Ca²⁺ marks: Miniature calcium signals in single mitochondria driven by ryanodine receptors

To investigate transmission of RyR-mediated localized [Ca²⁺]_c signals to the mitochondria, we first carried out simultaneous fluorescence imaging measurements of [Ca²⁺]_c and [Ca²⁺]_m in permeabilized cardiac myotubes by using fura-C18 anchored to cellular membranes and compartmentalized rhod2, respectively (21, 22). The permeabilized myotube preparation has the advantage that it allows direct access to the cytosol and mitochondria and the cytosolic component of rhod2 is eliminated. In response to suboptimal caffeine (0.5 mM), the myotubes displayed [Ca²⁺]_c spiking that was confined to small regions of the cells, whereas subsequent application of maximal caffeine yielded elevations in [Ca²⁺]_c throughout the cells (Fig. 1). During the localized [Ca²⁺]_c spikes, elevations of [Ca²⁺]_m were synchronized to the rising phase of the [Ca²⁺]_c signal in the discrete regions of Ca²⁺ release, whereas no [Ca²⁺]_m increase was observed at other locations (Fig. 1A, images ix–xiii, and B). By contrast, a global [Ca²⁺]_m increase was coupled to the diffuse Ca²⁺ release mediated by maximally activated RyR (Fig. 1A, image xiv, and B). [Ca²⁺]_m increases never preceded the [Ca²⁺]_c rise and showed slower decay than the [Ca²⁺]_c transients (Fig. 1B). Furthermore, the [Ca²⁺]_c and [Ca²⁺]_m responses displayed region-specific amplitude and temporal patterns during [Ca²⁺]_c spiking restricted to subdomains of the cells (Fig. 1). Localized coupling of [Ca²⁺]_c and [Ca²⁺]_m spikes were also observed when addition of ryanodine or Ca²⁺ was used to activate the RyR (not shown). The amplitude of the locally restricted [Ca²⁺]_c and [Ca²⁺]_m spikes typically were smaller than those during global responses (Fig. 1B). The magnitude of the F_rhod2 rise from [Ca²⁺]_m signals confined to <10-μm-diameter regions was 21.7 ± 2.3% of the F_rhod2 increase obtained during global [Ca²⁺]_c signals (n = 16). These results suggest that in permeabilized myotubes, suboptimal stimulation of RyR triggers localized [Ca²⁺]_c spikes that are coupled to localized, submaximal [Ca²⁺]_m transients.

Based on the area of activation, the localized [Ca²⁺]_m signals shown in Fig. 1 could be derived from small groups of mitochondria. To discriminate the contributions of individual mitochondria, we used confocal imaging of [Ca²⁺]_c and [Ca²⁺]_m with fluo3 added to the permeabilized cells and compartmentalized rhod2, respectively. First, we recorded two-dimensional images during sequential stimulation of the cells with submaximal and maximal doses of caffeine. When 1 and 10 mM caffeine was added, both doses yielded transient [Ca²⁺]_c increases that appeared as global [Ca²⁺]_c responses, and each was coupled to a [Ca²⁺]_m elevation that manifested in discrete structures (Fig. 2A Top). The spatial distribution of the [Ca²⁺]_c and [Ca²⁺]_m responses was visualized by using a three-dimensional presentation (Fig. 2A Middle and Bottom). This presentation underscores that [Ca²⁺]_c elevations took place throughout the cytoplasm during each response (Fig. 2A, xi and xiii; the nucleus exhibited above-average fluo3 labeling), whereas the [Ca²⁺]_m increases were manifest as discrete peaks corresponding to individual mitochondria or clusters of a few mitochondria (Fig. 2A, vii and ix). Based on electron microscopic evaluation of the ultrastructure of myotubes, many of the mitochondria are oval-shaped (diameter of <1 μm), whereas some mitochondria appear as long tubular structures (21). As shown by comparison of peak heights between images vii and ix, the [Ca²⁺]_m response was maximal during the first step of stimulation for some mitochondria, whereas other mitochondria exhibited a considerably larger response to maximal caffeine. Because mitochondrial Ca²⁺ uptake sites do not appear to exhibit intrinsic heterogeneity (13, 24), subcellular heterogeneity in the mitochondrial sensitivity to RyR-mediated [Ca²⁺]_c signal may reflect differences in Ca²⁺ release flux in the immediate vicinity of the mitochondria.

A more striking attenuation of the [Ca²⁺]_m signal exhibited by the mitochondria occurred when a localized RyR-mediated [Ca²⁺]_c elevation was elicited with caffeine (0.5 mM; as in Fig. 1). The only mitochondria displaying a [Ca²⁺]_m rise were at the center of [Ca²⁺]_c signals (Fig. 2B, ii; three-dimensional presentation in x), whereas the other mitochondria in the vicinity remained silent (ii vs. iv). Time courses for the region of the responsive mitochondrion revealed a [Ca²⁺]_c spike lasting a few seconds and a relatively small and more slowly decaying [Ca²⁺]_m increase (Fig. 2B, ix). Taken together, the two-dimensional confocal imaging studies revealed the fundamental structures exhibiting the [Ca²⁺]_m rise during RyR-mediated Ca²⁺ release. These studies demonstrated that the number of responding mitochondria, as well as the amplitude of single mitochondrial [Ca²⁺] increases, increase with the extent of RyR activation. Thus, the question arises of whether activation of a few release sites during a single [Ca²⁺]_c spark with the typical small amplitude and 100- to 200-ms decay (2), is sufficient to increase the activity of mitochondrial Ca²⁺ uptake and sufficient to elevate [Ca²⁺]_m in a mitochondrion.

Because of the temporal limitations of two-dimensional confocal imaging, we used high-speed scanning along a single line to visualize the elementary events of the cytosolic and mitochondrial calcium signal. First, we carried out linescan imaging of [Ca²⁺]_c in fluo3-loaded intact myotubes. In cells exposed to a low dose of caffeine (0.25 mM), which failed to cause propagating [Ca²⁺] response, spatially localized and short-lived [Ca²⁺]_c signals were recorded (Fig. 3A). As described in other systems (25, 26), similar events were seen at a lower frequency in the absence of caffeine. These [Ca²⁺]_c events were smaller and much shorter than the events shown in Figs. 1 and 2 and displayed properties very similar to Ca²⁺ sparks in other systems [width at half-maximal amplitude, 2.86 ± 0.04 μm, amplitude (F/F₀) 1.67 ± 0.02 and duration at half-maximal amplitude, 51.5 ± 1.5 ms; Fig. 4C]. Consistent with recent observations (27–29), we could also record miniature Ca²⁺ release events in permeabilized myotubes with fluo3 added to the cytosolic buffer (Fig. 4B). Based on their spatiotemporal pattern, these events represent RyR-mediated [Ca²⁺]_c sparks in the H9c2 cardiac myotubes. Therefore, we used the conditions in which [Ca²⁺]_c sparks were generated in this preparation to investigate whether elementary Ca²⁺ release events are delivered to the mitochondria.

Confocal linescan imaging revealed the presence of highly localized (<1 μm), subsecond F_rhod2 transients in the permeabilized myotube model (Fig. 3B). These fluorescence responses were dependent on Ca²⁺ release from sarcoplasmic reticulum and also on mitochondrial Ca²⁺ uptake, because predepletion of the sarcoplasmic reticulum store with a Ca²⁺ pump inhibitor, thapsigargin (2 μM, n = 5, not shown), or elimination of the driving force of mitochondrial Ca²⁺ uptake by uncoupler (FCCP; 5 μM, n = 5, not shown) prevented the F_rhod2 transients. In contrast to Ca²⁺ sparks, the F_rhod2 transients did not show lateral spreading, which can be explained by the mitochondrial membrane acting as a Ca²⁺ diffusion barrier (Fig. 3B). Properties of the highly localized F_rhod2 transients recorded in myotubes are summarized in Fig. 4A. The diameter of these events was 0.59 ± 0.02 μm at half-width (Fig. 4A, ii), suggesting that the [Ca²⁺] rise took place in individual mitochondria. The magnitude of the single mitochondrial [Ca²⁺]_m increases was estimated to be 200–400 nM (see Experimental Procedures). The mean duration was 174 ± 8.4 ms at half-height (Fig. 4A, iii), illustrating the relatively slow decay of the miniature [Ca²⁺]_m elevations. However, we also observed events that showed rapid decline similar to the sparks (Fig. 3B, v; broad distribution of duration is shown in Fig. 4A, iii). Collectively, these data suggest that [Ca²⁺]_m increases restricted to single mitochondria are coupled to elementary Ca²⁺ release events in myotubes. Because the miniature [Ca²⁺]_m increases represent the mark left by [Ca²⁺]_c sparks on [Ca²⁺]_m, we refer to these fundamental mitochondrial events as “Ca²⁺ marks.” The spatiotemporal differences between the elementary cytosolic and mitochondrial signals are underscored by three-dimensional plots of the individual events (Fig. 3C). Specifically, [Ca²⁺]_c sparks were relatively broad in the spatial domain, but dissipate rapidly in time, reflecting diffusion of released Ca²⁺ from the spark site (Fig. 3C Upper). By contrast, [Ca²⁺]_m marks occupy a small area with sharp edges, because of restriction of diffusion of Ca²⁺ by the mitochondrial membrane (Fig. 3C Lower). [Ca²⁺]_m marks were also more prolonged than [Ca²⁺]_c sparks. Interestingly, [Ca²⁺]_m marks are confined to a considerably smaller area than the localized mitochondrial depolarizations reported by Duchen et al. (19) in myocytes. Because mitochondria have been proposed to form a functional syncytium (7) and an electrically coupled network (30), depolarization could propagate from one mitochondrion to the next. However, based on photobleaching experiments, Park et al. (13) have proposed that mitochondria are not necessarily luminally connected, and our data suggest that small [Ca²⁺]_m elevations do not display luminal spreading.

To better understand how Ca²⁺ sparks leave their marks on [Ca²⁺]_m, we evaluated the spatial distribution of Ca²⁺ marks. In the majority of linescan frames examined (397 of 509, 3,072 ms each), the [Ca²⁺]_m signal appeared to derive from only one mitochondrion (<1 μm in half-width). In the other linescan images in which Ca²⁺ marks were characterized, either the spatial separation of marks was >2.5 μm (110 of 509) or more closely apposed mitochondria were activated with >100-ms temporal separation (2 of 509). In light of these data, we conclude that a spark is not likely to result in more than one Ca²⁺ mark. To determine the fraction of sparks that propagates to the mitochondria, we calculated the number of regions in permeabilized myotubes that displayed [Ca²⁺]_c sparks and [Ca²⁺]_m marks, respectively. The frequency of [Ca²⁺]_c sparks was 60% higher than the frequency of [Ca²⁺]_m marks (0.36 ± 0.05/frame, 15 experiments; 0.22 ± 0.08/frame, 13 experiments). We also attempted to record [Ca²⁺]_m marks simultaneously with [Ca²⁺]_c sparks. However, these experiments were not successful because of the relatively low fluorescence signal associated with marks and reduced signal-to-noise ratio in our multiparameter measurements. Other factors also may limit the occurrence of Ca²⁺ marks. Not all mitochondria are immediately adjacent to the sarcoplasmic reticulum surface (e.g., refs. 7 and 21) and mitochondrial Ca²⁺ uptake sites exhibit low affinity to Ca²⁺, suggesting that [Ca²⁺]_c sparks may occur unrecognized by mitochondria (as reported for [Ca²⁺]_c puffs in HeLa cells; ref. 31). Because mitochondrial Ca²⁺ uptake sites display Ca²⁺-induced sensitization (32), during repetitive sparks, the first event may sensitize uptake sites such that only subsequent sparks are delivered into the mitochondrion. Despite the above-mentioned limitations of [Ca²⁺]_c spark propagation to mitochondria, after the addition of mitochondrial inhibitors, we observed a 2-fold increase in the frequency of sparks (see below). This result also suggests activation of mitochondrial Ca²⁺ uptake by a substantial fraction of [Ca²⁺]_c sparks and implies that mitochondria may directly accumulate Ca²⁺ from activated clusters of RyR, preventing the generation of [Ca²⁺]_c sparks.

The variations in decline, the relatively long mean half-time, and restricted lateral spreading of [Ca²⁺]_m marks suggest that mechanisms other than diffusion are involved in the falling phase. Recently, we have shown that deactivation of [Ca²⁺]_m spikes during global [Ca²⁺]_c oscillations use the mitochondrial Ca²⁺ exchanger, whereas no major contribution of the permeability transition pore was found (22). Thus, we studied the effect of CGP37157, an inhibitor of the Ca²⁺ exchanger, on the duration of [Ca²⁺]_m marks to determine whether carrier-mediated mitochondrial Ca²⁺ efflux is important in recovery of the elementary events. In cells displaying [Ca²⁺]_m marks, CGP37157 (10 μM) resulted in prolongation of the elementary events (215.9 ± 19.5 and 374.3 ± 40.9 ms at half-height, before and after addition of CGP37157, n = 64 and 63, P < 0.001). By contrast, CGP37157 failed to change the duration of [Ca²⁺]_c sparks (47.9 ± 0.8 vs. 44.0 ± 0.6 ms in the presence and absence of CGP37157, n = 616 and 814). Taken together, these data suggest that decay of [Ca²⁺]_m marks is dependent on carrier-mediated Ca²⁺ extrusion from the mitochondria. Although the Ca²⁺ exchanger has been demonstrated to control the decay of [Ca²⁺]_m increases in many cell types, our result shows that in contrast to cytosolic and nucleoplasmic elementary Ca²⁺ signals, [Ca²⁺]_m marks do not simply decay by diffusion. In addition to the role of the Ca²⁺ exchanger demonstrated in this study, CGP37157-insensitive efflux pathways as well as slow intramitochondrial Ca²⁺ buffering may also contribute to the decay of marks. Regulation of the life-time of [Ca²⁺]_m marks by mitochondrial transporters yields another difference between [Ca²⁺]_c sparks and [Ca²⁺]_m marks and allows these efflux pathways to modulate the mitochondrial targets of the [Ca²⁺]_m mark as well as to exert feedback on [Ca²⁺]_c.

Several mitochondrial dehydrogenases are sensitive to [Ca²⁺]_m changes in the submicromolar range (14), and imaging studies on the coupling between gradual increases in [Ca²⁺]_m and NAD(P)H fluorescence have suggested that relatively small increases in [Ca²⁺]_m may be sufficient to stimulate mitochondrial metabolism (8). However, with current imaging technology, we could not resolve an NAD(P)H increase resulting from propagation of a single [Ca²⁺]_c spark to an individual mitochondrion. Recent studies also have demonstrated multiple feedback effects of mitochondrial Ca²⁺ uptake on [Ca²⁺]_c signaling during the course of global calcium signals (for review, see refs. 33–37) and that mitochondria may sequester 25–50% of mobilized Ca²⁺ (21). To evaluate the physiological significance of the mitochondrial Ca²⁺ uptake during [Ca²⁺]_m marks in elementary [Ca²⁺]_c signaling, we studied the effect of mitochondrial Ca²⁺ uptake blockers on spontaneous [Ca²⁺]_c sparks in intact myotubes. Treatment with uncoupler (5 μM FCCP) or antimycin A (5 μM), in the presence of oligomycin to prevent cellular ATP depletion, resulted in a considerable increase in the number and size of localized [Ca²⁺]_c increases recorded in two-dimensional images (Fig. 5 Upper). Furthermore, the [Ca²⁺]_c sparks displayed increased frequency, duration, and width in linescan images (Fig. 5 Lower). On average, mitochondrial inhibitors caused a 2-fold increase in the frequency of [Ca²⁺]_c sparks (control, 3.2 ± 0.3; FCCP + oligomycin, 6.5 ± 0.6; and antimycin + oligomycin, 7.7 ± 0.8 mm⁻¹ × s⁻¹; n = 814, 477, and 299). The mean duration and spatial size of the [Ca²⁺]_c sparks was also increased by 20–30% in FCCP or antimycin-pretreated cells [FCCP + oligomycin, 119%, and antimycin + oligomycin, 130% of control (duration at half-height); FCCP + oligomycin, 124% and antimycin + oligomycin, 125% of control (half-width)]. These changes could arise, in part, from an increase in global [Ca²⁺]_c. However, FCCP caused only a small rise in [Ca²⁺]_c under these conditions (data not shown). Thus, the data demonstrate that delivery of [Ca²⁺]_c sparks to the mitochondria also may result in an important feedback regulation of the elementary [Ca²⁺]_c signals. As a consequence, mitochondria could serve to suppress the excitability of the excitation–contraction coupling system by reducing spark frequency and size.

Our data reveal that [Ca²⁺]_c sparks permit delivery of Ca²⁺ to neighboring mitochondria, giving rise to [Ca²⁺]_m marks. Thus, elementary [Ca²⁺]_c signaling may allow for asynchronous regulation of functionally discrete mitochondria. Calcium marks are confined to individual mitochondria and are likely to have smaller amplitude than the [Ca²⁺]_m elevations occurring when numerous elementary events are summated to give rise to a regenerative [Ca²⁺]_c wave. This may be because the open time of a given cluster of RyR is longer during a regenerative [Ca²⁺]_c wave than that during a [Ca²⁺]_c spark, or each mitochondrion may sense [Ca²⁺]_c sparks originated from more than one cluster of RyR. Nevertheless, elementary mitochondrial Ca²⁺ uptake events appear to contribute to the organization of elementary calcium signaling in the cytosol and may also establish control over the mitochondrial Ca²⁺ targets. Importantly, we also demonstrated that [Ca²⁺]_m marks dissipate relatively slowly and use exchanger-mediated Ca²⁺ export in their decay phase. The prolongation of marks compared with cytosolic sparks may facilitate activation of the Ca²⁺-sensitive targets in the mitochondrial matrix and also may support localized recharging of the sarcoplasmic reticulum Ca²⁺ store after the RyR cease to fire. Thus, we propose that the elementary events of calcium signaling involve mitochondrial contributions, which establish precise, localized control of [Ca²⁺]_m-dependent mitochondrial functions.

This paper was submitted directly (Track II) to the PNAS office.

We thank Drs. T. Pozzan, S. Györke, S. K. Joseph, and J. B. Hoek for valuable discussions and Paul Anderson for his help with software development. This work was supported by grants from the American Heart Association and National Institutes of Health (to G.H.). G.H. is a recipient of a Burroughs Wellcome Fund Career Award. P.P. is a recipient of a Juvenile Diabetes Foundation Postdoctoral Fellowship.