Sustained Activity of Calcium Release-activated Calcium Channels Requires Translocation of Mitochondria to the Plasma Membrane*

A rise of the intracellular Ca2+ concentration has multiple signaling functions. Sustained Ca2+ influx across plasma membrane through calcium release-activated calcium (CRAC) channels is required for T-cell development in the thymus, gene transcription, and proliferation and differentiation of naïve T-cells into armed effectors cells. Intracellular Ca2+ signals are shaped by mitochondria, which function as a highly dynamic Ca2+ buffer. However, the precise role of mitochondria for Ca2+-dependent T-cell activation is unknown. Here we have shown that mitochondria are translocated to the plasma membrane as a consequence of Ca2+ influx and that this directed movement is essential to sustain Ca2+ influx through CRAC channels. The decreased distance between mitochondria and the plasma membrane enabled mitochondria to take up large amounts of inflowing Ca2+ at the plasma membrane, thereby preventing Ca2+-dependent inactivation of CRAC channels and sustaining Ca2+ signals. Inhibition of kinesin-dependent mitochondrial movement along microtubules abolished mitochondrial translocation and reduced sustained Ca2+ signals. Our results show how a directed movement of mitochondria is used to control important cellular functions such as Ca2+-dependent T-cell activation.


Cells-Human
Jurkat T-cell lines were isolated and grown as described previously (24,25).
To visualize mitochondria, cells were incubated with 100-200 nM MitoTracker for 30 min at 22Ϫ23°C. Excess dye was removed by washing twice. Cells were illuminated at 490 nm. DCLP 500 (UV) was used as a dichroic mirror and LP 515 as an emission filter.
Bead Stimulation-As described previously (27), we followed the standard procedure for absorbing proteins on polystyrene microparticles (size Ͼ0.5 m) established by Polysciences Europe GmbH Company (www.polysciences.com). Two alterations in the procedure were required to optimize our results. In step 1, we employed 100 l of a 2.5% suspension of beads (ϳ 2.1 ϫ 10 7 beads), and in step 7, we added 50 g of the protein to be absorbed. Chemical composition and pH of buffers were the same as recommended in the protocol. Azid-free anti-human CD3 mAbs (Euroclone) were passively coupled to microparticles (diameter ϭ 5.83 m). They were stored at 4°C in the specified storage buffer until use. Beads were washed twice with phosphatebuffered saline before resuspending them in the Ringer's solution used for the experiments.
Confocal Microscopy-For confocal imaging, a Nipkow discbased scanning head (QLC-100, VisiTech International) was attached to an upright microscope (Eclipse 600, Nikon) equipped with a 100ϫ water lens (numerical aperture 1.1, Nikon). The light source was a 488-nm solid-state laser (Saphire 488 -30, Coherent). A dichroic mirror between the microlens and pinhole disc reflects the emission light that passes a 500-nm long-pass barrier filter. For detection, a charge-coupled device camera (OrcaER, Hamamatsu Photonics) in the 2 ϫ 2 binning mode, resulting in a 672 ϫ 512 pixel resolution, was used. Acquisition was controlled by the VoxCellScan Software (VisiTech International). Two-dimensional time lapse data at a maximum rate of 20 images/s were analyzed using ImageJ (Wayne Rasband, National Institute of Mental Health). Stacks were deconvolved using AutoDeblur X (AutoQuant).
2-Photon Microscopy-The same experimental set-up as previously described was used (28). The emission beam splitter contained a dichroic mirror at 565 nm followed by a channel 1 band pass at 610/75 nm and a channel 2 long pass at 500 nm. The 2-photon excitation setting was 800 nm for dyes with 1-photon excitation near 488 nm. Complete pictures of the sample were taken every 5 min with a resolution of 512 ϫ 512 or 2048 ϫ 2048 pixels.
To stain mitochondria, cells were loaded with 200 nM MitoTracker for 1 h. The plasma membrane was stained with a mouse anti-human CD45-Alexa Fluor488 mAb. Cells were incubated for 30 min with antibodies.
Electrophysiology-Patch clamp experiments were done exactly as described previously (26 Electroporation of Antibodies-Electroporation was carried out as described previously (27) with RPMI 1640 medium containing 0.1 mg/ml anti-kinesin 5B mAb or 0.1 mg/ml anti-hu-  DECEMBER 29, 2006 • VOLUME 281 • NUMBER 52 man IgG Ab. Cells were pulsed with 300 V and 400 microfarads and then cultivated until use (30 min after transfection).

Mitochondria Localization and CRAC Activity
Data Analysis and Statistics-Data were analyzed using TILL Vision (TILL Photonics), Igor Pro (Wavemetrics), and Microsoft Excel. All values are given as mean Ϯ S.E.; n ϭ number of cells. In case data points were normally distrib-uted, a paired or unpaired two-sided Student's t test was used. If normal distribution could not be confirmed, a non-parameterized test (Mann-Whitney) was carried out. Levels of significance are indicated in Figs. 1-5 (* refers to p Ͻ 0.05, ** refers to p Ͻ 0.01, and *** refers to p Ͻ 0.001).

RESULTS
Ca 2ϩ Influx Induces Mitochondrial Translocation toward the Plasma Membrane-To test the hypothesis of whether or not directed mitochondrial movement was important for Ca 2ϩ signaling and Ca 2ϩ -dependent T-cell activation, we used different microscopy techniques to analyze the intracellular localization of mitochondria. Because CRAC channels are stimulated by Ca 2ϩ store depletion and inactivated by store refilling (2), we stimulated T-cells with TG, which has been shown to maximally activate CRAC channels by blocking the sarcoendoplasmatic reticulum Ca 2ϩ ATPase pump (29). The intracellular localization of mitochondria was analyzed before (resting condition) and after T-cell stimulation by the application of TG in the absence and in the presence of extracellular Ca 2ϩ , a standard protocol to activate Ca 2ϩ influx through CRAC channels (Fig. 1A). We found that sustained Ca 2ϩ influx (Ca 2ϩ plateau) was correlated with mitochondrial translocation toward the plasma membrane. This is evident when comparing the changes in a subplasma membrane MitoTracker fluorescence after stimulation between concentric circles, the outer of which reflects the position of the plasma membrane, the inner one placed 0.99 m apart from the outer one ( Fig. 1, B and C). The analysis with two concentric circles was chosen because unstimulated T-cells were fairly round and did not change their shape significantly during thapsigargin treatment. Directed movement toward the plasma membrane was only seen during the sustained Ca 2ϩ plateau (at 1000 s after cell stimulation) but not during store depletion (at 500 s after cell stimulation) or during the very first seconds of the Ca 2ϩ influx phase. A similar mitochondrial translocation toward the plasma membrane was also observed in CD3 ϩ peripheral blood lymphocytes during the Ca 2ϩ plateau (data not shown). We next analyzed the complete mitochondrial population by spacing five concentric circles within the cells with a constant distance of 0.99 m (Fig. 1D). The average distance between mitochondria and the plasma membrane did not change during store depletion compared with resting conditions; however, during the Ca 2ϩ plateau, the fluorescence between the outer circles was significantly increased. In parallel, the Mito-Tracker fluorescence decreased between the inner circles, indicating that the complete mitochondrial population had moved toward the plasma membrane.
To analyze whether mitochondrial movement toward the plasma membrane also occurred in T-cells in the case of T-cell receptor stimulation, we used anti-CD3-coated beads to stimulate the cells. Fig. 1E shows a typical bead experiment in which the contact between bead and cell induces a rise of the intracellular Ca 2ϩ concentration ([Ca 2ϩ ] i ), as previously reported (27).
In parallel, mitochondrial movement toward the plasma membrane was also observed after bead stimulation, as evident from Fig. 1F.
[Ca 2ϩ ] i signals and the amount of mitochondrial movement were comparable with the cells stimulated with TG. In the analysis, we only included cells that did not change their shape following bead stimulation. We observed also, however, translocation of mitochondria to the plasma membrane in cells that did change their shape significantly following bead contact.
Because Ca 2ϩ depletion did not induce any mitochondrial translocation toward the plasma membrane, we tested mitochondrial movement in the following experiments by inducing Ca 2ϩ influx through TG in 1 mM Ca 2ϩ Ringer's solution, as shown in the inset of Fig.  2A. A translocation of mitochondria toward the plasma membrane was already measured 2 min after the onset of Ca 2ϩ entry, suggesting the existence of a fast Ca 2ϩ -dependent mitochondrial movement mechanism ( Fig. 2A). The dependence of mitochondrial movement on Ca 2ϩ influx was supported by the correlation between the translocation of mitochondria toward the plasma membrane and the increase of the extracellular Ca 2ϩ concentration (Fig. 2B), which enhances the driving force for Ca 2ϩ entry, as previously reported (24). Even in the presence of 0.1 or 0.25 mM external Ca 2ϩ , which generates only very little Ca 2ϩ entry through CRAC channels, a small movement of mitochondria toward the plasma membrane was observed. The movement appears to be saturated with 1 mM external Ca 2ϩ present. Another argument for the dependence of the mitochondrial movement on Ca 2ϩ influx can be gathered from the observation that mitochondrial translocation toward the plasma membrane was completely reversed after removing the extracellular Ca 2ϩ solution (Fig.  2C) and was abolished in T-cells loaded with the Ca 2ϩ chelator BAPTA (Fig. 2D). Because Ca 2ϩ influx in T-cells depends strictly on the activity of CRAC channels (24,30), we analyzed the intracellular localization of mitochondria in T-cells treated with the CRAC channel blocker 3,5-bis(trifluoromethyl)pyrazole derivate BTP2 (31). 100 nM BTP2 significantly reduced mitochondrial movement toward the plasma membrane (Fig.  2E). BTP2 did not completely prevent the movement, which was expected, because in the presence of 100 nM BTP2, Ca 2ϩ entry through the CRAC channel is reduced to a level similar to  Fig. 2B). Directed mitochondrial movement toward the plasma membrane in T-cells following Ca 2ϩ influx through CRAC channels was also confirmed by confocal microscopy and by 2-photon microscopy (Fig. 2F). In both cases, the mitochondrial population was translocated to areas close to the plasma membrane after the Ca 2ϩ signal reached the plateau following Ca 2ϩ influx induced by the re-addition of 1 mM Ca 2ϩ solution in TG-pretreated T-cells, an effect not observed after the re-addition of 1 mM Ca 2ϩ solution without TG pretreatment. In addition, we also confirmed mitochondrial movement toward the plasma membrane in T-cells transfected with mitochondrial green fluorescent protein (data not shown). The results shown in Figs. 1 and 2 demonstrate that the directed movement of mitochondria toward the plasma membrane depends on Ca 2ϩ influx through CRAC channels in T-cells.
Mitochondria Translocation toward the Plasma Membrane Is Required for Sustained Ca 2ϩ Signals-The mitochondrial movement toward the plasma membrane following the activation of Ca 2ϩ influx through CRAC channels was almost com-pletely abolished by pre-incubation of the cells with nocodazole, a microtubule inhibitor (Fig. 3, A and  B) but only slightly with latrunculin B, a microfilament inhibitor (data not shown). The inhibition of mitochondrial movement toward the plasma membrane by nocodazole greatly reduced the sustained Ca 2ϩ plateau compared with control conditions (Fig. 3, C and D). The disruption of microtubules induced by nocodazole is time-dependent, because the percentage of subplasma membrane MitoTracker fluorescence between the two concentric circles as well as the sustained Ca 2ϩ signal decrease with the nocodazole incubation time (Fig. 3,  C and D). The Ca 2ϩ re-addition protocol usually generates very high Ca 2ϩ influx signals, because the slow up-modulation of the plasma membrane Ca 2ϩ ATPases cannot prevent the "overshoot" of the intracellular Ca 2ϩ concentration (32). To avoid this [Ca 2ϩ ] i overshoot, we also compared the effect of nocodazole to control conditions by directly activating Ca 2ϩ influx in the presence of external Ca 2ϩ . Under these conditions, mitochondrial movement toward the plasma membrane was completely abolished by nocodazole (Fig. 3E) and the concomitant [Ca 2ϩ ] i rise was also drastically reduced (Fig. 3F). Nocodazole did not interfere with mitochondrial Ca 2ϩ uptake or plasma membrane Ca 2ϩ ATPase activity (supplemental Fig. 1). We therefore postulate that the sustained high [Ca 2ϩ ] i , which depends on Ca 2ϩ influx through CRAC channels, is maintained through stimulation-dependent mitochondrial movement toward the plasma membrane. A testable prediction of this model is that inhibition of mitochondrial movement into the vicinity of the plasma membrane should decrease CRAC current amplitudes.
Mitochondrial Movement toward the Plasma Membrane Controls CRAC Channel Activity-To analyze CRAC channel activity, we used the whole-cell configuration of the patch clamp technique. CRAC channels were activated through depletion of the endoplasmic reticulum Ca 2ϩ store using a combination of InsP 3 and TG. In a first series of experiments, we compared CRAC channel inactivation using a small concentration of the Ca 2ϩ chelator EGTA in the pipette, which is not sufficient to prevent Ca 2ϩ -dependent inactivation of CRAC channels (16). Under these conditions, mitochondrial Ca 2ϩ uptake close to the channels is required to maintain CRAC channel activity by reducing local accumulation of Ca 2ϩ near sites that govern Ca 2ϩ -dependent CRAC channel inactivation ( Fig. 4A; compare also Ref. 16). Because pre-incubation of T-cells with nocodazole inhibits the fast movement of mi-tochondria along microtubules toward the plasma membrane after the Ca 2ϩ influx through CRAC channels, our model predicts that CRAC currents should be inactivated faster under these conditions. We indeed found a significant inactivation of CRAC channels in nocodazole-treated T-cells compared with control conditions (Fig.  4, B and E). To verify that this is not a direct effect of nocodazole on the CRAC channels or an unspecific side effect, we repeated the experiments in the presence of 10 mM EGTA in the pipette solution, which is sufficient to buffer all incoming Ca 2ϩ , thereby preventing CRAC channel inactivation. Under these conditions, nocodazole had no effect on CRAC channel inactivation (Fig. 4, compare C and D; statistics in E). Because mitochondrial Ca 2ϩ uptake close to the plasma membrane is not required to prevent Ca 2ϩ -dependent CRAC channel inactivation in the presence of 10 mM EGTA (Fig. 4C), we can exclude that potential unspecific side effects of nocodazole were responsible for the inactivation of CRAC currents observed in Fig. 4B. The time course of CRAC inactivation (Fig. 4B) is comparable with the onset of mitochondrial movement toward the plasma membrane within the first 2 min following activation of Ca 2ϩ influx ( Fig. 2A). From the experiments in Figs. 3 and 4, we conclude that mitochondrial movement toward the plasma membrane is required to maintain CRAC channel activity and sustained high [Ca 2ϩ ] i signals.
Kinesin Controls Mitochondrial Translocation toward the Plasma Membrane and Sustained Ca 2ϩ Signals-Kinesin motor proteins control the anterograde movement of different organelles and cargos along microtubules (33). Because the translocation of mitochondria toward the plasma membrane is an anterograde motion, we introduced 0.1 mg/ml anti-kinesin 5B mAb into T-cells using electroporation and analyzed in-parallel mitochondrial translocation and [Ca 2ϩ ] i signals following TG stimulation. Anti-kinesin 5B mAb  completely blocked the anterograde mitochondrial translocation toward the plasma membrane ( Fig. 5A; statistics in B) compared with control conditions (electroporation with anti-human IgG mAb), and this movement inhibition correlated with a significant reduction of the sustained Ca 2ϩ signal (Fig. 5C), stressing again the importance of mitochondrial localization beneath plasma membrane for Ca 2ϩ signals and Ca 2ϩ -dependent signal transduction in T-cells. The introduction of the anti-kinesin 5B mAb induced a drastic reduction of the initial amplitude of the Ca 2ϩ signal upon Ca 2ϩ re-addition, which was not observed in cells treated with nocodazole (Fig. 3C). We believe that this discrepancy may be due to the finding that mitochondria collapse close to the nucleus once the function of kinesin has been disrupted (21,34,35), whereas nocodazole may simply freeze the current position of mitochondria. The collapse of mitochondria close to the nucleus has been shown to reduce store-dependent Ca 2ϩ signals (34).
The data with the anti-kinesin 5B mAb explain the fast translocation of mitochondria toward the plasma membrane observed following the Ca 2ϩ rise in T-cells, because kinesin motor proteins move along microtubules by a hand-over-hand mechanism that allows rapid transport of cargoes over long distances (36).

DISCUSSION
We have shown that stimulation of Ca 2ϩ entry in T-cells facilitates translocation of the complete mitochondrial population toward the plasma membrane in a Ca 2ϩ /kinesin-dependent manner along microtubules (Fig. 6). Ca 2ϩ influx through CRAC channels triggers directed mitochondrial motility. The exact mechanism of the directed movement, however, remains unknown. The mitochondrial movement toward the plasma membrane in turn sustains CRAC channel activity and high [Ca 2ϩ ] i by reducing Ca 2ϩ -dependent CRAC channel inactivation. We refer to this as a double positive feedback loop that is initiated by CRAC channel activation through depletion of Ca 2ϩ stores; Ca 2ϩ influx through activated CRAC channels triggers mitochondrial translocation to the plasma membrane, which in turn keeps CRAC channels active. There are numerous examples of such double positive feedback loops; for example, InsP 3 activates InsP 3 -R, which releases Ca 2ϩ , which binds to InsP 3 -R, which in turn releases more Ca 2ϩ . According to a model proposed by Hajnoczky and co-workers (20), the high submembranous [Ca 2ϩ ] i (in our case generated through CRAC activity) would decrease mitochondrial motility thereby immobilizing them close to the plasma membrane. There are, however, some differences between the model of Hajnoczky and co-workers (20) and our data. Hajnoczky and co-workers show that mitochondrial motility is reduced in a myoblast cell line (H9c2) after Ca 2ϩ release and recovered by 50% when the Ca 2ϩ signal reaches a plateau. They conclude that mitochondria are immobilized close to microdomains of high Ca 2ϩ near the endoplasmic reticulum. In T-cells, we do not observe such a high Ca 2ϩ release, and T-cell activation is largely determined by Ca 2ϩ influx through CRAC channels. Ca 2ϩ release in T-cells is usually initiated in large areas throughout the cytosol. Therefore, there would probably be no preferential direction for mitochondria movement. Hence, in the "release" case, Ca 2ϩ -dependent mitochondria movement toward the endoplasmic reticulum would be extremely hard to measure, because such movement would almost not differ from random movement. In the "influx" case, there is a preferential direction for a Ca 2ϩ -dependent movement, namely toward the plasma membrane, where they would be finally immobilized by the high [Ca 2ϩ ] i microdomains close to CRAC channels. Therefore, the data by Hajnoczky and co-workers (20) and our data may be reconcilable, because after an initial movement due to Ca 2ϩ signals (which may not be distinguishable from random movement in case the endoplasmic reticulum is the Ca 2ϩ source), the subsequent high [Ca 2ϩ ] i may immobilize mitochondria (in our case at the plasma membrane). In addition, in flat cells, such as adherent myoblasts, there may be no directed movement of mitochondria to the plasma membrane, because mitochondria are already close to the influx channel (in z dimension).
Because Ca 2ϩ influx through CRAC channels increases the accumulation of mitochondria close to the plasma membrane and this localization near CRAC channels sustains CRAC activity, we propose that CRAC channels and mitochondria control each other in a double positive feedback loop. The concomitant increase in global [Ca 2ϩ ] i shifts the transcription factor profile away from mostly NFAT-AP1 to a combination of NFAT, NFAT-AP1, and NFB ( Fig. 6; see also Refs. 3 and 7). Consequently, this transcriptional profile change leads to a differential protein expression. Thus, a small change in mitochondrial localization is likely translated into a large effect on T-cell activity.