Is Selective Late Na+ Current Inhibition Different From Class I/B Antiarrhythmic Action? Comparison of The Effects of GS967 to Mexiletine in Canine Ventricular Myocardium


 Enhancement of the late Na+ current (INaL) increases arrhythmia propensity in the heart, while suppression of the current is antiarrhythmic. GS967 is an agent considered as a selective blocker of INaL. In the present study, effects of GS967 on INaL, on L-type calcium current (ICa), and on action potential (AP) morphology were studied in canine ventricular myocytes by using conventional voltage clamp, action potential voltage clamp and sharp microelectrode techniques. These effects of GS967 were compared to tetrodotoxin (TTX) and to the class I/B antiarrhythmic compound mexiletine. 1 µM GS967, 40 µM mexiletine, and 10 µM TTX dissected largely similarly shaped inward currents under action potential voltage clamp conditions. In case of GS967 and mexiletine, the amplitude and integral of this current was significantly smaller when measured in the presence of 1 µM nisoldipine, while no difference was observed in case of TTX. Under conventional voltage clamp conditions, INaL was significantly decreased by 1 µM GS967 and 40 µM mexiletine (79.0±3.0% and 63.3±2.7% reduction of current integrals, respectively). The integral of ICa was moderately but significantly diminished by both drugs (reduction of 9.3±3.3% and 14.1±1.5%, respectively). These changes were associated with acceleration of inactivation of ICa. Drug effects on peak Na+ current (INaP) were also assessed by recording AP upstroke in multicellular preparations. Both GS967 and mexiletine showed fast onset and offset kinetics: 110 ms and 289 ms offset time constants, respectively, as determined from V+max measurements in right ventricular papillary muscles, while the onset kinetics was characterized by 5.3 AP and 2.6 AP, respectively, at 2.5 Hz. Effects on beat-to-beat variability of AP duration (APD) was studied in isolated myocytes. Beat-to-beat variability was significantly decreased by both GS967 and mexiletine (reduction of 42.1±6.5% and 24.6±12.8%, respectively) while their shortening effect on APD was comparable. It is concluded that the electrophysiological effects of GS967 are similar to those of mexiletine, but with somewhat faster offset kinetics of V+max block. However, since GS967 depressed V+max and INaL at the same concentration, the current view that GS967 represents a new class of drugs that selectively block INaL has to be questiond and it is suggested that GS967 should be classified as a class I/B (or I/B + IV) antiarrhythmic agent.


Introduction
Following the large Na + current surge associated with the AP upstroke (called I NaP ) a smaller but sustained current component (called I NaL ) remains active throughout the entire cardiac AP. Enhanced I NaL , due to diseases like heart failure [1][2][3] , hypertrophic cardiomyopathy [4] or LQT3 [5] , is known to increase arrhythmia propensity in the heart [1,[6][7][8] . It has been known for a long time that many of the class I antiarrhythmic drugs inhibit I NaL in addition to blocking I NaP [9][10][11][12][13][14] . This latter effect, however, was considered to enhance proarrhythmic risk and consequently the incidence of sudden cardiac death [15] . As a consequence, development of agents to suppress I NaL selectively was a straightforward strategy in the past decades [2,[16][17][18][19] . One of these agents, 6-(4-(tri uoromethoxy)phenyl)-3-(tri uoromethyl)-[l,2,4]triazolo[4,3-a]pyridine, known as GS967 or GS458967, was reported to be a particularly selective candidate when tested in rabbit ventricular cells [16] . It was also shown that important species-dependent variations existed in the electrophysiologic properties of myocardial preparations [20,21] , and canine ventricular myocytes are considered a reasonably good model for human ventricular cells [22][23][24] . Furthermore, it was previously shown using the action potential voltage clamp technique that the kinetic properties of I NaL are similar in dogs and humans [20] , while differing from those of other mammals including guinea pigs, rabbits and pigs [25][26][27] . Therefore, we studied the effects of GS967 on I NaL and I Ca and compared them to those of tetrodotoxin (TTX) and the class I antiarrhythmic agent mexiletine in canine ventricular preparations. In this work, based on experimental evidence we challenge the present concept that GS967 exerts a selective I NaL blocking effect, however, it has very similar class I/B antiarrhythmic properties as mexiletine.

Isolation of cardiomyocytes
Single canine myocytes were obtained by enzymatic dispersion using the segment perfusion technique, as previously described [28] . Brie y, a wedge-shaped section of the ventricular wall supplied by the LAD coronary artery was cannulated, dissected and perfused with a nominally Ca 2+ -free Joklik solution (Minimum Essential Medium Eagle, Joklik Modi cation, Sigma-Aldrich Co. St. Louis, MO, USA) for a period of 5 min. After this, the tissue was perfused with Joklik solution supplemented with 1 mg/ml collagenase (Type II, Worthington Biochemical Co., Lakewood, NJ, USA; representing nal activity of 224 U/ml) and 0.2% bovine serum albumin (Fraction V., Sigma) containing 50 µM Ca 2+ for 30 min.
Finally, the normal external Ca 2+ concentration was gradually restored and the cells were stored at 15 o C in Minimum Essential Medium Eagle until use. The chemicals used in the experiments were obtained from Sigma-Aldrich Co. (St. Louis, MO, USA).

Electrophysiology
Cells were placed in a plexiglass chamber under an inverted microscope, allowing for continuous superfusion with a modi ed Tyrode solution by gravity ow at a rate of 1-2 ml/min. The modi ed Tyrode solution contained (in mM): NaCl 121, KCl 4, CaCl 2 1.3, MgCl 2 1, HEPES 10, NaHCO 3 25, glucose 10 at pH = 7.35, which was supplemented according to the actual experimental design. The osmolarity of this solution was 300 ± 3 mOsm, measured with a vapor pressure osmometer. In all experiments, the bath temperature was set to 37 ºC using a temperature controller (Cell MicroControls, Norfolk, VA, USA).
Electrical signals were ampli ed and recorded (MultiClamp 700A or 700B, Molecular Devices, Sunnyvale, CA, USA) under the control of a pClamp 10 software (Molecular Devices) following analogue-digital conversion (Digidata 1440A or 1332, Molecular Devices). Electrodes, having tip resistances of 2-3 MΩ when lled with pipette solution, were manufactured from borosilicate glass. Transmembrane currents were recorded in whole-cell voltage clamp con guration. The series resistance was typically 4-8 MΩ, and the measurement was discarded if it changed substantially during the experiment.

Action potential voltage clamp
Action potential voltage clamp experiments were performed according to the methods described [29,30] . A previously recorded midmyocardial canine ventricular AP was applied as command signal and the current traces were recorded continuously in modi ed Tyrode solution before and after 5 min superfusion with the Na + channel inhibitor applied. The drug-sensitive current was obtained by subtracting the postdrug trace from the reference pre-drug trace. These measurements were performed either in the presence or absence of 1 µM nisoldipine added to the Tyrode solution. The pipette solution contained (in mM): Kaspartate 120, KCl 30, MgATP 3, HEPES 10, Na 2 -phosphocreatine 3, EGTA 0.01, cAMP 0.002, KOH 10 at pH = 7.3 with an osmolarity of 285 mOsm. The amplitude of the dissected current (I NaL in the presence of nisoldipine) was evaluated at half-duration of the command AP. When determining the current integral, the initial 20 ms after the AP upstroke was excluded from evaluation in order to eliminate the contribution of I NaP . In each experiment 20 consecutive current traces were averaged and analyzed in order to reduce the noise and the trace-to-trace uctuations of action potential con guration. Ion currents were normalized to cell capacitance, determined in each cell by applying hyperpolarizations from + 10 to − 10 mV for 15 ms.

Conventional voltage clamp
Conventional voltage clamp experiments, using rectangular command pulses, were performed to study the effects of GS967 and mexiletine on I NaL and I Ca at stable test potentials. 2). Test pulses were clamped to − 20 mV for 2 s from the holding potential of -120 mV before and after application of GS967 or mexiletine, while the total amount of I NaL was determined by pharmacological subtraction performed by a nal superfusion with 20 µM TTX. The amplitude of I NaL was evaluated at 50 ms after beginning the pulse. For determination of current integral the initial 20 ms was excluded from evaluation in order to minimize the contribution of I NaP .
In the case of I Ca measurements the bath solution was modi ed Tyrode solution supplemented with 3 mM 4-aminopyridine to suppress K + currents. The pipette solution contained (in mM): K-aspartate 120, KCl 30, MgATP 3, HEPES 10, Na 2 -phosphocreatine 3, EGTA 0.01, cAMP 0.002, KOH 10 at pH = 7.3. Test pulses to + 5 mV, lasting for 200 ms, arose from the holding potential of -80 mV, while a prepulse to − 40 mV for 15 ms was interposed between the holding potential and the test pulse to inactivate Na + channels. In this case the current integral contained the total amount of current carried by I Ca from the beginning to the end of the test pulse.

Recording of APs from multicellular preparations
Multicellular preparations (right ventricular papillary muscles) were selected to prevent the limitations of isolated myocyte studies, like the absence of intercellular clefts or potential damage to channel proteins, allowing better representation of in vivo conditions.
The experiments were performed as it was previously described [31] . Brie y, transmembrane potentials were recorded using 3 M KCl lled sharp glass microelectrodes having tip resistance between 10 and 20 MΩ. These electrodes were connected to the input of a high impedance electrometer (MDE GmbH, Heidelberg, Germany). Preparations were paced by a pair of platinum electrodes using 1 ms wide rectangular current pulses with twice the threshold amplitude at 37 °C. The pacing cycle length was set to 1 s for at least 60 min allowing the preparations to equilibrate before starting the experiment.
Following equilibration at 1 s cycle length, the cycle length was sequentially varied between 0.3 and 5 s. At each cycle length the 25th AP was recorded, and the cycle length was then changed. Under these conditions a quasi steady-state rate-dependence could rapidly be obtained. APs were digitized at 100 kHz using an ADA 3300 data acquisition board (Real Time Devices Inc., State Collage, PA, USA) and stored for later analysis. After taking control records at each cycle length the preparations were superfused with either GS967 or mexiletine for 20 min and then the protocol was repeated. Efforts were made to maintain the same impalement throughout each experiment. If, however, an impalement became dislodged, adjustment was attempted, but when the parameters of the re-established impalement deviated by more than 5% from the previous record, the experiment was discarded.
Restitution kinetics of the maximum rate of depolarization (V + max ) is considered as the indicator of offset time constant. To determine the restitution time constant for V + max the preparations were paced using a train of 20 basic stimuli delivered at a basic cycle length of 1 s. Each train was followed by a single extra stimulus applied with successively longer coupling intervals. The train of basic stimuli was reinitiated following the delivery of the extra stimulus. In this way, each 20th basic AP was followed by a single extra AP occurring at gradually increasing diastolic intervals. The diastolic interval was de ned as the time from APD 90 of the last basic member of the train to the upstroke of the extra AP. Recovery curves were generated by plotting the V + max of each extra AP as a function of the respective diastolic interval and data were tted to a single exponential function.
Onset kinetics of drug action on V + max were determined by stimulating the preparation at a cycle length of 0.4 s following a few min period of rest and the initial 40 APs were recorded and data were plotted against the number of the analyzed AP within the train. The rate of development of block was obtained by monoexponential tting of the V + max values.

Determination of beat-to-beat variability of APD in isolated myocytes
Since beat-to-beat variability of APD is relatively small in multicellular preparations due to the balancing effect of the neighboring cells, these experiments were performed in isolated myocytes. Series of 50 consecutive action potentials were analyzed to estimate the beat-to-beat variability of APD, de ned as short term variability (SV), according to the following formula: where SV is short term variability, APD n and APD n+1 indicate the durations of the i th and i + 1 th APs, respectively, at 90% level of repolarization and n beats denotes the number of consecutive beats analyzed [32] . Changes in SV were presented as Poincaré plots where 50 consecutive APD values are plotted, each against the duration of the previous AP.

Statistics
Results are expressed as mean ± SEM values, n denotes the number of myocytes or multicellular preparations studied. Statistical signi cance of differences was evaluated using one-way ANOVA followed by Student's t-test for paired or unpaired data as pertinent. Differences were considered signi cant when p was less than 0.05.

Effects of GS967, mexiletine and TTX under action potential voltage clamp conditions
Under action potential voltage clamp conditions, 10 µM TTX, 1 µM GS967 and 40 µM mexiletine dissected similarly shaped inward current pro les dominated by I NaL (Fig. 1). These concentrations were chosen because the densities, measured at half-duration of the AP (50% of APD 90 ), and integrals of the dissected currents were largely comparable in size. TTX was applied as a reference, since 10 µM TTX was shown to suppress more than 80% of I NaL in canine ventricular cells [20] . Importantly, in all action potential voltage clamp experiments performed with mexiletine, the bath solution contained 1 µM E4031 and 100 µM chromanol 293B in order to eliminate any interference from K + currents [33] . Since I Ca was not suppressed in these experiments, the measurements were repeated in the presence of 1 µM nisoldipine (NISO) and the results were compared to those obtained without nisoldipine (NO NISO) as presented in Fig. 2. In the case of GS967 and mexiletine, the half-duration current densities were signi cantly smaller in the presence (− 0.27 ± 0.02 and − 0.25 ± 0.02 A/F) than in the absence (− 0.42 ± 0.03 and − 0.45 ± 0.04 A/F) of nisoldipine. Similar differences were found when comparing the current integrals: −48 ± 4 and − 43 ± 4 mC/F in the absence, while − 68 ± 5 and − 74 ± 10 mC/F in the presence of nisoldipine, respectively. Importantly, there was no signi cant difference between the NISO and NO NISO data in the case of TTX. These results suggest that both GS967 and mexiletine inhibit I Ca in addition to I NaL blockade.

Effects of GS967 and mexiletine on I NaL under conventional voltage clamp conditions
The effects of 1 µM GS967 and 40 µM mexiletine on I NaL were also studied under conventional voltage clamp conditions by applying 2 s duration depolarizations to − 20 mV from the holding potential of − 120 mV (Fig. 3). 1 µM GS967 signi cantly reduced the density of I NaL , measured at 50 ms after the beginning of the pulse (from − 0.313 ± 0.05 to − 0.062 ± 0.01 A/F, corresponding to an 80.4 ± 2.2% reduction on the average of 6 myocytes). For comparison, this parameter was also signi cantly decreased by 40 µM mexiletine (from − 0.385 ± 0.036 to − 0.156 ± 0.014 A/F, reduction of 59.1 ± 1.8%, n = 12). Similar results were obtained when comparing the current integrals, i.e. the charge carried by the current with the exclusion of the initial 20 ms. As determined from the same experiments, the current integrals were signi cantly decreased from − 69.1 ± 7.9 to − 15.4 ± 3.9 mC/F (79.0 ± 3.1% inhibition) by 1 µM GS967 and from − 76.4 ± 7.6 to − 26.7 ± 2.6 mC/F (63.3 ± 2.7% reduction) by 40 µM mexiletine.

Effects of GS967 and mexiletine on I Ca under conventional voltage clamp conditions
Since action potential clamp experiments suggested some inhibition of I Ca by both GS967 and mexiletine, this was clari ed using conventional voltage clamp, as shown in Fig. 4. Although peak I Ca was not altered signi cantly by 1 µM GS967 (− 4.92 ± 0.16 and − 4.76 ± 0.04 A/F), the current density, measured at 50 ms after the beginning of the pulse, was signi cantly reduced (from − 0.66 ± 0.05 to − 0.62 ± 0.05 A/F). Accordingly, the current integral (measured from the beginning to the end of pulse) was also signi cantly decreased by GS967 in the 7 myocytes studied (from − 111.9 ± 5.0 to − 100.7 ± 3.1 mC/F, reduction of 9.3 ± 3.3%). The suppressive effect of 40 µM mexiletine on I Ca was somewhat more pronounced. It was signi cant for the reduction of peak current (from − 4.9 ± 0.4 to − 4.4 ± 0.4 A/F), and for the current measured at 50 ms (from − 0.86 ± 0.08 to − 0.70 ± 0.07 A/F), as well as the current integral (from − 114 ± 11 to − 98 ± 9 mC/F, corresponding to 14.1 ± 1.5% reduction on average of 5 experiments). Both drugs accelerated the decay time of I Ca , determined between 90% and 10% of current amplitudes, signi cantly: from 48.8 ± 2.5 to 44.2 ± 2.9 ms by GS967, and from 50.8 ± 5.4 to 44.9 ± 5.0 ms by mexiletine. This acceleration of current decay may account for the decreased current integrals.

Effects of GS967 and mexiletine on action potential upstroke
After determining the effects on I NaL and I Ca , the actions of GS967 and mexiletine on I NaP were studied and compared. Since direct measurement of cardiac I NaP is di cult at 37 o C, the maximum velocity of depolarization during the action potential upstroke (V + max ) was used as an approximate, although not linear, measure of I NaP [34][35][36] . Due to the more physiological situation (e.g higher stability) in multicellular cardiac preparations, these experiments were performed in right ventricular papillary muscles using sharp microelectrodes. As demonstrated in Fig. 5

Effects of GS967 and mexiletine on APD
Although APD in right ventricular trabeculae was shortened by both GS967 and mexiletine in a reverse rate-dependent manner, this effect was statistically signi cant only at the longest constant cycle lengths above 1.5 s (Fig. 6).

Effects of GS967 and mexiletine on beat-to-beat variability of APD
Elevated beat-to-beat QT interval variability is a good predictor of ventricular arrhythmia in a wide variety of patients, while at a single cell level it translates into beat-to-beat variability of APD (also called short term variability, SV) -both parameters are considered as indicators of proarrhythmic risk [37][38][39][40] . In contrast to multicellular preparations, where the neighboring cells may effectively balance the individual differences in APD, SV is more pronounced in single cells, therefore the effects of GS967 and mexiletine were studied on SV in isolated ventricular cells, paced at a constant cycle length of 1 s. Under these conditions SV was signi cantly decreased by 1 µM GS967 and 40 µM mexiletine (reduction of 1.01 ± 0.19 and 0.63 ± 0.02 ms, respectively, corresponding to 42.1 ± 6.5% and 24.6 ± 12.8% decrease), while APD was moderately shortened by 32.5 ± 6.9 and 41.4 ± 4.1 ms, respectively, (all p < 0.05 and n = 8). All these effects were largely reversible upon washout (Fig. 7). Since APD itself is also known to affect the magnitude of SV, the term of relative SV (RSV) was introduced [41,42] . Accordingly, RSV = dSV / dAPD, i.e. the change in SV is normalized to that of APD. The value of RSV was signi cantly higher for GS967 than mexiletine (0.039 ± 0.007 versus 0.015 ± 0.007) predicting a better antiarrhythmic effectivity of GS967 than mexiletine.

Discussion
We studied and compared the effects of GS967 (1 µM) to those of the class I/B antiarrhythmic drug mexiletine (40 µM) on I Ca , I NaL and I NaP in canine ventricular myocardium by combining the conventional microelectrode, voltage clamp and action potential voltage clamp techniques. It was found that GS967, which is generally considered as a selective blocker of I NaL [16] , inhibits I Ca and I NaP as well, similarly to the class I/B antiarrhythmic drug mexiletine [33] but with higher potency. The I Ca blocking effects were more prominent under action potential voltage clamp than under conventional voltage clamp conditions in the case of both drugs (compare Figs. 2 and 4). The reason for this difference is unknown at present, however, it highlights the advantage of the action potential voltage clamp technique in cardiac cellular electrophysiology and pharmacology.
Based on the hypothetically selective I NaL blocker nature of GS967, the drug was previously mentioned as a novel class VI antiarrhythmic agent [43] . However, without questioning the theoretical possibility of the concept for selective I NaL inhibition, an alternative approach should also be considered, since the concept whether I NaL can be blocked selectively is an interesting and still unresolved issue. There is evidence that sodium current in cardiac tissues may also be conducted by sodium channels other than the cardiac speci c Na v 1.5 channels, such as Na v 1.8 [44] . Also, relatively high mRNA expression levels of Na v 2.1 were reported in human ventricle [45] but proper functional evidence for its role is still lacking. If these channels play a role in I NaL but not in I NaP , pharmacological inhibition of these channels may result in selective I NaL blockade. However, GS967 has not been shown to inhibit these latter types of sodium channels but it was reported to inhibit cardiac type Na v 1.5 channels [9,10] . Na v 1.5 channels have complex and multiple open and closed states [46] with different drug binding properties governing active, inactive and resting channel states. Accordingly, drugs interacting with the binding sites of the channels depending on their actual open or closed channel states may produce variable effects on I NaP and I NaL . For example, when a drug binds rapidly and with high a nity to open and inactivated channel states, and it dissociates rapidly from closed resting channel states, it would not inhibit I NaP but I NaL . This would be due to complete drug dissociation from its binding site in the resting closed states unless frequency was very high or at least the cycle length would be shorter than its dissociation from the channel. Consequently, whether a drug inhibits I NaP or I NaL selectively or both of them, largely depends on the stimulation protocol but not on existing speci c I NaP or I NaL binding sites. Present results and other recently published data [9,10] are consistent with this suggestion and do not support a mechanism which is based on speci c I NaL inhibition that is distinctly different from class I/B antiarrhythmic actions described for drugs such as mexiletine [14,46] , lidocaine [9,10,14] , amiodarone [11] and ranolazine [47] . This approach is also in line with the reported high (38-fold) selectivity of ranolazine on I NaL over I NaP [2] , which is intermediate (13-fold) for amiodarone [3] and much lower (only 3-fold) for ecainide [48] .
In our experiments, both GS967 and mexiletine signi cantly depressed V + max at high stimulation rates. It was previously established that changes of V + max and I NaP are not linear and a relatively modest decrease of V + max can represent robust depression of I NaP [35,36] . Accordingly, the 20-30% reduction of V + max measured in the present study in papillary muscle preparations (at pacing cycle lengths of 0.3-0.4 s) following GS967 and mexiletine application can represent a similar degree of I NaP depression as the measured 60-80% reduction of I NaL obtained in ventricular myocytes. Therefore, in theory, neither GS967 nor mexiletine can be considered as "selective" I NaL inhibitors. However, when therapy is concerned -assuming that decreased I NaP is proarrhythmic, while reduced I NaL is antiarrhythmic -GS967, which has about 3-fold faster offset kinetics, can be more bene cial than mexiletine since I NaP would be affected in a lesser extent than I NaL by GS967 at normal or moderately enhanced heart rates.
On the basis of present and other results it is clear that GS967 affects I NaP in a strongly rate-and moderately species-dependent fashion. Similarly to our results, V + max was reduced by 0.3 µM GS967 in murine myocytes [49] , while the same concentration of the drug failed to modify V + max in canine Purkinje strands [50] . On the other hand, GS967 shortened APD in a variety of preparations, including rat [51,52] , murine [49] , rabbit [16,53] and human [54] ventricular cells within a wide concentration range (0.1-1 µM)similarly to the present observations in isolated canine ventricular cells. In our multicellular preparations, however, a signi cant APD shortening effect appeared only at cycle lengths longer than 1.5 s. This can be explained by the well-known higher drug-sensitivity of single cells comparing to multicellular preparations. The moderate reduction of I Ca by GS967, in addition to its decrease of I NaL , is not expected to impair A-V conduction. However, this effect can contribute to the GS967-induced shortening of the elongated APD, reduction of the enhanced dispersion and short term variability of repolarization, changes often preceding torsade de pointes arrhythmias [38][39][40] .
Taken together the present results and the literature, it is likely that a compound having kinetic properties similar to GS967 would be a very promising new antiarrhythmic agent, since several in vitro [16][17][18][19][49][50][51][52][53][54] and in vivo [51,55] studies support the potent antiarrhythmic activity of GS967. Its kinetic properties are better than those of mexiletine, as shown in this study, and also than those of ranolazine [9] , agents known to suppress I NaL . Although GS967 had high brain penetration and caused a profound use-dependent block on all sodium channel isoforms studied, making the compound prone for possible central nervous system side effects [56] , a new agent exhibiting the same kinetic properties of GS967 without CNS side effects would represent a promising candidate for future development.
In summary, GS967 -similarly to mexiletine -inhibited both the peak and late components of Na + current, suppressed Ca 2+ current and decreased beat-to-beat variability of APD. Based on its kinetic properties, GS967 should be classi ed as a new potent class I/B (or probably I/B + IV) antiarrhythmic agent. The results of the present study also suggest that investigations of "selective" I NaL inhibitors should be carried out through a wide range of stimulation rates since the effect of drugs like GS967 or mexiletine, that possess fast offset kinetics of I NaP inhibition, can be misinterpreted.
Declarations Figure 1 Effects of TTX, GS967 and mexiletine in isolated canine ventricular myocytes under action potential voltage clamp conditions. A: representative membrane current records dissected by 10 µM TTX, 1 µM GS967 and 40 µM mexiletine, respectively. The command AP is shown above the current traces. Dashed lines indicate zero voltage and current levels. B: Current densities measured at half-duration of the command AP. C: Current integrals from which the initial 20 ms period was excluded. Columns and bars denote mean ± SEM values, numbers in parentheses indicate the number of myocytes studied.   C,D: Average ICa densities measured as peak currents or 50 ms after beginning the pulse, and ICa integrals. Columns and bars are mean ± SEM, numbers in parentheses indicate the number of myocytes studied, asterisks denote signi cant differences between the pre-drug control and the GS967-or mexiletine-treated data.