Allitridi Inhibits Multiple Cardiac Potassium Channels Expressed in HEK 293 Cells

Allitridi (diallyl trisulfide) is an active compound (volatile oil) from garlic. The previous studies reported that allitridi had anti-arrhythmic effect. The potential ionic mechanisms are, however, not understood. The present study was designed to determine the effects of allitridi on cardiac potassium channels expressed in HEK 293 cells using a whole-cell patch voltage-clamp technique and mutagenesis. It was found that allitridi inhibited hKv4.3 channels (IC50 = 11.4 µM) by binding to the open channel, shifting availability potential to hyperpolarization, and accelerating closed-state inactivation of the channel. The hKv4.3 mutants T366A, T367A, V392A, and I395A showed a reduced response to allitridi with IC50s of 35.5 µM, 44.7 µM, 23.7 µM, and 42.4 µM. In addition, allitridi decreased hKv1.5, hERG, hKCNQ1/hKCNE1 channels stably expressed in HEK 293 cells with IC50s of 40.2 µM, 19.6 µM and 17.7 µM. However, it slightly inhibited hKir2.1 current (100 µM, inhibited by 9.8% at −120 mV). Our results demonstrate for the first time that allitridi preferably blocks hKv4.3 current by binding to the open channel at T366 and T367 of P-loop helix, and at V392 and I395 of S6 domain. It has a weak inhibition of hKv1.5, hERG, and hKCNQ1/hKCNE1 currents. These effects may account for its anti-arrhythmic effect observed in experimental animal models.

A recent report demonstrated that allitridi selectively inhibited the transient outward potassium current I to and had no significant effect on the ultra-rapidly delayed rectifier potassium current I Kur and L-type calcium current (I Ca.L ) in human atrial myocytes [12]. However, the anti-arrhythmic effect and the prolongation of cardiac action potential duration and effective refractory period reported previously with allitridi and/or garlic constituents [11,13,14] can not fully interpreted by the inhibition of cardiac I to , because the 4-aminopyridine-sensitive I to was not expressed in cardiac myocytes in some species (e.g. guinea pigs and pigs) [15,16]. On the other hand, it is unknown the molecular determinants of allitridi for inhibiting cardiac I to . The present study was therefore designed to determine the molecular determinants of allitridi for blocking Kv4.3 channels (coding human cardiac I to ) [17], and to investigate whether allitridi would inhibit other cardiac potassium channels stably expressed in HEK 293 cells, including hERG (coding human cardiac I Kr , rapidlydelayed rectifier potassium current) [18], hKCNQ1/hKCNE1 (coding human cardiac I Ks , slowly-delayed rectifier potassium current) [19], hKv1.5 (coding human cardiac I Kur ) [20], and hKir2.1 channels (coding human cardiac I K1 , inward rectifier potassium current) [21] stably expressed in HEK 293 cells using mutagenesis and whole-cell patch voltage-clamp techniques. Our results demonstrated that allitridi preferably blocked hKv4.3 channels by interaction with the sites of P-loop helix and S6 domain of the channel, and it also inhibited hKv1.5 channels, hERG channels, and hKCNQ1/hKCNE1 channels expressed in HEK 293 cells with a relatively weak effect.

Electrophysiology
Cells on a coverslip were transferred to an open cell chamber (0.5 ml) mounted on the stage of an inverted microscope and superfused with Tyrode solution at ,2 ml/min. The whole cell patch-clamp technique was used as described previously [22][23][24][25][26][27]. The whole-cell membrane currents were measured using an EPC-10 amplifier and Pulse software (Heka Elektronik, Lambrecht, Germany). Borosilicate glass electrodes (1.2-mm OD) were pulled with a Brown/Flaming puller (model P-97, Sutter Instrument, Nato, CA) and had resistances of 2-3 MV when filled with the pipette solution. A 3-M KCl agar bridge was used as the reference electrode. The tip potential was zeroed before the patch pipette contacted the cell. After the giga-Ohm seal was obtained, the cell membrane was ruptured by applying gentle pressure to establish a whole-cell configuration. Series resistance (Rs) was 4-6 MV and was compensated by 50-70% to minimize voltage errors. The liquid junction potential (14.7 mV) calculated with the software Clampex was not corrected in the experiment and data analysis. Cell membrane capacitive transient was electrically compensated with the Pulse software. Current and voltage signals were low-pass filtered at 5 kHz and stored in the hard disk of an IBM compatible computer. All experiments were conducted at room temperature (22-23uC).

Statistical Analysis
The data are expressed as mean6SEM. Paired and/or unpaired Student's t-test were used as appropriate to evaluate the statistical significance of differences between two group means, and ANOVA was used for multiple groups. Values of P,0.05 were considered to be statistically significant.

Inhibition of hKv4.3 Current by Allitridi
The previous study reported that the IC 50 of allitridi for inhibiting human atrial I to was about 45 mM [12]. Here we initially used a concentration of 30 mM allitridi to determine the potential inhibition of hKv4.3 current stably expressed in HEK 293 cells. Considering that allitridi is a volatile sulfate compound [28], the experimental working bath solution with 30 mM allitridi was immediately prepared before the drug application. The hKv4.3 current recorded with 300-ms voltage step to +60 mV from a holding potential of 280 mV was rapidly inhibited by 30 mM allitridi. The inhibitory effect reached a steady-state level within 130 s application, and was gradually reversed by washout ( Fig. 1A). Significant inhibition of voltage-dependent hKv4.3 current could also be observed with 10 mM allitridi (Fig. 1B). Figure 1C illustrates the current-voltage relationships of mean values of hKv4.3 current in the absence and presence of 10 and 30 mM allitridi. Allitridi significantly inhibited the current at test potentials of 210 to +60 mV (n = 16, P,0.05 or P,0.01 vs. control). It is interesting to note that the current (at +60 mV) was inhibited by 46% and 78% with 10 and 30 mM allitridi and the inhibitory efficacy of hKv4.3 current by allitridi is much stronger than that in human atrial I to reported previously [12]. We also observed a weak blocking effect of hKv4.3 current with a preprepared allitridi working solution (40-50% inhibition with 30 mM allitridi, data not shown). The different efficacy suggests that the immediate preparation of experimental working bath solution is crucial for obtaining the accurate pharmacological profile of this volatile compound.
In addition to the reduction of current amplitude, allitridi induced a facilitation of hKv4.3 current inactivation ( Fig. 1A   the current traces, measured the time to peak of hKv4.3 current, and fitted the inactivation process with a mono-exponential equation in the absence and presence of 10 and 30 mM allitridi. The normalized current at +60 mV showed a quick inactivation and a reduced time to peak of current activation in a representative cell ( Fig. 2A). Figure 2B and 2C illustrate the mean values of the time to peak of activation and the time constant (tau) of inactivation. The time to peak of hKv4.3 activation and the time constant of hKv4.3 current inactivation were significantly reduced at all test potentials (210 to +60 mV) by 10 and 30 mM allitridi (P,0.01 vs. control). The acceleration of the activation (time to peak) and the inactivation time constant indicates that allitridi inhibits hKv4.3 current by blocking the open channel. Figure 3 displays the effect of allitridi on kinetics of hKv4.3 current. Figure 3A and 3B shows the representative current and voltage protocol used for determining the availability of hKv4.3 current and the activation with tail current. Figure 3C illustrates the mean values of the variables of availability (I/I max ) of hKv4.3 current using the protocol as shown in Fig. 3A and the variables of activation conductance (G/G max ) measured from the current tail as shown in Fig. 3B in absence and presence of 10 mM. The variables of I/I max and G/G max were fitted to a Boltzmann function in individual cells as described previously [29]. The V 1/2 of hKv4.3 channnl availability was negatively shifted by 11.9 mV (from 236.761.1 mV in control to 248.660.9 mV in 10 mM allitridi, n = 12, P,0.01 vs. control), while the V 1/2 of activation conductance of the current was not altered (3.161.2 mV in control, 20.261.5 mV in allitridi, n = 10, P = NS vs. control).  Protocol and current traces used to assess availability (I/I max , steadystate inactivation) of hKv4.3 current. B. Protocol and tail current traces used to assess activation conductance (G/G max , steady-state activation) of hKv4.3 current. C. Mean values of hKv4.3 current (I/I max ) variables and conductance (G/G max ) variables before and after 10 mM allitridi were fitted to the Boltzmann function: g = 1/(1+exp((V 1/2 -V t )/K)), where V 1/2 is the voltage of 50% channel availability or maximal activation of the channel, V t is the test potential, and K is slope factor. C. Mean values of recovery time course of hKv4.3 current from inactivation, determined with protocol as shown in the inset before and after 10 mM allitridi, were fitted to a mono-exponential function. doi:10.1371/journal.pone.0051550.g003 Figure 3D shows the mean values of recovery time course of hKv4.3 current from inactivation determined by a paired pulse using a 300-ms step to +50 mV from a holding potential of 280 mV with variable P 1 -P 2 interval as shown in the inset. The recovery time course was fitted to a mono-exponential function in individual cells before and after 10 mM allitridi application. The recovery time constant (t) was 132.164.1 ms in control, and 126.765.1 ms in 10 mM allitridi (n = 10, P = NS vs. control). In another group of experiments, we found that inhibition of hKv4.3 current by allitridi (10 mM) was use-or rate-independent from 0.2 Hz to 3.3 Hz (n = 6, data not shown). The results suggest that allitridi has no effect on the recovery of hKv4.3 channels from inactivation, and is use-or rate-independent inhibition of hKv4.3 current.

Effect of allitridi on closed-state inactivation of hKv4.3 current
The previous study reported that the steady-state inactivation of Kv4.3 channels occurs predominantly from the closed state [30], we therefore determined whether allitridi would affect the development kinetics of closed-state inactivation of hKv4.3 channels. Figure 4A illustrates the closed state inactivation current traces of hKv4.3 channels in control and after application of 10 mM allitridi. The current was recorded with a double pulses (300-ms) protocol. A progressively increasing duration of a closed state potential of 250 mV (below activation threshold) was applied for second pulse. Allitridi (10 mM) clearly accelerated the closedstate inactivation of hKv4.3 channels. The normalized second pulse current was plotted against the time duration of closed-state potential. The closed-state inactivation time course was well fitted to a mono-exponential function before and after application of 10 mM allitridi (Fig. 4B). The mean values of the inactivation time constant was 1305625 ms in control, and 713615 ms in 10 mM allitridi (n = 6, P,0.01 vs. control). The result suggests that allitridi significantly accelerates the kinetics of closed-state inactivation of hKv4.3channels.

Molecular determinants of hKv4.3 channel blockade by allitridi
The molecular determinant of the block of hKv4.3 channels by allitridi was investigated using hKv4.3 mutants (see Materials and Method). These mutants are located in the pore-forming area. T366A and T367A are located in the P-loop helix, while V392A, I395A, and V399A are located in the S6 transmembrane domain. Figure 5A shows the representative current traces of wild type (WT), T366A, T367A, V392A, I395A, and V399A hKv4.3 channels activated with a 300-ms voltage step to +50 mV from a holding potential of 280 mV in the absence and presence of 30 mM allitridi. This concentration of allitridi remarkably inhibited the WT and V399A currents. A less inhibition was observed for the T366A, T367A, V392A, and I395A currents. The concentration-dependent response to allitridi was evaluated in WT and hKv4.3 mutant currents (at +50 mV), and the concentration-response curves were fitted to a Hill equation as in Fig. 5C. The IC 50 s of allitridi in inhibiting the hKv4.3 channels were 11.4 mM for WT, 35.5 mM for T366A, 44.7 mM for T367A, 23.7 mM for V392A, 42.4 mM for I395A, and 11.2 mM for V399A, respectively. The Hill co-efficient was in between 1.2 and 1.9 in WT and mutant Kv4.3 channels. These results suggest that T366, T367, V392, and I395, but not V399, are likely the major molecular determinants of channel blocking by allitridi.
To determine the relationship between the potential changes in kinetics of mutant hKv4.3 channels and the drug blocking sensitivity, we analyzed the availability (I/I max ) and activation conductance (G/G max ) of mutant hKv4.3 channels as in Fig. 3. Figure 6 shows the mean values of the variables of I/I max and (G/ G max ) of mutant hKv4.3 channels. The variables were fitted to a Boltzmann function [29] in individual cells, and the data were summarized in Table 1. The V 1/2 of I/I max was significantly shifted to depolarization potentials in the mutants T366A, T367A, V392A, and I395A (P,0.01 vs. WT), but not V399A. Allitridi 30 mM only significantly shifted the V 1/2 of I/I max in WT hKv4.3, but not in the mutant channels (Table 1). These results suggest that the residue position is important in determining the availability of the channel and the sensitivity of hKv4.3 to block by allitridi. The V 1/2 of G/G max was slightly shifted to depolarization potentials in the mutant hKv4.3 channels, which seems not related to the sensitivity of allitridi for blocking the channel.

Effect of Allitridi on hKv1.5 Current
The effect of allitridi on hKv1.5 channels expressed in HEK 293 cells was determined also with the immediately prepared working solution. Figure 7A shows the hKv1.5 current traces elicited by 300-ms voltage steps to between 240 and +60 mV from a holding potential of 280 mV in a representative cell the absence and presence of allitridi. Allitridi at 30 mM (5 min exposure) induced a significant increase of the current inactivation, a typical of open channel blockade. The inhibition was partially reversed by washout. Figure 7B illustrates the time course of hKv1.5 current recorded in a typical experiment with a 300-ms voltage step to +50 mV from a holding potential of 280 mV with accumulated application of 10, 30, and 100 mM allitridi. Allitridi at 10 mM induced a slight inhibition, while it at 30 and 100 mM remarkably suppressed the current. The inhibitory effect was partially reversed by washout. Figure 7C displays the I-V relationships of mean values of hKv1.5 current in the absence and presence of allitridi. Significant inhibition of hKv1.5 current was observed with 30 and 100 mM allitridi at test potentials of 210 to +60 mV (n = 6, P,0.05 or P,0.01 vs. control). The concentration-response curve (Fig. 7D) of allitridi for inhibiting hKv1.5 current (+50 mV) was fitted to a Hill equation. The IC 50 of allitridi for inhibiting hKv1.5 current was 40.3 mM with a Hill co-efficient of 2.1.

Effect of Allitridi on hERG Channels
The effect of allitridi on hERG channels was determined in HEK 293 cells stably expressing KCNH2 gene [23,25]. Figure 8A shows the voltage-dependent hERG current recorded in a typical experiment with 3-s voltage steps to between 240 and +60 mV, then to 250 mV from a holding potential of 280 mV in the absence and presence of allitridi. Allitridi at 30 mM (5 min exposure) remarkably inhibited hERG step and tail currents, and the inhibition was partially reversed by washout. Figure 8B illustrates the I-V relationships of hERG step current (I hERG.step ) and tail current (I hERG.tail ) in the absence (control) and presence of 3, 10, and 30 mM allitridi. The step current and tail current of hERG channels were significantly inhibited by 10 and 30 mM allitridi at test potentials of 220 to +60 mV (n = 7, P,0.05 or P,0.01 vs. control). The concentration-response curve of allitridi for inhibiting I hERG.tail (Fig. 8C) was fitted to a Hill equation. The IC 50 of allitridi for inhibiting I hERG.tail was 19.6 mM with a Hill coefficient of 1.5. Effect of Allitridi on Cardiac hKCNQ1/hKCNE1 Channels The effect of allitridi on human cardiac I Ks was determined in HEK 293 cells stably hKCNQ1/hKCNE1 [26,31]. Figure 9A displays the voltage-dependent I Ks current recorded in a representative cell with 3-s voltage steps to between 260 and +60 mV (20-mV increment), then to 250 mV from a holding potential of 280 mV in the absence and presence of allitridi. Allitridi at 30 mM (5 min exposure) remarkably inhibited the step and tail currents of I Ks , and the inhibition was partially reversed by washout. Figure 9B shows the I-V relationships of I Ks step current in the absence (control) and presence of 3, 10, and 30 mM allitridi. Significant inhibition of the current was observed with 10 and 30 mM allitridi at test potentials of 0 to +60 mV (n = 6, P,0.01 vs. control). The concentration-response curve of allitridi for inhibiting I Ks is illustrated in Fig. 9C, which was fitted to a Hill equation. The IC 50 of allitridi for inhibiting human cardiac I Ks was 17.7 mM with a Hill co-efficient of 1.3.

Discussion
The present study demonstrates that allitridi preferably blocks hKv4.3 channels (coding human cardiac I to ) expressed in HEK 293 cells (IC 50 = 11.4 mM) by binding to T366 and T367 of the Ploop helix, and V392 and I395 of the S6 domain of the channel. In addition, allitridi may also suppress other human cardiac potassium channels expressed in HEK 293 cells with a relatively weak effect, including hKv1.5 channels (coding human atrial I Kur , IC 50 = 40.3 mM), hERG channels (coding human cardiac I Kr , IC 50 = 19.6 mM), hKCNQ1/hKCNE1 channels (coding human cardiac I Ks , IC 50 = 17.7 mM). It slightly decreases hKir2.1 channels (coding human cardiac I K1 , by 9.8% with 100 mM allitridi). Therefore, the efficacy of allitridi for inhibiting human cardiac potassium currents is I to .I Ks .I Kr .I Kur ..I K1 . In addition, we have shown that the inhibitory efficacy of allitridi on hKv4.3 and hKv1.5 currents is stronger than that observed previously in human atrial I to and I Kur [12], which suggests that the fresh preparation of the experimental working solution is important for the accurate pharmacological effect of this volatile compound. It may be non-reliable for the relative high concentration for inhibiting human atrial I to observed in the previous study [12] and for blocking Kv4.3 current using a pre-prepared allitridi working solution. In addition, the previous report demonstrated that allitridi is a major biologically active volatile organosulfur compound in garlic (with a concentration of 1.1 mg/ g) [32]. Whether the concentration of allitridi in garlic will have an impact on cardiac K + channels function remains to be studied.
In addition to the anti-microbial effects [1,33,34], garlic and its constituents including allitridi were reported to have anti-cancer effects by inducing G2/M phase cell cycle arrest and apoptosis via inhibiting PI3K/Akt activation and modulating Bcl-2 family proteins [2,35,36]. The in vitro and in vivo studies showed that garlic and its bioactive compounds have multiple cardiovascular beneficial effects via inhibiting enzymes involved in lipid synthesis, reducing blood platelet aggregation and cholesterol, lowering blood pressure, and increasing antioxidant status [37]. Earlier studies demonstrated that garlic and its extract had antiarrhythmic effects in ventricular tachycardia/fibrillation induced  Table 1. Effects of allitridi (30 mM) on midpoint potential (V 1/ 2 ) of availability (I/I max ) and activation (G/G max ) of WT and mutant hKv4.3 channels. by ischemia/reperfusion in rats [38,39], and ventricular arrhythmias induced by ouabain or isoproterenol in pigs and atrial arrhythmia in rats [11]. Rat atrial refractory period was prolonged by garlic dialysate in a concentration-dependent manner [11]. Recent studies showed that garlic extract improved defibrillation efficacy, and significantly decreased the inducibility of ventricular arrhythmia in a dose-dependent manner in a pig model [40,41]. A more recent study has reported that allitridi has cardioprotective action against cardiac ischemia/reperfusion injury in a mouse model via releasing H 2 S which exerts a preconditioning effect [42]. However, the prolongation of cardiac effective refractory period by allitridi in isolated cardiac tissues/hearts [11,13,14] can not be fully interpreted by the in vivo H 2 S release mechanism. The blockade of multiple cardiac potassium channels by effective concentrations of allitridi observed in the present study may account for the alteration of cardiac electrophysiology and the anti-arrhythmias.
The transient outward potassium current I to plays an important role in repolarization of atrial and ventricular repolarization in rodent hearts [43]. Blockade of I to would prolong cardiac action potential duration and effective refractory period. Therefore, inhibition of I to by allitridi may account for the increase of cardiac effective refractory period and anti-arrhythmia in rats [11,13]. However, no I to channel expression was observed in pig [15] and guinea pig [16] hearts, the anti-arrhythmic effect is likely related to the inhibition of I Kr and I Ks in the later species [11,13,14]. Previous studies demonstrated that I Ks is positively regulated by Akt/PI3K kinases [44], allitridi releases H 2 S [8] and therefore induces an inhibition of Akt/PI3K [45]. Thus, the inhibition of  Akt/PI3K may contribute at least in part to the decrease of I Ks by allitridi.
It is well recognized that I to plays an important role in the repolarization of action potentials in atrial myocytes [46,47,48] and also the phase 1 fast repolarization of ventricular action potentials, especially in ventricular epicardium in humans [29,49] and in dogs [50]. Therefore, preferable blockade of I to /Kv4.3 and a weak inhibition of I Kur /Kv1.5, I Kr /hERG, and I Ks by allitridi would prolong human atrial action potential and may be effective in anti-atrial fibrillation in humans.
It has been documented that the shift in cardiac repolarizing current due to a decrease in sodium or calcium channel currents or an increase in I to , I K.ATP , I K.ACh , or other outward currents may induce J-wave syndromes that involve in Brugada syndrome and early repolarization syndrome, which may trigger life-threatening arrhythmia [51,52]. The increase of I to amplitude by gain-offunction mutations in KCND3-encoded Kv4.3 channels is the molecular pathogenesis for the lethal arrhythmia in patients with Brugada syndrome [53]. The I to blocker 4-aminopyridine restored the epicardial action potential dome, reduced both transmural and epicardial dispersion of repolarization, normalized the ST segment, and prevented phase 2 reentry and ventricular tachycardia/ventricular fibrillation in experimental Brugada syndrome [51,54]. Allitridi has a strong inhibition of Kv4.3 current. It, as 4aminopyridine [51,54], is likely effective in suppressing Brugada syndrome-related arrhythmias. However, it remains to be studied in the future.
We found that the allitridi inhibited Kv4.3 channels by shifting the availability voltage to more negative potentials and accelerating the closed-state inactivation of the channel. It significantly reduced the time to peak of current activation and the time constant of Kv4.3 current inactivation. This suggests that allitridi may quickly bind the channel when they open. However, allitridi, as rosiglitazone [55], blocked the open channels of hKv4.3 in a useor rate-independent manner. The open channel blocking effect was also observed in Kv1.5 current.
Alanine-scanning mutagenesis is a method of systematic alanine substitution and has been particularly used for the identification of functional epitopes [56]. This technique is usually used for identifying the drug binding sites of ion channel blockers [24,57,58]. With this technique, we demonstrated that the inhibitory efficacy of allitridi on the hKv4.3 mutants T366A and T367A at the P-loop of the pore helix was significantly reduced. This implies that allitridi may be trapped into the channel pore and block the open channel. Moreover, the mutants V392A and  I395A, but not V399A, of the S6 domain exhibit a significantly reduced response to allitridi, indicating that in addition to binding to the P-helix filter, allitridi may interact with V392 and I395 of the S6 domain. Therefore, these four residues (T366, T367, V392, and I395) of the channel are likely critical for allitridi inhibition of hKv4.3 current.
Collectively, the present study demonstrates that allitridi blocks hKv4.3 channels by interacting with T366 and T367 of the P-loop helix, and V392 and I395 in the S6 domain, and also has a relatively weak inhibition of hKv1.5, hERG, and I Ks . These effects may count for anti-arrhythmias observed in experimental arrhythmic animal models.