Isoproterenol Increases the Fraction of Spark-Dependent RyR-Mediated Leak in Ventricular Myocytes

Abstract

Recent research suggests that the diastolic ryanodine-receptor-mediated release of Ca²⁺ (J_leak) from the sarcoplasmic reticulum of ventricular myocytes occurs in spark and nonspark forms. Further information about the role(s) of these release manifestations is scarce, however. This study addresses whether the fraction of spark-mediated J_leak increases due to β-adrenergic stimulation. Confocal microscopy was used to simultaneously image Ca²⁺ sparks and quantify J_leak in intact rabbit myocytes, either in the absence or in the presence of 125 nM isoproterenol. It was found that isoproterenol treatment shifts the spark-frequency-J_leak relationship toward an increased sensitivity to a [Ca²⁺] trigger. In agreement, a small but significant increase in spark width was found for cells with matched baseline [Ca²⁺] and total SR [Ca²⁺]. The reconstruction of release fluxes, when applied to the average sparks from those selected cells, yielded a wider release source in the isoproterenol event, indicating the recruitment of peripheral ryanodine receptors. Overall, the results presented here indicate that β-adrenergic stimulation increases the spark-dependent fraction of J_leak. Working together, the increased Ca²⁺ sensitivity and the greater spark width found during isoproterenol treatment may increase the probability of Ca²⁺ wave generation.

Introduction

Ventricular arrhythmias are an important cause of death in heart failure patients. Some arrhythmias correlate with sudden surges of the β-adrenergic tone. Diastolic sarcoplasmic reticulum (SR) Ca²⁺ release with resultant Ca²⁺ waves may contribute to such arrhythmogenesis by causing a depolarizing Na⁺/Ca²⁺ exchanger (NCX) current (1). Such a current may initiate a delayed afterdepolarization (DAD) that may trigger an action potential. When the above sequence synchronizes within a large enough cluster of ventricular myocytes, a premature beat may occur (2). Thus, unstable Ca²⁺-induced Ca²⁺ release (CICR) during diastole has emerged as a major player in the generation of and predisposition to arrhythmias. Therefore, much effort has been devoted to understanding the factors leading to wave occurrence.

Intact myocytes from hearts with chronic failure display an increased diastolic ryanodine-receptor (RyR)-mediated Ca²⁺ release (J_leak) at a given total SR [Ca²⁺] compared to myocytes from healthy control hearts (3,4). At the local level, intact diseased cells presented a heightened spark frequency and amplitude for a given total SR [Ca²⁺] (5). Therefore, both global and local measurements of diastolic RyR activity suggest that RyRs have increased activity in diseased cells.

An increased J_leak has also been observed in isoproterenol-treated myocytes (6). Local photolysis of Ca²⁺-containing compounds unmasked an enhanced RyR responsiveness in isoproterenol-treated cells, despite their lower total SR Ca²⁺ compared to control cells (7). Additional confocal measurements showed that Ca²⁺/calmodulin-kinase-II-treated (8), but not protein-kinase-A (PKA)-treated (9), permeabilized cells greatly increase their spark activity, size, and amplitude for a given total SR [Ca²⁺]. Voltage-clamped myocytes also showed an increased spark activity in isoproterenol for a given cytosolic [Ca²⁺] ([Ca²⁺]_i) and total SR [Ca²⁺] (10).

The above data indicate that the diastolic release of Ca²⁺ may be increased in cells with high adrenergic tone. However, the character of that release may play a major part in determining whether such an increase in release may be arrhythmogenic.

Recently, it has been shown that a sizable fraction of the physiological J_leak does not occur as sparks (4,11–13). This nonspark release may be the result of short-lived single RyR openings, which would not be as effective at triggering CICR. The relationship between the global magnitude of nonspark diastolic Ca²⁺ release and the generation of Ca²⁺ waves is not known, but nonspark release would be expected to have a much lower likelihood of generating waves than Ca²⁺ sparks. A Ca²⁺ spark, by definition, is a relatively large, detectable signal, of magnitude consistent with the simultaneous opening of multiple RyRs (12,14). Such an abrupt release will result in a local increase of [Ca²⁺]_i as the signal spreads from the initiation site to the neighboring junctional clefts. In the case of a nonspark event, release would instead be brief and the amplitude and size of the signal low. Moreover, a spark might be more likely to cause CICR from any RyR outside the junctional or corbular SR. The existence of such channels is still hypothetical (11).

In this study, we tested whether isoproterenol increases the fraction of spark-dependent J_leak. Our data show that at the same total J_leak, isoproterenol treatment of intact rabbit ventricular myocytes increases the spark frequency and the amount of Ca²⁺ released per spark. The findings show that β-adrenergic stimulation increases the potential for the RyRs to release a given amount of Ca²⁺ abruptly, as sparks.

Methods

Cell isolation, indicator loading, and experimental solutions

Left ventricular myocytes were isolated from New Zealand White rabbits (2.5–3 kg) in accordance with the standards established and controlled by the Institutional Animal Care and Use Committee at Rush University. The procedures for cell isolation and loading of the Ca²⁺ indicator, Fluo 4, are described in the Supporting Material.

The composition of the experimental solutions was as follows (in mM): Normal Tyrode (NT)—140 NaCl, 4 KCl, 1 MgCl₂, 5 HEPES, 10 glucose, 2 CaCl₂, pH 7.4 with NaOH, with 125 nM isoproterenol added as indicated; 0 Na⁺/0 Ca²⁺—140 LiCl, 4 KCl, 1 MgCl₂, 5 HEPES, 10 glucose, 10 EGTA, pH 7.4 with LiOH, with 1 mM tetracaine or 10 mM caffeine added as indicated.

Data acquisition

The total RyR-dependent release in diastolic conditions was measured in Fluo-4-loaded intact rabbit myocytes, as described previously (15). See text and Fig. S1 in the Supporting Material for more details. Before the J_leak measurement, myocytes from the the control and treated groups were brought to a similar steady-state total SR [Ca] ([Ca]_SRT) by field stimulation at 0.5 Hz for control cells and 0.25 Hz for isoproterenol-treated cells. [Ca²⁺]_i was subsequently measured in the absence of extracellular Na⁺ and Ca²⁺, first in the presence and later in the absence of 1 mM of the RyR antagonist tetracaine. The shift of cytosolic Ca²⁺ from the cytosol to the SR with RyR block is proportional to the underlying RyR-mediated release before blockade (15). Ca²⁺ sparks were measured during the portion of the protocol in which no tetracaine was present. No sparks were ever detected in the presence of tetracaine.

Data analysis

The approach used for determining the spark-dependent J_leak is the one from Santiago et al. (12), with minor modifications. The quantitative details can be found in the Supporting Material.

Briefly, the spark-dependent J_leak was calculated for control and isoproterenol by determining the Ca²⁺ released by each spark (on average) and multiplying by the spark frequency. Average sparks were constructed from individual events centered at their peaks and then scaled up to account for the effects of out-of-focus sparks (see Supporting Material for details). The Ca²⁺ flux that generated the average spark (a function of time) was calculated and the amount of Ca²⁺ released per spark was subsequently determined. This was multiplied by the spark frequency per unit cell volume to yield spark-mediated J_leak with the dimensions of rate of concentration change.

The values of the diastolic [Ca²⁺] at 0.5 Hz, which serves as a fundamental variable in the method, were assumed to be 120 nM in both conditions. This is based on previous experimental measurements from our lab, using Fura 2 (see Supporting Material for further details).

Statistics

Results are presented as the mean ± SE. A two-tailed t-test with Welch correction (P < 0.05) was used to check for differences in sample averages. The comparison of linear slopes and y-intercepts was made by methods similar to an analysis of covariance, described in Zar (16).

Results

Spark-frequency-J_leak relationship in control myocytes

Fig. 1 displays the response of a myocyte to the experimental protocol. Each line-scan image spans 1 s at 2 ms/line. The first two line-scan images (Fig. 1, upper left) contain the steady-state response to pacing at 0.5 Hz. Each subsequent image from left to right displays the response when the perfusate is replaced by 0 Na⁺/0 Ca²⁺ + 1 mM tetracaine. 0 Na⁺/0 Ca²⁺ blocks the sarcolemmal Ca²⁺ transport (17), whereas tetracaine antagonizes the RyR-mediated Ca²⁺ release (18). Because SERCA continues operating while all other quantitatively important fluxes remain blocked, a net transport of cytosolic Ca²⁺ into the SR occurs. This can be visualized as a decreased Fluo-4 signal in the absence of sparks. The bottom row of line-scan images starts when the perfusate is replaced by 0 Na⁺/0 Ca²⁺ alone. With the restoration of the leak-uptake balance, some of the SR Ca²⁺ shifts back to the cytosol, resulting in increased Fluo-4 signal and the appearance of sparks. The last image displays a measurement of the total SR [Ca²⁺] using 10 mM caffeine. The tetracaine-dependent steady-state shift of Ca²⁺ from the SR to the cytosol is proportional to the steady-state J_leak. Ca²⁺ sparks belonging to the steady state in 0 Na⁺/0 Ca²⁺ were further considered for analysis.

Figure 1.

Measurement of J_leak and Ca²⁺ sparks in a control rabbit myocyte. The figure illustrates two series of line-scan (xt) images, showing the response to 0 Na⁺/0 Ca²⁺ with (upper) and without (lower) tetracaine. The first two images in the upper row illustrate the last steady-state [Ca²⁺]_i transients before switching to tetracaine solution. The last image in the lower row shows the response to caffeine, which gain was modified (and calibrated at upper left) to increase the dynamic range of the measurements. Numbers indicate Ca²⁺ sparks (not all sparks are shown). The graph at lower right displays the global [Ca²⁺]_i during perfusion with 0 Na, 0 Ca²⁺ ± tetracaine. Signal values during tetracaine perfusion were corrected for a 5% quenching.

A summary of results from 57 cells is shown in Fig. 2. Cells were divided according to whether sparks were detected (solid bars) or not (open bars). Only in seven cells did sparks remain undetectable, with an average J_leak of 6.65 ± 1.5 μM s⁻¹. Sparking cells displayed a significantly higher average J_leak, 12.12 ± 0.87 μM s⁻¹ (P = 0.01; Fig. 2 B). A linear correlation was observed between J_leak and spark frequency (Fig. 2 C, r² = 0.33, P < 0.0001).

Figure 2.

Summary of results in control myocytes, displayed in the same format as the mouse results from Santiago et al. (12). (A) Total SR [Ca²⁺] was similar in sparking and nonsparking cells (mean ± SE, n = 50 sparking vs. 7 nonsparking cells). (B) J_leak was different among cell subpopulations. (C) J_leak correlated relatively well with spark frequency. The best linear regression is shown only for sparking cells. The linear regression including all cells is displayed in Fig. 3. (D) The average spark frequency increased with the average total SR [Ca²⁺] (r² = 0.1; n = 16, 14, 8, 9, and 3 for the bins, from left to right).

The difference in J_leak between sparking and nonsparking cells may relate to differences in total SR [Ca²⁺], since a lower total SR [Ca²⁺] would be expected to reduce the number of RyR channel openings. However, the total SR [Ca²⁺] was not different among the groups (129.7 ± 11.65 μmol (l cytosol)⁻¹ in nonsparking versus 136.1 ± 5.64 μmol (l cytosol)⁻¹ in sparking cells; Fig. 2 A). On the other hand, a correlation was found when the total SR [Ca²⁺] ([Ca]_SRT) was plotted versus the spark frequency (Fig. 2 D, r² = 0.1, P = 0.03). We conclude that Ca²⁺ sparks are more frequent at higher [Ca]_SRT. However, since the sparking and nonsparking cells had, on average, the same [Ca]_SRT, we further conclude that additional regulatory factors (i.e., other than the [Ca]_SRT) determine how much of the total J_leak takes the form of Ca²⁺ sparks in intact cells.

Spark-frequency-J_leak relationship in isoproterenol-treated myocytes

Isoproterenol is known to increase SERCA activity, leading to an increase in [Ca²⁺]_SRT. Therefore, to constrain the total SR [Ca²⁺] in isoproterenol-treated cells so that the [Ca]_SRT would be similar to that of control myocytes, the SR Ca leak protocol was performed with a reduced field-stimulation frequency of 0.25 Hz. Isoproterenol had the expected effects upon the pretetracaine cellular [Ca²⁺] transients (Figs. S2 and S3).

In contrast to control cells, the entire population of isoproterenol-treated cells showed Ca²⁺ sparks. Also in contrast to control cells, one-third of the cells studied displayed at least one wave after the removal of tetracaine and before attaining a stable steady-state level of fluorescence in 0 Na⁺/0 Ca²⁺ solution (11 of 34 isoproterenol-treated cells versus 1 of 57 control cells).

Fig. 3 displays a summary of the results for 34 isoproterenol-treated cells versus the 57 control myocytes from Fig. 2. A significant correlation was observed between J_leak and spark frequency in both groups (Fig. 3 A), either when expressed linearly (r² = 0.18 and P = 0.0126 vs. r² = 0.33 and P < 0.0001 for isoproterenol-treated and control cells, respectively) or nonparametrically (r = 0.46 and P = 0.0067 vs. r = 0.58 and P < 0.0001 for isoproterenol-treated and control cells, respectively). Results are in agreement with the notion that J_leak and Ca²⁺ sparks are different manifestations of a similar underlying phenomenon. Though the slopes were not statistically different, the higher y-intercept in isoproterenol-treated cells (P < 0.001) reflected an upward shift in the spark-frequency-J_leak relationship (i.e., there was a higher spark frequency at a given J_leak).

Figure 3.

Control versus isoproterenol-treated myocytes. (A) The spark-frequency-versus-J_leak relationship was shifted in isoproterenol-treated cells. Binning is 0.8 sparks (100 μm)⁻¹ s⁻¹ for control cells and 1.2 sparks (100 μm)⁻¹ s⁻¹ for isoproterenol-treated cells. Bin sizes are 19, 23, 11, 2, and 2 for control cells(n = 57) and 8, 9, 5, 8, 3, and 1 for isoproterenol-treated cells (n = 34). (B and C) Distributions of J_leak and observed spark frequency in histogram form. (D) The total SR [Ca²⁺] was significantly lower in isoproterenol-treated cells. (E) The spark frequency doubled in isoproterenol-treated cells throughout the range of total SR [Ca²⁺].

The average J_leak was not different (11.45 ± 0.82 μM s⁻¹ in control vs. 9.99 ± 1.31 μM s⁻¹ in isoproterenol-treated) despite a significantly lower total SR [Ca²⁺] in the isoproterenol-treated cells with the lower-frequency loading protocol (135.3 ± 5.12 μmol/l cytosol in control vs. 117.2 ± 4.94 μmol/l cytosol in isoproterenol-treated; P = 0.013). There was also a substantial change in spark frequency (1.267 ± 0.16 vs. 2.595 ± 0.32 sparks (100 μm)⁻¹ s⁻¹ in control versus isoproterenol-treated cells; P < 0.05), and the result was the same when nonsparking control cells were excluded (data not shown). Though not statistically significant, isoproterenol-treated cells nearly doubled their spark activity throughout the entire range when correlated with the total SR [Ca²⁺] (P = 0.06; Fig. 3 B). This observation is in agreement with studies in guinea pig cells (10).

Spark morphology at similar [Ca²⁺]_i and total SR [Ca²⁺]

Control sparking cells with [Ca²⁺]_i values matching those in isoproterenol-treated cells were selected for further analysis (93.34 ± 2.88 nM in n = 31 control cells vs. 91.37 ± 2.96 nM in n = 34 isoproterenol-treated cells, P = 0.635) (Fig. 4, inset). In this data subset, there was no difference in total SR [Ca²⁺] (132.8 ± 7.66 μmol/l cytosol vs. 117.2 ± 4.94 μmol/l cytosol, P = 0.094; Fig. 4, inset).

Figure 4.

Ca²⁺ sparks in selected cells, particularly cells with a similar [Ca²⁺]_i baseline (and, coincidentally, total SR [Ca²⁺]). This reduction in the sample size is necessary to ensure that the baseline spark fluorescence is similar among the groups being compared. Sample sizes are 79 control sparks and 187 isoproterenol sparks. (Inset) Baseline [Ca²⁺]_i and total SR [Ca²⁺] in the cells selected for the study.

The observed spark amplitudes were similar in both groups (P = 0.23). The spatial size (full width at half-maximum) at peak time was 200 nm wider in the isoproterenol event (P = 0.044) and the signal mass at peak time showed an upward trend (P = 0.14). Temporally, the sparks for isoproterenol-treated cells had a rising phase that was 2.5 ms longer (P = 0.014) plus longer times to 50%, 70%, and 90% decay (P = 0.0007–0.0111). These results show that isoproterenol sparks are slightly larger than control sparks for a given [Ca²⁺]_i and total SR [Ca²⁺]. Both the longer rising phase and longer time of decay suggest the possibility that channel openings are longer in isoproterenol-treated cells compared to those in control cells.

The average sparks and their detection volumes

Fig. 5 displays the raw average sparks from the subset of cells used to study spark morphology. In agreement with the statistics of single events, the average spark in isoproterenol-treated cells was found to be slightly wider than that in control cells. Similar results were obtained for sparks centered at their beginning (data not shown). Because out-of-focus sparks have a lower observed amplitude than in-focus sparks, a correction was applied to the average spark to account for this factor and increase the magnitude of fluorescence to the appropriate value (19). The corrected spark amplitudes for the average control and isoproterenol-treated cells were not significantly different. Control sparks were scaled up 46% and isoproterenol sparks by 58% using this method (see Fig. S6 and the accompanying text for details).

Figure 5.

(A and C) Raw average sparks in cells with similar [Ca²⁺]_i and total SR [Ca²⁺]. The number of constituent events and their average resting [Ca²⁺]_i are indicated. (B and D) Comparison of the temporal and spatial profiles (average of 5 lines).

Detectability was studied as in Cheng et al. (20). For each group, 100 scaled-down versions of its average spark were inserted into 10 spark-free images filled with white noise at the average experimental signal/noise ratio. There was a slight increase in detectability in isoproterenol due to the greater spark width. The detectability threshold allowed calculation of the cell volume within which a detectable spark must originate (12). The detection volumes for a 100-μm line in an xt scan were 0.51 pL for control and 0.63 pL for isoproterenol (see Fig. S6 and the accompanying text for details).

Spark-dependent J_leak at similar [Ca²⁺]_i and total SR [Ca²⁺]

Spark-dependent J_leak was determined as the Ca released by one spark per unit volume multiplied by the spark frequency. The spark frequencies doubled in the isoproterenol-treated cells regardless of whether they were observed directly (1.027 ± 0.16 vs. 2.595 ± 0.32 sparks (100 μm)⁻¹ s⁻¹; P < 0.0001) or after correction for the difference in detection volume (2.007 ± 0.31 vs. 4.10 ± 0.51 sparks pL⁻¹ s⁻¹; P = 0.0009; Fig. S6). Release flux was reconstructed from the average sparks, accounting for isoproterenol-induced differences in buffering and uptake (Table S2).

High-frequency noise was eliminated from the averaged sparks by fitting the experimental input using mathematical expressions (Fig. 6, A and B, idealized sparks) (Fig S7 and accompanying text). This preprocessing is useful, as reconstruction of release fluxes from Ca²⁺ transients exhibits a sensitivity to noise that compromises its performance (21). Fig. 6, C and D, displays [Ca²⁺] calculated after a partial deblurring of the idealized spark. The isoproterenol spark is wider at the peak (Fig. 6 D, inset). Fig. 6, E–I, provides an explanation for this difference. The release-flux densities show a wider source in isoproterenol, implying the recruitment of peripheral RyRs. As a result, the peak release current in isoproterenol-treated cells was double that in control cells (Fig. 6 J; note that the volume integral grows as a function of the cube of the radius). The amount of Ca²⁺ released per spark was therefore increased in isoproterenol-treated cells compared to control cells (0.61 and 0.31 attomoles, respectively).

Figure 6.

Spark-dependent J_leak when using the backward calculation of the release flux. (A and B) To remove high-frequency noise, the input sparks are idealized versions of the scaled-up average sparks. (C and D) The [Ca²⁺](r,t) profile was wider in the isoproterenol spark. (Inset) Profiles at time of maximum (average of 3 lines). (E and F) Release-flux densities. (G–I) Profiles of the release flux densities (average of 3 lines). Note the wider source in the isoproterenol condition. (J) Release current, obtained by volume integration of the release-flux densities. (K and L) Global RyR-mediated release (spark + nonspark), shown at left, is compared to the spark-dependent component, shown at right.

Knowing the amount of Ca/spark, the volume, and the time, we were able to calculate the Ca leak in the form of sparks. For the total of 8.92 ± 0.86 μM s⁻¹ Ca²⁺ that leaked through the RyR in the control condition (from Fig. 1), only 0.62 ± 0.097 μM s⁻¹ (8%) did so in the form of sparks (P < 0.0001; Fig. 6 K). This is in contrast to 2.50 ± 0.31 μM s⁻¹, or 28% of the total J_leak of 9.99 ± 1.31 μM s⁻¹ in the isoproterenol condition (P < 0.0001; Fig. 6 L).

Overall, the current results support the hypothesis that the fraction of spark-dependent J_leak increases upon isoproterenol treatment (independent of the amount of Ca in the SR and the cytosol). This increase appears to develop at two levels: globally, by an increased spark frequency for a given J_leak; and locally, by an increased amount of Ca²⁺ released per spark.

Long-lasting events, secondary recruitment at neighboring sites, and repetitive firing sites at similar [Ca²⁺]_i and total SR [Ca²⁺]

We further studied the sparks in isoproterenol-treated cells for more qualitative changes in the morphology of release. A trend toward more propagated release events (whereby one spark triggers a secondary event at a neighboring Ca²⁺ release unit) was found in isoproterenol-treated cells (5.5 ± 2.6% of the control sparks vs. 11.76 ± 3.4% of those in the isoproterenol condition; P = 0.15).

A similar frequency of long-lasting, nonpropagating events was found in isoproterenol versus control (8.5 ± 2.46% and 7.42 ± 3% of the sparks, respectively, P = 0.78). Likewise, 8.95 ± 4% of the sparks in control cells and 13.08 ± 3.23% of those in isoproterenol-treated cells were associated with repetitive firing sites. These values also were not different (P = 0.43). However, the interspark intervals (beginning-to-beginning) displayed a continuous distribution in control and a clear bimodal distribution in isoproterenol-treated cells (Fig. 7). The first mode was composed of intervals similar to those of the control group, and the second of shorter intervals lasting <250 ms. When considered together, the intervals during isoproterenol treatment were not significantly shorter than those of the control group (P = 0.06). However, when the isoproterenol group was considered as representing two separate classes, short intervals were different from both the control and the isoproterenol-treated long intervals (P = 0.0007 versus control and P < 0.0001 versus long intervals in isoproterenol). We interpret these data as indicating an accelerated recovery from refractoriness at some, but not all, repetitive firing sites.

Figure 7.

(A) The interspark interval was defined as the amount of time between the beginning of two consecutive sparks at the same site. (B and C) The interspark intervals followed a continuous distribution in control cells but a bimodal distribution in isoproterenol-treated cells.

Discussion

Our study focused on characterizing the spark-frequency-J_leak relationship and, for selected cells at similar [Ca²⁺]_i baselines and total SR [Ca²⁺]s, the fraction of spark-mediated J_leak. We quantitated J_leak, the spark frequency, and the amount of Ca²⁺ released per spark (and therefore the total spark-dependent J_leak) separately. The data suggest that the amount of Ca released in the spark form, which is potentially more arrhythmogenic, is higher in isoproterenol when both cytosolic and SR [Ca] are the same in each group. The difference is, therefore, related to changes in the elemental release process, likely at the RyR.

Technical considerations

The fraction of spark-dependent J_leak in control cells was low compared to estimates from permeabilized rabbit cells (4), and definitely lower than our previous estimate from intact mouse cells (12). The difference in the latter case may indicate a species dependence.

One potential source of error lies in the oscillations in the release-flux waveform at times >52 ms after time initiation (Fig. 6 I). These oscillations mainly arise from the deblurring correction and the calculation of Laplacians (21) and can cause under-estimation of the spark Ca flux. This said, the results agree well with the notion that little diastolic spark activity is observed in healthy intact rabbit cells, despite a sizable leak rate (4,15).

A forward simulation (spark reconstruction from a defined current waveform of variable magnitude and duration) and fit were also attempted to determine the release flux, as in Santiago et al. (11). A 200-nm-radius spherical release source and a trapezoidal release current (to account for intra-SR depletion) were used. However, the model could not fit the sparks, because it could not simultaneously account for their spatial widths and amplitudes (data not shown). These results constitute instances of the well-known spark-width paradox (22) and were not useful in providing information about the hypothesis under consideration.

The analysis assumes that the same stereotypical spark (i.e., the average spark) occurs in all cells. Ideally, the spark-dependent leak should be quantified for every cell individually, from its own sparks. This is, however, beyond the limits of the methodology presented here in intact rabbit cells under physiological conditions, and the averages would be too noisy for the results to be interpretable.

A global perspective

The goal of this study was to gain insight into the effects of β-adrenergic stimulation on sparks versus global diastolic J_leak. It was found that isoproterenol increases the fraction of spark-dependent J_leak relative to that of the control, a result that may be useful in the study of arrhythmogenesis during heart failure. First, isoproterenol treatment shifted the spark-frequency-J_leak relationship so that there were more sparks at a given leak rate (Figs. 3 and 4). Second, isoproterenol increased the average width of individual sparks for similar values of [Ca²⁺]_i and total SR [Ca²⁺] compared to control cells (Fig. 4). Third, isoproterenol widened the release source, implying an increased recruitment of RyRs (Fig. 6) due, perhaps, to increased sensitivity to the trigger Ca²⁺.

The relationship between Ca²⁺ sparks and waves is one of causality (23). Therefore, factors that promote the appearance of Ca²⁺ sparks and the propagation of release might be considered potential Ca²⁺-wave promoters as well. In addition to the effects described here, high total SR [Ca²⁺] increases spark width and frequency (4,24). We would therefore predict a nonlinear boost in the probability of Ca²⁺ waves (25). Indeed, the work of Hilliard et al. (26), in which flecainide doubled the frequency of events of smaller spatial size, resulting in fewer Ca²⁺ waves, suggests that spark size is an extremely important determination of further propagation.

The results of this study therefore suggest that isoproterenol may increase the tendency to trigger Ca²⁺ waves. Cells release more Ca²⁺ per spark more abruptly, thus increasing the chances of Ca²⁺ propagation, in contrast to a more gradual, more spatially uniform nonspark release. The results reinforce those reported in the literature on diastolic RyR activity (however, see Kashimura et al. (27)), particularly the notion that the so-called RyR hyperactivity may be tuned by a therapeutical strategy. These principles may partially explain the increased life expectancy of heart failure patients treated with β-blockers.

J_leak as nonspark events

Clustered RyRs

The observation of nonspark J_leak is not new (28,29), although only recently has its physiological meaning been explored (12,13,30). The current consensus is that Ca²⁺ sparks represent the concerted gating of multiple RyRs and that short-lived openings of individual RyRs are undetectable by confocal microscopy. The distribution of open times for isolated RyRs is exponential, with a mean open time of 1.7 ms for calsequestrin-stripped RyRs in near-physiological cis conditions (13). The dyadic [Ca²⁺] gradients dissipate in <0.5 ms (31), and therefore, the [Ca²⁺] stimulus from an isolated RyR would last <2.2 ms on average. This time frame is insufficient for the [Ca²⁺] rise to be detectable.

The tendency of each RyR cluster to spark is likely a cluster-specific property. Factors to consider include the posttranslational modification of the RyRs and their association with accessory proteins. The nature of the association between the SR and the T-tubular structures and the perimeter/size ratio of the cluster may also be of importance. Local high SR [Ca²⁺] in the relevant junction may increase RyR Ca sensitivity, and variable spacing between RyRs (32) must also be considered.

The mechanism(s) of nonspark production within RyR clusters may be partially gleaned from bilayer experiments. For instance, Zahradniková et al. (33) studied the effects of Mg²⁺ during the application of spikes of photoreleased Ca²⁺ (10–100 μM) for submillisecond timescales. They found that in near-physiological conditions most of the spikes did not induce RyR openings. When openings occurred, the kinetics was indistinguishable from that in the absence of Mg²⁺. It was concluded that a [Ca²⁺] stimulus would induce RyR openings only if the stimulus lifetime was longer than the lifetime of the Mg/RyR complex. Because four activation sites must be free of Mg²⁺ for a single RyR to open (34), this occurs very infrequently at physiological [Mg²⁺]. An additional mechanism for nonspark J_leak generated at an RyR cluster may be a variable degree of allosteric coupling (that is, a lower degree to which multiple channels open and close as a single unit would lead to more nonspark J_leak). This phenomenon (35) has been observed in clusters of two to three channels in bilayers (36).

Nonspark J_leak is also consistent with the involvement of isolated nonjunctional (rogue) RyRs, as proposed by Sobie et al. (11). However, because rogue RyRs, by definition, are isolated channels, and because physiological CICR resulting in the generation of a Ca²⁺ spark is a highly cooperative process that requires nanometer-tight clustering of RyRs, it follows that if there is any contribution of rogue RyRs to the isoproterenol-induced effects, it must be minimal.

Isoproterenol increases the fraction of spark-dependent J_leak

Shifted spark-frequency-J_leak relationship

The most parsimonious explanation for the shifted spark-frequency-J_leak relationship is that isoproterenol increased the effective sensitivity to a given Ca²⁺ trigger. The obvious mechanism would be promotion of CICR by an increased affinity of the RyR activation sites for Ca²⁺. However, diastolic CICR may also be facilitated if isoproterenol treatment resulted in perturbed cooperativity among the RyR subunits, lowering the Hill coefficient of the single-channel [Ca²⁺] open-probability curve.

Conversely, the Mg²⁺ affinity of the RyR may decrease, resulting in an apparent increased affinity for Ca²⁺ (34). It has been shown that the RyR's responsiveness to Mg²⁺ is lost (37) or altered (38) upon PKA phosphorylation in bilayers. However, those results are difficult to reconcile with the absence of change in the spark properties after PKA activation in intact cells (39). Given that our study documents an increase in spark width, duration, and frequency (Figs. 5 and 6), the data presented here are more consistent with the Ca²⁺/calmodulin-kinase-II-dependent changes observed in phospholamban knockout mice (8) and intact guinea pig cells (10). It has been shown also that a sizable Mg²⁺ extrusion occurs after a few minutes of β-adrenergic stimulation in ventricular myocytes (40), and it is possible that this Mg²⁺ loss plays a role in the observed effects.

Bovo et al. (30) recently reported an involvement of reactive oxygen species in isoproterenol-induced leak, in addition to the involvement of phosphorylation. Future studies in this area are necessary to ascertain the relationship between reactive oxygen species and our results.

Increased recruitment of RyRs during a single spark

The release-flux reconstruction yielded a wider source during isoproterenol treatment. This result implies the recruitment of additional peripheral RyRs (Fig. 6), much in line with the expectations from an increased sensitivity to a Ca²⁺ trigger. The effects of isoproterenol on the morphology of sparks were nearly the same as those of caffeine, a Ca²⁺-sensitivity enhancer, on sparks of skeletal muscle (42). The result also explains how it is possible for the spatial size of isoproterenol-induced events to be greater than that of the control sparks without a change in amplitude (Figs. 4 and 5). The additional, recruited RyRs may be, but are not necessarily, allocated outside the classical junctions (i.e., those adjacent to the Z-disks). In this sense, electron microscopy studies in rat ventricular myocytes have revealed the presence of axial junctions, which would contain as much as 20% of a cell's RyRs. Such axial junctions have lengths of 510 nm on average, but they can occasionally span an entire sarcomere (43).

Recovery from refractoriness

No change was observed in the cell-wide fraction of sparks associated with repetitive firing sites. However, the distribution of interspark intervals shifted from a unimodal distribution in control conditions to a bimodal distribution in isoproterenol-treated cells (Fig. 7). The isoproterenol mode with the longest intervals had properties similar to the control distribution, whereas the second mode represented intervals of much shorter duration. We interpret these data as indicating an increased recovery of refractoriness for many of the Ca²⁺ release units. The rate of recovery at any individual site depends upon the connectivity of its local SR network, which determines the speed of Ca²⁺ reloading (44). The extent to which this variable connectivity affected the recovery of refractoriness is unknown.

The interspark periods reported by Ramay et al. (45) are on the order of 60 ms, which is much shorter than the intervals reported here. We believe that those results are very compatible with ours, because Ramay et al. used 50 nM ryanodine in their solutions. At those concentrations, ryanodine-bound channels work as pacemakers that originate the repetitive events.

Summary

This study provides a framework to investigate spark versus nonspark RyR-mediated diastolic release. Our results suggest that the fraction of spark-mediated J_leak can be increased by isoproterenol even when there are no changes in [Ca]_SRT and [Ca]_i. Therefore, in addition to these factors, high adrenergic tone may make triggered arrhythmias more likely in diseased patients by a mechanism that is independent of changes in [Ca²⁺].