Sodium-Calcium Exchange Is Essential for Effective Triggering of Calcium Release in Mouse Heart

Spontaneous Ca²⁺ spark frequency is reduced in NCX KO myocytes

We have established that calcium efflux in NCX KO mice is reduced compared to that in WT mice (19). To determine whether this reduced efflux influences spark properties such as size, we measured the characteristics of spontaneous sparks in patch-clamped mouse ventricular myocytes loaded with the Ca²⁺ indicator fluo-3, held at a constant potential of  −75 mV, and imaged with laser scanning confocal microscopy in line-scan mode (2 ms/line) (22). Spark amplitude and width (full width at half-maximum (FWHM)) were larger in KO than WT mice (Fig. 1, B and C), but spark frequency was lower in KO mice. This is evident in the representative line-scan images shown in Fig. 1 A, where seven sparks are visible in the WT image but only three can be seen in the KO image. Spark amplitude depends upon the flux of Ca²⁺ released from clusters of RyRs in a couplon and the rate at which this Ca²⁺ is removed from the vicinity of the couplon by diffusion, SR uptake by SERCA, and transsarcolemmal efflux by the plasma membrane Ca²⁺ pump and NCX. Since NCX is absent in KO mice, efflux is slow (19), and thus sufficient to explain the higher spark amplitude. The reduced frequency of resting sparks in KO myocytes could either be caused by altered triggers (e.g., lack of NCX or altered LCCs) or by altered release mechanisms (e.g., RyRs or SR Ca²⁺). We have demonstrated previously that there is a 50% reduction in I_Ca in KO cells (16,18). This reduction in I_Ca is caused by accumulation of subsarcolemmal Ca²⁺, which increases Ca²⁺-dependent inactivation of LCCs (18). Since our previous studies have shown that SR Ca²⁺ content is identical in WT and KO cells (19), the reduction in I_Ca is a leading candidate for the lower spark frequency in KO cells.

To examine the influence of the sarcolemma on resting Ca²⁺ sparks, we compared spark characteristics in resting WT and KO ventricular myocytes permeabilized with the detergent saponin and loaded with fluo-3. Permeabilization mitigates the influence of sarcolemmal processes on RyRs, and exposes the RyRs from both WT and KO cells to exactly the same cytoplasmic Ca²⁺ concentration. We found that sparks from WT and KO cells were indistinguishable after permeabilization with regard to amplitude, duration, and frequency (Fig. 2). These data suggest that sarcolemmal processes regulating diadic cleft Ca²⁺, rather than compensatory changes in RyR activity or SR Ca²⁺, are primarily responsible for the differences in WT and KO Ca²⁺ spark properties.

Action potential-evoked Ca²⁺ sparks in NCX KO myocytes

We next investigated sparks evoked by APs in WT and NCX KO myocytes. We have observed previously that Ca²⁺ transients in WT and KO cells are similar (16,19), despite a smaller I_Ca in the KO mice. Thus, EC coupling gain (defined as the ratio of Ca²⁺ release to the magnitude of I_Ca) is significantly higher in KO mice. Since there is no evidence of enlarged SR Ca²⁺ content in KO myocytes, we propose that coupling fidelity, i.e., the fraction of single LCC openings that successfully trigger sparks (23), is higher in KO than in WT myocytes. This would enable a reduced trigger within a KO couplon to match the probability of spark activation in the WT (with its larger trigger), accounting for the increased gain. We hypothesize that the elevated Ca²⁺ concentration in the diadic cleft of KO myocytes due to reduced Ca²⁺ efflux (18) accounts for the higher coupling fidelity and gain. Elevated cleft Ca²⁺ should increase the probability that an LCC opening triggers Ca²⁺ release, because further Ca²⁺ loading, or priming, of the diadic cleft by NCX is not required. In WT cells, which have a lower cleft Ca²⁺ concentration to begin with, NCX can help prime the diadic cleft before activation by LCCs and therefore increase coupling fidelity (8).

We investigated this possibility by studying Ca²⁺ spark formation using high-speed line-scan confocal imaging during APs in WT and KO myocytes. EGTA (3 mM) was included in the pipette solution along with 1.3 mM Ca²⁺ to set free Ca²⁺ at 100 nM. EGTA is often used in confocal imaging to investigate the spatiotemporal characteristics of Ca²⁺-dependent phenomena (22,24–27), because it constrains visualization of Ca²⁺ to the source of release, and the peak of the resulting spike measures the flux of Ca²⁺ from a couplon (28). In these experiments, we also used EGTA to lower diadic cleft Ca²⁺ to expose the intrinsic effectiveness of I_Ca in activating a couplon. For convenience, we will continue to refer to local Ca²⁺ release under these highly buffered conditions as sparks.

When EGTA was included in the pipette, APs activated more Ca²⁺ release sites in WT cells than in KO cells. In a representative WT myocyte, an AP evoked 20 Ca²⁺ release events (Fig. 3 left). In contrast, only 11 Ca²⁺ release events were evoked in a typical KO myocyte (Fig. 3 right). NCX KO myocytes exhibited a greater number of regions where Ca²⁺ release events were not evoked by APs along the scan line, i.e., fluorescence values were close to background. These are seen clearly when the timescale is expanded (Fig. 3 B, arrowheads). Locations where sparks are absent in KO myocytes indicate a total failure of couplon activation. These temporally silent regions were observed in all NCX KO myocytes equilibrated with 3 mM EGTA. In summary, there were a total of 135 Ca²⁺ release events in eight WT cells (17 ± 1/cell) but only 56 Ca release events in eight KO cells (7 ± 1/cell, P < 0.0001).

We also found a significant difference in the latency of Ca²⁺ release events between WT and KO myocytes. In WT cells, all Ca²⁺ release events occurred within 15 ms of the AP stimulus (Fig. 3 C, left), whereas in KO cells, spark activation latencies were distributed throughout the recording period (Fig. 3 C, right). In both cases, examination of the line-scan images and their corresponding APs on an expanded timescale (scale bar, 10 ms) show that the leading edge of the Ca²⁺ transient begins just after the onset of the repolarization phase of the AP (dashed lines). Only 77% of release events occurred within 15 ms of the AP stimulus in KO cells compared with 100% in WT (Fig. 3 C). We termed this group of Ca²⁺ sparks the early pool, because they appeared proximate to the AP depolarization. We termed the other 23% of sparks that occurred after the initial 15 ms the late pool. The average latency of the early pool was (in ms) 2.4 ± 0.2 for WT and 2.7 ± 0.3 for KO (P = 0.3; see Fig. 5 B), whereas the average latency (including both pools) was (in ms) 2.4 ± 0.2 for WT and 21.0 ± 2.0 for KO (P < 0.0001; see Fig. 5 A). Thus, in NCX KO myocytes, AP-evoked Ca²⁺ spark triggering in the presence of EGTA was both reduced and poorly synchronized compared to that in WT myocytes. Rapid caffeine application (5 mM) revealed no reduction in SR Ca content in KO compared to WT myocytes that might account for these changes (ΔF/F₀ of 1.7 ± 0.1 in WT and 1.8 ± 0.2 in KO myocytes, P = 0.73). We suggest that loss of NCX prevents normal activation of couplons when EGTA is present to buffer Ca²⁺ and prevent priming of the diadic cleft. Under these conditions, we speculate that coupling fidelity is less in KO than in WT, where priming of the diadic cleft may still occur via reverse NCX in response to activation of I_Na and depolarization.

Improved EC coupling in NCX KO myocytes in the absence of EGTA

We further investigated coupling fidelity by repeating the above experiments without EGTA in the pipette. Fluo-3 was still present at 1 mM to help resolve Ca²⁺ sparks during the high-speed line scan. In the absence of pipette EGTA (Fig. 4), there was no difference in the total number of evoked sparks in WT and KO cells, though the distribution of spark latencies remained different between the two cell types. Fig. 4 shows the temporal relationship between APs (A) and Ca²⁺ release (B) in representative WT and NCX KO myocytes patch-clamped in the absence of EGTA. The line-scan image in the left panel of Fig. 4 B, recorded from a WT cell, exhibits activation of 19 Ca²⁺ release sites recorded from a 41-μm-wide region of a cell. These release sites are clearly triggered in a synchronous fashion with a very short latency of 1.8 ± 0.2 ms. The AP duration at 90% repolarization was 24 ms.

The line-scan image in Fig. 4 B (right), recorded from an NCX KO myocyte, shows a total of 18 evoked Ca²⁺ release sites, similar to WT myocytes. However, unlike WT myocytes, only 14 of these were triggered at the beginning of the Ca²⁺ transient. The latency of these early sparks was 2.2 ± 0.1 ms. The other four Ca²⁺ sparks were triggered 15, 84, 112, and 140 ms after the AP peak, raising the total averaged Ca²⁺ spark latency to 21.3 ± 10.4 ms in this cell. The expanded timescale in Fig. 4 B, right (scale bar, 10 ms) shows that, similar to WT, the leading edge of the Ca²⁺ transient starts during the repolarization phase of the AP. However, as with the high-EGTA group, fewer sparks were observed at the onset of the transient. Once again, there were temporally silent locations in the KO cells where AP-evoked Ca²⁺ release sites were absent or substantially delayed (Fig. 4 B, arrowheads).

These findings were typical of our observations in all cells studied when EGTA was omitted from the pipette solution. Overall a similar number of sparks was triggered in both WT and KO cells during the total 230-ms recording period (17 ± 2 and 16 ± 2, respectively; n = 8 in each group; P = 0.8). However, the latency of activation differed considerably between the two groups. Frequency histograms of Ca²⁺ spark latencies for eight WT and eight KO cells (Fig. 4 C) show that in WT cells, 100% of sparks (N = 133) were evoked within the first 15 ms of the AP (highly synchronized), whereas spark occurrence was broadly distributed (poorly synchronized) in KO cells. Only 66% (N = 81) of sparks from KO cells occurred within 15 ms of the AP (early pool), whereas 34% (N = 43) occurred later. The average latency of the early pool was (in ms) 1.9 ± 0.2 for WT and 2.2 ± 0.2 for KO (P = 0.26; Fig. 5 B), but when considering both pools, the average latency was (in ms) 1.9 ± 0.2 for WT and 33.4 ± 6.6 for KO (P < 0.001; Fig. 5 A). Thus, omitting EGTA restored the number of triggered calcium sparks in KO cells to WT cell levels, but did not reduce the latency of these Ca²⁺ release events in KO cells to WT values, as shown in Fig. 5. It is noteworthy that in these experiments we used a high concentration of fluo-3 so that we could reliably image the local release events. Therefore, even though we removed EGTA, we did not completely eliminate Ca²⁺ buffering. Nevertheless, the results are consistent with the idea that Ca²⁺ in the diadic cleft of KO cells influences their coupling fidelity and hence the number of initial release events. These changes were not caused by decreased SR content, which was the same in the two groups (ΔF/F₀ = 4.0 ± 0.7 in WT and 4.4 ± 1.0 in KO cells, P = 0.63). We will address the issue of delayed sparks and increased latencies in the Discussion section. Finally, we reported previously that Ca²⁺ transients in KO and WT cells are identical (16,19). This is under circumstances where there is no buffering by EGTA and where the concentration of Ca²⁺ indicator is low to avoid Ca²⁺ buffering, in contrast to the experiments described here, where we deliberately buffered Ca²⁺.

Effect of EGTA on the spatially averaged Ca²⁺ transient

The reduction in Ca²⁺ release events in KO cells loaded with EGTA should result in a reduction in the spatially averaged Ca²⁺ transient compared with WT. To demonstrate this, we measured spatially averaged fluorescence transients from line scans recorded in the presence and absence of EGTA. Fig. 6 A shows superimposed traces of the evoked Ca²⁺ transient recorded from representative WT and KO myocytes loaded with 1 mM fluo-3 and 3 mM EGTA. In both cases, there was a rapid rise of the transient, as seen in the normalized traces (Fig. 6 A, inset). Consistent with the reduced spark activation, the KO trace did not reach as high a peak as that seen in the WT trace. On average, peak fluorescence in KO myocytes was reduced 62% compared to WT myocytes (ΔF/F₀ = 1.02 ± 0.12 for WT and 0.40 ± 0.15 for KO myocytes, P < 0.001; Fig. 6 B). Both transients decayed quickly to baseline after the peak, consistent with EGTA reducing the diffusion of bound dye (29). However, the KO transients decayed more slowly than the WT transients (τ = 48 ± 3 ms for KO and 37 ± 2 for WT transients, P < 0.05), most likely because the lack of NCX reduces efflux rate.

In the complete absence of EGTA, differences in the fluorescence transients between WT and KO transients were less apparent. Transients from representative WT and KO cells are shown in Fig. 6 C. Both WT and KO have an initial rapid rising phase, followed by a very slow decay in the WT and a continued slow rise in the KO. Pooled data from multiple experiments (N = 8 for each condition) show that there is no significant reduction in peak fluorescence in KO (ΔF/F₀ = 1.04 ± 0.23) compared with WT transients (ΔF/F₀ = 1.28 ± 0.14; P = 0.4; Fig. 6 D). Transients decline very slowly in both cases, an expected consequence of the high concentration of fluo-3 in the cell (30).

T-tubular system remains normal in NCX KO myocytes

A disruption of the t-tubule system has been implicated in the pathogenesis of asynchronous SR Ca²⁺ release in models of heart failure (31–33). To investigate whether t-tubule spatial remodeling could explain the increased dispersion of latencies in KO myocytes, we examined the t-tubule network in isolated WT and KO cells using the potentiometric dye di-8 ANEPPS and 2-D confocal microscopy (27,33). We found that both WT and KO cells exhibited the classical striated pattern of t-tubule distribution. Plots of fluorescence intensity profiles along the long axis of these cells revealed a peak and valley distribution that was comparable in WT and KO cells. This suggests that the t-tubule network is preserved in cells lacking NCX. Average sarcomere lengths estimated from Gaussian fits to the fluorescence profiles were similar in both groups of cells (1.81 ± 0.02 and 1.79 ± 0.02 μm for WT (N = 7) and KO (N = 10), respectively). T-tubule densities were also similar (25.7 ± 0.7% for WT and 25.0 ± 0.9% for KO). Thus, there is no alteration in the 2-D appearance of the t-tubular network morphology in KO myocytes. Representative images and fluorescence profiles can be viewed in Fig. S1 in the Supporting Material.

Ca²⁺ sparks in resting myocytes from WT and NCX KO mice

We find that knocking out NCX results in brighter and larger spontaneous sparks in unstimulated patch-clamped myocytes held at a resting potential of −75 mV. In addition, the frequency of spontaneous sparks is lower in KO than in WT mice. There are two broad possible explanations for these differences: 1), the properties of the SR are modified in the KO mice—for example, the open-time distribution of RyRs is decreased or the SR is depleted of Ca²⁺; or 2), some process integral to the sarcolemmal membrane affects spontaneous spark properties. For this reason, we investigated spontaneous sparks after permeabilizing WT and KO myocytes with saponin to mitigate the influence of the sarcolemma while retaining the influences of RyR, SERCA, and SR Ca²⁺. Permeabilization eliminated all differences in resting spark properties. We conclude that the differences in spontaneous spark properties in WT and KO mice are attributable to changes in some sarcolemmal process. There are two processes associated with the sarcolemmal membrane that could influence resting spark production; these include 1), spontaneous LCC activity, and 2), NCX activity. At the moment, we cannot distinguish between these processes. However, it seems plausible that if the reason for a reduced I_Ca in KO myocytes (18) is a reduction in LCC open probability, and if the probability of LCC openings has a major influence on the frequency of resting spark activity, then we should expect a reduced frequency of resting sparks in KO cells. It has been argued that spontaneous openings of RyRs, not LCC open probability, determine resting spark frequency (5). However, several studies support the idea that LCCs may influence resting spark frequency (34–36). Regardless of the mechanism of this interaction, it is plausible that the reduced LCC activity caused by relatively higher cleft Ca²⁺ reduces spontaneous Ca²⁺ sparks.

Although we did not test SR content under conditions identical to those used in this experiment, we have consistently found that SR Ca²⁺ is similar in patch-clamped WT and KO myocytes (16,19) This is also the case in the experiments shown in Figs. 3 and 4, where EGTA and/or fluo-3 were used to buffer Ca²⁺. If there were any difference in SR content, we would have predicted that WT cells would be relatively depleted of Ca²⁺ compared with KO, since forward NCX in the WT cells would tend to remove Ca²⁺ under our experimental conditions (−75 mV holding potential and 10 mM Na⁺ in the pipette). Of course, there is no NCX in KO cells to remove Ca²⁺. A depleted SR in WT would not be consistent with the higher spark frequency we observed in WT compared to KO cells. Thus, we believe our results are not adequately explained by differences in SR load.

We have previously reported that in patch-clamped rabbit ventricular myocytes, resting spark frequency increases immediately when NCX is blocked abruptly by removing external Na⁺ (in the absence of pipette Na⁺) (20). How can we reconcile these results with those from the experiment shown in Fig. 1? We suggest that in rabbit cells, the abrupt removal of Na⁺ to block NCX does not lead to a significant reduction in LCC activity. Instead, LCC activity remains steady and spark frequency increases, because the rise in cleft Ca²⁺ due to ongoing SR leak increases the probability of RyR triggering. In contrast, resting spark frequency in NCX KO is low despite relatively elevated ambient cleft Ca²⁺, apparently because spark frequency is dominated by the reduced LCC activity caused by Ca²⁺-induced inactivation (18).

How can we explain the difference in Ca²⁺ spark size in KO versus WT cells? The maximum size of a spark is determined by two basic processes, the rate at which Ca²⁺ is released to form the spark and the rate at which Ca²⁺ is removed from the vicinity of the spark. NCX falls into the latter category, and its absence in KO myocytes likely accounts for the greater amplitude and width of the spontaneous sparks.

In aggregate, these arguments support the importance of junctional microdomains in determining, among other things, the characteristics of spontaneous sparks. In particular, it seems likely to us that the extent to which a junctional microdomain is filled with Ca²⁺ will influence the likelihood that a spontaneous spark is activated by I_Ca, i.e., the coupling fidelity of I_Ca is increased by diadic cleft priming. We do not exclude the possibility that there are spontaneous openings of RyRs that produce sparks, particularly in nonjunctional RyRs, but our preliminary data suggest that their occurrence is low (N. Torres, F. Sachse, J. Goldhaber, and J. H. B. Bridge, unpublished observations).

Defective spark recruitment in NCX KO myocytes revealed by Ca²⁺ buffering

The fundamental unit of EC coupling in heart, the couplon, is composed of a cluster of LCCs in the sarcolemmal membrane of the t-tubule opposed to a cluster of RyRs on the SR membrane (3,37). Upon depolarization during an AP, LCCs in a couplon open and allow Ca²⁺ influx with consequent triggering of RyRs. RyRs are likely triggered by Ca²⁺ entering with the earliest opening of an LCC in a cluster that has a sufficient open time to deliver enough Ca²⁺ across the cleft (22,38). Since most LCC openings in response to depolarization occur with a short latency, they activate RyRs and produce sparks with a short latency (38). LCC openings with longer latencies also occur, but these are less frequent and ordinarily are obscured by the earlier openings within a cluster.

In NCX KO myocytes, there is a reduced number of Ca²⁺ sparks triggered with a short latency. There is also an increased number of temporally silent areas where triggering fails (Fig. 3 B, arrows). We found no evidence for structural changes in terms of t-tubule distribution (Fig. S1) that might explain the loss of spark production in NCX KO myocytes, though it is possible that 3-D reconstructions might reveal differences in t-tubular structure that we cannot appreciate in 2-D (39). However, we did observe that the extent of the reduction in early spark triggering was clearly influenced by the degree of Ca²⁺ buffering. Strong buffering (fluo-3 + EGTA) was associated with a more profound defect in triggering in KO compared to WT than weak buffering (fluo-3 alone). Since SR content is the same in KO and WT myocytes, and since our permeabilization experiments demonstrate that RyR function is similar in KO and WT myocytes, we must look to other processes for an explanation of spark failure in the KO myocytes.

In KO myocytes, Ca²⁺ is elevated in the subsarcolemmal space (18). Elevated subsarcolemmal Ca²⁺ promotes Ca²⁺-dependent inactivation, thereby reducing the amount of Ca²⁺ entering through LCCs (18) and reducing triggering of RyRs. However, this tendency is counteracted by two factors: 1), the more rapid repolarization of the AP in KO myocytes, which enhances coupling fidelity (19,40,41); and 2), the elevated cleft Ca²⁺, which permits RyR triggering with less additional Ca²⁺ entry through LCCs. This means, in effect, that the increased Ca²⁺ in the cleft increases the coupling fidelity of I_Ca. Thus, in KO cells, the cleft space is already primed with Ca²⁺, and RyRs are readily triggered by incremental increases in cleft-space Ca²⁺ by LCC openings. When we deplete Ca²⁺ in the cleft by including high concentrations of a Ca²⁺ chelator in the patch pipette, I_Ca likely returns to WT levels (18). However, this restored Ca²⁺ flux through LCCs cannot trigger RyRs with the same efficiency in the KO, because the diadic cleft space is no longer primed in the presence of EGTA. Thus, the recovery of I_Ca magnitude cannot overcome the lack of priming of the cleft with Ca²⁺. As a consequence, in unbuffered KO cells, Ca²⁺ entering through LCCs is reduced, but paradoxically, it is more likely to trigger SR Ca²⁺ release because of the increased baseline cleft Ca²⁺.

Why, then, are WT myocytes able to trigger calcium release even when the subsarcolemmal space is buffered with EGTA? We suggest that in WT cells, NCX operating in reverse mode primes the diadic cleft with Ca²⁺ upon depolarization, so that the incremental Ca²⁺ entering through LCCs can now trigger RyRs. It is well known that spikelike local release events can be elicited in the presence of EGTA (22,29,42), and even 10 mM EGTA cannot eliminate the influence of SR Ca²⁺ release on inactivation of I_Ca (43). We conclude that EGTA is too slow to prevent a transient accumulation of Ca²⁺ in the cleft produced by reverse NCX. We emphasize that the function of NCX here is to prime the cleft but not necessarily to directly trigger release. This effect of NCX is consistent with the close anatomical relationship between NCX, LCCs, and RyRs in the diadic cleft, as previously argued (15,44). We propose that the activity of NCX occurs immediately upon depolarization and proceeds with a latency that is significantly shorter than that exhibited by any Ca²⁺ channel. To summarize, in the absence of either elevated subsarcolemmal Ca²⁺ in the diadic cleft or reverse NCX to prime the diadic cleft with Ca²⁺, LCCs by themselves are incapable of reliably triggering RyRs. Thus, couplon recruitment is less effective in KO myocytes buffered with EGTA, because cleft Ca²⁺ is low and the coupling fidelity of LCCs is also low. We cannot exclude the possibility that the shorter AP duration in KO myocytes compared to WT contributes to the reduced spark recruitment in EGTA. However, since the KO AP duration is the same in both the absence and presence of EGTA, the critical difference must be the coupling fidelity. The most likely mechanism for influencing the coupling fidelity is the extent to which the diadic cleft is primed with Ca²⁺.

The mechanism we have outlined is intrinsically nonlinear. If we consider the sigmoid dependence of RyR open probability on gating Ca²⁺ concentration (45), the open probability at rest is set by the level of cleft Ca²⁺. This resides at the far left of the sigmoid curve. Priming of the cleft by NCX might be expected to shift Ca²⁺ concentration along the flat portion of this sigmoid curve to its point of initial inflection, where we would not expect to see much increase in the open probability of RyRs. However, any further increase in cleft Ca²⁺ at the point of inflection would produce a rapid increase in the open probability of RyRs. In the absence of priming, Ca²⁺ channels will first have to elevate Ca²⁺ to the point of inflection before significant triggering will occur.

Origin of sparks with long latency

In WT rabbit myocytes, we have reported that the latencies of sparks triggered by depolarization reflect the latencies of LCC openings (22). The histograms in Figs. 3 and 4 show that activation of the late pool of sparks in KO myocytes occurs well after repolarization of the AP, when the open probability for LCCs is expected to be very low. It seems unlikely that these late Ca²⁺ sparks (occurring >30 ms after the stimulus) are triggered events that reflect the distribution of LCC open probability and open times. It is more likely that in the setting of slowed efflux due to the absence of NCX, Ca²⁺ released by RyRs within the first few milliseconds of the Ca²⁺ transient raises the concentration of Ca²⁺ in nearby diadic clefts to levels high enough for late sparks to be triggered by stochastic openings of LCCs. The probability of these sparks subsequently declines during the diastolic interval, presumably because the plasma membrane Ca²⁺ pump continues to reduce cleft Ca²⁺ during this period. Additional mechanisms, e.g., diffusion of Ca²⁺ out of the cleft and uptake of cytoplasmic Ca²⁺ by SERCA, also contribute to the decline of spark occurrence in diastole. Stochastic openings of RyRs (e.g., independent of LCCs) are unlikely to explain the late sparks. This is because the flux of Ca²⁺ passing through an RyR is significantly larger than that passing through an LCC and would therefore be expected to fill the cleft sufficiently to produce regenerative activation of a couplon even in the absence of cleft priming. The addition of 3 mM EGTA restricts Ca²⁺ diffusion and reduces priming of the diadic cleft by SR Ca²⁺, which reduces the occurrence of late Ca²⁺ sparks by 70% compared to recordings in the absence of EGTA (Fig. 5).

When regarding the spatially averaged Ca²⁺ transient, late sparks contribute to its slowed relaxation in KO compared to WT cells (Fig. 6). The explanation for this slowed relaxation is the same as that given above for slowed efflux from the diadic cleft, and is primarily a consequence of the elimination of NCX. KO cells compensate for this reduced Ca²⁺ removal rate by reducing Ca²⁺ influx through an upregulation of I_to and a reduction in I_Ca (19).