Cytosolic Ca²⁺ triggers early afterdepolarizations and torsade de pointes in rabbit hearts with type 2 long QT syndrome

Abstract

The role of intracellular Ca²⁺ (${\text{Ca}}_{\text{i}}^{2+}$) in triggering early afterdepolarizations (EADs), the origins of EADs and the mechanisms underlying Torsade de Pointes (TdP) were investigated in a model of long QT syndrome (Type 2). Perfused rabbit hearts were stained with RH327 and Rhod-2/AM to simultaneously map membrane potential (V_m) and ${\text{Ca}}_{\text{i}}^{2+}$ with two photodiode arrays. The I_Kr blocker E4031 (0.5 μM) together with 50% reduction of [K⁺]_o and [Mg²⁺]_o elicited long action potentials (APs), V_m oscillations on AP plateaux (EADs) then ventricular tachycardia (VT). Cryoablation of both ventricular chambers eliminated Purkinje fibres as sources of EADs. E4031 prolonged APs (0.28 to 2.3 s), reversed repolarization sequences (base→apex) and enhanced repolarization gradients (30 to 230 ms, n = 12) indicating a heterogeneous distribution of I_Kr. At low [K⁺]_o and [Mg²⁺]_o, E4031 elicited spontaneous ${\text{Ca}}_{\text{i}}^{2+}$ and V_m spikes or EADs (3.5 ± 1.9 Hz) during the AP plateau (n = 6). EADs fired ‘out-of-phase’ from several sites, propagated, collided then evolved to TdP. Phase maps (${\text{Ca}}_{\text{i}}^{2+}$ vs. V_m) had counterclockwise trajectories shaped like a ‘boomerang’ during an AP and like ellipses during EADs, with V_m preceding ${\text{Ca}}_{\text{i}}^{2+}$ by 9.2 ± 1.4 (n = 6) and 7.2 ± 0.6 ms (n = 5/6), respectively. After cryoablation, EADs from surviving epicardium (∼1 mm) fired at the same frequency (3.4 ± 0.35 Hz, n = 6) as controls. At the origins of EADs, ${\text{Ca}}_{\text{i}}^{2+}$ preceded V_m and phase maps traced clockwise ellipses. Away from EAD origins, V_m coincided with or preceded ${\text{Ca}}_{\text{i}}^{2+}$. In conclusion, overload elicits EADs originating from either ventricular or Purkinje fibres and ‘out-of-phase’ EAD activity from multiple sites generates TdP, evident in pseudo-ECGs.

Class III anti-arrhythmic agents inhibit delayed rectifying K⁺ currents (I_Kr or I_Ks) resulting in a prolongation of the cardiac action potential (AP) and its effective refractory period. In theory, the prolongation of refractory period is anti-arrhythmic but in clinical practice, these agents have been found to be pro-arrhythmic producing an abnormal prolongation of the QT interval with syncope caused by a polymorphic ventricular tachycardia (VT) called Torsade de Pointes (TdP) (Dessertenne, 1966). The general consensus is that long QT-related arrhythmias are elicited by early afterdepolarizations (EADs) and that an enhanced dispersion of repolarization may be required to provide a substrate to promote reentrant arrhythmias (Surawicz, 1989).

Ca²⁺ overload and a spontaneous rise of intracellular free Ca²⁺ (${\text{Ca}}_{\text{i}}^{2+}$) have been implicated in the long QT syndrome (LQTS) as mechanisms that trigger EADs and TdP (for review see Volders et al. 2000). AP prolongation may increase Ca²⁺ entry via L-type Ca²⁺ current (I_Ca,L) during the long plateau phase causing an excessive accumulation of Ca²⁺ in the sarcoplasmic reticulum (SR) and spontaneous SR Ca²⁺ release (Szabo et al. 1995; Verduyn et al. 1995; Patterson et al. 1997; Volders et al. 1997, 2000). The elevation of cytosolic Ca²⁺ (${\text{Ca}}_{\text{i}}^{2+}$) may depolarize myocytes via Ca²⁺-dependent chloride currents and/or the electrogenic Na⁺-Ca²⁺ exchange current (I_Na/Ca) thereby eliciting EADs. Therefore, can reach a level sufficient to cause an inward current and suprathreshold depolarization, resulting in the firing of EADs, impulse propagation and the initiation of an arrhythmia.

Several lines of evidence suggest that the spontaneous release of Ca²⁺ from the sarcoplasmic reticulum (SR) triggers EADs. (1) EADs often appear jointly with delayed afterdepolarizations, in conditions that favour spontaneous SR Ca²⁺ release (Volders et al. 1997). (2) An increase in adrenergic tone enhances Ca²⁺ entry in cardiac cells, increasing SR Ca²⁺ load and the incidence of EADs (Volders et al. 1997). (3) Experimental conditions that reduce I_Na/Ca also inhibit EADs suggesting that the Ca²⁺-dependent I_Na/Ca serves as the inward current that triggers EADs (Szabo et al. 1995). (4) The blockade of SR Ca²⁺ release channels or ryanodine receptors prevents the induction of EAD-dependent TdP (Verduyn et al. 1995). (5) EADs are prominently observed in hearts treated with nifedipine, which blocks I_Ca,L, suggesting that the I_Ca,L window current is not a likely trigger of EADs and that SR Ca²⁺ release and I_Na/Ca elicit EADs (Patterson et al. 1997).

Given the likely importance of ${\text{Ca}}_{\text{i}}^{2+}$ as a trigger of EADs and VT, several studies measured both ${\text{Ca}}_{\text{i}}^{2+}$ and voltage in isolated myocytes. Volders et al. (1997) reported that the elevation of ${\text{Ca}}_{\text{i}}^{2+}$ measured with the Ca²⁺ indicator fura-2 AM and the shortening of canine myocytes preceded the upstroke of EADs. Using Ca²⁺ indicator dyes and video imaging techniques, the spontaneous rise of ${\text{Ca}}_{\text{i}}^{2+}$ associated with delayed afterhyperpolarizations (DADs) was found to originate from the centre of myocytes; in contrast, with EADs ${\text{Ca}}_{\text{i}}^{2+}$ appeared to be released synchronously throughout the cell. This suggests that different mechanisms underlie the initiation of DADs and EADs (Miura et al. 1993, 1995; De Ferrari et al. 1995). An acknowledged limitation of these studies was the slow sampling rate of the camera (30 frames sec⁻¹), which obscures fast changes in and might fail to detect a rise of ${\text{Ca}}_{\text{i}}^{2+}$ preceding the upstroke of an EAD (Miura et al. 1995).

Isolated myocyte studies are limited by the lack of intercellular coupling. Cell-cell coupling in the myocardium may suppress or facilitate EADs as shown in a study of Purkinje-ventricular cell interactions using a rabbit Purkinje fibre coupled to a passive model cell (Huelsing et al. 2000). In the latter study, an intercellular resistance as small as 297 MΩ suppressed EADs in Purkinje fibres because of electrotonic interactions with surrounding ventricular myocytes. Conversely, when cell-cell coupling was reduced (as might occur in infarcted hearts), the reduced Purkinje- ventricular coupling promoted EADs in Purkinje fibres and led to arrhythmias.

Although the general consensus is that EADs are closely associated with TdP, whether reentry or focal activity underlies TdP remains controversial. The original hypothesis of Dessertenne (1966) is that two or more automatic foci firing impulses at slightly different frequencies are responsible for TdP. This was substantiated experimentally when stimulation of the right and left ventricles at slightly different rates produced ECG patterns classically associated with TdP (D'Alnoncourt et al. 1982). An alternative explanation of TdP is that the QT prolongation is a manifestation of prolonged APDs and that an enhanced dispersion of repolarization is required to increase the vulnerability to reentrant arrhythmias (Kuo et al. 1985). Reentry occurs when a premature impulse is blocked upon encountering a region of refractory tissue but conducts though regions with short refractory periods (unidirectional propagation). In this mechanism, EADs launched primarily by Purkinje or M-cells serve merely to initiate reentry (El-Sherif et al. 1988; Antzelevitch et al. 1998). Reentrant waves (or spirals) may be unstable due to prolonged APDs (Starmer et al. 1995) or dispersion of repolarization (Abildskov & Lux, 1993, 1994, 2000). Optical mapping studies using a CCD camera showed that reentrant waves can meander across the heart, resulting in undulating ECG waves (Gray et al. 1995).

Despite the importance of ${\text{Ca}}_{\text{i}}^{2+}$ as a trigger of VT, no studies have measured both ${\text{Ca}}_{\text{i}}^{2+}$ and voltage at high spatial and temporal resolution and no data are available to compare the kinetics of APs or ${\text{Ca}}_{\text{i}}^{2+}$ transients during EADs in intact hearts where cells are electrically coupled.

We investigated the role of ${\text{Ca}}_{\text{i}}^{2+}$ handling in the initiation of EADs and TdP by simultaneously mapping voltage and ${\text{Ca}}_{\text{i}}^{2+}$ in a rabbit model of long QT. Hearts were stained with a voltage-sensitive dye (RH237) and loaded with a Ca²⁺ indicator dye (rhod-2 AM) and both parameters were simultaneously mapped with two 16 × 16 photodiode arrays viewing the anterior surface of the hearts. As previously shown, simultaneous recordings of voltage and ${\text{Ca}}_{\text{i}}^{2+}$ from multiple (252) sites makes it possible to accurately measure the temporal relationship between the membrane depolarization and the release of Ca²⁺ from the SR and their interaction during electromechanical alternans (Choi & Salama, 2000). The technique was applied here to investigate the role of dispersion of repolarization and alternans overload in the initiation of LQT-related arrhythmias. A preliminary report of this work has been presented in abstract form (Choi & Salama, 1999).

Methods

This investigation conformed to the current Guide for Care and Use of Laboratory Animals published by the National Institutes of Health (NIH Publication No. 85-23, revised 1996).

Heart preparations

New Zealand White rabbits (female, 1.6-2.5 kg) were injected with pentobarbital (35 mg kg⁻¹, i.v.) plus heparin (200 U kg⁻¹), then the heart was excised and retrogradely perfused through the aorta with (mm): 130 NaCl, 24 NaHCO₃, 1.0 MgCl₂, 4.0 KCl, 1.2 NaH₂PO₄, 20 dextrose, 1.25 CaCl₂, at pH 7.4, gassed with 95 % O₂-5 % CO₂. Temperature was maintained at 37.0 ± 0.2 °C and perfusion pressure was adjusted to ≈70 mmHg with a peristaltic pump (Choi & Salama, 2000). The atrioventricular node was ablated with an injection of formaldehyde (37 % solution) to slow down the heart rate. The heart was stained with a voltage-sensitive dye (RH 237 (Molecular Probes, Eugene, OR, USA), 10 μl of 1 mg ml⁻¹ in dimethyl sulfoxide, DMSO) and was loaded with a Ca²⁺ indicator (rhod-2 AM (Molecular Probes), 300 μl of 1 mg ml⁻¹ in DMSO) delivered through the bubble trap, above the aortic cannula. Pseudo-ECGs and perfusion pressure were continuously monitored to insure that the dyes did not produce lasting pharmacological effects. Hearts were placed in a chamber to reduce movement artifacts. Long QT and TdP were induced without electrical stimulation by adding the I_Kr blocker, E4031 (0.5 μM) and lowering K⁺ and Mg²⁺ concentrations in the perfusate by 50 % (Zabel et al. 1997; Eckardt et al. 1998).

Optical apparatus

The optical apparatus (Fig. 1A) was previously described (Choi & Salama, 2000). Briefly, light from two 100 W tungsten-halogen lamps was collimated, passed through 520 ± 30 nm interference filters and centred on the anterior surface of the heart. Fluorescence emitted from the stained heart was collected with a camera lens, passed through a 45 deg dichroic mirror (630 nm, Omega Optical, Brattleboro, VT, USA) to reflect the rhod-2 fluorescence while transmitting the RH 237 fluorescence. Fluorescence images from each dye were refocused on two 16 × 16 photodiode arrays (Hamamatsu Corp., Hamamatsu City, Japan; no. C4675-103) such that each diode on the ‘voltage array’ was in exact register with a diode on the ‘calcium array’. Outputs from the arrays were amplified, digitized at 2000 frames s⁻¹ and stored in computer memory. Surface electrograms and stimulus pulses were simultaneously recorded with optical signals.

Figure 1. Optical apparatus and symbolic maps of APs and ${\text{Ca}}_{\text{i}}^{2+}$.

A, light from two 100 W tungsten-halogen lamps was collimated, passed through 520 ± 30 nm interference filters, and focused on the heart. Fluorescence from the stained heart was collected by a camera lens and passed through a dichroic mirror to split emission wavelengths below and above 645 nm. Wavelengths below 645 nm were passed through a 585 ± 20 nm interference filter and those above through a 715 nm cutoff filter and the two images of the heart were focused on two 16 × 16 photodiode arrays. B, diagram shows the anterior surface of the rabbit heart and the region of the heart viewed by the array throughout the study. The top edge of the array is aligned with the base of the heart; the bottom edge with the apex. RA: right atrium, RV: right ventricle, LA: left atrium, LV: left ventricle. C and D, traces from two photodiode arrays. Each box represents a diode on the array with APs and transients drawn in their respective location on the V_m and arrays. Signals were recorded from a female rabbit heart perfused with E4031 (0.5 μM).

Data analysis

Activation time at each site was determined from the local AP upstroke, (dF_v/dt)_max. APD at each site was the interval from (dF_v/dt)_max to the inflection point of the AP downstroke, (d²F_v/dt²)_max which has been shown to coincide with ≈97 % repolarization to baseline and recovery from refractoriness (Efimov et al. 1994). Isochronal maps of activation and repolarization were generated as previously described (Choi & Salama, 2000). However, the rise of fluorescence during an EAD upstroke was too slow to map propagation using the first derivative technique. Instead, the cross correlation of EAD upstrokes were calculated between pixels and the time lag at maximum cross correlation was used to map the propagation of EADs. Diodes detecting pronounced movement artifacts were excluded from the analysis.

Estimates of membrane potential (V_m) and ${\text{Ca}}_{\text{i}}^{2+}$

The fractional fluorescence changes (ΔF_v/F_v) of voltage-sensitive dyes have been shown to vary linearly with changes in V_m (Salama & Morad, 1976). V_m(t) was deduced from fluorescence (F_v) APs:

(1)

where F_v,min and F_v,max are minimum and maximum fluorescence. The equation is based on the assumption that V_m at diastole is equal to −80 mV and at the peak of the upstroke V_m = +20 mV.

Fluorescence signals from rhod-2 (F_Ca) were used to estimate ${\text{Ca}}_{\text{i}}^{2+}$, as previously described (Del Nido et al. 1998) using the following equation:

(2)

where K_d = 710 nm, F_Ca,min is the fluorescence of the free dye in zero Ca²⁺, F_Ca,max is the fluorescence of the dye-Ca²⁺ complex saturated with Ca²⁺. F_Ca,max was measured after perfusing the heart with the Ca²⁺ ionophore A23187 (10 μM), dithiodipyridine (100 μM to trigger SR Ca²⁺ release) and high external free Ca²⁺ (5 mm), in the presence of diacetyl monoxime (DAM, 15 mm) to block movement. ${\text{Ca}}_{\text{i}}^{2+}$ calibrations before and after cryoablation were not possible because the method requires that the position of the heart remains fixed relative to the optical apparatus throughout the experiment. Thus, instead of direct measurements of F_Ca,max, the peak ${\text{Ca}}_{\text{i}}^{2+}$ value during ${\text{Ca}}_{\text{i}}^{2+}$ transients was set to 1100 nm (obtained from calibrations of non-cryoablated hearts) and free Ca²⁺ was estimated from this arbitrarily derived F_Ca,max value.

Two groups of hearts were studied: (1) hearts (n = 24) treated with E4031 first in normal Tyrode solution then with half the normal concentration of K⁺ and Mg²⁺ to induce LQT; and (2) hearts (n = 4) where the endocardium, Purkinje fibres of the conductile system and the septum were cryoablated with liquid nitrogen to ensure that EADs could originate only from the ≈1 mm-thick layer of surviving epicardium, as previously described (Allessie et al. 1989). Briefly, two glass tubes were inserted in the left and right ventricular chambers to fill the cavities with liquid N₂ for 4 min and 30 s, respectively. At the end of each experiment, the heart was perfused with tetrazolium solution to stain and delineate the layer of surviving epicardium. Hearts were sectioned transversely from apex to base and digital images were taken of all the sections to measure the thickness of the surviving epicardial layer (≈1 mm thick).

Pilot studies were carried out to determine the concentration of E4031 that produced the maximum prolongation of APDs (n = 3). DAM (5 mm, n = 3) or cytochalasin D (2 μM, n = 4) was added to the perfusate to block contractions, test their effects on EADs and to ensure that ${\text{Ca}}_{\text{i}}^{2+}$ and V_m during EADs were devoid of movement artifacts.

Results

Effects of I_Kr blockers on AP and transients

Figure 1B illustrates the anterior surface of the heart mapped by the voltage and calcium arrays. Unless stated, the same region of the heart was observed throughout this study. After labelling the heart with RH237 and rhod-2 AM, maps of APs and ${\text{Ca}}_{\text{i}}^{2+}$ transients were simultaneously recorded. Maps of voltage (Fig. 1C) and calcium (Fig. 1D) transients are shown for a heart perfused with E4031 (0.5 μM) where each AP and ${\text{Ca}}_{\text{i}}^{2+}$ transient recorded by a diode is drawn in its respective location. Each diode on the ‘voltage array’ is aligned with and is in register with a diode on the ‘calcium array’ that views the same area of epicardium (Choi & Salama, 2000). Perfusion with E4031 for 10–15 min prolonged APDs in the range of 2–3 s but failed to elicit EADs or to alter the rise times and delays between ${\text{Ca}}_{\text{i}}^{2+}$ and AP upstrokes. However, after the AP upstroke, ${\text{Ca}}_{\text{i}}^{2+}$ declined to a lower steady-state level (30-50 % of peak ${\text{Ca}}_{\text{i}}^{2+}$) while the voltage remained depolarized during the plateau phase.

Figure 2 compares the time course and shape of APs and ${\text{Ca}}_{\text{i}}^{2+}$ transients recorded from the base (top traces) or apex (bottom traces) of the heart before (panel A) and after (panel B) E4031 for ≈2 min. The addition of E4031 (0.5 μM) prolonged APDs from 277 ± 65 to 2115 ± 67 ms, with a diastolic interval of ≈500 ms (n = 14 hearts). Such long APDs were routinely recorded after perfusion with E4031 (n = 24) but the I_Kr blocker also produced a marked prolongation of cycle length making it difficult to compare APDs before and after E4031 at the same cycle length. The diastolic interval in control heart can be longer than a second but during E4031 perfusion, diastolic interval showed marked differences at the same basic cycle length. However, comparisons of APDs were made at similar diastolic intervals (≈500 ms) to ensure that the time to recover from the previous AP was the same. Although E4031 prolonged APD by more than 2 s, ${\text{Ca}}_{\text{i}}^{2+}$ and EAD-like V_m oscillations and VT were not observed (n = 22/24 hearts), except in two hearts where extra beats fired around the late repolarization phase without progressing to salvos of depolarization or VT.

Figure 2. APD and ${\text{Ca}}_{\text{i}}^{2+}$ prolongation by E4031.

A, superposition of APs and transients recorded from the base (top traces) and apex (bottom traces) of the heart. B, recordings from the same sites as in A, but after perfusion with 0.5 μM E4031. APDs prolonged to 1100 ms in the presence of E4031. ${\text{Ca}}_{\text{i}}^{2+}$ recovered partially after the AP upstroke then remained elevated during the plateau phase. C, superposition of AP and ${\text{Ca}}_{\text{i}}^{2+}$ transients at fast sweep speed to compare the kinetics of the two upstrokes. The sampling rate (0.5 ms) was sufficiently fast to quantitatively measure the rise times of APs and ${\text{Ca}}_{\text{i}}^{2+}$. V_m traces are drawn as continuous lines and ${\text{Ca}}_{\text{i}}^{2+}$ traces as dotted lines.

In Figure 2C, expanded traces of AP and ${\text{Ca}}_{\text{i}}^{2+}$ show that the high temporal resolution (0.5 ms) made it possible to capture the detailed kinetics and relative delays between AP and ${\text{Ca}}_{\text{i}}^{2+}$ upstrokes. In rabbit hearts, the rise times of AP and ${\text{Ca}}_{\text{i}}^{2+}$ upstrokes were 8.0 ± 1.4 and 25.8 ± 5.7 ms, respectively (n = 6). ${\text{Ca}}_{\text{i}}^{2+}$ followed APs with a delay of 9.2 ± 1.4 ms (n = 6), similar to the delay measured in guinea-pig hearts (Choi & Salama, 2000). Here, V_m and ${\text{Ca}}_{\text{i}}^{2+}$ signals were measured from 0.9 mm × 0.9 mm areas of epicardium and represent the spatial average of hundreds of cells.

Most interesting was the time course of ${\text{Ca}}_{\text{i}}^{2+}$ and APs after perfusion with E4031. The ${\text{Ca}}_{\text{i}}^{2+}$ upstroke rose 8.6 ± 1.9 ms after the AP upstroke (a similar delay as that before E4031) then ${\text{Ca}}_{\text{i}}^{2+}$ followed a complex time course during the long plateau phase (Fig. 2B). (1) After reaching a peak, ${\text{Ca}}_{\text{i}}^{2+}$ declined by ≈50 % during the early phase of the AP. (2) ${\text{Ca}}_{\text{i}}^{2+}$ then rose to a new steady-state of variable amplitude in different regions of the heart and remained at this level. Steady-state V_m and ${\text{Ca}}_{\text{i}}^{2+}$ during the plateau phase were estimated (see Methods) to be ≈0.0 mV and 710 nm, respectively. (3) Finally, ${\text{Ca}}_{\text{i}}^{2+}$ recovered back to baseline after the AP downstroke. The kinetics of V_m and ${\text{Ca}}_{\text{i}}^{2+}$ indicated that they remained tightly coupled during long APs.

To determine the spatial heterogeneity of I_Kr on the epicardium, the spatial distribution of APDs was measured before and after E4031. In sinus rhythm, activation traversed the epicardium in ≈4 ms (Fig. 3A) and repolarization began near the apex of the heart and spread towards the base (Fig. 3B). Similar findings were reported for guinea-pig hearts (Salama et al. 1987; Kanai & Salama, 1995). The repolarization pattern was the result of an intrinsic gradient of APDs, which were 12 ± 3 ms shorter at the apex compared to the base (n = 14). In the presence of the I_Kr blocker E4031, activation did not change, remaining as shown in Fig. 3A. However, the direction of repolarization was reversed: repolarization started from the base and slowly spread to the apex region (Fig. 3C). This reversal was due to the greater prolongation of APD by E4031 at apical vs. basal sites (Fig. 3C), leading to a dramatic gradient of APDs, with ΔAPD = 152 ± 47 ms (n = 6). The reversal of gradients of APDs and of repolarization was observed with other selective inhibitors of I_Kr such as d-sotalol consistent with a heterogeneous distribution of I_Kr on the epicardium (Fig. 3D).

Figure 3. Maps of activation and repolarization ± E4031.

A, activation map from spontaneously beating heart (contour lines are 2 ms apart). Sites of earliest activation are coded as white and latest as black. B, repolarization map of AP (2 ms isochrones) from a control heart. Repolarization begins at the apex and spreads basally towards the left and right ventricles. APDs are in the range of 175 (white) to 197 (black) ms. C, repolarization map of the heart shown in B after perfusion with E4031 for ≈2 min. The repolarization gradients become considerably steeper (isochrones are 10 ms apart). APDs at the apex are markedly prolonged and APDs are in the range of 954 (white) to 1100 (black) ms. The dispersion of repolarization is enhanced and reversed going from base to apex. D, repolarization map after perfusion with d-sotalol (100 μM; Bristol-Myers Squibb, New York). Blockade of I_Kr with 100 μM d-sotalol produced a similar prolongation of APDs and reversal of repolarization as with E4031 (panel C) indicating that other selective blockers of the current can expose the functional gradient of I_Kr.

The superposition of APs recorded from a column of 14 diodes running from base to apex showed that E4031 produced a systematic spread of APDs, with shorter APs at the base and longer APs at the apex (Fig. 4A). By superposing APs and their first derivative (dF/dt) from the base and apex, the rate of repolarization was found to be considerably slower at the base than the apex (Fig. 4B). Plotting mean rate of repolarization as a function of position from base to apex revealed a gradual increase in downstroke velocity (|-dF/dt|) in going from the base (0 = first diode in a column) to the apex (15 = last diode in the column) (see Fig. 1B for orientation). The reversal of repolarization pattern and the systematic change in downstroke velocity from base to apex caused by E4031 were highly reproducible (n = 6/6) indicating a functional gradient of the delayed rectifying current, I_Kr across the epicardium, from high at the apex to low at the base.

Figure 4. Spatial heterogeneities of AP downstroke.

A, APs from a column of 16 diodes are superimposed after aligning their upstrokes (dV_m/dt)_max, in time. After perfusion with E4031 (0.5 μM), there is a systematic prolongation of APDs from base to apex. B, two APs (top traces), one from the apex, the other from the base are superimposed and their first derivative (bottom traces) are used to compare the rate of repolarization (R_rep) between sites on the apex and the base. Repolarization at the apical site is both faster and more delayed compared to that at the basal site. C, rate of repolarization as a function of location along a vertical axis oriented base to apex. The rate of repolarization was measured (as in panel B) for six columns oriented along the base to apex axis. The mean rate of repolarization, R_rep (± s.d.) was calculated for each of the 16 rows (or diodes) of APs in each column, normalized relative to the rate of rise of the AP upstroke and plotted as a function of location, base to apex. In hearts treated with E4031, the rate of repolarization -d V_m/dt was almost twice as fast at the apex than the base.

Despite remarkably long APDs (> 2 s), no EADs or TdP were observed. Furthermore, EADs could not be induced by applying electrical shocks during the plateau phase (1.5 to 2 × threshold) or by burst pacing at a cycle length of 25 or 50 ms (not shown).

V_m and ${\text{Ca}}_{\text{i}}^{2+}$ oscillations induced by E4031 in low [K⁺]_o and [Mg²⁺]_o

In a rabbit model of long QT induced by d-sotalol, lowering the extracellular concentration of K⁺ and Mg²⁺ by 50 % was required to induce EADs and polymorphic VTs (Zabel et al. 1997; Eckardt et al. 1998). In this E4031 model of long QT, lowering K⁺ and Mg²⁺ in hearts perfused with E4031 (0.5 μM) elicited EAD-like voltage oscillations which progressed from one or two spikes per AP plateau to salvos > 15 then to VT. Figure 5A illustrates typical V_m and ${\text{Ca}}_{\text{i}}^{2+}$ traces, after 5 min of perfusion with E4031 (0.5 μM) at low K⁺ and Mg²⁺.

Figure 5. Effect of cryoablation on the incidence of V_m and oscillations.

A, traces of V_m and are superimposed for two diodes recording from the same site on a heart, which was perfused with E4031 and low [K⁺]_o and [Mg²⁺]_o. Initially, APs plateaux were simply prolonged; later brief (3-4) and long (10-15) salvos of V_m oscillations appeared during each plateau phase. B, comparable V_m and ${\text{Ca}}_{\text{i}}^{2+}$ traces were recorded after cryoablation of the right and left ventricles, abolishing electrical activity in the endocardium and Purkinje fibres and eliminating these cells as a possible source of EADs (n = 6). Cryoablation did not change the effects of E4031and low K⁺ and Mg²⁺ on the incidence of EADs or the vulnerability to LQT-related arrhythmias. C, at the end of experiments, hearts were perfused with tetrazolium solution to stain the live tissue and delineate the surviving region of epicardium. Note the narrow region of surviving epicardium and the effective ablation of the conductile system and the endocardium. In addition, most of the midwall was ablated implying a significant reduction in overall M-cell activity. V_m traces are black and ${\text{Ca}}_{\text{i}}^{2+}$ traces are red.

Incidence and kinetics of EADs in cryoablated hearts

EAD-like oscillations recorded at the surface of the heart result from EADs originating in deeper cells such as Purkinje fibres, endocardial or M-cells and propagating transmurally to the epicardium. Such V_m oscillations are not necessarily EADs. They could arise from the firing of short durations APs at sites below the surface that break through to the epicardium producing salvos of depolarizations on the plateau phase of long APs. The likelihood that deeper cells were the source of EADs or EAD-like oscillations was investigated by cryoablation of the heart resulting in a thin layer (1-2 mm thick) of surviving epicardium. In cryoablated hearts, EADs and APs could not be initiated by Purkinje or endocardial cells but only by surviving epicardial cells. Figure 5C shows a rabbit heart that was sectioned and stained with tetrazolium following the cryoablation of the right and left ventricular chambers. As evident from cross sections of the heart taken from base to apex, the endocardium and the Purkinje fibres are effectively ablated. In cryoablated hearts perfused with E4031, low [K⁺]_o and [Mg²⁺]_o, V_m and ${\text{Ca}}_{\text{i}}^{2+}$ had similar kinetics and the incidence of EADs (Fig. 5B) was the same in control (3.5 ± 1.9 Hz; n = 4) and cryoablated hearts (3.4 ± 0.35 Hz; n = 6). Therefore, in this model of drug-induced LQT, V_m oscillations are likely to arise from EADs originating from ventricular cells at or near the surface since they can no longer arise from deeper layers. The similarity of V_m oscillations in control and cryoablated hearts suggests that in both cases they reflect the firing of EADs rather than short duration APs originating from below the surface.

Possible effects of movement artifacts on voltage and kinetics

A concern regarding measurements of V_m and ${\text{Ca}}_{\text{i}}^{2+}$ kinetics and their temporal relationship during EADs is that oscillations of muscle contractions during long AP plateaux could produce movement artifacts that alter the time course of V_m and ${\text{Ca}}_{\text{i}}^{2+}$ signals. Simultaneously recorded electrograms confirm that optically measured V_m depolarizations were coincident with ECG signals. Pilot studies tested two chemical uncouplers of excitation- contraction, cytochalasin D (cyto-D, 2–10 μM, n = 4) and diacetyl monoxime (DAM, 5–15 mm, n = 3) in the perfusate to block contractions and investigate the effects of uncouplers on the incidence of V_m oscillations. DAM at 5 mm did not fully block movement yet it abolished V_m oscillations and polymorphic VTs (n = 3). In contrast, cyto-D (2 μM) blocked more than 90 % of force generation without significantly changing the incidence of V_m oscillations and polymorphic VTs (n = 14). As shown in Fig. 6B, DAM abolished the incidence of V_m oscillations elicited by E4031 and low [K⁺]_o and [Mg²⁺]_o whereas cyto-D had little effect on APDs, the incidence and kinetics of V_m and ${\text{Ca}}_{\text{i}}^{2+}$ oscillations (Fig. 6C).

Figure 6. Effect of chemical uncouplers on the incidence of EADs.

A, an AP recorded at a single site from a heart treated with E4031 (0.5 μM) and low external K⁺ and Mg²⁺. B, APs recorded as in A, but in the presence of DAM (5 mm). DAM reduced the action potential duration and abolished the firing of EADs. C, AP measured as in A, but after perfusion with 2 μM cytochalasin D, which had minimal effect on either APD or the incidence of EADs.

Kinetic relationship between ${\text{Ca}}_{\text{i}}^{2+}$ and V_m in cryoablated hearts

Figure 7 illustrates a set of simultaneous V_m and ${\text{Ca}}_{\text{i}}^{2+}$ traces recorded after 1, 2 and 5 min of perfusion with E4031 (0.5 μM) at low K⁺ and Mg²⁺. Within 1 min, APDs increased (> 1 s) with V_m and exhibiting a complex multiphasic time course (A). In phase 1, the AP upstroke was followed by a rapid rise in ${\text{Ca}}_{\text{i}}^{2+}$, as expected for a normal AP and ${\text{Ca}}_{\text{i}}^{2+}$ transient. In phase 2, even though V_m remained in the plateau phase, ${\text{Ca}}_{\text{i}}^{2+}$ decreased rapidly towards baseline. In phase 3, ${\text{Ca}}_{\text{i}}^{2+}$ drifted up, perhaps thwarting the full repolarization of the AP. In phase 4, a slight depolarization was associated with a slight decline of ${\text{Ca}}_{\text{i}}^{2+}$. In phase 5, a more abrupt rise of V_m and ${\text{Ca}}_{\text{i}}^{2+}$ occurred synchronously, which could be an EAD. In phase 6, the AP downstroke was associated with the recovery of ${\text{Ca}}_{\text{i}}^{2+}$ to baseline. After 2 min, V_m and ${\text{Ca}}_{\text{i}}^{2+}$ oscillations of variable amplitudes were present in the plateau phase. After 5 min, the plateau phase of each AP contained salvos of rapid depolarization that were followed by rapid rises of ${\text{Ca}}_{\text{i}}^{2+}$ (n = 5/6). The combination of E4031 plus low [K⁺]_o and [Mg²⁺]_o consistently produced salvos of EADs or rapid depolarization riding on top of plateau phases lasting 1–3 s.

Figure 7. Time course of V_m and ${\text{Ca}}_{\text{i}}^{2+}$ oscillation elicited by E4031 in low K⁺ and Mg²⁺.

In hearts perfused with E4031, low K⁺ and Mg²⁺, oscillations of V_m and ${\text{Ca}}_{\text{i}}^{2+}$ developed in a time dependent manner in both normal (not shown) and cryoablated hearts. In this cryoablated heart, long APDs plateaux with 1–2 EADs appeared within 1 min (A), oscillations increased to 3–4 EADs per plateau in 2 min (B) and to 10–15 salvos of EADs in 5 min (C). While both V_m and ${\text{Ca}}_{\text{i}}^{2+}$ oscillated, ${\text{Ca}}_{\text{i}}^{2+}$ had consistently higher amplitudes of oscillation implying that ${\text{Ca}}_{\text{i}}^{2+}$ overload in the SR network and spontaneous SR Ca²⁺ release generated the large Ca²⁺ oscillation. V_m traces are drawn above ${\text{Ca}}_{\text{i}}^{2+}$ traces.

A consistent observation was that at the origins of EADs, the rise of ${\text{Ca}}_{\text{i}}^{2+}$ preceded the voltage rise (n = 6/6). At sites more distant from the origin of the EAD, the rise of ${\text{Ca}}_{\text{i}}^{2+}$ became synchronous then followed the EAD upstroke.

Figure 8 compares the shift in the kinetic relationship between V_m and ${\text{Ca}}_{\text{i}}^{2+}$ during EADs recorded from the origin of the EAD and more distant sites. V_m and ${\text{Ca}}_{\text{i}}^{2+}$ traces were normalized and the time-base was reduced to compare the rising phases of voltage and ${\text{Ca}}_{\text{i}}^{2+}$. Figure 8A shows a recording of V_m and during an EAD (top trace) and the corresponding electrical depolarizations recorded with bipolar electrogram (bottom trace). EAD activation patterns were obtained by mapping the time-lag that produced the maximum cross correlation between an EAD at one site and EADs at every other site. Time-lag cross correlation maps identified the origin and activation sequence of EADs (panel B). Panels C-E show traces of V_m and ${\text{Ca}}_{\text{i}}^{2+}$ from three sites, the origin of EADs (panel E) and more distant sites (panel D then C). At the origin of the EAD (panel E), precedes V_m suggesting that spontaneous ${\text{Ca}}_{\text{i}}^{2+}$ elevation caused the depolarization. At more distant sites (>3 mm), V_m and ${\text{Ca}}_{\text{i}}^{2+}$ became synchronous (panel D) and at still farther sites V_m preceded ${\text{Ca}}_{\text{i}}^{2+}$ (panel C) suggesting that EADs were actively propagating.

Figure 8. Time delay between voltage and during EADs.

A, V_m, ${\text{Ca}}_{\text{i}}^{2+}$ and bipolar electrograms recorded from a cryoablated heart perfused with E4031, low K⁺ and Mg²⁺. B, activation map of the EAD labelled with an arrow in panel A was determined from the time-lag that produces the maximum spatial cross correlation of EADs over the array (252 pixels). C and D, EAD and ${\text{Ca}}_{\text{i}}^{2+}$ traces are superimposed to relate temporally the rise ${\text{Ca}}_{\text{i}}^{2+}$ to the upstroke of the EAD measured at sites away from the origin of the EAD. At those sites, V_m is synchronous with ${\text{Ca}}_{\text{i}}^{2+}$ (panel D) or precedes ${\text{Ca}}_{\text{i}}^{2+}$ (panel C). E, voltage and ${\text{Ca}}_{\text{i}}^{2+}$ traces recorded at the first site to fire the EAD. Here, the rise of ${\text{Ca}}_{\text{i}}^{2+}$ precedes the rise of V_m. V_m traces are black and ${\text{Ca}}_{\text{i}}^{2+}$ traces are red.

To better visualize the temporal relationship between V_m and ${\text{Ca}}_{\text{i}}^{2+}$, phase plots of ${\text{Ca}}_{\text{i}}^{2+}$vs. V_m were generated for control APs, after perfusion with E4031 and at the origin of an EAD (DuBell et al. 1991; Schlotthauer & Bers, 2000). Figure 9A shows a set of three APs and transients from a normal cardiac beat and plots their phase map. In a normal beat, ${\text{Ca}}_{\text{i}}^{2+}$ vs. V_m phase maps had a counterclockwise trajectory in the shape of a ‘boomerang’ that delineated five phases (labelled a-e) of the ${\text{Ca}}_{\text{i}}^{2+}$-V_m relationship. Starting the trajectory from the resting potential at point a, the AP upstroke fires first preceding the rise of ${\text{Ca}}_{\text{i}}^{2+}$. V_m reaches a maximum at point b and ${\text{Ca}}_{\text{i}}^{2+}$ reaches its maximum at point c. The early phase of the AP plateau (c-d) falls below the centre line (dotted line) on the phase map indicating that the slope of recovery is steeper than that for V_m. At point d, repolarization during the downstroke becomes steeper than the recovery of ${\text{Ca}}_{\text{i}}^{2+}$ (d-e) and ${\text{Ca}}_{\text{i}}^{2+}$ returns to baseline after V_m has returned to baseline (e-a). The ‘boomerang’ trajectory was highly reproducible in various locations on the epicardium as well as from beat-to-beat and from heart-to-heart. After perfusion with E4031, the phase map of ${\text{Ca}}_{\text{i}}^{2+}$ vs. V_m had an altered trajectory due to the altered time course ${\text{Ca}}_{\text{i}}^{2+}$ during the long AP plateau. AP and ${\text{Ca}}_{\text{i}}^{2+}$ upstrokes followed similar trajectories as in controls (a-b) but ${\text{Ca}}_{\text{i}}^{2+}$ declined while voltage remained depolarized during the plateau phase resulting in an abrupt drop in the trajectory (b-c) then a smooth recovery during the plateau phase (d) and the downstroke (e) (Fig. 9B). In contrast, phase maps plotted during EADs had a clockwise elliptical trajectory near the origin of EADs, (Fig. 9C). The reversed trajectory demonstrated that ${\text{Ca}}_{\text{i}}^{2+}$ preceded V_m at sites that fired EADs first. Slight variations in the trajectories of the phase maps recorded during a salvo of EADs probably reflect the kinetic changes of the cellular milieu at the onset of TdP. At sites more distant from the origins of EADs, the phase maps had the opposite, counterclockwise orientation (not shown) indicating a shift in temporal relationship where V_m preceded ${\text{Ca}}_{\text{i}}^{2+}$.

Figure 9. Phase plots of vs. voltage.

V_m and ${\text{Ca}}_{\text{i}}^{2+}$ were normalized for each set of diodes before plotting phase maps of ${\text{Ca}}_{\text{i}}^{2+}$ vs. V_m. A, phase maps of ${\text{Ca}}_{\text{i}}^{2+}$ vs. V_m, for a heart in sinus rhythm (right panel) are derived from a sequence of three APs and ${\text{Ca}}_{\text{i}}^{2+}$ transients (left panel). V_m and ${\text{Ca}}_{\text{i}}^{2+}$ start at (0, 0) on the phase plot and point (a) on the first AP. The trajectory on the phase map travels to the right (see arrow) at the onset of the AP upstroke or V_m depolarization (point a→b). After a time delay (≈10 ms), the trajectory shifts up along the ${\text{Ca}}_{\text{i}}^{2+}$ axis (b→c) upon the release of Ca²⁺ from the sarcoplasmic reticulum. ${\text{Ca}}_{\text{i}}^{2+}$ levels decrease faster than the decline of V_m during the plateau phase such that the trajectory falls below the centre line (y = x line) (c→d) then V_m recovers to resting potential while ${\text{Ca}}_{\text{i}}^{2+}$ remains slightly elevated (d→e). Normal ${\text{Ca}}_{\text{i}}^{2+}$ vs. V_m phase maps had a counterclockwise trajectory in the shape of a ‘boomerang’. B, phase maps of ${\text{Ca}}_{\text{i}}^{2+}$ vs. V_m after E4031 perfusion. During ${\text{Ca}}_{\text{i}}^{2+}$ and V_m upstrokes, phase maps had a similar ‘boomerang’ shape, as in panel A but during the plateau phase, the trajectory fell rapidly below the centre line (due to fast recovery of ${\text{Ca}}_{\text{i}}^{2+}$), reaching an intermediate value at point c where ${\text{Ca}}_{\text{i}}^{2+}$ and V_m varied slowly. Then during the recovery of V_m and ${\text{Ca}}_{\text{i}}^{2+}$ back to baseline, the trajectory followed the path of c→d→e. C, phase maps of ${\text{Ca}}_{\text{i}}^{2+}$ vs. V_m at the origin of an EAD. During EADs, phase maps have elliptical densely packed trajectories near the centre line. The direction of rotation is clockwise or opposite to the trajectory during a normal AP and ${\text{Ca}}_{\text{i}}^{2+}$ transient. The trajectories of the phase maps indicate that ${\text{Ca}}_{\text{i}}^{2+}$ precedes V_m at the origins of EADs. V_m traces are black and ${\text{Ca}}_{\text{i}}^{2+}$ traces are red.

In non-cryoablated hearts, a kinetic analysis of V_m and ${\text{Ca}}_{\text{i}}^{2+}$ oscillations from normal heart showed that V_m preceded ${\text{Ca}}_{\text{i}}^{2+}$ with a delay ranging from 3 to 8 ms, at various sites on the epicardium (n = 5/6). In one heart, ${\text{Ca}}_{\text{i}}^{2+}$ was found to precede V_m by 8.1 ± 3.4 ms at 3.7 Hz (n = 1/6), implying that most EADs in intact hearts originate from below the surface.

Activation patterns during EADs and polymorphic VT

The nature of long-QT related arrhythmias was investigated by mapping the activation patterns produced during salvos of V_m oscillations and their progression to polymorphic VT. The heart was perfused with E4031 (0.5 μM) plus low [K⁺]_o and [Mg²⁺]_o but was not cryoablated. As shown in Fig. 10A, a long AP with a salvo of depolarizations was recorded along with a bipolar electrogram recording from the posterior surface of the heart. Isochronal maps of activation are displayed for the AP upstroke (1), a salvo of spikes (2-6), the last depolarization (7) and the next AP upstroke (8). The activation map for the AP upstroke (1) during sinus rhythm showed a typical activation pattern across the anterior surface of the heart (4 ms) (1). During the salvo of spikes (2-6), activation maps showed that two sites fire out-of-phase (⋆-). Spikes that fired from one site (labelled ) were synchronous over an increasingly larger area of epicardium (maps 2-6) but failed to propagate to the adjacent tissue (either depolarizations are subthreshold or encountered refractory tissue) resulting in a functional line of block. In contrast, spikes emanating from a second site (⋆) captured and propagated around the zone of depolarized tissue produced by the first spike. The last spike (7) propagated from the base of the right ventricle (top left) to the apex of the left ventricle before the termination of the AP. The activation pattern of the next AP upstroke (8) was again similar to the sinus rhythm pattern recorded in the previous beat (1) with a rapid activation from the right to the left ventricle in 4 ms.

Figure 10. Maps of activation during V_m oscillations and polymorphic VTs.

A, example of V_m and bipolar electrogram (BE) traces from the left ventricle during a salvo of depolarizations from a non-cryoablated heart. B, activation maps of the AP upstroke (spike 1) and the sequence of spikes (2-7) are derived from V_m shown in panel A. The activation pattern of spike 1 is typical for activation on the anterior surface during sinus rhythm. Activation patterns for spikes 2–6 show that there are two sites on the field-of-view that depolarize out-of-phase. The first site to fire is labelled with an open star () and the second with a filled star (⋆). The activation wave originating from the filled star (⋆) spreads and surrounds the region activated by the open star (). In the sequence of spikes labelled 2-6, the area of tissue activated by the depolarization at the open star () enlarges and appears to encompass a larger zone of activation breakthrough while activation originating from the filled star (⋆) becomes increasingly smaller as it encounters depolarized tissue. The large synchronous activation labelled with the open star () suggests that those depolarizations originate below the surface emerging on the epicardium in an increasingly larger zone. By spike 7, depolarization from the filled star (⋆) has disappeared. Spike 8 has the normal activation patterns seen in sinus rhythm. C, V_m and an electrogram (BE) recorded during a polymorphic VT from a cryoablated heart. Oscillations of V_m exhibit undulating phases of low (spikes 1-3) and high (spikes 4-6) amplitude depolarizations typical of TdP. D, sequence of activation maps for spikes 1–6 during the polymorphic VT shown in panel C. Activation maps of low amplitude spikes 1–3 indicate that there are two or three foci that fire out-of-phase (one from the base and 1 or 2 from the apex), spread and collide. Activation maps of the high amplitude spikes exhibit a site with dominant focal activity suggesting that multi-focal activity maintains VT.

In cryoablated hearts perfused with E4031 and low [K⁺]_o and [Mg²⁺]_o for more than 10 min, salvos of EADs during the long plateau phases progressed to polymorphic VT (Fig. 10C). Optical traces of V_m (top left) and bipolar electrograms (bottom left) exhibited the characteristic undulating pattern of TdP. Activation patterns during polymorphic VT are displayed for a sequence of three VT beats during low (1-3) and high amplitude (4-6) depolarization (panels 1-6). Activation maps of beats 1–3 show that activation was initiated from two locations, one at the base and the other at the apex, for three consecutive beats. The activation frequencies at these two sites were calculated by FFT analysis. In this episode recorded in the presence of cyto-D, the higher frequency occurred at the base at 2.86 Hz and the lower frequency was at the apex at 2.45 Hz. The frequency of EADs was highly reproducible at 2.7 ± 0.2 Hz with cyto-D (2 μM) (n = 8) and 3.4 ± 0.4 Hz, in the absence of chemical uncouplers of excitation- contraction (n = 8). The results show that in this model of long QT, polymorphic VT with the characteristics of TdP are generated by out-of-phase, multifocal activity, as originally proposed by Dessertenne (1966).

Discussion

The long QT syndrome (LQTS) is a repolarization disorder characterized by marked prolongation of the QT interval. A hallmark of long QT syndrome is the recurrent syncope during episodes of polymorphic VT, called Torsade de Pointes (TdP) (Dessertenne, 1966). The current consensus is that LQT-related arrhythmias are initiated by the firing of EADs, which in the presence of an enhanced dispersion of repolarization results in the initiation and maintenance of reentry.

In turn, intracellular Ca²⁺ overload due to long APD has been implicated as a trigger of spontaneous Ca²⁺ release from the sarcoplasmic reticulum network causing membrane potential oscillations via Ca²⁺ sensitive currents, I_Na/Ca and/or I_Cl. Moreover, it has been suggested that Purkinje fibres and/or M-cells are the primary source of EADs in the LQTS rather than ventricular cells. Evidence for these hypothetical mechanisms is lacking because of difficulties in measuring voltage and intracellular Ca²⁺ simultaneously in intact hearts. To this end, we applied a novel optical mapping technique to measure transmembrane potential and intracellular Ca²⁺ transients simultaneously from multiple sites on a Langendorff perfused rabbit heart.

Simultaneous maps of AP and ${\text{Ca}}_{\text{i}}^{2+}$ showed that ${\text{Ca}}_{\text{i}}^{2+}$ preceded V_m at sites where the EAD was initiated. As EADs captured and propagated across the epicardium, the rise of ${\text{Ca}}_{\text{i}}^{2+}$ followed V_m upstrokes. Phase maps of ${\text{Ca}}_{\text{i}}^{2+}$ vs. V_m showed that at EAD initiation sites, ${\text{Ca}}_{\text{i}}^{2+}$ increased before voltage suggesting that intracellular Ca²⁺ overload elicited membrane potential depolarizations via Ca²⁺-sensitive currents and the firing of EADs. Cryoablation of the heart demonstrated that the elimination of electrical activity from Purkinje and endocardial cells did not significantly alter the incidence of V_m oscillations or the propensity to polymorphic VT. Thus, ventricular epicardial cells were at least as likely to fire EADs as cells of the conduction system. Salvos of EADs produced undulating ECG patterns associated with TdP, which progressed to VT.

Temporal relationship between V_m and ${\text{Ca}}_{\text{i}}^{2+}$

Two hypotheses have been proposed to explain the cellular mechanisms responsible for triggering EADs: (1) the reactivation of I_Ca,L window current and (2) spontaneous Ca²⁺ release from the sarcoplasmic reticulum. The complex dynamic interaction of ionic fluxes during long APs has been a major obstacle to our understanding of EAD mechanisms because experimental manipulations designed to alter one of mechanisms influences the other. Hence, computer models of EADs were developed to investigate the contribution of ionic currents to EADs. Zeng & Rudy (1995) reproduced the shape and time course of EADs based on the reactivation of I_Ca,L with little contribution from I_Na/Ca. The same mechanism was supported by a simulation of rate-dependent EAD generation (Viswanathan & Rudy, 1999). In contrast, Nordin and his coworkers (Nordin & Ming, 1995; Nordin, 1996, 1997) performed computer simulations with more realistic Ca²⁺ handling equations consisting of two intracellular Ca²⁺ pools to mimic the local control of SR Ca²⁺ release. This model predicted that an elevation of intracellular Ca²⁺ causes an initial depolarizing current via an inward, I_Na/Ca that leads to the activation of the larger I_Na or I_Ca currents and the firing of EADs (Nordin & Ming, 1995; Nordin, 1996, 1997).

Simultaneous measurements of voltage and ${\text{Ca}}_{\text{i}}^{2+}$ have been extensively applied in single cell studies using microelectrodes and Ca²⁺ indicators (Miura et al. 1993; Balke et al. 1994; Cannell et al. 1994; Cheng et al. 1994; Han et al. 1994; Lopez-Lopez et al. 1994; De Ferrari et al. 1995; Miura et al. 1995). We developed an approach to record AP and ${\text{Ca}}_{\text{i}}^{2+}$ simultaneously with two photodiode arrays (Choi & Salama, 2000) using a combination of voltage-sensitive and Ca²⁺ indicator dyes which have similar excitation but different emission spectra so that two fluorescence images could be separated to avoid cross talk between V_m and ${\text{Ca}}_{\text{i}}^{2+}$ recordings.

Movement artifacts due to cardiac contractions can distort optical recordings and may interfere with measurements of the magnitude and kinetics of V_m and ${\text{Ca}}_{\text{i}}^{2+}$. As previously demonstrated (Kanai & Salama, 1995), mounting the perfused heart in a chamber was effective in abating movements due to contractions such that movement artifacts became a minor component compared to the large V_m and ${\text{Ca}}_{\text{i}}^{2+}$ optical responses. Although cryoablation (n = 6) reduced force, movement artifacts could still distort signals recorded from the edges of the preparation. In pilot studies, chemical uncouplers were added to reduce movement artifacts. Diacetyl monoxime, at a concentration too low to eliminate contractions (5 mm) blocked V_m oscillations during long APs (n = 3) and therefore could not be used with this model of LQT. Cyto-D (2 μM) blocked contractions without interfering with the firing of V_m oscillations and polymorphic VTs, except for a slight reduction in frequency (from 3.4 to 2.7 Hz). The temporal relationship between V_m and ${\text{Ca}}_{\text{i}}^{2+}$ during oscillations was essentially the same in the presence or absence of cyto-D in cryoablated hearts, with a tendency to slightly reduce the rate of rise and to prolong the duration of EADs. Movement artifacts also tend to reduce the rate of rise and to prolong the duration of electrical signals such that the elimination of movement artifacts should have produced the opposite effects. The effects of cyto-D on EAD shape and time course gave us further confidence that movement artifacts did not interfere with kinetic measurements of V_m and ${\text{Ca}}_{\text{i}}^{2+}$, during EADs, in the absence of chemical uncoupler. Phase maps of V_mvs. ${\text{Ca}}_{\text{i}}^{2+}$ showed that ${\text{Ca}}_{\text{i}}^{2+}$ precedes V_m during the firing of EADs in cryoablated hearts but only at the sites that initiated EADs. Once an EAD propagated across the epicardium V_m preceded ${\text{Ca}}_{\text{i}}^{2+}$ by 4.7 ± 2.8 ms. This study provides the first direct measurements of the temporal relationship between ${\text{Ca}}_{\text{i}}^{2+}$ and V_m during EADs in perfused hearts. We have shown that the elevation of ${\text{Ca}}_{\text{i}}^{2+}$ precedes the depolarization of EADs indicating that V_m no longer controls ${\text{Ca}}_{\text{i}}^{2+}$, as it does in a normal cardiac beat.

Dispersion of APD by I_Kr blocker, E4031

There is growing evidence that ion channels are heterogeneously distributed in different region of mammalian ventricular muscle. In ferret hearts, immunofluorescence images of the heart revealed an increasing distribution of ERG (the channel protein responsible for the rapid component of the delayed rectifier K⁺ current) from apex to base (Brahmajothi et al. 1997). Consistent findings were observed in voltage-clamp studies on rabbit ventricular myocytes that revealed a greater ratio of I_Kr/I_Ks currents at the apex than at the base (Cheng et al. 1999). APDs of guinea pig hearts have been extensively demonstrated by optical mapping studies (Salama et al. 1987; Efimov et al. 1994; Kanai & Salama, 1995) to be shorter at the apex than the base. However, the spatial distribution of the K⁺ currents responsible for these APD gradients was not investigated. The present study on rabbit hearts showed that APDs are slightly shorter at the apex than the base (Fig. 3B) and that blockade of I_Kr by E4031 (0.5 μM) prolonged APDs to > 1 s, producing a marked enhancement of the base to apex dispersion of repolarization (Fig. 3C). The data demonstrate that there is a functional difference in the level of I_Kr along the epicardium that can be blocked by E4031.

Possible effects of low [K⁺]_o and [Mg²⁺]_o

Normally, V_m must be in the range of −40 to 0 mV to reactivate I_Ca,L window currents (January et al. 1988). This may explain why EADs were absent in hearts treated with E4031 but normal [K⁺]_o and [Mg²⁺]_o, where V_m was estimated (see Methods) to be > 0 mV (see Fig. 2) during the plateau phase. Another possibility is that the reactivating I_Ca,L window current dissipated electrotonically via cell-to-cell coupling and was too low to capture and fire EADs. Alternatively, ${\text{Ca}}_{\text{i}}^{2+}$ oscillations which are required to elicit EADs were not observed in hearts treated with E4031 at normal [K⁺]_o and [Mg²⁺]_o.

In contrast, when E4031 was introduced with low K⁺ and Mg²⁺, hearts exhibited a high incidence of EADs with oscillations of ${\text{Ca}}_{\text{i}}^{2+}$ and V_m (Fig. 5). The exact effects of low K⁺ and Mg²⁺ on ${\text{Ca}}_{\text{i}}^{2+}$ handling are not known. Low K⁺ has been shown to reduce potassium currents and enhance the efficacy of agents that block of K⁺ channels (Sanguinetti & Jurkiewicz, 1992; Yang & Roden, 1996). Mg²⁺ has been known to suppress the amplitude and incidence of EADs (Bailie et al. 1988; Kaseda et al. 1989; Fazekas et al. 1993) through a link to Mg²⁺-dependent channel opening of HERG potassium current (Po et al. 1999).

Origins of EADs

Previous studies have suggested that EADs originate preferentially from Purkinje cells because Purkinje cells have longer APDs and a lower density of repolarizing currents (El-Sherif et al. 1988; Nattel & Quantz, 1988; Carlsson et al. 1992, 1993, 1996; Li et al. 1992). To eliminate the possibility of EADs emanating from Purkinje fibres, hearts (n = 8) were cryoablated and the temporal relationship between V_m and ${\text{Ca}}_{\text{i}}^{2+}$ was investigated. The data showed that the elimination of > 80 % of the tissue including the Purkinje fibres, endocardium and most of the midwall did not alter the incidence of EADs. Therefore, in this model of LQT, ventricular epicardial cells can fire EADs with at least equal likelihood as the specialized fibres of the conductile system. Nevertheless, is not known whether ventricular cells are the primary origin of EADs in intact hearts with functional Purkinje fibres.

The question that remained unanswered is whether or not the rise of ${\text{Ca}}_{\text{i}}^{2+}$ linked to the firing of EADs originates from Ca²⁺ release by subcellular organelles (e.g. the sarcoplasmic reticulum) and precedes the depolarization or, alternatively, the L-type Ca²⁺ current is reactivated during long APD due to a ‘window’ current and follows the depolarization. We addressed this question by recording V_m and ${\text{Ca}}_{\text{i}}^{2+}$ simultaneously from Langendorff perfused hearts using a well-accepted model of long QT that is known to result in EADs and TdP. The data indicate that ${\text{Ca}}_{\text{i}}^{2+}$ triggers EADs because we obtained direct recordings of preceding V_m at sites that fire EADs. Moreover, the rise of could not be attributed to ${\text{Ca}}_{\text{i}}^{2+}$ influx through voltage-gated Ca²⁺ channels because that influx is expected to be too low to be detected by Ca²⁺ indicator dyes. In mammalian hearts, the contribution of Ca²⁺ influx via L-type Ca²⁺ channels is a small percentage of the total ${\text{Ca}}_{\text{i}}^{2+}$ during detected with Ca²⁺ indicators dyes during a contraction. For instance, in a heart treated with ryanodine to deplete Ca²⁺ from the sarcoplasmic reticulum, the rise of ${\text{Ca}}_{\text{i}}^{2+}$ depends on influx through L-type Ca²⁺ channels and is less than 10 % of the normal rise of ${\text{Ca}}_{\text{i}}^{2+}$ (Choi & Salama, 2000). Hence, the elevation of ${\text{Ca}}_{\text{i}}^{2+}$ preceding EADs must be coming from internal stores of Ca²⁺.

We also showed that during an EAD, the rise of ${\text{Ca}}_{\text{i}}^{2+}$ preceded that of the AP indicating that spontaneous SR Ca²⁺ release might trigger EADs via Ca²⁺-dependent currents (e.g. probably I_Na/Ca) in agreement with the Ca²⁺ overload hypothesis. In hearts perfused with E4031 and normal [K⁺]_o and [Mg²⁺]_o, AP prolonged to more than 2 s, with stable ${\text{Ca}}_{\text{i}}^{2+}$ (350-550 nm) and V_m (-10 to +10 mV) values that did not oscillate or elicit the firing of EADs. This indicates that a long APD is not in itself sufficient to elicit EADs.

The role of EADs during Torsade de Pointes

EADs are widely thought to initiate TdP but it is unclear whether subsequent beats of the arrhythmia are the result of rapid firing of EADs from several foci (Dessertenne, 1966; D'Alnoncourt et al. 1982) or from sustained reentrant circuits with different exit points (Coumel et al. 1985; Surawicz, 1989; Haverkamp et al. 1995; El-Sherif et al. 1996; Abildskov & Lux, 2000). The original hypothesis of ‘focal impulse formation’ by Dessertenne was based on two automatic foci firing from opposite sides of the heart with slight different frequencies. Such a mechanism predicted the undulating waveform of the QRS complex, as observed in TdP. D'Alanoncourt et al. (1982) showed that stimulation of the right and left ventricles at slightly different rate generated TdP like pattern of the ECG, which supports the hypothesis of Dessertenne that TdP is caused by the interaction of two ectopic ventricular foci. Another argument against reentry comes from clinical studies, which show that TdP is not generally inducible by programmed electrical stimulation.

Evidence from activation maps in the present study suggests that EADs firing repetitively generate polymorphic VTs. However, this mechanism does not necessarily apply to other models of LQT because the size of rabbit hearts may be too small to sustain several reentrant circuits.

El-Sherif et al. (1996) showed the change in QRS morphology during TdP induced by anthopleurin-A (AP-A) in dog was due to shifting sites of ectopic activity, resulting in varying activation patterns and QRS configurations. Different results between reentry and multifocal activity could be due to the size of hearts, gender and/or species differences and different effects of AP-A and E4031. In another model of LQT, Asano et al. (1997) investigated quinidine and E4031 induced EADs and polymorphic VTs using a CCD camera and found that polymorphic VTs were maintained by beat to beat change of foci firing EADs and occasionally observed meandering spiral waves. Our results show that E4031 produced more pronounced AP prolongation and more stable EAD activation maps (5-10 EADs) than those observed by Asano et al. (1997) but are consistent with the notion that multifocal activity underlies VTs.

Analysis of simultaneous maps of V_m and ${\text{Ca}}_{\text{i}}^{2+}$ has shed new light on the kinetics of intracellular free Ca²⁺ during the plateau phase of APs in LQT2 and during the firing of EADs. I_Kr blockade with E4031 was not sufficient to elicit EADs despite the marked prolongation of the AP and the enhanced dispersion of repolarization. A reduction of external K⁺ and Mg²⁺ concentrations is essential to trigger EADs and VT, as in the sotalol model of LQT2 (Zabel et al. 1997; Eckardt et al. 1998). Cytosolic ${\text{Ca}}_{\text{i}}^{2+}$ overload appears to be responsible for the ${\text{Ca}}_{\text{i}}^{2+}$ elevation and oscillations that precede EAD upstrokes, at the origins of EADs. Variations in the trajectories of the phase maps recorded during a salvo of EADs probably reflect the kinetic changes of the cellular milieu at the onset of TdP. One can speculate that instabilities in phase maps of ${\text{Ca}}_{\text{i}}^{2+}$ vs. V_m during EADs could be related to when and how EADs spontaneously stop firing or conversely convert the heart to TdP. Cryoablation experiments demonstrated that Purkinje fibres are not necessarily the origin of EADs and ventricular cells are also equally likely to initiate EADs and focal EAD activity. These findings are limited to the specific animal model of LQTS elicited by an I_Kr blocker plus low [K⁺]_o and [Mg²⁺]_o in the young female rabbit. Further studies are needed to compare various models of LQTS and to investigate the factors that influence the initiation and maintenance of LQT-related arrhythmias.