, Earm YE (2004) Role ofstretch-activated channels in stretch-induced changes of electricalactivity in rat atrial myocytes. Kor J Physiol Pharmacol 8:33–41

1134 Pflugers Arch - Eur J Physiol (2013) 465:1121–1134