Local recovery of Ca 2+ release in rat ventricular myocytes

J Physiol 565.2 (2005) pp 441 447 441 RPID REPORT Local recovery of Ca 2+ release in rat ventricular myocytes Eric. Sobie, Long-Sheng Song and W. J. Lederer Medical iotechnology Center, University of Maryland iotechnology Institute, altimore, MD, US Excitation contraction coupling in the heart depends on the positive feedback process of Ca 2+ -induced Ca 2+ release (CICR). While CICR provides for robust triggering of Ca 2+ sparks, the mechanisms underlying their termination remain unknown. t present, it is unclear how a cluster of Ca 2+ release channels (ryanodine receptors or RyRs) can be made to turn off when their activity is sustained by the Ca 2+ release itself. We use a novel experimental approach to investigate indirectly this issue by exploring restitution of Ca 2+ sparks. We exploit the fact that ryanodine can bind, nearly irreversibly, to an RyR subunit (monomer) and increase the open probability of the homotetrameric channel. y applying low concentrations of ryanodine to rat ventricular myocytes, we observe repeated activations of individual Ca 2+ spark sites. Examination of these repetitive Ca 2+ sparks reveals that spark amplitude recovers with a time constant of 91 ms whereas the sigmoidal recovery of triggering probability lags behind amplitude recovery by 80 ms. We conclude that restitution of Ca 2+ sparks depends on local refilling of SR stores after depletion and may also depend on another time-dependent process such as recovery from inactivation or a slow conformational change after rebinding of Ca 2+ to SR regulatory proteins. (Resubmitted 10 March 2005; accepted 6 pril 2005; first published online 7 pril 2005) Corresponding author W. J. Lederer: Medical iotechnology Center, 725 W. Lombard Street, altimore, MD 21201, US. Email: lederer@umbi.umd.edu The triggering of Ca 2+ sparks, localized increases in intracellular Ca 2+ concentration ([Ca 2+ ] i )(Chenget al. 1993), is the key event linking electrical excitation to contraction in cardiac muscle cells (Guatimosim et al. 2002; Wier & alke, 1999). Each spark represents the release of calcium from a cluster of Ca 2+ -sensitive RyRs in the sarcoplasmic reticulum (SR) membrane. The synchronized triggering of numerous ( 10 000) Ca 2+ sparks upon membrane depolarization results in an increase in average [Ca 2+ ] i from 100 nm to 1 µm. These Ca 2+ ions then bind to myofilaments to initiate contraction. Ca 2+ entry through L-type Ca 2+ channels in the cell membrane has been established as the primary trigger of Ca 2+ sparks (Cannell et al. 1995; Collier et al. 1999; Lopez-Lopez et al. 1995). However, much remains unknown about how these events terminate and how release recovers after termination. Each Ca 2+ spark is locally a positive feedback event in that Ca 2+ passing through an open RyR can activate that channel and the other channels in the cluster. In theory, then, Ca 2+ sparks could continue indefinitely; however, they last roughly 30 ms under normal conditions. This brevity emphasizes the strength of the mechanism that overcomes the intrinsic positive feedback of CICR to terminate Ca 2+ sparks. The mechanism that underlies Ca 2+ spark termination can be probed by examining the time course of recovery after the spark ends, a process called restitution. If probabilistic closing (i.e. stochastic attrition, see Stern, 1992) of RyRs were solely responsible, restitution would have no time dependence. If, however, sparks terminate due to a shift in RyR gating properties (e.g. RyR inactivation), then an interval would have to elapse before sparks could again be triggered. Stated alternatively, at the moment of termination the cluster of RyRs would be refractory to the Ca 2+ stimulus that had been keeping the channels open. However, little is known about the factors that control Ca 2+ spark restitution. limited number of studies have demonstrated refractoriness of Ca 2+ release at the subcellular level (Shamet al. 1998; Tanaka et al. 1998), but progress has been hampered by the technical difficulty of triggering consecutive Ca 2+ sparks from the same cluster of RyRs. The experimental strategy we use to study restitution of Ca 2+ sparks in rat ventricular myocytes is illustrated in Fig. 1. t diastolic levels of [Ca 2+ ] i,stochastic openings of RyRs are infrequent, and spontaneous sparks are relatively rare. Thus, over a time scale of hundreds of milliseconds, the probability of observing two consecutive Ca 2+ sparks DOI: 10.1113/jphysiol.2005.086496

442 E.. Sobie and others J Physiol 565.2 from a single site is infinitesimally small. However, low concentrations of the RyR agonist ryanodine can induce more frequent openings of the channel (uck et al. 1992; idasee et al. 2003), as in Fig. 1. Moreover, the extremely high affinity of RyRs for ryanodine (Zucchi & Ronca-Testoni, 1997; Fill & Copello, 2002) ensures that a ryanodine molecule bound to a single subunit of an RyR homotetramer can influence the channel s gating for a long period of time (seconds). We have taken advantage of these features of RyR gating by perfusing cells with a very low concentration of ryanodine (50 100 nm). s shown below, under these conditions only 0 10 ryanodine molecules bind tightly to individual RyR monomers within a cell, and these channels serve as periodic Ca 2+ sources to repetitively trigger Ca 2+ sparks. y analysing Ca 2+ sparks that occur consecutively at the same spark site, we have obtained new information about how the amplitude and the triggering of Ca 2+ sparks recover with time. Methods Cell isolation dult rats were killed by lethal injection of pentobarbital (100 mg kg 1 ), and ventricular myocytes were prepared by standard enzymatic dissociation methods (Cannell et al. 1994). Rats were maintained and experiments were conducted in accordance with the guidelines of the Institutional nimal Care and Use Committee of the University of Maryland. Experiments were performed at room temperature (22 C). Confocal recording and solutions Isolated cells were superfused with Tyrode solution containing (mm): NaCl 140, KCl 5, Hepes 5, NaH 2 PO 4 1, MgCl 2 1, CaCl 2 1.8, glucose 10 (ph 7.4). To allow for confocal imaging of [Ca 2+ ] i,cellswereloaded for 30 min with 5 µm fluo-3 M (Molecular Probes, Eugene, OR, US), then washed and stored for 20 min in Tyrode solution to enable dye de-esterification. Confocal imaging in line-scan mode was performed on a Zeiss 510 microscope. Cells were scanned with light at 488 nm from an argon-ion laser, and fluorescence above 505 nm was recorded. To generate repeated Ca 2+ sparks at a limited number of locations, 50 nm ryanodine (Calbiochem, San Diego, C, US) was added to the external solution. C Figure 1. Experimental rationale, schematic of RyR gating and representative line-scan images under different conditions. In control, RyR openings and spontaneous Ca 2+ sparks are infrequent at diastolic [Ca 2+ ]., low concentrations of ryanodine cause frequent openings, and 50 nm ryanodine applied to a quiescent cell can induce repetitive sparks. C, delays between successive sparks and their relative amplitudes were analysed to examine Ca 2+ spark restitution. nalysis of Ca 2+ spark pairs Each spark pair was analysed as follows (see Fig. 1C). Line scan images were converted to units of F/F 0 on a pixel-by-pixel basis, then Ca 2+ spark time courses were derived by averaging over ±0.5 µm from the centre of the spark. Spark-to-spark delay was computed as the duration from the first to the second spark upstroke (maximum df/dt). The first spark amplitude was calculated relative to the baseline value of 1 F 0 whereas the amplitude of the second spark in each pair was defined as the difference between the peak of the spark and the fluorescence level immediately before the upstroke. This measure was used because sparks occurring at very short delays occasionally appeared on a baseline of elevated fluorescence (see, e.g. Fig. 2).

JPhysiol 565.2 Ca 2+ spark restitution 443 Results typical line scan image acquired under control conditions is shown in Fig. 1, revealing relatively infrequent spontaneous Ca 2+ sparks (Cheng et al. 1993). dding 50 nm ryanodine generally caused markedly different behaviour (Fig. 1). In this example, five consecutive spontaneous Ca 2+ sparks occur in the same location, and each spark in the series begins within 1 s of the previous spark s end. Repetitive sparks such as this at a single spatial location were never observed in line scans from six control cells totalling over 50 s. If sparks normally occur stochastically at a rate of 100 cell 1 s 1, and each cell contains 10 000 RyR clusters, this implies that the average spark-to-spark delay at a single cluster should be roughly 100 s. Thus, we expect the probability of three consecutive spark-to-spark delays < 1s occurring by chance to be (1 e 0.01 ) 3 10 6.Inaddition, the extremely slow rate of ryanodine unbinding from the high-affinity binding site on the RyR (McGrew et al. 1989) means that a bound ryanodine molecule is likely to remain attached for several seconds. Thus, in the example shown, we are confident that the repetitive Ca 2+ sparks shown after application of 50 nm ryanodine originated from the same RyR cluster. Moreover, because a very low concentration of ryanodine was added to the outside of the cells, we think it is likely that the repetitive Ca 2+ spark site shown results from a single ryanodine molecule binding to an individual RyR monomer within the cluster. In some cells, we observed a dramatic increase in Ca 2+ spark activity upon ryanodine application such that it was difficult to determine whether repetitive sparks originated from one or several different clusters. Results from these cells have been excluded from the data presented below. Example Ca 2+ spark pairs, with different delays between the two, are displayed in Fig. 2.These results indicate that the relative amplitude of the second spark tends to be small when the spark-to-spark delay is short and larger as this delay increases. scatterplot of all spark pairs analysed (Fig. 2) confirms the impression given by the examples and shows that spark amplitude recovery can be fitted with a single exponential with a time constant of 91 ms. The relationship between RyR trigger events (i.e. openings of the ryanodine-bound channel) and the delay following a previous trigger should theoretically be described by a decaying exponential, similar to the re-opening of any non-inactivating channel. If the probability that a given opening will trigger a spark remains constant with time, the shape of the second spark delay histogram should be identical, i.e. also a decreasing exponential. In contrast, if restitution involves a time-dependent recovery from refractoriness, the histogram of spark delays should have a biphasic shape. Figure 3 shows that the histogram of spark-to-spark delays is indeed biphasic. In these experiments, sparks were quite unlikely to occur at short delays (zero sparks from 0 to 60 ms) and most likely to occur with a delay of around 240 ms. The two largest bins in the histogram are those which represent 180 240 ms and 240 300 ms. Thus, triggering of Ca 2+ sparks displays refractoriness to activation that recovers with time. The decaying phase of the histogram in Fig. 3 (delays > 240 ms) can be well Figure 2. Ca 2+ spark amplitude recovery, four typical spark pairs. The relative amplitude of the second spark increases as the delay increases., scatterplot of spark-to-spark delay versus second spark relative amplitude (126 pairs from 14 cells) and single exponential fit to the data (τ = 91 ms).

444 E.. Sobie and others J Physiol 565.2 fitted with a single exponential (red time; time constant 189 ms), suggesting that spark triggering probability is constant over this range of delays. We can derive an estimate of how this triggering probability evolves with time by dividing the values in the histogram by the fit and normalizing. This quantity, Derived P TRIG, is displayed in Fig. 3. This increases with time, indicating that the RyR cluster is refractory to activation early after the initial spark event. The time course of Derived P TRIG appears to include an initial delay of 80 ms and then to rise with a time constant of roughly 80 ms. Discussion We have investigated Ca 2+ spark restitution in intact heart cells by exploiting the ability of low doses of ryanodine to alter the gating properties of RyRs and trigger repeated Ca 2+ sparks at the self-same site. Under these conditions we found that Ca 2+ sparks occurring at short intervals tend to be smaller than their predecessors, Figure 3. nalysis of spark-to-spark delays, histogram of delays between Ca 2+ sparks (160 pairs from 9 cells). The descending portion of the histogram can be fitted by a decaying exponential (red line; τ = 189 ms)., Ca 2+ spark triggering probability was derived by dividing the values in the histogram by the exponential fit. This triggering recovery function is sigmoidal with a half-time of 150 ms. with spark amplitude recovery following a single exponential with a time constant of 91 ms. Recovery of spark triggering probability, derived from the histogram of spark-to-spark delays, follows a similar time course after a delay. These results suggest that local refilling of SR stores is an important factor in both processes. ecause the process by which Ca 2+ release recovers is likely to be related to that by which it terminates, our data may also offer insight into this important unresolved issue. Under our experimental conditions, the primary effect of 50 nm ryanodine was to generate repeated Ca 2+ sparks at a limited number of RyR clusters within the cell. Usually, zero or one repetitive site could be detected in a line scan that spanned the cell length, and sparks from these sites were morphologically similar to those originating at other sites. Previous reports (Cheng et al. 1993) observed extremely long lasting (hundreds of milliseconds) Ca 2+ sparks after application of higher concentrations of ryanodine (100 nm), consistent with the long-lasting subconductance states seen in planar lipid bilayer studies of RyR gating. However, these bilayer experiments have also demonstrated that very low ryanodine doses can cause an increase in RyR open probability with no change in conductance and only a small increase in mean open time (ull et al. 1989; uck et al. 1992; idasee et al. 2003). These observations led to the hypothesis that the latter effect results from the binding of a single ryanodine molecule to the channel whereas long-lasting subconductance states only occur when two molecules are bound. We occasionally observed long-lasting Ca 2+ sparks in our experiments; however, by adding a very low dose of ryanodine externally and by acquiring all of our recordings within the first 10 min, we minimized the opportunities for RyRs gating in the long opening mode to confound the interpretation of the results. In that we added an agent to increase Ca 2+ spark probability, our strategy is similar to that previously employed by Terentyev et al. (2002), who collected delays between repeated sparks after adding imperatoxin to permeabilized cells. However, we have significantly extended the analysis performed by these authors by examining restitution of Ca 2+ spark amplitude and deriving an estimate of the recovery of Ca 2+ spark triggering probability. This study represents the most complete examination of the local recovery of Ca 2+ release in heart cells, an important issue that, primarily due to technical difficulties, has only been addressed in a handful of previous reports. Cheng et al. (1996) recorded confocal line scan images of Ca 2+ transients evoked by field stimulation and documented restitution of stimulus-induced Ca 2+ release when stimuli were given soon after regenerative Ca 2+ waves. Their derived recovery function is slower than Ca 2+ spark amplitude recovery measured here, a consistent result since that measurement reflected a composite of

JPhysiol 565.2 Ca 2+ spark restitution 445 amplitude recovery, triggering probability recovery, and possibly also recovery of the L-type Ca 2+ current trigger. In another study, Tanaka et al. (1998) observed a tendency for evoked Ca 2+ transients to not contain sparks at locations where a spontaneous Ca 2+ spark had recently (< 25 ms) occurred. When Ca 2+ sparks did occur at these locations, little change in the Ca 2+ transient amplitude was seen. Two factors can explain the apparent discrepancy with our results. One is that the majority of their experiments were performed in the presence of isoproterenol, an agent that will speed SR refilling and may lead to faster recovery of Ca 2+ release, as was recently observed at the whole cell level (Szentesi et al. 2004). second important point is that the Ca 2+ transient amplitude measured at any given location will include contributions from sparks originating outside the plane of focus and will therefore be a relatively insensitive measure of the true amount of local SR Ca 2+ release. DelPrincipe et al. (1999) used two-photon laser pulse trains to trigger local Ca 2+ release by uncaging Ca 2+ bound to the light-sensitive Ca 2+ buffer DM-nitrophen and found no refractoriness in the ability of these pulse trains to trigger Ca 2+ sparks. However, since two-photon triggers were delivered with an interpulse delay of at least 5 fc 2 p 5 ms 20 RyRs 1.5 4 RyRs F/F 0 C 5 ms 1 D 10 ms Spark amplitude ( F/F 0 ) 1.5 1.0 0.5 % Recovery 100 50 spark amplitude channel availability 0 5 10 15 20 # open channels 200 400 600 800 1000 time (ms) Figure 4. Computational analysis of Ca 2+ spark amplitude recovery Two computer models were used to examine the factors controlling Ca 2+ spark amplitude recovery. One model simulated the balance between JSR emptying via Ca 2+ release and refilling via SERC pumps and diffusion from network SR (NSR). One-dimensional diffusion (D Ca = 250 µm 2 s 1 ) was simulated between a disk of JSR (300 nm diameter, 75 nm thick) and a tube of NSR. The 120 nm diameter NSR had a branched shape, with 1.25 µm branches attached to the main trunk at 250, 500, 750, and 1000 nm from the JSR. The total length of the NSR was 7.5 µm. JSR and NSR contained 15 mm and 1 mm calsequestrin (K d = 0.8 mm), respectively. In a 100 nm transition zone, SR diameter and calsequestrin concentration varied linearly. The rate of Ca 2+ exit from the JSR was set to match recent estimates of Ca 2+ flux through RyRs under physiological conditions (Kettlun et al. 2003). With an initial [Ca 2+ ] JSR of 1 mm, each open RyR passed 1560 ions ms 1, equivalent to a single channel current of 0.5 p. Up to 20 RyRs could contribute to a Ca 2+ spark. SR Ca 2+ efflux over an assumed RyR open time of 20 ms was used as the input to a previously published model (Sobie et al. 2002) to compute the resulting Ca 2+ spark., predicted SR release currents and integrated fluxes (inset) for 20 (red) and 4 (black) open RyRs., Ca 2+ sparks resulting from the fluxes shown in. C,asthe number of channels increases, Ca 2+ spark amplitude rapidly reaches a plateau level. D, the model predicts that if SR refilling occurred instantaneously after Ca 2+ sparks, spark amplitude would recover with a much faster time course (red line) than channel availability (black line).

446 E.. Sobie and others J Physiol 565.2 200 ms, this result is not necessarily inconsistent with the kinetics observed here. fourth approach was employed by Sham et al. (1998), who documented local refractoriness of Ca 2+ release by noting a negative correlation between the local quantities of Ca 2+ released upon depolarization and those triggered by tail currents upon repolarization. These authors also observed slow recovery of whole-cell Ca 2+ release when a two-pulse protocol was applied (half time of 500 ms). While this Ca 2+ release recovery is considerably slower than we have observed here, this result is not necessarily inconsistent with our findings. Sham et al. (1998) performed experiments with a high concentration of exogenous buffer (4 mm EGT) added to the patch pipette so that Ca 2+ spikes (Song et al. 1998) could be recorded, and this excess buffer would have acted to slow SR refilling. Therefore, if refilling of local SR stores plays a key role in the recovery of release from refractoriness, as we hypothesize, the slow recovery observed is to be expected. Our results complement these prior efforts and provide new information of the time course of Ca 2+ release restitution. In particular, this study has generated simultaneous estimates of how both Ca 2+ spark amplitude and spark triggering probability recover with time. Two primary factors can presumably contribute to the recovery of Ca 2+ spark amplitude with time: (1) greater local SR content due to refilling after depletion, and (2) increased availability of RyRs due to recovery from a process such as inactivation. We hypothesize that refilling plays a much greater role in the recovery of spark amplitude, due to the extremely small SR volumes that appear to provide the Ca 2+ ions released during sparks. Ultrastructural studies suggest that a junctional SR (JSR) release unit is a disk approximately 400 nm in diameter and perhaps 15 nm thick (Inui et al. 1988). If such a volume ( 2 al) contained 1 mm free Ca 2+ and 15 mm of Ca 2+ -bound buffer (primarily calsequestrin), it would hold only 18 000 Ca 2+ ions. Since a 1 p current is equivalent to roughly 3000 Ca 2+ ions per millisecond, depletion of local JSR should occur quickly during a Ca 2+ spark, and this depletion would result in a decreasing Ca 2+ efflux function. This idea was explored through computer modelling, as illustrated in Fig. 4. Two separate models were used to simulate SR Ca 2+ release currents and the resulting Ca 2+ sparks (see figure legend for model details). Figure 4, which plots the simulated Ca 2+ efflux from the SR during an assumed 20 ms period of release, shows that depletion occurs much more quickly when 20 (red line, top) RyRs are open than when only 4 RyRs (black line, bottom) are open. s a result, the integrated release, plotted in the inset in units of femtocoulombs, is only 39.5% greater (20.9 versus 14.9 fc) when 20 RyRs release Ca 2+.saconsequence, the Ca 2+ spark resulting from 20 open RyRs is only 38.7% larger than the 4 RyR spark (Fig. 4). Figure 4C displays this less-than-proportional increase in amplitude with an increase in the number of open RyRs. dditional simulations (not shown) confirmed that this non-linear relationship was primarily due to faster depletion rather than saturation of the fluo-3 indicator. The net result of this non-linearity is that if SR refilling were instantaneous and Ca 2+ spark amplitude recovery resulted solely from recovery of RyRs from inactivation, spark amplitude would recover far faster than channel availability, as shown in Fig. 4D. ForCa 2+ spark amplitude to recover with a half-time of 63 ms, as measured, channel availability would have to recover with atime constant of 758 ms, a rate that is unrealistically slow. The above argument suggests that the time course of Ca 2+ spark amplitude recovery approximates the recovery of local SR stores ([Ca 2+ ] SR ). Given this, it is not surprising that the probability of Ca 2+ spark triggering (Fig. 3) recoverssomewhat more slowly. s the local SR refills, the Ca 2+ flux through any open RyR will increase roughly proportionally. Thus, the periodic openings of the ryanodine-bound RyR will become more likely to trigger the other RyRs in its cluster. If the chance that an opening will trigger a spark is proportional to the flux passing through the channel, then the composite spark triggering probability will recover with the same time course as the local SR load. If, however, RyRs also become more sensitive to trigger Ca 2+ due to greater local [Ca 2+ ] SR (Ching et al. 2000; Gyorke & Gyorke, 1998), then Ca 2+ spark triggering recovery will lag behind the recovery of amplitude. Thus, the results presented here are consistent with the hypothesis that Ca 2+ spark refractoriness is due in part to reduced local [Ca 2+ ] SR immediately after Ca 2+ sparks, which is to be expected if partial SR depletion plays akey role in terminating the spark, as recent computer modelling (Sobie et al. 2002) and experimental (Terentyev et al. 2002; Terentyev et al. 2003) studies have suggested. lternatively, the lag between Ca 2+ spark amplitude and Ca 2+ spark triggering recovery could result primarily from recovery of the RyRs from a process such as inactivation. t present, we cannot distinguish between these hypotheses, and, indeed, both mechanisms may contribute to Ca 2+ release restitution. Nonetheless, by allowing for simultaneous estimates of Ca 2+ spark amplitude and Ca 2+ spark triggering recovery, this study has provided new insight into the local regulation of Ca 2+ release in cardiac myocytes. References idasee KR, Xu L, Meissner G & esch HR Jr (2003). Diketopyridylryanodine has three concentration-dependent effects on the cardiac calcium-release channel/ryanodine receptor. J iol Chem 278, 14237 14248. uck E, Zimanyi I, bramson JJ & Pessah IN (1992). Ryanodine stabilizes multiple conformational states of the skeletal muscle calcium release channel. J iol Chem 267, 23560 23567.

JPhysiol 565.2 Ca 2+ spark restitution 447 ull R, Marengo JJ, Suarez-Isla, Donoso P, Sutko JL & Hidalgo C (1989). ctivation of calcium channels in sarcoplasmic reticulum from frog muscle by nanomolar concentrations of ryanodine. iophys J 56, 749 756. Cannell M, Cheng H & Lederer WJ (1994). Spatial non-uniformities in [Ca 2+ ] i during excitation-contraction coupling in cardiac myocytes. iophys J 67, 1942 1956. Cannell M, Cheng H & Lederer WJ (1995). The control of calcium release in heart muscle. Science 268, 1045 1049. Cheng H, Lederer WJ & Cannell M (1993). Calcium sparks: elementary events underlying excitation-contraction coupling in heart muscle. Science 262, 740 744. Cheng H, Lederer MR, Lederer WJ & Cannell M (1996). Calcium sparks and [Ca 2+ ] i waves in cardiac myocytes. m J Physiol 270, C148 C159. Ching LL, Williams J & Sitsapesan R (2000). Evidence for Ca 2+ activation and inactivation sites on the luminal side of the cardiac ryanodine receptor complex. Circ Res 87, 201 206. Collier ML, Thomas P & erlin JR (1999). Relationship between L-type Ca 2+ current and unitary sarcoplasmic reticulum Ca 2+ release events in rat ventricular myocytes. JPhysiol 516, 117 128. DelPrincipe F, Egger M & Niggli E (1999). Calcium signalling in cardiac muscle: refractoriness revealed by coherent activation. Nat Cell iol 1, 323 329. Fill M & Copello J (2002). Ryanodine receptor calcium release channels. Physiol Rev 82, 893 922. Guatimosim S, Dilly K, Santana LF, Jafri MS, Sobie E & Lederer WJ (2002). Local Ca 2+ signaling and EC coupling in heart: Ca 2+ sparks and the regulation of the [Ca 2+ ] i transient. JMolCellCardiol 34, 941 950. GyorkeI&GyorkeS(1998). Regulation of the cardiac ryanodine receptor channel by luminal Ca 2+ involves luminal Ca 2+ sensing sites. iophys J 75, 2801 2810. Inui M, Wang S, Saito & Fleischer S (1988). Characterization of junctional and longitudinal sarcoplasmic reticulum from heart muscle. J iol Chem 263, 10843 10850. Kettlun C, Gonzalez, Rios E&Fill M (2003). Unitary Ca 2+ current through mammalian cardiac and amphibian skeletal muscle ryanodine receptor channels under nearphysiological ionic conditions. JGeneralPhysiol 122, 407 417. Lopez-Lopez JR, Shacklock PS, alke CW & Wier WG (1995). Local calcium transients triggered by single L-type calcium channel currents in cardiac cells. Science 268, 1042 1045. McGrew SG, Wolleben C, Siegl P, Inui M & Fleischer S (1989). Positive cooperativity of ryanodine binding to the calcium release channel of sarcoplasmic reticulum from heart and skeletal muscle. iochem 28, 1686 1691. Sham JS, Song LS, Chen Y, Deng LH, Stern MD, Lakatta EG & Cheng H (1998). Termination of Ca 2+ release by a local inactivation of ryanodine receptors in cardiac myocytes. Proc Natl cad SciUS95, 15096 15101. Sobie E, Dilly KW, Dos Santos CJ, Lederer WJ & Jafri MS (2002). Termination of cardiac Ca 2+ sparks: an investigative mathematical model of calcium-induced calcium release. iophys J 83, 59 78. Song LS, Sham JS, Stern MD, Lakatta EG & Cheng H (1998). Direct measurement of SR release flux by tracking Ca 2+ spikes in rat cardiac myocytes. JPhysiol 512, 677 691. Stern MD (1992). Theory of excitation-contraction coupling in cardiac muscle. iophys J 63, 497 517. Szentesi P, Pignier C, Egger M, Kranias EG & Niggli E (2004). Sarcoplasmic reticulum Ca 2+ refilling controls recovery from Ca 2+ -induced Ca 2+ release refractoriness in heart muscle. Circ Res 95, 807 813. Tanaka H, Sekine T, Kawanishi T, NakamuraR&Shigenobu K (1998). Intrasarcomere [Ca 2+ ]gradients and their spatiotemporal relation to Ca 2+ sparks in rat cardiomyocytes. JPhysiol 508, 145 152. Terentyev D, Viatchenko-Karpinski S, Gyorke I, Volpe P, Williams SC & Gyorke S (2003). Calsequestrin determines the functional size and stability of cardiac intracellular calcium stores: Mechanism for hereditary arrhythmia. Proc Natl cad SciUS100, 11759 11764. Terentyev D, Viatchenko-Karpinski S, Valdivia HH, Escobar L &GyorkeS(2002). Luminal Ca 2+ controls termination and refractory behavior of Ca 2+ -induced Ca 2+ release in cardiac myocytes. Circ Res 91, 414 420. Wier WG & alke CW (1999). Ca 2+ release mechanisms, Ca 2+ sparks, and local control of excitation-contraction coupling in normal heart muscle. Circ Res 85, 770 776. ZucchiR&Ronca-Testoni S (1997). The sarcoplasmic reticulum Ca 2+ channel/ryanodine receptor: Modulation by endogenous effectors, drugs and disease states. Pharmacol Rev 49, 1 51. cknowledgements This work was supported by grants from the National Institutes of Health and the merican Heart ssociation.