Elevation of Mitochondrial Calcium by Ryanodine-sensitive Calcium-induced Calcium Release*

Calcium is an important regulator of mitochondrial function. Since there can be tight coupling between inositol 1,4,5-trisphosphate-sensitive Ca2+ release and elevation of mitochondrial calcium concentration, we have investigated whether a similar relationship exists between the release of Ca2+ from the ryanodine receptor and the elevation of mitochondrial Ca2+. Perfusion of permeabilized A10 cells with inositol 1,4,5-trisphosphate resulted in a large transient elevation of mitochondrial Ca2+ to about 8 μm. The response was inhibited by heparin but not ryanodine. Perfusion of the cells with Ca2+ buffers in excess of 1 μm leads to large increases in mitochondrial Ca2+ that are much greater than the perfused Ca2+. These increases, which average around 10 μm, are enhanced by caffeine and inhibited by ryanodine and depletion of the intracellular stores with either orthovanadate or thapsigargin. We conclude that Ca2+-induced Ca2+ release at the ryanodine receptor generates microdomains of elevated Ca2+ that are sensed by adjacent mitochondria. In addition to ryanodine-sensitive stores acting as a source of Ca2+, Ca2+-induced Ca2+release is required to generate efficient elevation of mitochondrial Ca2+.

Mitochondrial ATP synthesis is vital to all but a few primitive eukaryotic cells, and Ca 2ϩ has been shown to be a key regulator of mitochondrial function (1)(2)(3)(4). Although the mitochondria have long been recognized to take up substantial amounts of Ca 2ϩ , the relationship between [Ca 2ϩ ] m 1 and [Ca 2ϩ ] c has been far from clear. The development of luminescent and fluorescent indicators that enable selective measurement of [Ca 2ϩ ] m allowed these relations to be examined anew (5,6). Initial studies using targeted aequorin revealed rapid and large transient elevations of [Ca 2ϩ ] m in response to Gprotein-coupled agonists. Both the transient nature and the large amplitude of the [Ca 2ϩ ] m responses are explained by a hypothesis in which the mitochondria sense microdomains of highly elevated Ca 2ϩ adjacent to the InsP 3 release sites in the ER (7). In support of this hypothesis, we found that in HeLa cells the ER and mitochondria were in close apposition, whereas in ECV304 cells (where influx had a greater influence on [Ca 2ϩ ] m signaling) the mitochondria were more closely associated with the plasma membrane (8). More recently, experiments using green fluorescent protein mutants targeted to the ER and mitochondria revealed points of near contact between the mitochondria and the ER (9).
Experiments using fluorescent indicators also show [Ca 2ϩ ] m responses to be dynamic; they follow [Ca 2ϩ ] c oscillations in hepatocytes, myocytes, and astrocytes for example (6, 10 -12). Since the mitochondria can take up Ca 2ϩ during [Ca 2ϩ ] c responses, it is not surprising to find increasing evidence for mitochondrial Ca 2ϩ uptake and release to be influencing [Ca 2ϩ ] c responses (13)(14)(15)(16). Recent publications suggest that the mitochondria influence Ca 2ϩ release from InsP 3 -sensitive stores and Ca 2ϩ influx by modulating [Ca 2ϩ ] c adjacent to Ca 2ϩ release sites (17)(18)(19)(20)(21).
The InsP 3 receptor is not the only conduit for Ca 2ϩ release. In muscle cells, neurones, and even some nonexcitable cells, RyRs also act as the Ca 2ϩ release channels (22). Evidence points to the RyRs being closely associated with the mitochondria. In muscle tissue mitochondria and SR can be viewed in close proximity (23 (10), whereas the data from smooth muscle cells imply that, whereas the SR acts as a source of Ca 2ϩ , [Ca 2ϩ ] m is not elevated by the generation of privileged microdomains of released Ca 2ϩ (25).
Smooth muscle proves to be an interesting model to investigate the influence of Ca 2ϩ release on [Ca 2ϩ ] m . It possesses both InsP 3 and ryanodine-sensitive Ca 2ϩ release channels, thus enabling the influence of both channels on [Ca 2ϩ ] m to be compared (21,25 6 was diluted to 100 l with sterile Dulbecco's modified Eagle's medium containing 1% antibiotic and no serum. 1.5 l of FuGENE 6 containing 0.5 g of plasmid DNA was added to each well. Prior to experimentation, the culture medium was replaced with PS, containing 145 mM NaCl, 2.5 mM KCl, 1 mM Na 2 HPO 4 , 1 mM MgSO 4 , 10 mM HEPES, 10 mM glucose, 1 mM CaCl 2 , 10 mg/ml bovine serum albumin). To reconstitute functional aequorin, the cells were then incubated with 6 M coelenterazine for 2 h. The coverslips were placed in a purpose-built perfusion chamber, perfused with PS buffer held at 37°C. Vasopressin was added to PS as indicated. For permeabilization PS was replaced with a high K ϩ intracellular buffer (IB), containing 130 mM KCl, 10 mM NaCl, 5 mM K 2 HPO 4 , 5.6 mM glucose, 1 mM MgSO 4 , 5 mM Tris-succinate, 20 mM HEPES (pH 7.0 at 37°C), 1 mM ATP, supplemented with 75 M EGTA ([Ca 2ϩ ], 50 nM). For permeabilization, digitonin was added as indicated. To manipulate [Ca 2ϩ ] p , total EGTA was increased to 2.6 mM and CaCl 2 added to give the indicated [Ca 2ϩ ] p . [Ca 2ϩ ] p was verified using fura-2 (5 M). Reagents were added to the perfusion buffer as indicated in the figures. InsP 3 was perfused in IB containing 75 M EGTA. To measure [Ca 2ϩ ] m , the coverslips were placed in a superfusion chamber in close proximity to a photon-counting photomultiplier tube (27), the output of which was fed to a computer and analyzed using OSCAR (PTI Inc.) and Excel (Microsoft) software. After each experimental protocol, the cells were perfused with distilled water containing 10 mM Ca 2ϩ to consume all the remaining aequorin. [Ca 2ϩ ] m was then calculated from the fractional rate of consumption of the aequorin (28).

Measurement of [Ca 2ϩ ] c -Cytosolic
Ca 2ϩ was measured essentially as described by Lawrie et al. (8) using the Ca 2ϩ indicator fura-2 (29). A10 cells were grown on 22-mm diameter coverslips, and loaded with 2 M fura-2/AM for 45 min at 37°C in PS in the presence of 0.0125% pluronic F127. Coverslips were then placed in a thermostated sample chamber mounted on the stage of a Nikon Diaphot inverted microscope. The cells were excited alternately at 340 and 380 nm using a PTI Deltascan illuminator. Emission was monitored using an intensified charge-coupled device (Photonic Science Ltd). The data were subsequently analyzed using PTI Imagemaster software.

RESULTS
In intact A10 smooth muscle cells, we found that in the presence of the G-protein-coupled receptor agonist vasopressin, there was a transient increase in [Ca 2ϩ ] m to 9.9 Ϯ 1.8 M (n ϭ 13). The increase in [Ca 2ϩ ] m is much greater than the average increase in [Ca 2ϩ ] c (2.45 Ϯ 0.7 M, n ϭ 7) that occurred in the bulk cytosol (30).
In cells permeabilized with 10 M digitonin and in the presence of the Ca 2ϩ chelator EGTA, addition of 10 M InsP 3 evoked a similar rise in [Ca 2ϩ ] m to a mean of 7.6 Ϯ 0.85 M (n ϭ 23) (Fig. 1a). As seen with other cells, these results suggest that the mitochondria are able to sense Ca 2ϩ released from the InsP 3 -sensitive channels. When the permeabilized cells were subsequently superfused with a buffer containing Յ500 nM free Ca 2ϩ , a [Ca 2ϩ ] m rise was observed that was comparable to the [Ca 2ϩ ] p (Fig. 1a). These results show that, when the free [Ca 2ϩ ] c is Յ500 nM, the mitochondria in the smooth muscle may sense the mean [Ca 2ϩ ] c . In contrast, when superfusing permeabilized cells with a buffer containing Ն1 M free Ca 2ϩ , a [Ca 2ϩ ] m rise was observed that was approximately 3 times greater than [Ca 2ϩ ] p (Fig. 1b). In addition, when superfusing these cells with 500 g/ml heparin, an InsP 3 channel inhibitor, the InsP 3 response was abolished but heparin had no effect on the large [Ca 2ϩ ] m response induced by perfusion of 1.5 M free Ca 2ϩ (Fig. 1c). Likewise, the Ca 2ϩ -triggered response remained in permeabilized cells that were refractory to the addition of InsP 3 (Fig. 1d). This rules out the possibility that endogenous generation of InsP 3 accounts for the large increase in [Ca 2ϩ ] m triggered by Ca 2ϩ . Collectively, these data indicate that the Ca 2ϩ -induced increase in [Ca 2ϩ ] m is due to some additional process other than InsP 3 -mediated Ca 2ϩ release. When InsP 3 was superfused after obtaining responses to 1.5 or 3 M  (Fig. 3a). The mean response was 3.85 Ϯ 0.73 M (n ϭ 5). InsP 3 was still effective after this response, indicating that the InsP 3 -sensitive stores had not been depleted. In the presence of 1 M Ca 2ϩ , 20 mM caffeine increased [Ca 2ϩ ] m to 33.8 Ϯ 3.8 M (n ϭ 4). This potentiation of the CICR response is illustrated in Fig. 3b. These data imply that not only does caffeine increase the sensitivity of the response but that the amount of Ca 2ϩ available for mitochondrial uptake is increased as well.
In order to verify the source of Ca 2ϩ that leads to this large elevation of [Ca 2ϩ ] m , we investigated the actions of both ryanodine and the integrity of the Ca 2ϩ stores on the Ca 2ϩ -induced elevation of [Ca 2ϩ ] m . Addition of 100 M ryanodine had no effect on the peak [Ca 2ϩ ] m response evoked by 10 M InsP 3 , but it inhibited the large [Ca 2ϩ ] m response triggered by Ca 2ϩ buffers (Fig. 3c). After the addition of ryanodine, [Ca 2ϩ ] m reflected [Ca 2ϩ ] p . The data are summarized in Fig. 4. We then used two approaches to inhibit the SR Ca 2ϩ /ATPase pump and thereby deplete the intracellular stores. In the first we added 2 mM sodium orthovanadate prior to the trigger Ca 2ϩ buffer, and in the second we superfused the cells with 1 M thapsigargin. Under these conditions the increase in [Ca 2ϩ ] m produced by [Ca 2ϩ ] p in excess of 1 M was equivalent to that superfused (Fig. 4). Similarly, the [Ca 2ϩ ] m response to InsP 3 was completely abolished in the presence of either orthovanadate (n ϭ 6) or thapsigargin (n ϭ 4). Taken together, our results indicate that a CICR gated by RyRs occurs when the permeabilized cells are superfused with appropriate Ca 2ϩ buffers. DISCUSSION The RyRs and InsP 3 Rs are the two main conduits for the regulated release of Ca 2ϩ from intracellular stores in most cells. Although one or the other may be predominant in a specific tissue, some cells such as smooth muscle utilize both. A special relationship between InsP 3 Rs and mitochondria is now apparent. Although RyRs are abundant in the brain, heart, and skeletal muscles in addition to smooth muscles, very little is known about the relationship between mitochondria and Ca 2ϩ release from ryanodine-sensitive Ca 2ϩ stores. Our findings indicate that, in conjunction with InsP 3 Rs, RyRs must also be in close apposition and functionally coupled to mitochondria. When these observations are considered together with earlier findings (5,8,26,31), it suggests that the mitochondria are preferentially located adjacent to specific sources of Ca 2ϩ .
In intact cells vasopressin produced a large transient elevation of [Ca 2ϩ ] m , similar to that seen in bovine aortic endothelial and HeLa cells (5,7). When we subsequently perfused the permeabilized cells with InsP 3 , we found that it gave a large, rapid and transient rise in [Ca 2ϩ ] m , with a mean peak rise of around 8 M. The response, although shorter-lived, was similar to that evoked by vasopressin. Consistent with InsP 3 acting at its receptor on the SR, the response is abolished in the presence of heparin, a recognized antagonist of the InsP 3 receptor (32), and by orthovanadate and thapsigargin, inhibitors of the sarcoendoplasmic reticulum Ca 2ϩ ATPase pump (33,34).
The characteristic [Ca 2ϩ ] m response evoked by a G-proteincoupled agonist as reported here is transient. However, other studies using rhod-2 in rat pulmonary artery smooth muscle suggest that the elevation of [Ca 2ϩ ] m is sustained even when [Ca 2ϩ ] c declines (25). This apparent discrepancy may reflect the different modes of [Ca 2ϩ ] m measurement, although experiments performed by Hajnoczky et al. (6) using rhod-2 in hepatocytes are in broad agreement with aequorin-based measurements. It is important to remember that the relationship between mitochondria and the source of Ca 2ϩ influences the type of [Ca 2ϩ ] m responses that occur (8). Additionally, the apparent absence of Na ϩ /Ca 2ϩ exchange in the mitochondria of some smooth muscles (35) would certainly explain why sustained [Ca 2ϩ ] m responses can be seen in some instances.
Transient [Ca 2ϩ ] m responses occur even though the elevation of Ca 2ϩ in the cytosol is sustained (5, 7). These phasic [Ca 2ϩ ] m responses are attributed to the formation of microdomains of highly elevated [Ca 2ϩ ] c adjacent to the source of the Ca 2ϩ . Studies based on electron microscopy and fluorescence localization of mitochondria and ER support this hypothesis since they identify areas of close proximity between the two organelles where such microdomains could be generated (8,9). Calcium microdomains have been used to not only explain the large amplitude of the [Ca 2ϩ ] m responses but also their short duration. Diffusion of Ca 2ϩ into the bulk cytosol or re-uptake into intracellular stores will limit the lifetime of the microdomain and therefore the availability of Ca 2ϩ for mitochondrial uptake.
The permeabilized cells were superfused with Ca 2ϩ /EGTA buffers titrated to give free Ca 2ϩ values close to 500 nM, After elevation of [Ca 2ϩ ] m evoked by CICR, the InsP 3 response was significantly reduced. This may be caused by the InsP 3 -sensitive Ca 2ϩ pool being depleted, or as a result of Ca 2ϩ -dependent inactivation of the InsP 3 R. Such inactivation is well known to occur when [Ca 2ϩ ] c values increases above micromolar (38). In contrast, we can see that a preceding InsP 3evoked response does not prevent the subsequent CICR-induced [Ca 2ϩ ] m response, even though prior addition of InsP 3 abolishes a second response to InsP 3 . Heparin does not inhibit the CICR-mediated response, and ryanodine or caffeine do not prevent elevation of [Ca 2ϩ ] m in response to InsP 3 . The data imply that InsP 3 Rs and RyRs do not act in a co-ordinated manner, with Ca 2ϩ release from InsP 3 Rs neither triggering CICR nor depleting the CICR pool.
RyRs are also abundant in cardiac and skeletal muscle. Moreover, myocytes show dynamic (systolic) changes in [Ca 2ϩ ] m when paced (10,12). Since CICR is needed to elevate [Ca 2ϩ ] c in myocytes, it is reasonable to assume that a function of CICR is to elevate [Ca 2ϩ ] m , as well as simply providing Ca 2ϩ for the contractile apparatus. The ultrastructure of a myocyte is consistent with such a role. The local depolarizations in ⌿ m reported by Duchen et al. (24) are attributed to focal release of Ca 2ϩ from the SR. The authors were not certain of the magni- tude of the [Ca 2ϩ ] m responses that underlie these transient changes in⌿ m . Our evidence shows that CICR can lead to a substantial elevation of [Ca 2ϩ ] m . Since mitochondria take up Ca 2ϩ as a direct consequence of CICR, it is likely that in turn mitochondria influence the spread of SR Ca 2ϩ release (17). The total uptake of Ca 2ϩ by the mitochondria is calculated to be about 40-fold higher than the increase in free Ca 2ϩ (39). Thus, by local removal of Ca 2ϩ , the mitochondria may restrict the propagation of CICR.
Our data reveal that, in addition to the actions of InsP 3 , Ca 2ϩ release via CICR must also cause the generation of similar Ca 2ϩ microdomains adjacent to mitochondria. It appears that not only can ryanodine-sensitive stores act as a source of Ca 2ϩ for the mitochondria, but that the CICR process may be required to generate a sufficient local increase in [Ca 2ϩ ] c to enable efficient mitochondrial uptake of Ca 2ϩ . The mitochondrial uniporter is recognized as having a low affinity for Ca 2ϩ uptake and therefore a [Ca 2ϩ ] in excess of the 1-2 M that is normally found in the bulk cytosol, is needed to generate large increases in [Ca 2ϩ ] m (peak [Ca 2ϩ ] 5-10 M). Although a fast Ca 2ϩ uptake mode has been proposed for the mitochondria (40), it is not clear how this might operate in situ. It remains a possibility that the transient nature of the [Ca 2ϩ ] m response is mediated in part by the activation of a short-lived rapid uptake mode of the uniporter. Predictions by Crompton (41) and measurements by McCormack et al. (2,42) suggest that the mitochondria accumulate Ca 2ϩ in excess of the bathing [Ca 2ϩ ]. Such electrophoretic uptake of Ca 2ϩ is not sufficient to account for the large increases in [Ca 2ϩ ] m that we observe here, nor can it explain that the elevation of [Ca 2ϩ ] m in excess of [Ca 2ϩ ] p is dependent upon the presence of functional Ca 2ϩ stores.
It appears therefore that mitochondria are aligned to distinct sources of Ca 2ϩ in a tissue-dependent manner. These sources include Ca 2ϩ influx, InsP 3 -mediated Ca 2ϩ release, and, as shown here, ryanodine-sensitive CICR. In conclusion we demonstrate that in a cell type that possesses RyRs, CICR plays a crucial role in mitochondrial signal transduction. Since the distribution of RyRs is widespread among many tissues, CICR-mediated elevation of [Ca 2ϩ ] m is also likely to be widespread.