Spontaneous calcium release in tissue from the failing canine heart

Abstract

Abnormalities in calcium handling have been implicated as a significant source of electrical instability in heart failure (HF). While these abnormalities have been investigated extensively in isolated myocytes, how they manifest at the tissue level and trigger arrhythmias is not clear. We hypothesize that in HF, triggered activity (TA) is due to spontaneous calcium release from the sarcoplasmic reticulum that occurs in an aggregate of myocardial cells (an SRC) and that peak SCR amplitude is what determines whether TA will occur. Calcium and voltage optical mapping was performed in ventricular wedge preparations from canines with and without tachycardia-induced HF. In HF, steady-state calcium transients have reduced amplitude [135 vs. 170 ratiometric units (RU), P < 0.05] and increased duration (252 vs. 229 s, P < 0.05) compared with those of normal. Under control conditions and during β-adrenergic stimulation, TA was more frequent in HF (53% and 93%, respectively) compared with normal (0% and 55%, respectively, P < 0.025). The mechanism of arrhythmias was SCRs, leading to delayed afterdepolarization-mediated triggered beats. Interestingly, the rate of SCR rise was greater for events that triggered a beat (0.41 RU/ms) compared with those that did not (0.18 RU/ms, P < 0.001). In contrast, there was no difference in SCR amplitude between the two groups. In conclusion, TA in HF tissue is associated with abnormal calcium regulation and mediated by the spontaneous release of calcium from the sarcoplasmic reticulum in aggregates of myocardial cells (i.e., an SCR), but importantly, it is the rate of SCR rise rather than amplitude that was associated with TA.

heart failure (HF) is a serious public health problem that afflicts millions of people in the United States alone (32). HF is associated with impaired cardiac contractility and relaxation, as well as a high incidence of ventricular arrhythmias and sudden death. Abnormal calcium handling has been implicated as a source of both mechanical and electrical dysfunction observed in HF, making it a key target for investigation and clinical therapy.

At the myocyte level, impaired ventricular contractility and relaxation in HF have been attributed to decreased calcium transient amplitude due to diastolic sarcoplasmic reticulum (SR) calcium leak (12, 13, 35) and reduced SR calcium uptake (3, 8, 22, 24). Calcium dysregulation in human HF has been associated with a significant incidence of nonreentrant arrhythmias (28, 30, 33) that can occur as the result of early afterdepolarizations (EADs) or delayed afterdepolarizations (DADs). At the subcellular level, a DAD is caused by spontaneous calcium release from the SR that activates a transient inward current with a magnitude that depends on the amount of calcium flux (33). While studies in isolated myocytes have provided valuable insight into cellular pathophysiology, the translation of these results to arrhythmogenesis at the tissue level is not straightforward.

The factors that determine whether or not a DAD will induce a clinical arrhythmia in the whole heart are likely to be more complex than the factors identified in isolated myocyte studies (see Ref. 7). For example, it is unlikely that a spontaneous calcium release originating from a single cell can overcome the current sink of neighboring cells and generate a DAD that is sufficient to trigger a beat in well-coupled tissue. Rather, it is more likely that the current sink can be overcome by spontaneous calcium release in an aggregate of neighboring myocardial cells, which we have previously termed an SCR (9). We hypothesize that in HF, triggered activity (TA) is due to an SCR event, rather than an action potential arising from spontaneous calcium release within a single myocyte. We also hypothesize that what determines whether or not an SCR event will generate a DAD sufficient to trigger a beat is its peak calcium release (i.e., SCR amplitude). The novel findings of this study indicate that TA in HF is mediated by the spontaneous release of calcium from the SR in a relatively large aggregate of myocardial cells (i.e., an SCR), but surprisingly it is the rate of SCR rise rather than its amplitude that is associated with TA.

MATERIALS AND METHODS

Canine wedge preparation and optical mapping.

All experiments were reviewed and approved by an independent committee (Institutional Animal Care and Use Committee). HF was induced in 15 adult male mongrel canines (20–22 kg) by 4–6 wk of rapid, right ventricular pacing at 240 beats/min (1). Before the terminal study, all animals demonstrated clinical signs of end-stage HF including anorexia, lethargy, ascites, tachypnea, and muscle wasting. In our experience this model of HF is associated with a decrease in fractional shortening, on average, from 33% before implantation to 12% after 4–6 wk of rapid pacing. In addition, nine normal canines that did not undergo pacemaker implantation were studied as controls. Intracellular calcium and transmembrane potential (V_m) were mapped across the transmural surface of the canine left ventricular wedge preparation as described previously (14). Briefly, left ventricular wedges of cardiac tissue (∼40 mm height × 30 mm width × 12 mm depth) were taken near the base of the left ventricle and free-running Purkinje fibers were removed. The coronary artery of each wedge was cannulated and perfused (50–70 mmHg) with oxygenated (95% O₂-5% CO₂) Tyrode solution containing (in mmol/l) 135 NaCl, 0.9 NaH₂PO₄, 0.492 MgSO₄, 4.03 KCl, 5.5 dextrose, 1.8 CaCl₂, and 10 HEPES (pH 7.40). Wedges were loaded with indo-1 AM (Molecular Probes) at a final concentration of 10 μmol/l for 45 min at 36°C, followed by a 15-min washout period. In all experiments, 10–15 μmol/l of cytochalasin D (Sigma-Aldrich) was used to remove any motion artifact. The wedge, perfused with Tyrode solution (36 ± 1°C), was placed in a Lexan chamber containing Ag-AgCl electrodes to monitor global electrical activity (ECG). Physiological stability of the preparation for 2 to 3 h was ensured by monitoring the ECG, coronary pressure, coronary flow, and perfusion temperature continuously.

To determine the amplitude of calcium transients, ratiometric imaging of intracellular calcium was performed (n = 15) (11). Briefly, light (350 ± 10 nm) from an arc lamp (Thermo-Oriel) was used to excite indo-1 AM; emitted fluorescence at 485 and 405 nm was directed by a 445-nm dichroic mirror (Chroma Technology) to separate 16 × 16 element photodiode arrays. Ratiometric calcium transients were calculated by dividing the background-subtracted calcium transients at 405 nm by the background-subtracted calcium transients at 485 nm. In a subset of experiments (HF, n = 12), dual calcium-voltage recordings were performed to determine whether or not the observed SCR events were associated with changes in V_m (DADs). For these experiments, the voltage-sensitive dye pyridinium, 4-{2-[6-(dibutylamino)-2-naphthalenyl]ethenyl}-1-(3-sulfopropyl), hydroxide, inner salt (di-4-ANEPPS; Molecular Probes) at a final concentration of 15 μmol/l was excited by adding light (515 ± 5 nm) from a QTH lamp (Thermo-Oriel). To minimize cross talk between dyes, the fluorescence was measured at 485 nm for indo-1 AM and >695 nm for di-4-ANEPPS, which was directed by a 560-nm dichroic mirror (Chroma Technology) to separate 16 × 16 element photodiode arrays.

All signals (calcium and V_m) were multiplexed and digitized simultaneously with 12-bit precision at a sampling rate of 1,000 Hz per channel. For the present study, an optical magnification of ×1.24 was used, resulting in a total mapping field of 14 mm × 14 mm, with 0.9-mm spatial resolution and 0.8-mm² pixel size. To view, digitize, and store the position of the mapping array relative to anatomical features, the dichroic mirror was rotated to reflect an image of the preparation to a CCD video camera.

Experimental protocols.

To determine calcium handling properties, calcium transients were recorded during steady-state endocardial pacing at a cycle length (CL) of 600 ms. To elicit ectopic (unstimulated) beats caused by TA, a rapid pacing protocol was used in which the CL was briefly (10–15 s) decreased to 200–100 ms with one-to-one capture, followed by a halt in pacing. To produce β-adrenergic stimulation, isoproterenol (Iso, Sigma-Aldrich) was dissolved in DMSO and added to the perfusate at a final concentration of 0.2 μmol/l. Approximately 3–5 min were allowed for the administration of Iso to take effect (10).

Data analysis.

Calcium transient amplitude was measured as the difference between peak systolic and diastolic levels in ratiometric units (RU). Calcium transient duration was measured from the beginning of the upstroke of the calcium transient to 50% decay of peak amplitude (CaD₅₀). An analysis of ectopy induced by rapid pacing was limited to unstimulated beats that were preceded by an isoelectric period on the ECG after the halt in pacing that was no longer than 1 s to avoid reentrant activity and intrinsic (i.e., automatic) beats. Accordingly, all subsequent references to ectopic activity will be indicated as TA. To discriminate between triggered beats that originated in the mapping field from those that initiated off the mapping field, the upstroke timing of the earliest triggered action potential was compared with the earliest deflection on the ECG. A delay of ≤3 ms (corresponding to a distance of 1 mm, assuming a conduction velocity of 31.9 cm/s as reported previously; see Ref. 26) was considered to be within the mapping field. SCR events were defined as nonelectrically driven (no ECG activity) increases in intracellular calcium during diastole that exceeded 10% of the amplitude of a fully stimulated calcium transient and occurred over more than one pixel. SCR amplitude was defined as the difference between peak and diastolic ratiometric calcium levels. The rate of SCR rise was defined as the maximum slope of the rising phase of the SCR event and expressed as RU/ms. For SCR events that led to triggered beats, SCR amplitude was defined at the time just before the rapid increase associated with a full calcium transient.

Statistical analysis.

All numerical data are expressed as means ± SE. A comparison between groups of data was made using ANOVA with post hoc Student's t-tests, and statistical significance was considered for values of P < 0.05 with correction for multiple comparisons (Sidak adjusted α) where necessary. Categorical data from the occurrence rate analysis are expressed as percentages, the Fisher exact test was used to compare the HF to the normal group, and McNemar's case-control test was used to compare control and Iso conditions within the same preparation (with significance for P < 0.025 after correcting for multiple comparisons).

RESULTS

Calcium handling in the failing canine heart.

Intracellular calcium transients recorded at a CL of 600 ms in wedge preparations from normal (N) and failing hearts (HF) are shown in Fig. 1. Figure 1A, left, shows representative ratiometric calcium transients plotted on the same vertical scale to directly compare peak calcium transient amplitude. The traces demonstrate that the amplitude from the HF preparation is reduced compared with that of the normal preparation. Shown to the right are summary data for HF (n = 12) and normal (n = 9) preparations. Calcium transient amplitude was significantly reduced in HF relative to normal (135 ± 4 vs. 170 ± 3 RU, respectively, P < 0.05). In Fig. 1B, representative ratiometric calcium transients are shown with normalized amplitudes to compare differences in CaD₅₀. Summary data (Fig. 1B, right) demonstrate a significant prolongation of CaD₅₀ in HF (n = 12) compared with normal (n = 9) (252 ± 2 vs. 229 ± 2 ms, respectively, P < 0.05). As reported previously, maximum action potential duration (not shown) was significantly longer in HF (277 ± 30 ms) compared with normal (241 ± 11 ms, P < 0.01). Rapid pacing at a CL of 150 ms decreased calcium transient amplitude in normal and HF preparations by 14% and 6%, respectively, compared with baseline pacing (CL = 600 ms); however, this difference did not reach statistical significance. Rapid pacing also induced calcium transient amplitude alternans, similar to that reported previously in normal and HF canine wedge preparations (41).

Fig. 1.

Calcium (Ca²⁺) transient amplitude and duration in normal and failing wedge preparations. A: representative Ca²⁺ transients (left) and summary data (right) for ratiometric Ca²⁺ transients amplitude (Amp) from failing (HF, n = 12) and normal (N, n = 9) wedge preparations. Amplitude was significantly reduced in HF relative to normal [135 ± 4 ratiometric units (RU) vs. 170 ± 3 RU, respectively, P < 0.05]. B: Ca²⁺ transients with normalized amplitudes to emphasize differences in duration (left) and summary data (right) for ratiometric Ca²⁺ transients duration at 50% decay (CaD₅₀) from HF and N preparations. HF transients are prolonged with respect to N (252 ± 2 vs. 229 ± 2 ms, respectively, P < 0.05).

TA and SCR events in the failing heart.

TA was induced by rapid pacing at CLs between 200 and 100 ms. Shown in Fig. 2A are representative ECG, action potentials, and calcium transients recorded during the induction of TA in a HF wedge preparation during Iso administration. Below these three separate traces are superimposed traces for V_m and calcium on an expanded time scale during the triggered beat. Iso tended to increase calcium transient amplitude during rapid pacing in normal (+15%) and HF (+66%) preparations, but these differences did not reach statistical significance. Following the sudden halt in pacing (S1), the traces in Fig. 2A show a quiescent period of ∼300 ms before a full, unstimulated beat occurred. Over all the preparations (normal and HF), Iso administration tended to increase the occurrence of TA compared with control, but this difference was only significant for HF preparations (93% with Iso vs. 20% without Iso, P < 0.025).

Fig. 2.

Triggered activity (TA) and subthreshold SCR events in wedge preparations under control (CNTL) and isoproterenol (Iso) conditions. A: representative ECG, action potentials (see V_m trace), and Ca²⁺ transients recorded during the induction of a triggered beat (TA) by rapid pacing (S1) in a HF preparation during Iso administration. Below the 3 separate traces for ECG, transmembrane potential (V_m), and Ca²⁺ are superimposed traces for V_m and Ca²⁺ on an expanded time scale during the triggered beat. The occurrence of TA was significantly greater in HF-Iso conditions compared with HF-CNTL conditions. B: representative ECG, action potentials, and Ca²⁺ transients recorded during the induction of a subthreshold SCR and delayed afterdepolarization (DAD) in a HF preparation. Below the 3 separate traces for ECG, V_m, and Ca²⁺ are superimposed traces for V_m and Ca²⁺ on an expanded time scale during the SCR/DAD. Summary data show that SCR activity was significantly greater in HF compared with N. The numbers in parentheses are the number of preparations demonstrating the event by the total number of preparations.

At slower CLs, SCR activity could be seen in the absence of a triggered beat (i.e., a subthreshold SCR, Fig. 2B). In the traces shown in Fig. 2B, left, following the sudden halt in pacing (S1), intracellular calcium slowly increased and then decreased (SCR). The corresponding DAD recorded from the same site is shown in the V_m trace, importantly without evidence of action potential activity (see also ECG). The DAD and SCR are overlaid in the bottom trace on an expanded time scale. Over all the experiments (Fig. 2B, right), subthreshold SCR activity was significantly greater for HF preparations compared to normal with Iso (93% vs. 44%, P < 0.025) and without Iso (53% vs. 0%, P < 0.025). Interestingly, in the absence of Iso, a majority of HF wedges exhibited SCR activity. In contrast, no SCR activity was observed in normal wedges without Iso, suggesting that HF wedges exhibit abnormal SR calcium release, whereas normal wedges do not.

In HF preparations with Iso in which a subthreshold SCR event reproducibly occurred from the same region, we examined how SCR amplitude and SCR rate of rise varied with CL. We found that SCR amplitude and SCR rate of rise increased with decreasing CL, i.e., faster pacing rates (Fig. 3A). These relationships were fit to exponential curves with a doubling of percent change in amplitude and rate of rise for every 10-ms decrease in CL, suggesting that SCR amplitude and rate of rise are mediated by rate-dependent changes in SR calcium load. In addition, as shown in Fig. 3B, there was a strong, positive correlation between SCR rate of rise and SCR amplitude (R² = 0.84).

Fig. 3.

Rate dependence of SCR amplitude and SCR maximum rate of rise in HF preparations with Iso. A: in HF preparations in which SCRs occurred at multiple cycle lengths (CLs), decreases in CL resulted in increased SCR amplitude and SCR rate of rise measured at the same site on the transmural surface. The percent change in SCR amplitude and SCR rate of rise as a function of decreasing CL were fit with an exponential (solid curves, R² = 0.98 and R² = 0.97, respectively). B: linear regression of SCR rate of rise with SCR amplitude showing a strong, positive correlation (r = 0.92) with R² = 0.84.

Mechanism of TA in HF.

To investigate the causal relationship between SCR events and TA in HF preparations, we compared the precise location of SCR activity to the site of initiation of the ensuing triggered beat. The traces in Fig. 4 show the ECG, V_m, and intracellular calcium recorded simultaneously from the same location during a halt in rapid pacing and a subsequent induction of TA. The contour maps show the activation times derived from action potentials (V_m) and the cytoplasmic calcium levels (calcium in arbitrary units) derived from calcium transients at select time points as shown in the calcium trace associated with the last paced beat (Fig. 4, top) and the SCR and triggered beat (Fig. 4, bottom). During pacing, the activation emanated from the site of pacing and was associated with a rapid calcium release (see cytoplasmic calcium maps for S1) that occurred across the entire mapping field in ∼32 ms over the three contour maps shown.

Fig. 4.

Left: simultaneously recorded traces for ECG, V_m, and Ca²⁺ during rapid pacing-induced (S1) SCR (arrow) and TA in HF preparations with Iso. Right, top: transmural contour maps (14 × 14 mm) of action potential activation times (in ms) and cytoplasmic calcium levels [in arbitrary units (AU)] during select time points of the last paced beat (third S1, pacing symbol). All times are relative to the onset of electrical activity. Right, bottom: transmural contour maps of action potential activation times (in ms) and cytoplasmic calcium levels (in AU) during select time points of the triggered beat and preceding SCR, respectively. All times are relative to the onset of earliest SCR activity. The V_m activation map of the triggered beat shows that the beat originated focally, precisely at the same location as the SCR. Transmural contour maps are shown with the epicardium (Epi) on the left and the endocardium (Endo) on the right.

After the last paced beat, the next event is the onset of SCR activity as indicated by a slow, ramplike increase (arrow) in the calcium trace. The contour maps (Fig. 4, bottom) show the activation of the triggered beat (TA) beginning at 124 ms from the bottom of the mapping field where a detectable SCR had originated at 30 ms (see cytoplasmic calcium maps for SCR). Near its peak (91 ms), the SCR has expanded to a larger area with an increased amplitude in the absence of any action potential activity. By 143 ms, the rapid rising phase of a full calcium release occurred, corresponding to the origin of TA. The slow development of SCR activity is in sharp contrast to the rapid release of calcium observed during the S1 beat. Similar results were observed in all episodes in which dual calcium-voltage recordings were made during the initiation of a triggered beat within the mapping field. These data are consistent with SCR-mediated TA as the mechanism of ectopy and that spontaneous calcium release occurs over a broad aggregate of myocardial cells before action potential initiation.

SCR dynamics and development of TA.

Why do some SCR events trigger beats and others do not? Figure 5 depicts the spatial extent, amplitude, and time course of two SCR events in the same HF wedge preparation: one that triggered a beat (Fig. 5A) and a second, recorded at a slower CL, that did not (Fig. 5B). In Fig. 5A, the calcium trace depicts the last paced beat (S1) and an SCR event (arrow) followed by a full, unstimulated calcium release (TA). Above the calcium trace is a series of frames corresponding to cytoplasmic calcium levels at select time points during the last paced beat (S1), demonstrating a rapid and uniform calcium release emanating from the site of pacing. Below the trace are frames at select time points relative to the earliest onset of the SCR event (time of SCR onset defined as 0 ms). The color scale was created such that black corresponded to diastolic calcium levels, red corresponded to calcium levels associated with the SCR event, and green corresponded to calcium levels associated with a full, unstimulated calcium release. Therefore, the red-green transition corresponds to the “threshold” level at which the SCR triggered a full calcium release. At 50 ms, the SCR event (50 ms, area of red, Fig. 5A) is visible in a small cluster of pixels in the midmyocardium at the top of the mapping field. On average, the initial size of SCRs associated with TA was 3.9 mm², approximately five pixels in size. This SCR event continued to grow slowly in amplitude (brighter red) and spatial extent, until at 115 ms a beat initiated (area of green). At 127 ms, the green area at the top of the mapping field quickly expands, representing the rapid propagation of the triggered beat. Interestingly, a triggered beat also occurred at the same time from the bottom of the mapping field and collided with the first (133 ms, Fig. 5A). The last frame shows the partial recovery of calcium to diastolic levels (325 ms, Fig. 5A).

Fig. 5.

Shown are traces of 2 SCR events recorded from the same site in a single HF preparation: one that resulted in a triggered beat (A) and one that did not (B). In A, above the last paced beat (S1) are 3 frames showing transmural (14 × 14 mm) calcium level (amplitude) at select time points, demonstrating the rapid, uniform pattern of calcium release during pacing (pacing symbol). Below the trace are several frames of calcium levels from select time points during the SCR and subsequent triggered beat (TA). All times shown are relative to the earliest site of calcium release, during pacing (top) or the SCR (bottom). The Epi and Endo are shown on the left and right sides of each contour, respectively. For the color scale, black corresponds to diastolic calcium, red corresponds to subthreshold SCR, and the transition from red to green corresponds to the threshold for TA in A. Calcium release is much slower during the SCR compared with pacing (see text for details). In B, the exact same format is shown, except that frames of calcium levels during pacing are not shown since they are identical to A. The color scale created for A was also used in B. In A and B, SCRs occur in a relatively large aggregate of myocardial cells and achieved a similar amplitude; however, the rate of SCR rise is much greater in A when TA occurred compared with B when TA did not (see text for details).

In the same preparation, but at a slower CL, an SCR event failed to generate a triggered beat (Fig. 5B). To facilitate a comparison, the calcium trace in Fig. 5B is from the same location and is displayed on the same time and color scale as that in Fig. 5A. In general, the SCRs in Fig. 5B originate in the same location but develop more slowly. On average, the time delay from the last paced beat to the onset of SCR activity in HF was significantly less (i.e., the SCRs occurred earlier) during Iso administration (479 ± 7 ms) compared with control conditions (580 ± 15 ms, P < 0.017). Since a triggered beat did not occur, we expected that the amplitude of this subthreshold SCR would be less than the SCR in Fig. 5A. Surprisingly, by 325 ms, this SCR achieved the same amplitude as the SCR in Fig. 5A just before TA occurred, but this time it did not trigger a beat. If both SCRs achieved similar amplitudes in the same region of tissue, then why did one SCR trigger a beat but not the other? The most striking difference between the two events was the rising slope of the SCR (i.e., the rate of SCR rise). The SCR that caused a triggered beat had a rate of rise of 0.83 RU/ms, whereas the subthreshold SCR (Fig. 5B) had a maximum rate of rise of 0.36 RU/ms, less than half the value of the first SCR (Fig. 5A). In addition, the average initial SCR area of events that were not associated with TA was 1.5 mm², significantly less than SCRs associated with TA (3.9 mm², P < 0.01). Similarly, the rate of increase in SCR area over time was greater for those that triggered a beat (0.50 ± 0.02 mm²/ms) compared with those that did not (0.16 ± 0.01 mm²/ms, P < 0.05).

Figure 6 shows the amplitude (Fig. 6A) and maximum rate of SCR rise (Fig. 6B) of SCRs that triggered a beat (TA, green bars) and those that did not (no TA, red bars) in recordings made over an equivalent range of CLs. As seen in Fig. 6A, there is a complete overlap of the SCR amplitude histograms for TA and no TA (37 ± 1 vs. 36 ± 1 RU, P = not significant). In contrast, the histograms for the rate of SCR rise show that while there is some overlap, the two peaks are clearly distinct with a significantly greater mean rate of SCR rise in TA relative to no TA (0.41 ± 0.01 vs. 0.18 ± 0.01 RU/ms, P < 0.05). These data demonstrate that SCR events which caused TA had a faster maximum rate of rise but did not necessarily differ in amplitude compared with SCR events that did not cause TA.

Fig. 6.

Relationship between SCR amplitude (A) and SCR rate of rise (B) with the occurrence of triggered beats in HF preparations with Iso. Histograms of SCR amplitude (A) show that SCRs that lead to TA (TA, green) statistically do not have larger amplitudes than those SCR events that fail to trigger a beat (no TA, red). This is indicated by the overlapping histograms. In contrast, histograms of the rate of SCR rise (B) show that SCRs that result in TA (green) have a significantly faster (shifted to the right) rate of rise than SCRs that do not trigger beats (red).

DISCUSSION

In the present study we demonstrate a high incidence of TA mediated by spontaneous calcium release as measured from an aggregate of cells in intact failing myocardium (i.e., an SCR) that is enhanced under conditions of β-adrenergic stimulation. The novel findings are as follows: 1) in HF, SCR activity occurs over a large portion of myocardial tissue, and 2) in tissue, the rate of SCR rise appears to play a critical role in arrhythmogenesis.

Abnormal calcium handling in the failing heart.

Depressed cardiac contractility and impaired relaxation are key abnormalities contributing to morbidity and mortality in clinical HF (5, 18). Despite the fact that numerous studies from both human (3, 24) and animal (2, 22, 36) models of HF have attributed the mechanical dysfunction to changes in the calcium regulation observed in isolated myocytes (see Refs. 37 and 40 for reviews), few reports have made comparisons from intact tissue (16, 21, 23, 25, 34). We found in canine tachycardia-induced HF that calcium transient amplitude is reduced and the duration is prolonged compared with normal. Qualitatively, this is similar to what has been observed in isolated myocytes, but quantitatively the differences we have measured in intact tissue are much smaller in magnitude. Other studies in intact failing tissue report differences in calcium transients similar to our findings (16, 21).

Mechanisms of TA in the failing heart.

While a number of arrhythmia mechanisms exist in HF, it has been shown that nonreentrant mechanisms are a major source of ventricular arrhythmias (28–30), particularly in nonischemic HF (27). Our data are consistent with these findings. Furthermore, the occurrence of TA was closely associated with the occurrence of SCR (Fig. 2B) and DAD (Fig. 4) activity. Finally, with Iso, the occurrence of TA in HF was significantly enhanced compared with control conditions (Fig. 2A). These findings are consistent with studies in isolated myocytes (31) and ventricular trabeculae (39), which demonstrated that residual β-adrenergic responsiveness is a key factor in the genesis of triggered arrhythmias in HF. Taken together, these data demonstrate that in our model of HF, there is a substrate for nonreentrant arrhythmias caused by calcium-mediated TA. This is in addition to a substrate for reentrant excitation that is known to exist as well (1).

While there appears to be a consensus that spontaneous calcium release at the subcellular level is caused by increased open probability of ryanodine receptor type 2 (RyR2) resulting in SR calcium leak during diastole (12, 13, 35), the mechanisms behind leaky RyR2 in HF remain highly controversial (4, 8, 17, 42). With Iso, 14 of 15 HF preparations exhibited SCR activity. However, many HF preparations exhibited SCR activity in the absence of Iso compared with none in normal preparations. These data suggest that in our model of HF, RyR2 is leaky, and this leak becomes more arrhythmogenic under conditions of β-adrenergic stimulation. Venetucci et al. (38) reported in nonfailing isolated ventricular myocytes that a leaky RyR2 was not sufficient to induce spontaneous calcium release and that β-adrenergic stimulation was required to raise SR calcium load. We found that β-adrenergic stimulation promoted TA, but SCR activity alone (without TA) was observed in HF preparations in the absence of Iso. It is possible that the rapid pacing protocol we used to induce SCR activity was sufficient to raise SR calcium load or that HF is associated with additional factors that promote spontaneous calcium release. Further experiments would be necessary to determine the molecular mechanism underlying the increased RyR2 open probability and the subsequent enhancement under β-adrenergic stimulation.

Spontaneous calcium release in tissue.

To our knowledge, this is the first study to directly demonstrate spontaneous calcium release causing TA in intact failing myocardium. It has been suggested that the factors that determine the threshold for calcium-mediated triggered arrhythmias in an isolated cell are quite different than those in intact tissue (7). Therefore, the present study addresses an important gap in knowledge between isolated myocytes and the intact heart. Triggered arrhythmias occur when DAD amplitude is large enough to depolarize the membrane to a threshold potential. While DAD amplitude is known to be rate dependent, few studies have examined the rate dependence of SCR activity in intact ventricular myocardium. We have found in tissue that both SCR amplitude and SCR rate of rise are rate dependent (Fig. 3), similar to DADs. Schlotthauer and colleagues (31, 33) showed in both normal and failing isolated myocytes that SR calcium release induced by caffeine application increased in amplitude with an increase in pacing frequency, and they were able to assess the threshold amplitude required to trigger an action potential. Their data also suggest that the rate of spontaneous calcium release may be important as well. In tissue, even though SCR amplitude increased with decreasing CL (Fig. 3), it could not distinguish between spontaneous events that did and those that did not trigger a beat (Figs. 5 and 6A). Instead, what was noticeably different among SCR events was that a fast rise was associated with TA and a slow rise was not (Fig. 6B). This does not mean that the amount of calcium release from the SR is irrelevant but rather that the calcium must be released in a relatively rapid and synchronized fashion to trigger an arrhythmia. In a recent report by Fujiwara et al. (6), simultaneous confocal recordings of V_m and intracellular calcium in Langendorff-perfused rat hearts showed that single-cell calcium waves occurring in a sporadic fashion failed to depolarize the membrane, whereas synchronous calcium waves resulted in DADs and TA. From our study it is not clear why the rate of SCR rise modulates the threshold for TA, but a slow SCR rise, and thus reduced DAD slope, may inactivate sodium channels before the threshold is reached. Additional studies would be needed to test this hypothesis.

As we have previously reported in nonfailing models of TA (9, 10, 15), SCRs in intact failing myocardium occur over relatively large portions of tissue before electrical excitation occurs (Figs. 4 and 5). This may explain how in well-coupled tissue an SCR can overcome the current sink of neighboring cells and generate a DAD of sufficient magnitude to trigger a beat. Indeed, SCRs that triggered a beat were significantly larger in area than those that did not. In addition, the spatial synchronization of spontaneous calcium release may also play a role in the ability of an SCR to trigger a beat. For example, given that each recording site (0.9 × 0.9 mm) was an average response of many cells, it is possible that the rate of SCR rise which we measured is governed by the spatial synchronization of SR calcium release within each pixel as well as the rate of calcium release for individual SR release sites. Is an SCR just a calcium wave? It could be, since they are both nonelectrically driven (i.e., spontaneous) calcium release (19). However, additional studies may be needed to sort out any potential differences.

Limitations.

Human HF is a complex clinical condition with a wide range of etiologies and symptoms. One must exercise caution when extrapolating experimental findings to human HF since no single animal model can fully recreate the conditions of human HF. Despite this limitation, the canine tachycardia-induced model of HF has been widely used and has been demonstrated to produce many of the phenotypic traits of human HF (20). In addition, other factors such as reduced coupling (26) and expression of ion channels (31) will play a role in ectopy as well.

In our experiments, we mapped a two-dimensional (transmural) surface of a three-dimensional preparation. This leads to the possibility that events originating outside of the mapping field (offscreen) will either not be detected or propagate into our mapping field. We recorded all spontaneous events and then excluded offscreen events as described in Data analysis. In fact, many events were offscreen, which we took to be reassuring since this demonstrated that the cut (mapping) surface was not the sole source of TA. The three-dimensional preparation also makes it difficult to measure the maximum SCR area and its change over time.

GRANTS

This material is based on work supported under a National Science Foundation Graduate Research Fellowship (G. S. Hoeker); an American Heart Association, Ohio Valley Affiliate predoctoral fellowship 0415213B (R. P. Katra); a Career Development grant from the Emergency Medicine Foundation (L. D. Wilson); and National Heart, Lung, and Blood Institute Grant HL-168877 (K. R. Laurita).