Plasticity of Mitochondrial Calcium Signaling*

Evidence is emerging that a quasisynaptic local communication facilitates the calcium signaling between endoplasmic reticulum and mitochondria. However, it remains elusive whether the machinery of mitochondrial calcium signaling displays plasticity similar to the synaptic transmission. Here we studied the relationship between inositol 1,4,5-trisphosphate (IP 3 )-linked cytoso- lic [Ca 2 (cid:1) ] ([Ca 2 (cid:1) ] c ) oscillations and the associated rise in mitochondrial matrix [Ca 2 (cid:1) ] ([Ca 2 (cid:1) ] m ) in RBL-2H3 mast cells. We observed that the second [Ca 2 (cid:1) ] c spike is often associated with a larger rise in the [Ca 2 (cid:1) ] m than the first. It would appear that this phenomenon was not due to a change in the driving force for Ca 2 (cid:1) uptake and therefore must be due to an enhanced Ca 2 (cid:1) permeability of the mitochondrial Ca 2 (cid:1) uptake sites (uniporter). To investigate the activation and deactivation kinetics of the uniporter during IP 3 receptor-mediated Ca 2 (cid:1) mobilization, we established novel methods. Using these approaches, we demonstrated that the IP 3 -induced in- crease in the permeability of the uniporter lasted longer than the Ca 2 (cid:1) signal. The sustained increase in Ca 2 (cid:1) permeability was bidirectional. Furthermore, the addition of Ca 2 (cid:1) during the decay of the IP 3 effect evoked a large further use novel fluorescent time course of the uniporter Ca 2 (cid:1) permeability during IP 3 ER Ca 2 (cid:1) release and mitochondrial Ca 2 (cid:1) uptake in RBL-2H3 mast cells. We demonstrate delayed deactivation of the mitochondrial Ca 2 (cid:1) uptake sites and sensitization to Ca 2 (cid:1) during repetitive stimulation. Our studies provide evidence that a memory switch of the uniporter is triggered by exposure of the uniporter to high local Ca 2 (cid:1) and that the Ca 2 (cid:1) rise is sensed by a calmodulin/calmodulin-like element associated with the uniporter. Based on these data, we propose that the uniporter exhibits a memory function, providing a means to optimize the mitochondrial Ca 2 (cid:1) signal and increase in metabolism elicited by IP 3 and ryanodine receptor-dependent [Ca 2 (cid:1) ] c signals.

Since the mid-1990s, several lines of investigation have challenged the classic view that the only physiological role of mitochondria is in energy production. Mitochondria have been emerging as key players in cellular signaling. In particular, the role of mitochondria has been established in the release of apoptotic factors to the cytosol (reviewed in Refs. [1][2][3][4] and in intracellular calcium signaling (reviewed in Refs. [5][6][7][8][9]. Both in apoptosis and in generation of calcium signals, individual mitochondria may be recruited in a spatially and temporally coordinated manner (reviewed in Refs. 10 -12). However, it has also been reported that in different regions of polarized cells, mitochondria have discrete roles in the control of cellular function (13,14). A major reason for the specialization is that individual mitochondria not only sense and respond to changes in global [Ca 2ϩ ] c 1 but establish local communication with adjacent domains of the ER (reviewed in Refs. 7, 15, and 16) and with the plasma membrane (17,18). Thus, organization of the local control of mitochondria has become a major focus of interest.
Rizzuto et al. (19,20) demonstrated that mitochondria accumulate Ca 2ϩ during IP 3 -induced [Ca 2ϩ ] c signals and proposed that mitochondrial Ca 2ϩ uptake is promoted by the short lasting high [Ca 2ϩ ] c microdomain that occurs in the vicinity of the activated IP3Rs. Indeed, subdomains of the ER have been visualized in close association with the mitochondria (21), and more Ca 2ϩ puff sites were visualized in close association with the mitochondria than expected if Ca 2ϩ release sites were randomly distributed (22). Evidence has also been presented that the localized and short lasting Ca 2ϩ release during fundamental Ca 2ϩ release events (calcium sparks) results in depolarization (23) and a [Ca 2ϩ ] m rise (referred to as calcium marks) in a neighboring mitochondrion (24). Furthermore, we have shown that synchronized activation of the reticular Ca 2ϩ release sites evokes an almost maximal activation of the entire population of mitochondrial Ca 2ϩ uptake sites (25). Based on these results, we proposed that the reticular Ca 2ϩ release sites and the mitochondrial Ca 2ϩ uptake sites are concentrated at the ER/sarcoplasmic reticulum mitochondrial interface, providing the molecules for a "fast track" of calcium signal delivery between the two organelles (25,26). An intriguing possibility is that the local communication exhibits plasticity that may add a new dimension to decoding of the local Ca 2ϩ signals by the mitochondria.
Dynamic reorganization of the ER/sarcoplasmic reticulum and mitochondrial morphology (27) provides a means for rearrangement of the close associations between ER and mitochondria. In addition, the IP3Rs (28 -31) and ryanodine receptors (32) exhibit activity-dependent changes. Based on studies of Ca 2ϩ uptake by isolated liver mitochondria, Kröner (33) has also proposed a relatively slow, allosteric activation of the mitochondrial uniporter by Ca 2ϩ (reviewed in Ref. 34). Along this line, up-regulation of the mitochondrial response to [Ca 2ϩ ] c oscillations has been observed in hepatocytes (35) and in H9c2 myoblasts (36). However, in insulin-secreting cells, desensitization of the [Ca 2ϩ ] m signal was shown during repetitive stimulation with the mitochondrial substrate, methyl succinate (37). Furthermore, in HeLa cells, desensitization of the [Ca 2ϩ ] m signal was documented during IP 3 -linked [Ca 2ϩ ] c oscillations (38). Although the results with isolated mitochondria and various cells seem to be in conflict with each other, these studies raise the possibility that the uniporter-mediated Ca 2ϩ trans-port may exhibit a memory function during IP3R or ryanodine receptor-mediated Ca 2ϩ release. However, despite the exciting implications, it has not yet been possible to monitor the uniporter Ca 2ϩ permeability in cells.
Here we use novel fluorescent methods to determine the time course of the uniporter Ca 2ϩ permeability during IP 3 -induced ER Ca 2ϩ release and mitochondrial Ca 2ϩ uptake in RBL-2H3 mast cells. We demonstrate delayed deactivation of the mitochondrial Ca 2ϩ uptake sites and sensitization to Ca 2ϩ during repetitive stimulation. Our studies provide evidence that a memory switch of the uniporter is triggered by exposure of the uniporter to high local Ca 2ϩ and that the Ca 2ϩ rise is sensed by a calmodulin/calmodulin-like element associated with the uniporter. Based on these data, we propose that the uniporter exhibits a memory function, providing a means to optimize the mitochondrial Ca 2ϩ signal and increase in metabolism elicited by IP 3 and ryanodine receptor-dependent [Ca 2ϩ ] c signals.
After dye loading, the cells were rinsed with extracellular medium, and the coverslip was mounted to a heated (35°C) open incubator chamber in 1 ml of extracellular medium containing 0.25% bovine serum albumin and sulfinpyrazone. Fluorescence imaging of [Ca 2ϩ ] m (fura2FF) and [Ca 2ϩ ] c (fluo3) was carried out using a beam splitter (510 nm)/emission filter (520-nm long pass) combination optimized for simultaneous recording of fura and fluo type fluorophores (Chroma Technology Co., Brattleboro, VT) and a high quantum efficiency cooled CCD camera as described earlier (25) Image analysis was performed using custom-made software. Fluorescence (averaged from pixels covering whole cell areas) (Figs. 1A and 3B) was determined separately for each individual cell (ϳ35-50 cells/ field) on the field after subtraction of the background fluorescence measured at cell-free areas. In confocal images, pixels covering individual or small groups of mitochondria were averaged (Fig. 1B).

Fluorometric Measurements of [Ca 2ϩ ] c , [Ca 2ϩ ] m , Mn 2ϩ
Quench and ⌬⌿ m in Suspensions of Permeabilized Cells-RBL-2H3 cells were loaded with fura2FF for measurements of [Ca 2ϩ ] m , permeabilized with digitonin using a protocol that preserves the functional integrity of the calcium coupling between ER and mitochondria, and incubated as described earlier (25,39). Briefly, fura2FF-loaded cells (5.5 ϫ 10 6 cells/ml) were permeabilized in ICM supplemented with 25-35 g/ml digitonin (50% (w/w); Sigma) for 5 min at 35°C, followed by washout of the released cytosolic fura2FF (125 ϫ g for 4 min). Cell permeabilization was evaluated by Trypan blue exclusion, and after a 5-min incubation, Ͼ95% of the cells were Trypan-positive. Compartmentalized fura2FF has been shown to occur in the mitochondria of RBL-2H3 cells (25). Permeabilized cells were resuspended in ICM supplemented with 2 mM succinate and 0.25 M rhod2/FA or rhod5F/FA (Fig. 4) and maintained in a stirred thermostated cuvette at 35°C. Rhod2/FA or rhod5F/FA was added to monitor [Ca 2ϩ ] in the intracellular medium that exchanges readily with the cytosolic space. Fluorescence was monitored in a multiwavelength excitation dual wavelength emission fluorimeter using 340-and 380-nm excitation and 500-nm emission for fura2FF and 540-nm excitation and 580-nm emission for rhod2. In every experiment, five data triplets were obtained per second. In the Mn 2ϩ quench experiments, excitation of fura2FF was also carried out at 357 nm (5 data points/s), where the fura2FF-fluorescence was insensitive to Ca 2ϩ changes. At the end of each Mn 2ϩ quench measurement, a high concentration of Mn 2ϩ (500 M MnCl 2 ) was added in the presence of ionomycin to quench compartmentalized fura2FF completely. Before normalization to the initial fluorescence, the residual signal, autofluorescence of the cells was subtracted from the fluorescence signal. Calibration of the Ca 2ϩ signals was carried out at the end of the measurements as described previously (26). Although calibration of fura2FF and rhod2 fluorescence in terms of concentration (nM) of Ca 2ϩ was not feasible in the presence of Mn 2ϩ , the IP 3 -induced changes in Rfura2FF and Frhod2 were highly reproducible in every experiment and were translated to a [Ca 2ϩ ] m increase of 3.79 Ϯ 0.36 M and a [Ca 2ϩ ] c increase of 898 Ϯ 35 nM, respectively (n ϭ 4 -6) (26).
Fluorimetric measurements of ⌬⌿ m were carried out as described previously (39).
Experiments were carried out with 3-6 different cell preparations, and the data are shown as mean Ϯ S.E. Significance of differences from the relevant controls was calculated by Student's t test.  (Fig. 1A). Using a range of morphological, functional, and pharmacological assays, it has been shown previously that fura2FF is compartmentalized into the mitochondria and can be used to measure [Ca 2ϩ ] m in RBL-2H3 cells (25,26 Fig. 1A). Interestingly, correlation of IP 3 -induced Ca 2ϩ release directly with the activity of the uniporter indicated that sensitization oc-  (41,42). The following experiments were designed to test how the permeability of the mitochondrial uniporter changes during IP 3 -induced Ca 2ϩ mobilization.

Improvement of the ER-mitochondrial
Probing the Uniporter Ca 2ϩ Permeability by Mn 2ϩ Quench of Fura2FF Compartmentalized into the Mitochondria-We have used compartmentalized fura2FF to visualize the [Ca 2ϩ ] m rise evoked by IP 3 -induced Ca 2ϩ release in permeabilized RBL-2H3 cells (25) (Fig. 2B). Based on the following considerations, Mn 2ϩ quench of compartmentalized fura2FF can also be used to monitor the Ca 2ϩ permeability of the uniporter: 1) similar to fura2, fura2FF is insensitive to the changes in [Ca 2ϩ ] when excited at 360 nm (isosbestic point); 2) Mn 2ϩ quenches fura2FF fluorescence, and the Mn 2ϩ quench is relatively insensitive to changes in [Ca 2ϩ ] when fura2FF is excited at the isosbestic point ( Fig. 2A, bottom); 3) fura2FF is 20-fold less sensitive to Ca 2ϩ than fura2 (K d ϭ 4 -5 M versus 224 nM) ( Fig. 2A, top), but only 2-fold higher [Mn 2ϩ ] is required to quench fura2FF than fura2 ( Fig. 2A, middle); 4) Mn 2ϩ is transported through the mitochondrial uniporter (43) (reviewed in Refs. 34 and 44); 5) Mn 2ϩ can be added at any given time point during IP 3 -induced Ca 2ϩ release. Thus, Mn 2ϩ added to fura2FF-loaded permeabilized RBL-2H3 cells enters the mitochondria through the uniporter and quenches the fura2FF fluorescence in an essentially stoichiometric manner until the dye becomes saturated, providing a means to estimate the permeability of the uniporter. Importantly, the low concentrations of Mn 2ϩ required to induce the quench of mitochondrial fura2FF have been shown previously not to affect Ca 2ϩ release through the IP3R (45). The addition of Mn 2ϩ (50 M MnCl 2 ; [Mn 2ϩ ] c ϳ4 M) (46) to the suspension of nonstimulated fura2FF-loaded permeabilized RBL-2H3 cells caused an immediate drop in the fluorescence (ϳ20% of the total quenchable fluorescence when the dye was excited at the isosbestic point) followed by a slow quench of the residual fura2FF (Fig. 2B, lower left, hairline). The rapid quench was not sensitive to either blockers of the mitochondrial Ca 2ϩ uptake (FCCP, antimycin A, ruthenium red) (Fig. 2B, bottom) or an inhibitor of the ER Ca 2ϩ uptake (Tg; Fig. 2B, bottom) and was absent when the cytosol was completely removed after cell permeabilization (e.g. experiments with adherent permeabilized cells) (Fig. 3). Thus, the initial decrease in fluorescence reflected the quench of the rapidly accessible cytosolic fura2FF. Since several cycles of the washing of the permeabilized cell suspension was not feasible, Mn 2ϩ quench in nonstimulated cells was recorded in every To monitor [Ca 2ϩ ] c simultaneously with the fluorescence of fura2FF in permeabilized cell suspension, rhod2-free acid was added to the bathing medium that was continuous with the cytosol after cell permeabilization (25,26). Exposure of the permeabilized cells to IP 3 resulted in Ca 2ϩ release that appeared as a rapid [Ca 2ϩ ] c increase and was sensitive to Tg pretreatment but not to mitochondrial blockers (Fig. 2B, top). The IP 3 -induced [Ca 2ϩ ] c increase was associated with a rapid mitochondrial Ca 2ϩ uptake manifested as a large [Ca 2ϩ ] m increase (recorded with compartmentalized fura2FF), which was eliminated after Tg predepletion of the ER or after application of mitochondrial Ca 2ϩ uptake inhibitors (Fig. 2B, middle). Mn 2ϩ applied 10 s after IP 3 stimulation caused a rapid, large quench of fura2FF fluorescence (ϳ60% of the total quenchable fluorescence; Fig. 2B, lower left). The quench rate of the compartmentalized fura2FF was ϳ100-fold higher than in the time control (Fig. 2B, lower left). The IP 3 -dependent Mn 2ϩ quench was completely eliminated by Tg pretreatment, and it was also prevented by FCCP, antimycin A, or Ru360, suggesting that IP 3 -induced Ca 2ϩ release activated uniporter-mediated and mitochondrial membrane potential-driven Mn 2ϩ influx to the mitochondria. These data also indicate that fura2FF compartmentalized in compartments other than mitochondria such as ER or secretory vesicles is unlikely to contribute to the IP 3induced Mn 2ϩ quench in RBL-2H3 cells, since Tg pretreatment promotes the IP 3 receptor-mediated Mn 2ϩ flux to the ER (46) and antimycin A does not eliminate the driving force of the Mn 2ϩ uptake by the secretory vesicles.
IP 3 -induced Sustained Elevation of the Mn 2ϩ Permeability of the Uniporter in Suspensions of Permeabilized Cells-To determine the time course of the Ca 2ϩ permeability of the uniporter during IP 3 -induced Ca 2ϩ release, Mn 2ϩ was added at different intervals after IP 3 stimulation (Fig. 3A). Importantly, during IP 3 -induced Ca 2ϩ release, a stable mitochondrial membrane potential was recorded by JC1 (see Fig. 7C, bottom panel), and a constant amount of Mn 2ϩ was added in all conditions. Thus, both the membrane potential and the [Mn 2ϩ ] gradient components of the driving force for mitochondrial Mn 2ϩ uptake remained essentially unchanged under all tested conditions, so the Mn 2ϩ flux to the mitochondria was determined only by the changes in the permeability of the uniporter. Fig. 3A (left) shows that the fastest and largest quench occurred when Mn 2ϩ was added together with IP 3 . However, enhancement of the Mn 2ϩ quench was also recorded 10 s and even 90 s after the addition of IP 3 , suggesting a sustained increase in the permeability of the mitochondrial Ca 2ϩ uptake sites. This result was unexpected, since the IP 3 -induced increase in global [Ca 2ϩ ] c could induce only a very small activation of mitochondrial Ca 2ϩ uptake (25), and the IP 3 -induced Ca 2ϩ release that exposes the neighboring mitochondria to large local [Ca 2ϩ ] c elevations is completed in Ͻ3 s (Fig. 3A,  upper right). Also, the slow decay of the Mn 2ϩ quench response was observed when IP 3 was added together with Tg that abolished Ca 2ϩ recycling to the ER and so prevented the generation of persistent Ca 2ϩ hotspots in the vicinity of the IP3Rs (n ϭ 3, not shown). To clarify whether the IP 3 -induced sustained increase in Mn 2ϩ permeability could be evoked by the global [Ca 2ϩ ] c rise, Mn 2ϩ quench response evoked by the addition of CaCl 2 (10 M) to the cytosolic buffer was also measured (Fig.  3A, lower right). Despite the Ͼ2-fold higher global [Ca 2ϩ ] c rise, the Ca 2ϩ -induced initial mitochondrial Mn 2ϩ quench response was only 20% of the response evoked by the IP 3 addition. Collectively, these data suggest that the IP 3 -induced large increase in the permeability of the uniporter is relatively sustained. The sustained increase in Mn 2ϩ permeability cannot be a consequence of the IP 3 -induced prolonged and modest rise in global [Ca 2ϩ ] c but is likely to reflect a slowly decaying response of the mitochondrial Ca 2ϩ uptake sites to a short lasting and large local calcium signal. Thus, a low affinity Ca 2ϩ sensor seems to sense the Ca 2ϩ release and to trigger slowly decaying activation of the uniporter.  (25). To model the [Ca 2ϩ ] c spiking evoked by submaximal activation of IP 3 -linked cell surface receptors, a submaximal dose of IP 3 (125-175 nM) was used for stimulation. Mn 2ϩ added 60 s after IP 3 elicited a rapid quench of compartmentalized fura2FF in almost every cell (Fig. 3B, upper row), indicating a relatively homogenous sustained response of the uniporter in the population of cells. Furthermore, a small Ca 2ϩ pulse increased [Ca 2ϩ ] c to the level also measured 60 s after IP 3 addition but evoked only a small [Ca 2ϩ ] m and Mn 2ϩ quench response (Fig. 3B, lower row), consistent with the idea that a large perimitochondrial [Ca 2ϩ ] rise mediates the sustained effect of IP 3 on the Mn 2ϩ permeability of the uniporter.
Sustained Elevation of the Uniporter Ca 2ϩ Permeability in Cells Stimulated with IP 3 -Ca 2ϩ permeability of the uniporter was tested in its "reverse mode." Specifically, after stimulation with saturating IP 3 , uncoupler was added to induce Ca 2ϩ release from the mitochondria, and the rate of the [Ca 2ϩ ] c rise was determined. Ca 2ϩ release or uptake by the ER could not contribute to the uncoupler-induced [Ca 2ϩ ] c signal, because Tg (2 M) was added together with IP 3 , so complete Ca 2ϩ depletion and inhibition of ER Ca 2ϩ uptake was achieved in a few seconds. To prevent Ca 2ϩ release from the mitochondria through the permeability transition pore, the experiments were carried out in the presence of cyclosporin A (2 M). Cyclosporin A by itself had no effect on the mitochondrial Ca 2ϩ uptake or on the sustained phase of IP 3 -induced activation of Mn 2ϩ quench (not shown). To clarify the role of the uniporter, uncoupler-induced Ca 2ϩ release was tested in the absence and presence of Ru360 (10 M), and only the Ru360-sensitive component was considered as uniporter-mediated Ca 2ϩ transport. Importantly, calibration with known amounts of Ca 2ϩ showed ⌬[Ca 2ϩ ] c as a linear function of added Ca 2ϩ (nmol) throughout the working range of these experiments, so ⌬[Ca 2ϩ ] c was directly proportional to the amounts of Ca 2ϩ released from mitochondria to the cytosolic compartment.
In addition to the uniporter Ca 2ϩ permeability, the rate of Ca 2ϩ efflux is controlled by the Ca 2ϩ gradient. However the IP 3 -induced [Ca 2ϩ ] m rise peaked in a few seconds and was effectively unchanged for minutes (Fig. 3A). Furthermore, the total Ca 2ϩ content of the mitochondria at the time of uncoupler addition was also estimated from the Ca 2ϩ release evoked by ionomycin. As shown in Fig. 4A, very similar ionomycin-induced [Ca 2ϩ ] c signals were recorded at 60 and 180 s after IP 3 addition. By contrast, the [Ca 2ϩ ] c increase caused by FCCP was substantially faster at 60 s of IP 3 stimulation than at 180 s. Accordingly, the Ca 2ϩ release rates show a ϳ3-fold faster rate at 60 s than at 180 s (Fig. 4B). Importantly, even 180 s after IP 3 stimulation, the rate of FCCP-induced Ca 2ϩ release is significantly higher than that in the presence of Ru360 (22.4 Ϯ 0.57 versus 7.3 Ϯ 0.6%; n ϭ 4, p Ͻ 0.001). Taken together, these data demonstrate a sustained increase in the Ca 2ϩ permeability of uniporter during IP 3 -induced Ca 2ϩ release and are in good agreement with the data on the Mn 2ϩ quench time course. Furthermore, the mitochondrial Ca 2ϩ efflux studies also suggest that the sustained increase in the permeability of the uniporter after IP 3 stimulation does not have a directional preference.
Reactivation of the Uniporter during the Sustained Phase of IP 3 -induced Activation-The sustained increase in the permeability of the uniporter after IP 3 -induced Ca 2ϩ release may represent either a sensitized state that allows for an augmented response to a subsequent Ca 2ϩ exposure or a slow inactivation with an attenuated response during further Ca 2ϩ stimulation. To discriminate between these two possibilities, a "two-pulse" protocol was used. Specifically, a Ca 2ϩ pulse (5 M CaCl 2 ) was added to naive and IP 3 -pretreated (90 s) permeabilized cells, respectively (Fig. 5). As compared with naive cells, the [Ca 2ϩ ] c after the Ca 2ϩ addition was only 40% larger in IP 3 -pretreated cells, whereas the increase in Mn 2ϩ permeability was more than 400%. The rapid Mn 2ϩ quench involved a component that was due to IP 3 by itself; however, after subtraction of this component, the Ca 2ϩ pulse-induced component remained Ͼ2-fold larger than that of naive cells (12.7 Ϯ 2.8 versus 6 Ϯ 2.7%, p Ͻ 0.05, n ϭ 3-7). To determine whether the modest difference in [Ca 2ϩ ] c between naive and IP 3 -pretreated cells could result in the difference in Mn 2ϩ permeability, Ca 2ϩ pulse was also added to permeabilized cells that were preexposed to 10 M CaCl 2 for 90 s or to 2 M CaCl 2 for 10 s (Fig. 5). ] c were achieved upon the addition of the Ca 2ϩ test pulse in all three conditions; however, the Mn 2ϩ quench was substantially larger in the IP 3 -pretreated cells than in the other conditions. Thus, exposure of the mitochondria to the rapidly decaying IP 3 -in-duced Ca 2ϩ release resulted in a sustained increase in the response of the uniporter to a subsequent Ca 2ϩ exposure. Significantly, the exposure to 10 M CaCl 2 for 90 s also evoked a larger response to the test pulse than 2 M CaCl 2 for 10 s (p Ͻ 0.02, n ϭ 3), suggesting that a massive increase in bulk [Ca 2ϩ ] c can also cause preconditioning of the uniporter. Together, these data demonstrate that the IP 3 -induced sustained increase in the uniporter Mn 2ϩ permeability represents a state of the transporter sensitized to activation by Ca 2ϩ .
Potentiation of Mitochondrial Ca 2ϩ Uptake by IP 3 -induced Ca 2ϩ Mobilization-To evaluate whether facilitation of the uniporter Mn 2ϩ permeability by IP 3 -induced Ca 2ϩ mobilization is translated to potentiation of mitochondrial uptake, we used the two-pulse protocol and determined the second pulse-induced mitochondrial Ca 2ϩ uptake by measuring the decay rates of the [Ca 2ϩ ] c elevations. First, 5 M CaCl 2 was added 5 s after IP 3 -induced Ca 2ϩ release (Fig. 6A, left). The Ca 2ϩ -induced [Ca 2ϩ ] c rise displayed a rapid decline (thick line; 27.4 Ϯ 2.1 nM/s, n ϭ 6) that was prevented by ruthenium red (dotted line).
To match the magnitude of the IP 3 ϩ 5 M CaCl 2 -induced [Ca 2ϩ ] c signal, 8 M CaCl 2 was added to naive cells. Only a relatively slow decay occurred (Fig. 6A, left hairline; 9.9 Ϯ 0.5 nM/s, n ϭ 6). Thus, the 5-s pretreatment with IP 3 enhanced the Ca 2ϩ uptake activity. At 30 s after the IP 3 addition, the decay rate of the 5 M CaCl 2 -induced [Ca 2ϩ ] c spike was smaller than FIG. 4. Sustained increase in the uniporter Ca 2؉ permeability during IP 3 -induced Ca 2؉ release. To induce Ca 2ϩ efflux from the mitochondria at different intervals after IP 3 ϩ Tg stimulation, the ⌬⌿ m was dissipated by the addition of uncoupler. Ca 2ϩ efflux was measured as an increase in [Ca 2ϩ ] c . We have shown that the exchanger-mediated Ca 2ϩ efflux is little in RBL-2H3 cells (25) and added cyclosporin A (2 M) to prevent permeability transition-dependent Ca 2ϩ release. To determine the uniporter-mediated component of the uncoupler-induced Ca 2ϩ release, FCCP/oligomycin was applied in the presence or absence of Ru360 (FCCP ϩ Ru360 or FCCP; addition marked by the arrow). at 5 s but was still greater than the decay rate of the 8 M CaCl 2 (30 s of IP 3 ϩ 5 M CaCl 2 , 18.1 Ϯ 0.7 nM/s, n ϭ 6) (Fig. 6A, right,  thick line). In another control experiment, 3 M CaCl 2 was added first to mimic IP 3 -induced global [Ca 2ϩ ] c increase, and the 5 M CaCl 2 test pulse was applied 5 or 30 s later. In these conditions, the [Ca 2ϩ ] c decay rate was not different from the decay rate of the 8 M CaCl 2 (Fig. 6A, right). For each protocol described above, the mean increase in global [Ca 2ϩ ] c and the decay rate of the [Ca 2ϩ ] c rise are also shown in Fig. 6B. Both in IP 3 -and in CaCl 2 -pretreated cells, ruthenium red caused an essentially complete inhibition of the decay in [Ca 2ϩ ] c elevations, confirming that the decay was due to mitochondrial Ca 2ϩ uptake (Fig. 6A, shown as dotted or dashed linear regression  lines). Importantly, uptake of Ca 2ϩ to the ER was blocked by Tg (2 M, applied 5 s before IP 3 or CaCl 2 ) in all experiments shown in Fig. 6. Thus, the ER could not contribute to the uptake of the Ca 2ϩ test pulse. Collectively, these data demonstrate sensitization of the uniporter-mediated mitochondrial Ca 2ϩ uptake to Ca 2ϩ following IP 3 -induced Ca 2ϩ release. Sensitization may provide a mechanism underlying up-regulation of the [Ca 2ϩ ] m response during [Ca 2ϩ ] c oscillations (Fig. 1). The finding that up-regulation of the [Ca 2ϩ ] m response is not apparent in every cell may be because the first Ca 2ϩ release has already evoked optimal activation of the mitochondrial Ca 2ϩ uptake. Nevertheless, the phenomenon that we describe in ER-mitochondrial calcium signaling is similar to the synaptic facilitation, whereby successive action potentials release increasing amounts of transmitter.
Plasticity of Mitochondrial Ca 2ϩ Uptake Involves a Calmodulin dependent Mechanism-The results described above support the hypothesis that the Ca 2ϩ uniporter exhibits a memory switch that enables mitochondria to optimize Ca 2ϩ uptake during repetitive Ca 2ϩ spikes. A further important question is the mechanism of this memory switch. A major mechanism of Ca 2ϩ -dependent molecular memory utilizes the Ca 2ϩ -binding protein, CaM (recently reviewed in Ref. 47). In order to check whether Ca 2ϩ -CaM-mediated regulation was involved in the sustained activation of the uniporter induced by IP 3 stimulation, the effect of CaM antagonists was tested on the time course of the activation/deactivation of the uniporter using the same experimental protocol as used for Fig. 3A (Fig. 7). CaM inhibitors W7 (500 M), However, 90 s after IP 3 stimulation, the Mn 2ϩ quench was attenuated in the W7-pretreated cells (Fig. 7A, compare traces b at the left and right). Similar to W7, E 6 -berbamine, TFP, and calmidazolium failed to control the initial Mn 2ϩ quench (TFP and calmidazolium, no change; E 6 -berbamine, slight inhibition) (Fig. 7B), but exerted a large inhibition on the IP 3 -induced sustained increase in the uniporter Mn 2ϩ permeability (50 -90% inhibition at 90 s) (Fig. 7B). Notably, the sustained phase of the uniporter activation caused by the Ca 2ϩ addition was also sensitive to CaM inhibitors (W7, E 6 -berbamine, and TFP; n ϭ 2, data not shown). These data together suggest that a major component of the IP 3 -induced Mn 2ϩ quench is dependent on CaM or on a CaM-related factor. Some effects of CaM are independent of Ca 2ϩ ; however, the absence of modulation of the initial phase of the IP 3 effect by the CaM antagonists indicates that the [Ca 2ϩ ] rise is also required to establish the CaM-dependent control of the uniporter.
One potential mechanism utilized in CaM-dependent memory involves protein phosphorylation. CaM kinase II has been reported to undergo Ca 2ϩ -and CaM-dependent autophosphorylation that disables the kinase regulatory domain, thereby converting the enzyme to a Ca 2ϩ -independent form and entrapping the bound CaM (48). Thus, autophosphorylation of CaM kinase II may serve to prolong the effects of brief Ca 2ϩ transients (49). A similar mechanism could also play a role in the memory switch of the mitochondrial uniporter. However, a peptide that inhibits activation of the CaM kinase II (25 M) and wide spectrum protein kinase inhibitors KT5926 (10 M) and staurosporine (1 M) did not have significant inhibitory effect on the IP 3 -induced sustained increase of the Mn 2ϩ permeability of the uniporter (n ϭ 3; data not shown). Thus, Ca 2ϩ -CaM-dependent protein phosphorylation does not appear to be involved in facilitation of the uniporter-mediated Ca 2ϩ uptake. Another mechanism of short term plasticity involves permanent binding of apocalmodulin (apoCaM) to ion channels (L-, P/Q-, and R-type voltage-operated Ca 2ϩ channels, potassium channels, and sodium channels (for references, see Ref. 50); store-operated Ca 2ϩ channels (51); IP3R (52,53); ryanodine receptors (54)). The bound CaM senses the [Ca 2ϩ ] rise and, in turn, goes through a conformational change, leading to facilitation of the channel activity. Since the molecular identity of the uniporter is unknown, and it is not known whether CaM or a peculiar CaM-like domain is available in the mitochondria to interact with the uniporter, it is not feasible to study the mechanism of the facilitation by co-expression of the uniporter and wild-type or mutant CaM (50). However, our data indicating that a CaM kinase II-like inhibitory peptide (25 M) or purified CaM (25 M, human brain, n ϭ 2, not shown) did not change the IP 3 -induced sustained increase of the Mn 2ϩ permeability are compatible with Ca 2ϩ sensing by CaM or a CaM-like motif permanently bound to the uniporter. Further supporting this model, W7 and TFP, which inhibited sustained activation of the uniporter (Fig. 7), have also been described as interfering with the Ca 2ϩ regulation of other Ca 2ϩ channels that bind apoCaM (IP3R (53); L-type Ca 2ϩ channels (reviewed in Ref. 55)).
The uniporter traverses the inner mitochondrial membrane, so the Ca 2ϩ regulatory site should face either the intermembrane space or the mitochondrial matrix. Since the outer mitochondrial membrane is not permeable to particles of Ͼ1,500 Da, the lack of effect of the inhibitory peptide or externally added CaM could be explained by their relatively high molecular mass (2,300 and 17,000 Da, respectively). To evaluate this point, cells were pretreated with truncated Bid that induced permeabilization of the outer mitochondrial membrane and release of cytochrome c (14,000 Da) and Smac/DIABLO (25,000 Da) from the intermembrane space (39,56). In tBid-pretreated cells, potentiation of the uniporter by IP 3 -induced Ca 2ϩ release was preserved, and the inhibitory peptide or CaM failed to affect the sustained increase in Mn 2ϩ permeability (not shown). Although, in the presence of tBid, the outer mitochondrial membrane is likely to permit free passage of the peptide and CaM, one may further speculate that the Ca 2ϩ -CaM regulatory site is localized to the mitochondrial matrix, and the inner mitochondrial membrane barrier prevented the inhibitory peptide or CaM from entering the matrix space. However, the peak of the [Ca 2ϩ ] m rise evoked by the addition of 10 M CaCl 2 was as large as the effect of saturating IP 3 (25), whereas the 10 M CaCl 2 -induced sustained Mn 2ϩ entry was relatively small (Fig. 3). This observation indicates that the sustained response was not determined by the magnitude of the [Ca 2ϩ ] m rise. Notably, studies of Kröner (33) with isolated mitochondria also described an allosteric regulation of the uniporter-mediated Ca 2ϩ transport by a Ca 2ϩ regulatory site that readily exchanges Ca 2ϩ with the cytosolic buffer. Together, the present results are most consistent with a mechanism wherein the memory switch of the uniporter is mediated by sensing of a [Ca 2ϩ ] rise in the intermembrane space by apoCaM tightly bound to the uniporter or a CaM-like motif (EF-hand) of the uniporter. Preassociation of apoCaM to the uniporter may provide a means to ensure rapid and selective effect of the short lasting perimitochondrial [Ca 2ϩ ] signals established by the IP3R-mediated Ca 2ϩ release.
Based on ion flux measurements, the uniporter has been proposed to be a gated Ca 2ϩ channel (34). At the level of single channels, the memory switch might appear as promotion of channel opening by Ca 2ϩ . Along the line of the model proposed for the Ca 2ϩ -induced facilitation of the P/Q type Ca 2ϩ channels (50), the Ca 2ϩ binding to preassociated apoCaM may keep a putative inactivation gate from closing. Unfortunately, an electrophysiology study of the uniporter has not been feasible to directly evaluate the ion channel model.
Conclusion-The first important point in this study is the demonstration of an increase in the efficacy of calcium signal propagation to the mitochondria during [Ca 2ϩ ] c oscillations. We and others have shown that pulsatile and frequency-modulated [Ca 2ϩ ] c signals are superior to amplitude-modulated [Ca 2ϩ ] c responses in the control of mitochondrial metabolism, because the discrete [Ca 2ϩ ] c spikes that comprise the [Ca 2ϩ ] c oscillations are each delivered efficiently into the mitochondrial matrix (57)(58)(59)(60). This phenomenon has been attributed to a local communication between IP3R/ryanodine receptors and mitochondrial Ca 2ϩ uptake sites (19 -21, 25, 57, 61). The present work demonstrates that sensitization of mitochondrial Ca 2ϩ uptake induced by the initial [Ca 2ϩ ] c spike may also provide a fundamental mechanism to ensure optimal detection of [Ca 2ϩ ] c oscillations by the mitochondria. Although sensitization does not necessarily appear as a progressive increase in the size of the ⌬[Ca 2ϩ ] m associated with sequential [Ca 2ϩ ] c spikes, it may serve to maintain [Ca 2ϩ ] m spiking despite an increase in the mitochondrial matrix Ca 2ϩ buffering capacity and a decrease in the amplitude of the [Ca 2ϩ ] c oscillations.
Second, we established approaches that allowed us, for the first time, to monitor the permeability of the mitochondrial Ca 2ϩ uptake sites during IP 3 -induced Ca 2ϩ mobilization. Based on the time course of [Ca 2ϩ ] m , it has been difficult to discriminate between the respective contribution of mitochondrial Ca 2ϩ uptake, efflux, and buffering. Measurements of the Mn 2ϩ quench and uncoupler-induced Ca 2ϩ efflux provided evidence that IP 3 elicited a large and slowly decaying increase in the Ca 2ϩ permeability. After the decay of the [Ca 2ϩ ] c spike, the relatively sustained increase in Ca 2ϩ permeability results in uptake of only a small amount of Ca 2ϩ ; however, during the next [Ca 2ϩ ] c spike, Ca 2ϩ uptake may be largely facilitated. In the present study, this has been directly demonstrated by application of a "two-pulse" protocol. Sensitization of the mitochondrial Ca 2ϩ uptake could also be triggered by a large Ca 2ϩ prepulse, suggesting that facilitation of the uniporter during the IP 3 -induced calcium signal is mediated by a low affinity Ca 2ϩ sensor that responds to Ca 2ϩ released in the vicinity of the uniporter.
Third, the pharmacological experiments we conducted suggest that CaM is important for the Ca 2ϩ -induced facilitation. Our data are consistent with the model that CaM is tethered constitutively to the uniporter or a CaM-like motif is an auxiliary subunit of the uniporter. The interaction with CaM/CaMlike motif itself does not appear to be essential for Ca 2ϩ permeation through the uniporter; however, CaM seems to confer Ca 2ϩ dependence on the facilitation. As far as we know, this is the first example of Ca 2ϩ -CaM control of a mitochondrial ion transporter.
In a previous work, we have shown that several aspects of the Ca 2ϩ signal transmission from ER to the mitochondria are analogous to the functional organization of the synaptic transmission (25). This work demonstrates that the propagation of IP 3 -induced Ca 2ϩ signals to the mitochondria exhibits facilitation that is also a fundamental element of synaptic plasticity. One may also speculate that the continuous reorganization of the ER and mitochondria could result in formation of novel associations between ER and mitochondria, similar to the appearance of new synapses, another mechanism underlying synaptic facilitation. Notably, in HeLa and B cells, Ca 2ϩ -dependent desensitization of the mitochondrial Ca 2ϩ uptake toward agonist-induced Ca 2ϩ signals has also been reported (37,38), and this phenomenon is similar to synaptic depression. Thus, the ER-mitochondrial calcium signaling exhibits major mechanisms of memory and adaptation.