Quarky Calcium Release in the Heart

Abstract

Rationale

In cardiac myocytes, “Ca²⁺ sparks” represent the stereotyped elemental unit of Ca²⁺ release arising from activation of large arrays of ryanodine receptors (RyRs), whereas “Ca²⁺ blinks” represent the reciprocal Ca²⁺ depletion signal produced in the terminal cisterns of the junctional sarcoplasmic reticulum. Emerging evidence, however, suggests possible substructures in local Ca²⁺ release events.

Objectives

With improved detection ability and sensitivity provided by simultaneous spark-blink pair measurements, we investigated possible release events that are smaller than sparks and their interplay with regular sparks.

Methods and results

We directly visualized small solitary release events amidst noise: spontaneous Ca²⁺ quark-like or “quarky” Ca²⁺ release (QCR) events in rabbit ventricular myocytes. Because the frequency of QCR events in paced myocytes is much higher than the frequency of Ca²⁺ sparks, the total Ca²⁺ leak due to the small QCR events is approximately equal to that of the spontaneous Ca²⁺ sparks. Furthermore, the Ca²⁺ release underlying a spark consists of an initial high-flux stereotypical release component and a low-flux highly variable QCR component. The QCR part of the spark, but not the initial release, is sensitive to cytosolic Ca²⁺ buffering by EGTA, suggesting that the QCR component is attributable to a Ca²⁺-induced Ca²⁺ release mechanism. Experimental evidence, together with modeling, suggests that QCR events may depend on the opening of rogue RyR2s (or small cluster of RyR2s).

Conclusions

QCR events play an important role in shaping elemental Ca²⁺ release characteristics and the nonspark QCR events contribute to “invisible” Ca²⁺ leak in health and disease.

Introduction

The Ca²⁺ ion is the most versatile intracellular messenger involved in vital cellular processes that include contraction, secretion, apoptosis, and gene expression regulation underlying cell proliferation and differentiation¹. In cardiac cells, Ca²⁺ entry through voltage-gated Ca²⁺ channels in the plasma membrane during an action potential triggers intracellular Ca²⁺ release through the type 2 ryanodine receptor (RyR2) Ca²⁺ release channels in the sarcoplasmic reticulum (SR), a process known as Ca²⁺-induced Ca²⁺ release (CICR). Depending on species and developmental stage, the SR amplification of the Ca²⁺ influx varies from almost 0 (early development) up to 16 (adult rat or mouse heart)^{2, 3}.

As one continuous organelle⁴, the SR takes the form of an elaborated nanoscopic tubular and cisternal network that extends throughout the entire cytoplasm but occupies only ≈4% of the ‘‘cytoplasmic’’ volume in rabbit ventricular myocytes⁵. It is also connected to both the endoplasmic reticulum and the nuclear envelope Ca²⁺ stores that are implicated in non-contractile Ca²⁺ signaling⁴. Ca²⁺ release from the cardiac SR is controlled almost exclusively by RyR2 Ca²⁺ release channels. The SR membrane harbors ≈10⁶ RyR2s per myocyte, and ≈30–300 RyR2s form a two-dimensional paracrystalline array or Ca²⁺ release unit (CRU). A three-dimensional constellation of about 10⁴ CRUs are spangled regularly throughout the cell^{6, 7}. Most recent data suggest that some RyR2s within the regular organization of an array could be missing^{8, 9}, suggesting that clusters of RyR2s of different sizes (including rogue RyR2s¹⁰) could share the same junctional SR (jSR).

The prevalent view is that activation of a single CRU gives rise to a local, discrete Ca²⁺ release event – a Ca²⁺ spark¹¹. Summation of thousands of Ca²⁺ sparks that are activated during each heart beat gives rise to a whole-cell [Ca²⁺]_i transient during cardiac excitation-contraction coupling ^{12, 13}. Simply put, this view suggests that there are no other forms of Ca²⁺ release but Ca²⁺ sparks in cardiac myocytes under physiological conditions. Several lines of observation, however, have challenged this view. Lipp and Niggli initially reported that CICR dependent SR Ca²⁺ release may occur in a spatially uniform fashion under specific conditions¹⁴. They ascribed the unresolved fundamental Ca²⁺ release events to the subresolution entity dubbed “Ca²⁺ quarks”. Later they supported this hypothesis by showing that two-photon photolysis of Ca²⁺-caged compounds can trigger local Ca²⁺ release events that appears to be smaller than Ca²⁺ sparks¹⁵. Using loose-seal patch-clamp and confocal imaging techniques, Cheng and colleagues demonstrated that individual release events evoked from the same CRU differ significantly by virtue of amplitude, suggesting polymorphism of Ca²⁺ sparks¹⁶. Moreover, Sobie et al. hypothesized that rogue RyR2s may act to produce SR Ca²⁺ efflux at a lower level than seen with the stereotyped CRU activation that produces Ca²⁺ sparks¹⁰. Collectively, these reports question whether or not Ca²⁺ sparks are the sole "elementary" event underlying SR Ca²⁺ release. Alternatively, Ca²⁺ sparks may possess intriguing characteristics that are still unknown to us despite a 15-year intensive investigation.

As a Ca²⁺ spark appears, a local SR Ca²⁺ depletion signal, called a Ca²⁺ blink, can be observed⁵. Blinks and sparks are two manifestations of the same elementary Ca²⁺ release event, viewed from the SR lumen and the cytosolic space, respectively. The SR consists of special regions that include the jSR and the free SR (fSR) and each component is distinct with varying concentrations of lumenal Ca²⁺ buffers (eg calsequestrin, calreticulin), membrane-linked channels and transporters (eg RyR2s, SR Ca²⁺ ATPase [SERCA]2a) and regulatory proteins (eg triadin, junctin, phospholamban). The SR lumen does not appear to include cytosolic Ca²⁺ regulatory and buffering proteins (eg troponin, calmodulin). As such, characterization of Ca²⁺ blinks, particularly when done in parallel with Ca²⁺ sparks, should be able to provide unique insight into SR Ca²⁺ release mechanisms and more globally into intracellular Ca²⁺ signaling.

In the present study, we exploit Ca²⁺ spark-blink pairs to investigate local Ca²⁺ release events with improved detection ability and sensitivity. Our results reveal that low-flux quarky Ca²⁺ release (QCR) events coexist and interplay with regular Ca²⁺ sparks to provide rich Ca²⁺ signaling substructure and complexity (eg, dictating the decay of the sparks and the restoration of the blinks). This QCR component appears to be coupled to the high-flux initial release in a spark via the CICR mechanism, and may have arisen from activation of rogue RyR2s or complex regulation of RyR2s within an array. Our results also show that the Ca²⁺ leak mediated through the QCR mechanism appears to be at least as important as that through the Ca²⁺ spark mechanism.

Methods

To simultaneously visualize Ca²⁺ sparks and Ca²⁺ blinks, New Zealand White rabbits ventricular myocytes were first incubated for 2 hours at 37°C with the low affinity Ca²⁺ indicator fluo-5N-AM (20 µmol/L, Invitrogen) and then for 15 minutes at room temperature with the high affinity Ca²⁺ indicator rhod-2-AM (5 µmol/L, Invitrogen)⁵. The diastolic spark frequency being very low in rabbit ventricular myocytes, the SR Ca²⁺ load was increased by replacing extracellular Na⁺ (NaCl) by equimolar Li⁺ (LiCl) in the bath solution, unless otherwise noted (see Figure 3 and 5).

Figure 3. QCR in electrically paced cardiomyocytes.

A, Simultaneous measurement of diastolic QCR (rhod-2) (top) and QCD (fluo-5N) (middle) events in a paced (0.5 Hz) cardiomyocyte. Ticks to the left of the bottom image mark locations of the Z-disks. Traces on the bottom show time courses of 2 QCR-QCD events identified during diastole along with the transient. The traces of the QCR-QCD events have been enlarged in the inserts. B, Frequency of sparks and QCR events (* P<0.05). C, Averaged spark and QCR signal mass (* P<0.05). D, Summed Ca²⁺ leak of sparks or QCR events.

Figure 5. Effect of EGTA on kinetics and amplitudes of spark-blink pairs.

A, Linescan images of a spark (left) and its conjugate blink (center left) obtained in 0.5 mmol/L EGTA after background subtraction along with the blink automated detection (center right). The time courses of the spark and blink are also shown (right). The black lines between the images mark the positions of jSR located at the Z-disks. B, Same as A, except in a solution containing 2 mmol/L EGTA. C, Bar graph of spark and blink t_peak/t_nadir and t₆₇ in 0.5 and 2 mmol/L EGTA, * P<0.05. D, Scatter plot of spark and blink t₆₇ values in 0.5 and 2 mmol/L EGTA along with their regression lines. E, Spark and blink amplitudes in 0.5 and 2 mmol/L EGTA. * P<0.05.

An expanded Methods section is available in the Online Supplement at http://circres.ahajournals.org.

Results

Spark-Blink Pair Analysis Increased Detection Ability of In-Focus Ca²⁺ Release Events

Figure 1A showed a representative Ca²⁺ spark-blink pair in linescan images from a rabbit ventricular myocyte. Spatial profiles (Figure 1B) revealed that the blink aligned well with the jSR (J) and was sharply confined (full-width at half maximum, [FWHM]=1.01±0.05 µm, n=51), reflecting the nanometer-sized jSR ultrastructure and restricted intra-SR Ca²⁺ diffusion between the jSR and the fSR⁵. In contrast, the companion spark, which reflected the local Ca²⁺ release, Ca²⁺ diffusion and buffering in the cytosolic space, spanned the entire sarcomere and even spread into the neighboring junctions (J−1, J+1), displaying a FWHM of 2.27±0.12 µm (n = 51) and a volume 10-times greater than for a blink. Because of the sharp spatial delineation of Ca²⁺ blinks, we have exploited blinks as the guide to select in-focus release events, i.e., those sparks associated with discernible blinks. Under the present experimental conditions, ≈60% of Ca²⁺ sparks were rejected as out-of-focus events for the lack of companion blinks.

Figure 1. Spark-blink pairs.

A, Linescan images of simultaneous measurement of a Ca²⁺ spark (rhod-2) (left) and its companion Ca²⁺ blink (fluo-5N) (middle) in an intact rabbit ventricular myocyte are shown after background subtraction. Unsubtracted fluo-5N image (right) shows the enrichment for the fluo-5N dye in the jSR at the Z-disk. The spark-blink pair is centered on the jSR band which is labeled as “J”. B, Spatial profiles of the spark-blink pair. Arrows mark jSR locations (J, J−1, J−2, J+1 and J+2). C, Time courses of the spark-blink pair.

Simultaneous measurement of Ca²⁺ sparks and Ca²⁺ blinks provided us with the unprecedented ability to identify small local Ca²⁺ release fluxes amidst background noise. Using Gaussian noise simulated traces, we showed that dual channel simultaneous detection suppresses false positive events by ≈150-fold as compared to single-channel detection (Online Figure I), allowing the detection of a population of QCR events (see below).

Quarky Ca²⁺ Release: Novel Type of Local Ca²⁺ Signaling

Figure 2A–B showed an example of a succession of spark-blink pairs and QCR-quarky Ca²⁺ depletion (QCD) pairs on the same jSR. The kinetics of these events were very short (QCR: t_peak= 19.1±1.0 ms, t₆₇=20.1±1.1 ms; QCD: t_nadir= 19.2±1.3 ms, t₆₇=20.8±1.9 ms; n=42; Figure 2C). The amplitude of these reciprocal QCR-QCD events were only one-tenth to one-third of the regular spark-blink events (QCR: ΔF/F₀=0.069±0.006; QCD: ΔF/F₀=0.025±0.002; n=42; Figure 2D), suggesting that the QCR events may arise from a mode of Ca²⁺ release that is different from that which underlies the regular spark-blink pairs. This idea was substantiated by the amplitude histogram that suggests the existence of two distinct populations of Ca²⁺ release events (Figure 2E, n=176). Nevertheless, the properties of QCR events still appear larger than the isolated opening of a single RyR2 (i.e. the true Ca²⁺ “quarks” hypothesized by Lipp & Niggli¹⁴) since the Ca²⁺ flux of a single RyR2 Ca²⁺ quark is expected to be even smaller with a duration much shorter due to rapidly closing by stochastic attrition¹⁷.

Figure 2. Quarky Ca²⁺ Release (QCR).

A, Succession of Ca²⁺ sparks and QCR events on the same jSR (top), the corresponding Ca²⁺ blinks and quarky SR Ca²⁺ depletion (QCD) events (middle) and their time courses (bottom). Ticks to the left of the bottom image mark locations of the Z-disks. Arrows denote QCR and QCD on the images and time course plots. B, Enlarged view of the last succession of QCR and spark (top left) and the corresponding QCD and blink (top right) from panel A, the automated detection of QCD/blink (middle right) and the corresponding time courses (bottom). Arrows denote QCR and QCD on the images and time course plot. C&D, Kinetics (C) and amplitude (D) of QCR and QCD. * P<0.05. E, Histogram distribution of amplitude of Ca²⁺ release events (QCR and spark).

To evaluate the significance of these QCR events in more physiological conditions, we studied QCR when rabbit ventricular myocytes were electrically stimulated (0.5 Hz) in a regular HEPES buffer (Figure 3A). We found that QCR events were 10 times more frequent than spontaneous Ca²⁺ sparks (2.77±0.31 versus 0.24±0.06 [100 µm]⁻¹sec⁻¹; Figure 3B), while sparks displayed a 10 times greater signal mass (7.47±1.55 versus 0.62±0.13; Figure 3C). Consequently, the summed SR Ca²⁺ leak due to sparks was similar to that mediated by QCR events (1.48±0.37 versus 1.49±0.34). Taking into account that numerous QCR events may have gone undetected, we conclude that QCR represent an important, heretofore underappreciated SR Ca²⁺ leak mechanism (see below).

Spark-Blink Kinetics: Evidence for Low-Flux Continued Ca²⁺ Release

Careful examination of the kinetics of Ca²⁺ spark-blink pairs revealed substantial variation with respect to the decay of the sparks and the restoration of the blinks (Figure 4). Specifically, the spark decay time (t₆₇) exhibited a broad distribution with a mode at 45 ms, ranging from 25 (at 5% cutoff of the lower limit) to 95 ms (at 95 % cutoff of the upper limit, Figure 4A). Such a large variation was unexpected (see Discussion) but could not be dismissed as an artifact of confocal detection because we observed a similar large variation in blink recovery time (t₆₇) (Figure 4A), the latter being essentially undistorted by confocal sampling⁵.

Figure 4. Spark-blink characteristics.

A, Histogram distributions of spark decay time and blink recovery time (t₆₇). B, Scatter plot of spark and blink t₆₇ along with their regression line. C, Top, Average traces of sparks and blinks for t₆₇ (spark) <50 ms (24 events), 5<t₆₇<70 ms (19 events) and t₆₇>70 ms (7 events). Bottom, the same traces after normalization by the amplitude. D, Bar graphs of t_peak and t_nadir for sparks and blinks, respectively, for the same t₆₇ groups as in panel C. E, Peak amplitudes of sparks and blinks for the same t₆₇ groups as in panel C. F, Relationship between spark kinetics and amplitudes. Scatter plot of spark t₆₇ and amplitudes along with their regression line.

To investigate whether this variability could be simply attributed to regional differences in Ca²⁺ diffusion or buffering, we examined the relationship between Ca²⁺ sparks and their corresponding Ca²⁺ blinks, the latter reflecting local jSR depletion and refilling. Because Ca²⁺ diffusion and buffering differ markedly in the SR lumen versus the cytosol¹⁸, we had expected little correlation between the recovery of blinks and the decay of their companion sparks. However, both Ca²⁺ sparks and blinks exhibited similar average durations (t_peak=30.4±1.3 ms and t₆₇=54.3±2.8 ms for sparks versus t_nadir=29.6±1.4 ms and t₆₇=55.5±2.6 ms for blinks, n=51 from 31 cells). In addition, scatter plots showed a strong linear correlation between the t₆₇ of spark decay and the t₆₇ of blink recovery, with a slope of 0.88 and r²=0.91 (Figure 4B).

To further study the kinetics of the spark-blink pairs, we grouped spark-blink pairs based on t₆₇ of the sparks, which usually display better signal-to-noise ratios than the blinks. For events in the same group (t₆₇<50 ms, n=24 events; 50 ms<t₆₇<70 ms, n=19 events; or t₆₇>70 ms, n=7 events), we aligned the simultaneously recorded sparks and blinks using the spark peak as the reference point, and resultant averaged traces are shown in Figure 4C. The longer the spark decay, the slower the corresponding blink recovery. More strikingly, once normalized by the amplitude, sparks and blinks showed virtually overlapped time courses in every group (Figure 4C). The tight correlation between spark and blink kinetics indicates that a common mechanism underlies the variability both in the cytosol and the SR lumen. That the average t₆₇ value varied by more than twofold in the three groups suggests that this variability is not readily attributable to a change in local Ca²⁺ recycling by the SERCA2a, which plays only a minor role in shaping the spark kinetics¹⁹. Rather, these results indicate the presence of low-flux continued Ca²⁺ release that dictates the decay-recovery kinetics of the spark-blink pairs.

In contrast, the initial Ca²⁺ release appears to be stereotyped: there was little variation in the time-to-peak of the spark or time-to-nadir of the blink. The average value (≈30 ms) was similar for all different categories of spark-blink pairs (Figure 4D); sparks or blinks displayed similar amplitudes regardless of their duration (Figure 4E). Moreover, there was little correlation between the amplitude and t₆₇ of sparks (r²=0.0028, Figure 4F). This result indicates that, once triggered, the continued release component is self-sustained, independently of the size of initial release. Collectively, our data support the notion that the spark-blink pair consists of a stereotyped initial release component and a highly variable small flux trailing release component.

Effects of EGTA on Continued Ca²⁺ Release

To test the hypothesis that the trailing small-flux release was triggered by initial high flux release via local CICR, we examined Ca²⁺ spark-blink kinetics in the presence of two EGTA concentrations (0.5 or 2 mmol/L) in chemically permeabilized myocytes (Figure 5A–B). As an exogenous Ca²⁺ chelator, EGTA can effectively reduce the physical size and duration of Ca²⁺ sparks in the cytosol²⁰. However, EGTA was expected to exert no such effect on the companion Ca²⁺ blinks as long as Ca²⁺ release process remained unaltered. Should EGTA enter the SR, it should not materially affect the dynamics, either, as its high affinity for Ca²⁺ (K_d≈150 nmol/L at pH 7.2) renders it saturated at the high levels of [Ca²⁺]_SR normally observed (in the range of a few hundreds micromolar to a few millimolar).

We found that in the presence of 0.5 mmol/L EGTA, Ca²⁺ spark and Ca²⁺ blink t₆₇ values were 40.6±2.2 and 44.7±2.4 ms (n=58), respectively, which were 13.7 and 10.8 ms faster than the spark-blink pairs in intact cells in the absence of EGTA. Furthermore, in 2 mmol/L EGTA, Ca²⁺ spark and Ca²⁺ blink t₆₇ values were further shortened to 29.1±2.4 and 31.7±2.5 ms (n=26, p<0.005), respectively (Figure 5C). While the EGTA shortening of spark duration has been demonstrated before, the EGTA effect on Ca²⁺ blinks was totally unexpected for the reasons discussed above. The shortening of Ca²⁺ blinks suggested that the Ca²⁺ release process itself has been abbreviated by the inclusion of the Ca²⁺ buffer in the cytosol. Such EGTA sensitivity provided the first experimental evidence that a CICR-dependent mechanism participates in the release of an ongoing spark.

Analysis of the signal amplitudes showed that the Ca²⁺ spark amplitude decreased from 0.96±0.06 (n=58) in 0.5 mmol/L EGTA to 0.65 ± 0.06 (n=26, p<0.01) in 2 mmol/L EGTA, as expected. By contrast, the blink amplitude displayed no significant change (0.22±0.01, n=58 in 0.5 mmol/L EGTA versus 0.20±0.01, n=26 in 2 mmol/L EGTA, p>0.05, Figure 5E), suggesting that the amount of the initial SR Ca²⁺release was essentially unchanged. The time-to-nadir of the blink was only marginally shortened (26.6±0.81, n=58, and 23.5±0.95 ms, n=26, in 0.5 and 2 mmol/L EGTA, respectively, p<0.01, Figure 5C), indicating that the initial release, which determined the peak and nadir of the spark-blink pair, was largely EGTA-insensitive. Regardless of EGTA concentration, there was a linear correlation between spark and blink t₆₇ values (Figure 5D), as was the case in intact cells in the absence of EGTA (Figure 4B). Taken together, these results reinforce the idea that elemental SR Ca²⁺ release events consist of a large EGTA-resistant initial release flux followed by a trailing EGTA-sensitive lower-flux release of variable duration.

Substructures in Prolonged Spark-Blink Pairs

The aforementioned data suggested that Ca²⁺ release continues beyond the peak or the nadir of the spark-blink pair, which confers the variability to the decay-recovery kinetics. We further hypothesized that such continued low-flux Ca²⁺ release is of the same or similar nature than the solitary QCR events described above. Two questions arise from these observations: 1. how can we directly visualize QCR-QCD substructures in a spark-blink pair and 2. how does the presence of the QCR mechanism fit with robust Ca²⁺ spark (and Ca²⁺ blink) termination? In this regard, we noticed a subpopulation of prolonged Ca²⁺ sparks (t₆₇>mean+4 SD=135 ms) occurring spontaneously in intact rabbit ventricular myocytes, examples of which are shown in Figure 6.

Figure 6. QCR-QCD events during long spark-blink pairs.

A through C, Three examples of long sparks (left) and long blinks (middle), with the corresponding time courses on the right. Ticks next to the images denote the position of the jSR at the Z-disks.

Figure 6A shows a spontaneous long-lasting Ca²⁺ spark that rises to a peak and then develops a low long-lasting plateau before its termination, similar to the prolonged sparks induced by FK506, rapamycin or low-dose ryanodine in rat ventricular myocytes^{11, 17}. The companion blink displays a clear nadir at the peak of the spark and then returns towards a low plateau. The sustained low [Ca²⁺]_SR signal is consistent with continued Ca²⁺ efflux, and reflects a dynamic balance between the efflux and the refilling of the jSR. Presumably the prolonged or sustained Ca²⁺ sparks and blinks arise from sustained local CICR at a single jSR due to one or more RyR2s remaining active/open, a suggestion consistent with the low dose ryanodine (or other) treatment. More complex examples of sustained release are shown in Figures 6B and 6C, illustrating repetitive activation of the QCR mechanism during prolonged spark-blink pairs. In both examples, the kinetics of the [Ca²⁺]_SR closely mirrored the kinetics of the sparks. The succession of “bumps” (QCR events) of the prolonged Ca²⁺ spark are matched by “dips” (QCD events) of the prolonged Ca²⁺ blink, with an overall negative correlation coefficient r =−0.89 and −0.87, respectively. Interestingly, the long spark-blink pairs were not seen in 2 mmol/L EGTA (Figure 5), consistent with CICR triggering of QCR-QCD events during prolonged Ca²⁺ sparks.

Discussion

Novel features of Ca²⁺ release from the SR in heart cells have been identified and characterized here by dual imaging of Ca²⁺ signals in the cytosol and in the lumen of the SR. In addition to Ca²⁺ sparks, we report on the detection and characterization of small amplitude local Ca²⁺ release events that are spatially and temporally well-confined (i.e., QCR-QCD events). The Ca²⁺ leak mediated by these QCR events was at least as important as those by Ca²⁺ sparks. We also show that similar low-flux Ca²⁺ release events can occur in conjunction with a regular Ca²⁺ spark. The latter is vividly manifested by the trailing QCR-QCD events in a prolonged Ca²⁺ spark and may underlie the variability in the decay kinetics of a Ca²⁺ spark. The newly identified QCR and substructure of Ca²⁺ signals in a spark (see Figure 7) provoke a number of questions that are discussed below.

Figure 7. Schematic model of QCR activation.

Ca²⁺ spark arising from a CRU composed of 2 clusters of RyR2s and several rogue RyR2s (left). This Ca²⁺ spark may contain QCR events. B. QCR arising from rogue RyR2s on the same CRU as in A.

Local Ca²⁺ Release during a Spark

Although polymorphism of Ca²⁺ spark amplitude has been demonstrated previously,^{16, 21} here, we show for the first time that the kinetics of in-focus Ca²⁺ sparks display duration variation ascribed to the variability of the declining phase. Surprisingly, there is a tight correlation between kinetics of sparks and their corresponding blinks despite marked differences in calcium diffusion and buffering in the cytosol and the SR lumen. Moreover, increased cytosolic Ca²⁺ buffering by EGTA similarly abbreviates both Ca²⁺ sparks and Ca²⁺ blinks. These results suggest a number of important features of the spark release function. First, the SR Ca²⁺ release flux that contributes to the Ca²⁺ spark must last longer than initially hypothesized¹¹, i.e., it must continue after the peak of the spark. Second, the release fluxes that underlie Ca²⁺ sparks appear to be a decreasing function of time, as also suggested by derivation of the release function from spark measurement²² and in the modeling work of Sobie et al.¹⁷. Third, the SR Ca²⁺ flux is long enough and strong enough to dominate the effect of cytosolic Ca²⁺ diffusion on spark kinetics, and the effect of jSR refilling on the blink kinetics. These findings in cardiac cells are also in general agreement with reports on “Ca²⁺ embers” under various experimental conditions in skeletal muscles²³ although, unlike heart, square-wave release fluxes are thought to be the norm²⁴. It should also be noted that Zima et al.²⁵ recently reported a very long blink recovery time (≈161 ms on average) that was attributed to a slow intra-SR Ca²⁺ diffusion due to restrained connection between fSR and some jSR. However, our data suggests that the recovery of blinks can be very fast as evidenced by the population of blinks with t₆₇<70 ms in the absence and presence of millimolar EGTA. In light of the present findings, a more plausible explanation for Zima et al ²⁶ is that Ca²⁺ sparks can have variable kinetics and can display enhanced trailing SR Ca²⁺ release which would confer the unexpected slow recovery of the associated Ca²⁺ blink.

Furthermore, we show that the initial high Ca²⁺ release flux in a Ca²⁺ spark is stereotypical, in contrast to the trailing Ca²⁺ release flux. More importantly, the trailing, but not the initial, Ca²⁺ release flux is sensitive to cytosolic Ca²⁺ buffering by EGTA. Hence, the local SR Ca²⁺ release during a Ca²⁺ spark occurs in two modes - one that involves a high-flux Ca²⁺ release that operates in a largely all-or-none fashion; and a second one that involves low-flux Ca²⁺ release that is highly variable. The fact that EGTA abbreviates the spark and blink duration by selectively acting on the low-flux release mode is important as it indicates that the trailing release is activated and sustained by a CICR mechanism and may be more susceptible to modulation or perturbation by physiological and pathophysiological factors as well as changes in experimental conditions.

Possible Mechanism Underlying Subtler Local Ca²⁺ Release

Although QCR can occur at the same site as regular sparks (at the optical resolution), QCR events have distinct features including their small size and apparent lack of refractoriness, the latter is evidenced by repetitive activation of QCR events in long sparks. Our working hypothesis (Figure 7) is based on recent publications from Soeller⁸ and Hoshijima⁹ labs. Using optical super-resolution technique (PALM)⁸ or electron tomography⁹, they have shown that large clusters of RyR2s constituting the CRU were incompletely filled with RyR2. This leaves the possibility for some RyR2s to be rogue (single or in a small cluster) and in close proximity to the large cluster of RyR2s. Such rogue RyR2s may display higher CICR sensitivity (at low [Ca²⁺]_i levels) than the large cluster of RyR2s¹⁰. Using our previous numerical spark model²⁶, the differential effect of EGTA (between 0.5 and 2 mmol/L) on spark kinetics was modest as t₆₇ was only reduced by 5.5 ms in the presence of 2 mmol/L EGTA. In contrast, our experimental data showed that 2 mmol/L EGTA reduced t₆₇ by 11.5 ms. If we consider an opening duration of rogue RyR2s of 1 ms²⁷ and a release current of 5% of the initial release of a Ca²⁺ spark, the opening of 6 successive rogue RyR2s 10 ms after the peak of the spark would increase t₆₇ by ≈8 ms. The question whether or not these rogue RyR2s would be on the same jSR than larger clusters of RyR2s has also been further answered by the EGTA experiments. Our modeling suggested that 2 mmol/L EGTA can indeed exert substantial buffering effect over the distance of a half or full CRU (average width of 465 nm⁵, Online Figure II).

In our Ca²⁺ spark and QCR model, solitary QCR-QCD events arise from spontaneous activation of rogue RyR2s (Figure 7B), while the spark-commingled QCR events are activated by preceding CRU release via the CICR mechanism (Figure 7A). An important distinction of CRU and rogue release is that the former has a steeper [Ca²⁺]_i dependence due to the cooperativity of interconnected RyR2s¹⁰. This logic suggests two additional features: First, QCR-QCD events from rogue RyR2s would rarely trigger their nearby "master" CRU (in a retrograde manner). Second, repetitive openings of rogue RyR2 U (or CICR interplay among a few rogue RyR2 U) explain the high variability, and the EGTA sensitivity of the trailing release. As a cautionary note, we have assumed that isolated QCR and the straggling QCR on the tail of long sparks are of common mechanism, as judged by their phenotypical similarities. Nevertheless, the similarities may be only apparent and they may arise from distinct populations of rogue RyR2s. Future investigations are warranted to discriminate these possibilities.

Implications in SR Ca²⁺ Leak

All means by which Ca²⁺ exits the SR is "Ca²⁺ leak". The most common measurement of Ca²⁺ leak is Ca²⁺ spark. Larger SR Ca²⁺ releases like macrosparks, long sparks or Ca²⁺ waves are also easy to detect. However, so far, the small SR Ca²⁺ release events have been elusive until now with the new results described here. By using spark-blink pairs as a guide to identify events amidst noise, we have been able to show that QCR is a new form of active Ca²⁺ leak that otherwise would have been “invisible” or dismissed as non-significant. Since the real frequency of QCR events is difficult to estimate, we can only suggest that the real contribution of these small release events to Ca²⁺ leak is at least as important as Ca²⁺ leak due to sparks. In addition, when a QCR is occurring, the nearby jSR could be more likely to have a spontaneous Ca²⁺ spark or participate in a conducted wave. Overall, this component of Ca²⁺ leak may be part of an invisible Ca²⁺ leak that might become aggravated in disease. The existence of this invisible leak may also be of importance in setting the integrated "pump-leak balance" of the SR and thereby in the fine tuning of local and global Ca²⁺ signaling

In summary, we have demonstrated substructures in elementary SR Ca²⁺ release events, and identified a new population of subtler release events, the QCR which can be seen when the viewing is guided by "spark-blink pair" measurements. The cumulative Ca²⁺ leak associated with QCR appears to be as large as those associated with spontaneous Ca²⁺ sparks. Evidence suggests that the QCR may depend on the opening of rogue RyR2s or complex CRU regulation of RyR2s. Taken together, the understanding of coexistence and interplay of quarky and spark Ca²⁺ release events and their distinct regulation leads us to novel and important insight into cardiac Ca²⁺ signaling. If supported by additional independent evidence, our new understanding of QCR may lead to novel strategies for the control of SR Ca²⁺ leak and for the treatment of heart failure and Ca²⁺ dependent arrhythmias.

Novelty and Significance.

What Is Known?

• Ca²⁺ sparks represent the elemental units of Ca²⁺ release from the sarcoplasmic reticulum (SR) of the cardiac myocyte.

• Ca²⁺ depletion from the junctional (j)SR during a Ca²⁺spark can be measured and has been termed a “Ca²⁺ blink.”

• Clusters of type 2 ryanodine receptors (RyR2s) can be different sizes but could share the same jSR.

What New Information Does This Article Contribute?

• Imaging jSR and cytosolic Ca²⁺ simultaneously enables the detection of subtle Ca²⁺ release events that otherwise would be difficult to discriminate from noise.

• Using this method, we detected low amplitude, solitary Ca²⁺ release events, referred to as quarky Ca²⁺ release (QCR), which occurred either independently or during the declining phase of a full amplitude Ca²⁺ spark.

• QCR events, but not the primary spark-mediated Ca²⁺ release, were suppressed by the slow Ca²⁺ buffer EGTA, indicating that they were triggered by a Ca²⁺-induced Ca²⁺ release (CICR) mechanism. The stochastic recruitment of QCR events spawned by the Ca²⁺ spark plausibly explains the variability of spark duration.

• In paced myocytes, QCR events were so frequent that the SR Ca²⁺ leak from these events could be equal to that through Ca²⁺ sparks.

• QCR events not associated Ca²⁺ sparks could contribute to “invisible” Ca²⁺ leak in health and disease.

Ca²⁺ sparks represent the elemental unit of Ca²⁺ release. They result from the activation of large arrays of RyR2s. Recently, it has been shown that clusters of RyR2s of different sizes could share the same jSR, suggesting possible substructures in local Ca²⁺ release events. We tested this hypothesis by visualizing QCR events, which probably depend on the opening of small clusters of RyR2s (or rogue RyR2s). QCR events might also be responsible for the variability of the declining phase of a spark. Because the frequency of QCR events in paced myocytes is much higher than the frequency of Ca²⁺ sparks, the total Ca²⁺ leak caused by small QCR events is approximately equal to that of the spontaneous Ca²⁺ sparks. This new mode of Ca²⁺ release may be the mechanism behind the so-called “invisible” Ca²⁺ leak. These findings suggest new approaches for the treatment of heart diseases characterized by exaggerated SR Ca²⁺ leak, such as heart failure and catecholaminergic polymorphic ventricular tachycardia.

Non-standard Abbreviations and Acronyms

CICR

Ca²⁺-induced Ca²⁺ release

CRU

Ca²⁺ release unit

fSR

free SR

jSR

junctional SR

QCD

quarky Ca²⁺ depletion

QCR

quarky Ca²⁺ release

RyR2

type 2 ryanodine receptor

SERCA

SR Ca²⁺ ATPase

SR

sarcoplasmic reticulum