Deletion of mitochondrial calcium uniporter incompletely inhibits calcium uptake and induction of the permeability transition pore in brain mitochondria

Ca2+ influx into mitochondria is mediated by the mitochondrial calcium uniporter (MCU), whose identity was recently revealed as a 40-kDa protein that along with other proteins forms the mitochondrial Ca2+ uptake machinery. The MCU is a Ca2+-conducting channel spanning the inner mitochondrial membrane. Here, deletion of the MCU completely inhibited Ca2+ uptake in liver, heart, and skeletal muscle mitochondria. However, in brain nonsynaptic and synaptic mitochondria from neuronal somata/glial cells and nerve terminals, respectively, the MCU deletion slowed, but did not completely block, Ca2+ uptake. Under resting conditions, brain MCU-KO mitochondria remained polarized, and in brain MCU-KO mitochondria, the electrophoretic Ca2+ ionophore ETH129 significantly accelerated Ca2+ uptake. The residual Ca2+ uptake in brain MCU-KO mitochondria was insensitive to inhibitors of mitochondrial Na+/Ca2+ exchanger and ryanodine receptor (CGP37157 and dantrolene, respectively), but was blocked by the MCU inhibitor Ru360. Respiration of WT and MCU-KO brain mitochondria was similar except that for mitochondria that oxidized pyruvate and malate, Ca2+ more strongly inhibited respiration in WT than in MCU-KO mitochondria. Of note, the MCU deletion significantly attenuated but did not completely prevent induction of the permeability transition pore (PTP) in brain mitochondria. Expression level of cyclophilin D and ATP content in mitochondria, two factors that modulate PTP induction, were unaffected by MCU-KO, whereas ADP was lower in MCU-KO than in WT brain mitochondria. Our results suggest the presence of an MCU-independent Ca2+ uptake pathway in brain mitochondria that mediates residual Ca2+ influx and induction of PTP in a fraction of the mitochondrial population.

Mitochondrial Ca 2ϩ uptake and the ability of Ca 2ϩ to regulate mitochondrial functions have been established for many years (1)(2)(3)(4). Ca 2ϩ uptake by mitochondria plays an important role for the organelle and for the whole cell. It helps to maintain low cytosolic Ca 2ϩ when Ca 2ϩ influx into the cell is increased (e.g. following stimulation of ionotropic glutamate receptors in neurons) (5)(6)(7). The delivery of Ca 2ϩ into the mitochondrial matrix can activate mitochondrial dehydrogenases such as pyruvate, ␣-ketoglutarate, and isocitrate dehydrogenases and, thereby, can stimulate mitochondrial respiration and ATP production (4). Finally, excessive Ca 2ϩ uptake may lead to mitochondrial damage through induction of the permeability transition pore (PTP) 2 (3,8). Ca 2ϩ transport into mitochondria is mediated by the Ca 2ϩ uniporter, a Ca 2ϩ channel complex in the inner mitochondrial membrane (9,10), the molecular identity of which was recently revealed (11,12).
In recent studies, it was found that mitochondria contain several proteins involved in Ca 2ϩ transport into the organelle: MCU (mitochondrial calcium uniporter) (11,13), MCUb (14), MICU1 and MICU2 (15)(16)(17), EMRE (18), and MCUR (19). Among other components of mitochondrial Ca 2ϩ transport machinery, MCU is believed to be the Ca 2ϩ channel traversing the inner mitochondrial membrane (IMM) (13). In experiments with purified MCU reconstituted into a planar bilayer, ruthenium red, an inhibitor of the Ca 2ϩ uniporter (20,21), blocked MCU-mediated Ca 2ϩ currents (11). Genetic ablation of MCU completely prevented Ca 2ϩ uptake by skeletal muscle and heart mitochondria and inhibited induction of the PTP but failed to protect cells from cell death and failed to protect cardiac tissue from ischemia-reperfusion injury (22).
The Ca 2ϩ -dependent induction of the PTP is the major mechanism of Ca 2ϩ -induced mitochondrial damage (3,8,23,24). In the brain, PTP induction and consequent mitochondrial damage are the primary mechanisms contributing to glutamate cro ARTICLE excitotoxicity (25)(26)(27), a major deleterious factor in stroke, secondary traumatic brain injury, and various neurodegenerations. An induction of the PTP requires Ca 2ϩ influx into mitochondria (28). Correspondingly, knockdown of MCU reduced NMDA-induced increases in mitochondrial Ca 2ϩ in neurons, leading to milder mitochondrial depolarization and elevated resistance to excitotoxicity (29). On the other hand, MCU KO failed to protect brain from hypoxic-ischemic damage, despite strongly hindering mitochondrial Ca 2ϩ uptake and PTP induction (30).
One reason for this controversy could be an incomplete inhibition of Ca 2ϩ uptake in brain mitochondria from MCU-KO mice. However, brain mitochondria from MCU-KO mice are poorly characterized in terms of Ca 2ϩ uptake, membrane potential changes, respiratory activities, and susceptibility to PTP induction. In the present study, we investigated the effect of MCU KO on Ca 2ϩ uptake and membrane potential in isolated brain nonsynaptic and synaptic mitochondria as well as in liver, heart, and skeletal muscle mitochondria. We also compared respiratory activities of brain mitochondria isolated from WT and MCU-KO mice. Finally, we evaluated susceptibility of MCU-KO and WT mitochondria to PTP induction. Our results demonstrate that there is a unique MCU-independent, Ru360sensitive Ca 2ϩ uptake in brain mitochondria that results in PTP induction in a fraction of vulnerable organelles. Thus, these findings evince a potential mechanism by which deletion of MCU might be not very efficacious in protecting brain mitochondria from PTP induction and, consequently, in protecting the brain from ischemia-reperfusion insults.

The effect of MCU deletion on mitochondrial Ca 2؉ uptake
All MCU-KO mice used in our experiments were genotyped (Fig. 1A) and all mitochondrial preparations were tested for MCU by immunoblotting (Fig. 1B). The level of MCU expression in brain nonsynaptic mitochondria of MCU-KO mice was below the detection limit of Western blotting. Expression of MCU paralog, MCUb (18), was slightly but statistically significantly lower in brain nonsynaptic mitochondria from MCU-KO mice compared with mitochondria from WT animals. In our study, we used a high-resolution Ca 2ϩ -sensitive electrode that enabled us to evaluate, with high precision, changes in the rate of Ca 2ϩ uptake by mitochondria from MCU-KO and WT mice. To assess the rate and capacity of Ca 2ϩ uptake, multiple pulses of CaCl 2 were applied to mitochondria (Fig. 1, D and E). Mitochondrial Ca 2ϩ uptake was monitored by measuring a decrease in external Ca 2ϩ . Brain nonsynaptic mitochondria isolated from WT mice accumulated Ca 2ϩ faster than mitochondria from MCU-KO animals (Fig. 1D). Despite this, Ca 2ϩ uptake capacity, determined as the amount of Ca 2ϩ accumulated by mitochondria before they failed to accumulate additional Ca 2ϩ , was not significantly different. BSA significantly increased the Ca 2ϩ uptake capacity (Fig. 1E). The Ca 2ϩ uptake capacity is limited by an induction of the PTP (30). Because BSA binds free fatty acids (31), which are activators of the PTP (32), its inclusion in the incubation medium defers PTP induction and, consequently, increases Ca 2ϩ uptake capacity (33). In the presence of BSA, the Ca 2ϩ uptake capacities of MCU-KO and WT mitochondria remain similar (Fig. 1E). In parallel experiments, we tested the effect of Ca 2ϩ pulses on mitochondrial membrane potential (Fig. 1, F and G). The decreased rate of Ca 2ϩ uptake by MCU-KO mitochondria (Fig. 1, D and E) could result from these mitochondria being less polarized. However, under resting conditions, mitochondria from MCU-KO and WT mice accumulated comparable amounts of TPP ϩ , suggesting similar polarization of both MCU-KO and WT mitochondria. The repetitive Ca 2ϩ pulses produced fast, transient depolarizations in WT mitochondria (Fig. 1F), whereas in MCU-KO mitochondria, Ca 2ϩ pulses produced slower depolarizations with much smaller amplitude (Fig. 1G). At the end of these experiments, mitochondria were treated with the uncoupler 2,4-dinitrophenol (2,4-DNP) to produce maximal depolarization. A 70-kDa subunit of Complex II (CII) was used as a loading control. C, statistical analysis of immunoblotting. Data are mean Ϯ S.D. (error bars), n ϭ 3. *, p Ͻ 0.001 comparing MCU/Complex II in WT and MCU-KO mitochondria. D, Ca 2ϩ uptake by brain mitochondria from WT (blue trace) and MCU-KO (red trace) mice. Here and in E, mitochondria (0.18 mg of protein/ml) were incubated in the standard incubation medium supplemented with 3 mM pyruvate plus 1 mM malate, 0.1 mM ADP, and 1 M oligomycin. E, Ca 2ϩ uptake by WT (blue trace) and MCU-KO (red trace) brain mitochondria incubated in the presence of 0.1% BSA (free from fatty acids). F and G, the effect of Ca 2ϩ on mitochondrial membrane potential in WT and MCU-KO mitochondria, respectively. In these experiments, the standard incubation medium was supplemented with 0.1% BSA (free from fatty acids). D-G, Ca 2ϩ was added to mitochondria where indicated as 10 or 20 M CaCl 2 pulses, respectively. Numbers in parentheses indicate CaCl 2 concentrations in M. a.u., arbitrary units.

Ca 2؉ uptake and PTP in MCU-KO brain mitochondria
In contrast to brain nonsynaptic mitochondria, mitochondria isolated from liver, heart, and skeletal muscles of MCU-KO mice were not able to take Ca 2ϩ up and depolarize in response to Ca 2ϩ (Fig. 2). The MCU expression in heart mitochondria was the lowest compared with other tissues (Fig. 2J). The level of MCU in mitochondria from MCU-KO mice was below the detection limit of Western blotting. The level of MCUb varied among mitochondria from different tissues, with skeletal muscle mitochondria having the least amount of MCUb and heart mitochondria having the greatest amount of MCUb (Fig. 2K).
Expression of MCUb was distinctly lower in skeletal muscle mitochondria from MCU-KO mice compared with MCUb level in skeletal muscle mitochondria from WT animals.
ETH129, an electrophoretic Ca 2ϩ ionophore (34), can transport Ca 2ϩ across the membrane independently of the Ca 2ϩ uniporter (35). Ru360, an inhibitor of the Ca 2ϩ uniporter (36), strongly inhibited Ca 2ϩ uptake (Fig. 3B). When the Ca 2ϩ uniporter was blocked by Ru360, ETH129 restored Ca 2ϩ uptake by brain nonsynaptic mitochondria. Similarly, ETH129 significantly increased Ca 2ϩ uptake rate in brain nonsynaptic In A-C, liver mitochondria (1.5 mg of protein/ml) were incubated in the standard incubation medium supplemented with 3 mM pyruvate plus 1 mM malate, 0.1 mM ADP, 1 M oligomycin, 0.1% BSA (free from fatty acids). B and C, the effect of Ca 2ϩ on mitochondrial membrane potential in WT and MCU-KO liver mitochondria, respectively. Where indicated, Ca 2ϩ was added to mitochondria as 20 or 40 M CaCl 2 pulses, respectively. Here and in other panels, numbers in parentheses indicate CaCl 2 concentrations in M. In B and C, 60 M 2,4-DNP was used to depolarize mitochondria. D, Ca 2ϩ uptake by WT (blue trace) and MCU-KO (red trace) heart mitochondria (0.7 mg of protein/ml) incubated in the same medium as liver mitochondria. In E and F, the effect of Ca 2ϩ on mitochondrial membrane potential in WT and MCU-KO heart mitochondria, respectively. In E and F, 60 M 2,4-DNP was used to depolarize mitochondria. In G, Ca 2ϩ uptake by WT (blue trace) and MCU-KO (red trace) skeletal muscle mitochondria (1.3 mg of protein/ml) incubated in the same medium as liver mitochondria. H and I, effect of Ca 2ϩ on mitochondrial membrane potential in WT and MCU-KO skeletal muscle mitochondria, respectively. H and I, 60 M 2,4-DNP was used to depolarize mitochondria. J and K, Western blots demonstrating the lack of MCU in mitochondria from MCU-KO mice and the presence of MCUb in mitochondria of WT and MCU-KO animals, respectively. A 70-kDa subunit of Complex II (CII) was used as a loading control. In L and M, statistical analyses of immunoblotting. Data are mean Ϯ S.D. (n ϭ 3). L, **, p Ͻ 0.001 comparing MCU/Complex II in heart mitochondria with MCU/Complex II in liver and skeletal muscle mitochondria. M, *, p Ͻ 0.05 comparing MCUb/Complex II in skeletal muscle mitochondria from WT and MCU-KO mice; **, p Ͻ 0.001 comparing MCU/Complex II in WT or MCU-KO heart mitochondria with WT or MCU-KO liver and skeletal muscle mitochondria. a.u., arbitrary units.

Ca 2؉ uptake and PTP in MCU-KO brain mitochondria
mitochondria from MCU-KO mice (Fig. 3C). This is consistent with unaffected polarization of MCU-KO mitochondria and supports the notion that deletion of MCU was the main cause of inhibition of Ca 2ϩ uptake. On the other hand, Ru360 strongly inhibited Ca 2ϩ uptake by MCU-KO mitochondria (Fig. 3C). Fig. 3D shows statistical analysis of the data. Fig. 3A shows a calibration of the Ca 2ϩ -selective electrode with consecutive 100 M CaCl 2 pulses.
The intriguing finding that despite MCU deletion, MCU-KO brain nonsynaptic mitochondria were able to accumulate Ca 2ϩ (Fig. 1, D and E), suggested that either MCU is not a Ca 2ϩ channel essential for Ca 2ϩ uptake and serves a regulatory role or there are alternative pathways for Ca 2ϩ influx into mitochondria. Because reconstituted MCU was able to generate Ca 2ϩ currents in experiments with planar lipid bilayer (11), the former hypothesis seems unlikely. The alternative pathways for Ca 2ϩ uptake by MCU-KO brain mitochondria could include mitochondrial Na ϩ /Ca 2ϩ exchanger, operating in reverse (37,38), or ryanodine receptor, found in mitochondria (mRR) (39,40). However, neither CGP37157, an inhibitor of mitochondrial Na ϩ /Ca 2ϩ exchanger (41), nor dantrolene, an inhibitor of ryan-odine receptor that inhibited mRR in previous studies (39,40), suppressed the residual Ca 2ϩ uptake by MCU-KO mitochondria ( Fig. 4, C-F). Here again, Ru360, strongly inhibited Ca 2ϩ uptake by WT and MCU-KO mitochondria (Fig. 4, B and C).

The role of MCU in regulation of mitochondrial respiration
In the mitochondrial matrix, Ca 2ϩ activates pyruvate dehydrogenase as well as the Krebs cycle enzymes isocitrate dehydrogenase and ␣-ketoglutarate dehydrogenase (4), and therefore, inhibition of Ca 2ϩ influx into mitochondria might affect mitochondrial respiration. To test this possibility, we performed respirometry experiments using a Clark electrode and isolated brain nonsynaptic mitochondria fueled either with pyruvate plus malate (Complex I substrates) (Fig. 5, A and B) or with succinate (a Complex II substrate) (Fig. 5, C and D). In the latter case, the incubation medium was supplemented with 3 mM glutamate to remove oxaloacetate in transaminase reaction to prevent oxaloacetate accumulation and inhibition of succinate dehydrogenase (42,43). We analyzed basal respiration  Ca 2؉ uptake and PTP in MCU-KO brain mitochondria (V 2 ), governed predominantly by proton permeability of the IMM; ADP-stimulated respiration (V 3 ), determined primarily by activity of the oxidative phosphorylation system; the controlled respiration (V 4 ), determined by proton permeability of the IMM and possible contaminations with ATPase activities; and uncoupled respiration (V DNP ), governed by activity of the mitochondrial electron transport chain. In our experiments, we did not find a significant difference in respiration of mitochondria from MCU-KO and WT mice under any tested conditions. Fig. 5 (E and F) shows statistical analyses of respirometry data. Thus, deletion of MCU does not affect mitochondrial respiration under resting, ADP-stimulated, and uncoupled conditions. The Ca 2ϩ influx into mitochondria may induce the PTP (28), and this may affect mitochondrial respiration (44). Interestingly, this effect depends on the type of oxidative substrates used in these experiments. With WT mitochondria oxidizing pyruvate plus malate, Ca 2ϩ produced strong inhibition of respiration that could be rescued by succinate (Fig. 6A). Further improvement was achieved by adding a combination of 3 mM NADH and 30 g/ml cytochrome c (Cyt c) (Fig. 6B). With MCU-KO mitochondria, Ca 2ϩ -induced respiratory inhibition was also apparent but was much weaker (Fig. 6C). Similar to experiments with WT mitochondria, succinate and Cyt c plus NADH improved respiration of MCU-KO mitochondria (Fig.  6D), and the effect of succinate was stronger than with WT mitochondria. Glutamate added after succinate failed to further improve mitochondrial respiration (Fig. 6, A-D). Fig. 6E shows statistical analyses of these experiments.

The effect of MCU deletion on PTP opening
The Ca 2ϩ -induced inhibition of mitochondrial respiration fueled by Complex I substrates was demonstrated previously (44). This could be explained by an induction of the PTP and loss of mitochondrial NADH (45). On the other hand, stimulation of mitochondrial respiration fueled by succinate was also attributed to PTP induction leading to uncoupling of oxidative phosphorylation and the increased respiratory rate (44). Taken together, the effects of Ca 2ϩ on mitochondrial respiration suggested a decreased propensity of MCU-KO mitochondria for PTP induction. To test this hypothesis, we monitored simultaneously changes in light scattering of mitochondrial suspension, indicative of mitochondrial swelling, and distribution of lipophilic cation tetraphenylphosphonium (TPP ϩ ) across the IMM, indicative of changes in mitochondrial membrane potential (Fig. 7). An addition of CaCl 2 (100 M) to WT nonsynaptic mitochondria produced a rapid decrease in light scattering, indicating swelling of the organelles, and an increase of TPP ϩ concentration in the incubation medium, indicating mitochondrial depolarization (Fig. 7A). To quantify mitochondrial swelling, alamethicin, a pore-forming agent (46), was added to mito-

Ca 2؉ uptake and PTP in MCU-KO brain mitochondria
chondria at the end of experiments to produce maximal swelling. With MCU-KO nonsynaptic mitochondria, Ca 2ϩ produced less swelling and depolarization compared with WT mitochondria, suggesting suppressed PTP induction (Fig. 7B). Ru360, an inhibitor of the Ca 2ϩ uniporter (36), strongly inhibited PTP induction in MCU-KO mitochondria (Fig. 7C), consistent with strong inhibition of Ca 2ϩ uptake in these mitochondria produced by this agent (Fig. 3C). ETH129, an electrophoretic Ca 2ϩ ionophore (34) that accelerated Ca 2ϩ influx into MCU-KO mitochondria (Fig. 3C), restored Ca 2ϩ -induced mitochondrial swelling and depolarization (Fig. 7D). To attribute mitochondrial swelling to PTP induction, we applied a combination of cyclosporin A (CsA), ADP, and BSA, known inhibitors of the PTP (Fig. 7E). The PTP inhibitors completely prevented Ca 2ϩ -induced mitochondrial swelling, thus showing that it is attributable to PTP induction. Fig. 7F shows statistical analysis of these experiments. Overall, the presented data indicate that PTP induction in MCU-KO mitochondria is significantly but incompletely suppressed.
The expression of mitochondrial cyclophilin D (CyD) and the levels of ADP and ATP in mitochondria are major factors that regulate PTP induction (47). CyD sensitizes mitochondria to Ca 2ϩ -triggered PTP induction (48,49). Di-and triphosphate adenine nucleotides antagonize PTP induction by an incompletely understood mechanism (50). We assessed the CyD level in brain nonsynaptic mitochondria isolated from WT and MCU-KO mice and did not find a difference (Fig. 8). In parallel experiments, we assessed the levels of ADP and ATP in WT and MCU-KO brain nonsynaptic mitochondria (Fig. 8). ADP and ATP levels were comparable with those we reported previously (51). The difference in ATP levels was statistically insignificant, whereas ADP was slightly but statistically significantly decreased in mitochondria from MCU-KO mice. The reason for this difference is not clear. These data indicate that suppressed PTP induction in MCU-KO mitochondria was not due to a decrease in CyD expression or elevated ADP and/or ATP levels in these mitochondria.

Figure 8. CyD expression and ADP and ATP content in brain nonsynaptic mitochondria isolated from WT and MCU-KO mice.
A, Western blots demonstrating CyD levels in brain homogenates and nonsynaptic mitochondria isolated from brains of WT and MCU-KO mice. Note that despite the same protein loading for mitochondria and homogenate samples, there is enrichment of CyD and VDAC1 in the isolated mitochondria fraction compared with brain homogenates. MEK1/2 was used as a cytosolic marker to illustrate the purity of the mitochondrial preparation. VDAC1 was used as a loading control. B, statistical analysis of immunoblotting. Data represent a ratio of CyD to VDAC1 band intensities and are shown as mean Ϯ S.D. (error bars) (n ϭ 5). For ADP and ATP measurements (C and D, respectively), nonsynaptic mitochondria were incubated in the standard incubation medium supplemented with 3 mM pyruvate and 1 mM malate for 10 min at 37°C. Then ADP and ATP levels were measured as described under "Experimental procedures." Data are shown as mean Ϯ S.D. (error bars), n ϭ 6. *, p Ͻ 0.05 comparing ADP content in WT versus MCU-KO mitochondria. a.u., arbitrary units.

Ca 2؉ uptake and PTP in MCU-KO brain mitochondria
Brain nonsynaptic mitochondria used in our study originated from neuronal somata and glial cells, whereas synaptic mitochondria are from nerve terminals and are of purely neuronal origin. We tested synaptic mitochondria from MCU-KO mice and found properties similar to nonsynaptic mitochondria (Fig. 9). The level of MCU in synaptic mitochondria from MCU-KO mice was below the detection limit of Western blotting (Fig. 9A). The level of MCUb in synaptic mitochondria from MCU-KO mice was 2-fold lower than in synaptic mito-chondria from WT mice (Fig. 9, A and B). In synaptic mitochondria, deletion of MCU decreased the rate of Ca 2ϩ uptake but did not completely prevent Ca 2ϩ influx and membrane depolarization (Fig. 9, C and D). Consistent with this, MCU deletion in synaptic mitochondria inhibited Ca 2ϩ -dependent PTP induction but failed to completely prevent it (Fig. 9, E-G).

Discussion
In the present study, we discovered a unique MCU-independent Ca 2ϩ uptake mechanism existing in brain mitochondria isolated from MCU-KO mice (Figs. 1 and 9). We found that deletion of MCU significantly slows the rate of Ca 2ϩ influx into brain mitochondria, but does not completely abolish it. Consistently, the Ca 2ϩ -triggered induction of PTP in brain mitochondria from MCU-KO mice is hindered, but not completely prevented. Based on these findings, deletion of MCU may have limited ability to protect brain mitochondria from PTP induction and may incompletely protect neurons against glutamateinduced Ca 2ϩ dysregulation and excitotoxicity, in which an induction of PTP in mitochondria plays a key role (52).
Ca 2ϩ uptake by mitochondria plays an important role in regulation of cellular and mitochondrial functions. Mitochondria can accumulate large amounts of Ca 2ϩ , contributing to maintenance of low resting cytosolic Ca 2ϩ (9). In the mitochondrial matrix, Ca 2ϩ can activate pyruvate, ␣-ketoglutarate, and isocitrate dehydrogenases, thereby stimulating mitochondrial respiration and oxidative phosphorylation (4). In our study, however, MCU deletion failed to cause significant changes in mitochondrial respiration, suggesting that this kind of regulation might be not essential. This is in line with the notion that rapid Ca 2ϩ influx into mitochondria mediated by MCU does not regulate resting mitochondrial Ca 2ϩ levels and, consequently, MCU does not affect the resting mitochondrial respiration. On the other hand, the excessive accumulation of Ca 2ϩ in mitochondria may lead to the induction of PTP, manifested in mitochondrial depolarization and swelling (8). Both events were evaluated in our study, and both were significantly reduced in MCU-KO mitochondria compared with WT mitochondria.
The Ca 2ϩ -induced PTP damages mitochondria and, therefore, is inextricably involved in mitochondrial pathophysiology. The molecular identity of PTP is still a matter of debate (53). However, it is well established that Ca 2ϩ must enter mitochondria to induce the PTP (28). The Ca 2ϩ uniporter is the main pathway for Ca 2ϩ influx into mitochondria (10). Recently, the molecular identity of the Ca 2ϩ uniporter was revealed (11,12). It appears that the Ca 2ϩ uniporter is composed of several subunits, among which MCU serves as the Ca 2ϩ channel spanning the IMM (9). Consequently, genetic ablation of MCU in MCU-KO mice led to complete inhibition of Ca 2ϩ uptake in skeletal muscle and heart mitochondria (22). How MCU deletion affects Ca 2ϩ transport in brain mitochondria was not completely clear. Recently, Nichols et al. (54) reported that Ca 2ϩ influx in brain mitochondria from MCU-KO mice was inhibited and PTP induction was hindered. However, in this study, mitochondrial depolarization induced by Ca 2ϩ was not unequivocally attributed to PTP induction by preventing it with PTP inhibitors. In addition, it was not unambiguously proven

Ca 2؉ uptake and PTP in MCU-KO brain mitochondria
that inhibition of mitochondrial Ca 2ϩ uptake is indeed due to MCU deletion. Whether putative inhibition of PTP induction was the result of suppressed Ca 2ϩ influx into mitochondria or due to other factors remained unclear.
The lack of difference in CyD expression and comparable or higher levels of ADP and ATP in WT mitochondria are inconsistent with the idea that these factors might contribute to the decreased PTP induction in MCU-KO mitochondria. On the other hand, restoration of PTP induction in MCU-KO mitochondria in the presence of ETH129, an electrophoretic Ca 2ϩ ionophore that can deliver Ca 2ϩ into mitochondria (34,35) and accelerate Ca 2ϩ influx into MCU-KO mitochondria (Fig. 3C), suggests that inhibition of Ca 2ϩ uptake due to MCU deletion is the main cause of PTP inhibition in MCU-KO mitochondria. Nevertheless, the residual Ca 2ϩ uptake in MCU-KO mitochondria apparently could induce PTP in the fraction of vulnerable mitochondria. This explains why Ca 2ϩ could inhibit respiration of MCU-KO mitochondria fueled by Complex I substrates. Under these conditions, recovery of mitochondrial respiration with succinate is consistent with previously published results (44). Further improvement with Cyt c and NADH suggests the loss of Cyt c and depletion of mitochondrial NADH due to PTP induction (44,45) as the main mechanisms for respiratory inhibition.
The main unanswered question in our study concerns the mechanism of residual Ca 2ϩ uptake in MCU-KO mitochondria. Previously, it was reported that mitochondria may possess an mRR that may contribute to mitochondrial Ca 2ϩ uptake (39,40). In addition, it has been well established that mitochondria possess a Na ϩ /Ca 2ϩ exchanger (mNCX) responsible for Ca 2ϩ efflux from mitochondria (55). It is possible that mNCX, similarly to plasmalemmal NCX (56,57), may work in reverse under conditions of metabolic inhibition and/or elevated cytosolic Ca 2ϩ , bringing Ca 2ϩ into mitochondria (37,38). To test these possibilities, we used dantrolene, an inhibitor of ryanodine receptor, previously used in experiments with mitochondria (39,40), and CGP37157, an inhibitor of mNCX (41). We used these agents in concentrations that previously produced inhibitory effects on the corresponding targets, mRR or mNCX. In our experiments, however, neither of these agents affected the residual Ca 2ϩ uptake by MCU-KO mitochondria. On the other hand, Ru360, an inhibitor of the Ca 2ϩ uniporter (36), completely blocked Ca 2ϩ uptake by MCU-KO mitochondria. The question remains as to which protein(s) interacts with Ru360 in the absence of MCU and how this leads to complete inhibition of residual Ca 2ϩ uptake.
It is possible that mitochondria possess MCU-independent pathways for Ca 2ϩ uptake (58, 59). One of the possible candidates for the MCU-independent Ca 2ϩ uptake pathways is MCUb (14), which was detected in isolated brain mitochondria (Figs. 1B and 9A). Although Raffaello et al. (14) proposed that MCUb forms a nonconducting channel and this correlated with the lack of Glu-257 in MCUb, the structure study by Oxenoid et al. (60) argued against that conclusion. They found that the Glu-257 is dispensable for Ca 2ϩ uptake, thus implying that MCUb could also conduct Ca 2ϩ . In addition, it is known that Ser-259 of MCU is responsible for inhibition of the MCU-mediated Ca 2ϩ uptake by Ru360 (12, 60). This Ser-259 is con-served both in MCU and MCUb. These previous studies therefore raised the possibility that MCUb could be responsible for the MCU KO-resistant and Ru360-sensitive mitochondrial Ca 2ϩ uptake found in the present study. Alternatively, our data obtained with brain mitochondria do not rule out the possibility that MCU might not be a Ca 2ϩ channel, but instead could be an accessory, regulatory subunit of the mitochondrial Ca 2ϩ transport complex. However, our experiments with liver, heart, and skeletal muscle mitochondria as well as experiments with MCU reconstituted in the planar lipid bilayer, in which MCU ion channel activity has been detected (11), argue against these hypotheses.
Interestingly, a low expression of MCU and high expression of MCUb in WT heart mitochondria correlated with high Ca 2ϩ uptake capacity of heart mitochondria compared with liver and skeletal muscle mitochondria. The amount of Ca 2ϩ that can be accumulated by mitochondria is limited due to PTP induction (30). Consequently, our findings suggest that the levels of expression of MCU and MCUb may determine the propensity to PTP induction in mitochondria.
The failure to completely block Ca 2ϩ uptake by genetic ablation of MCU in brain mitochondria and, subsequently, incomplete prevention of PTP induction suggests that targeting MCU alone might not be very efficacious in protecting neurons from glutamate-induced Ca 2ϩ dysregulation and glutamate excitotoxicity. It is conceivable that combined deletion of MCU and CyD might be much more effective in neuroprotection. CyD deletion increases mitochondrial Ca 2ϩ uptake capacity, inhibits PTP induction, and augments resistance of neurons to deleterious effect of glutamate (52). However, this neuroprotection is limited. An exposure to higher glutamate concentrations that cause larger Ca 2ϩ influx in the cell and, subsequently, into mitochondria, may overcome neuroprotection achieved by CyD deletion (52). It would be very interesting to test in future studies whether a combination of CyD and MCU deletions would result in stronger neuroprotection and whether simultaneous pharmacological inhibition of CyD and MCU could be a valid approach in protecting brain cells from insults associated with elevation of Ca 2ϩ influx into neurons.

Animals
All procedures with animals were performed in accordance with the institutional animal care and use committeeapproved protocol. The breeders of MCU-KO mice were obtained from Dr. Toren Finkel (Center for Molecular Medicine, NHLBI, National Institutes of Health), and breeding col-Ca 2؉ uptake and PTP in MCU-KO brain mitochondria onies were established in the Laboratory Animal Resource Center at Indiana University School of Medicine (Indianapolis, IN). The mice were maintained on a CD1 (Charles River Laboratories, Wilmington, MA) background, and experimental homozygous mice were obtained by heterozygous crosses (22). Age-and sex-matched WT littermates were used as controls. The mice were housed under standard conditions with free access to water and food. In our experiments, we used animals of both sexes. However, in every individual experiment, we only used animals of the same sex. We did not find any significant difference in the results obtained with either sex, and therefore, the data were pooled together for statistical analysis.

Isolation of brain, liver, heart, and skeletal muscle mitochondria
Percoll gradient-purified brain nonsynaptic and synaptic mitochondria from homozygous MCU-KO mice and WT littermates were isolated as we described previously (61). Briefly, brains from three mice of each strain were harvested and processed simultaneously. Following brain tissue homogenization in a 15-ml glass Dounce homogenizer (10 strokes with pestle A, 30 strokes with pestle B) on ice, homogenates were diluted with 30 ml of Isolation Buffer 1 and centrifuged for 10 min at 2,400 rpm in the Beckman centrifuge Avanti J-26XP, rotor JA-25.50 (700 ϫ g). This and all other procedures and centrifugations were performed at 0 -2°C. Then supernatant was centrifuged for 10 min at 12,500 rpm (18,900 ϫ g) in the Beckman centrifuge Avanti J-26XP, rotor JA-25.50. Supernatant was discarded, and the pellet was resuspended in 35 ml of Isolation Buffer 2 and centrifuged for 10 min at 12,500 rpm (18,850 ϫ g) in the Beckman centrifuge Avanti J-26XP, rotor JA-25.50. Next, the pellet was resuspended in 5 ml of Isolation Buffer 3. The suspension was layered onto the top of the Percoll gradient (26%/ 40%) in Beckman Ultra-Clear centrifuge tubes and centrifuged for 28 min at 15,500 rpm (41,100 ϫ g) in the Beckman ultracentrifuge Optima L100K, bucket rotor SW41Ti. Nonsynaptic mitochondria were resuspended in Isolation Buffer 3 and centrifuged for 20 min at 15,500 rpm (41,100 ϫ g) in the Beckman ultracentrifuge Optima L100K, bucket rotor SW41Ti. The pellet was resuspended in Isolation Buffer 3 and centrifuged again for 20 min at 15,500 rpm (41,100 ϫ g) in the Beckman ultracentrifuge Optima L100K, bucket rotor SW41Ti. The pellet of nonsynaptic mitochondria was collected, resuspended in 0.15 ml of Isolation Buffer 3, and stored on ice. Isolation Buffer 1 con-tained 225 mM mannitol, 75 mM sucrose, 10 mM Hepes, pH 7.4, adjusted with KOH, 0.1% BSA, free from fatty acids, and 1 mM EGTA. Isolation Buffer 2 contained 225 mM mannitol, 75 mM sucrose, 10 mM Hepes, pH 7.4, adjusted with KOH, 0.1 mM EGTA. Isolation Buffer 3 contained 395 mM sucrose, 0.1 mM EGTA, 10 mM Hepes, pH 7.4. 26% Percoll in Percoll Buffer was prepared with 5.2 ml of Percoll (Sigma) and 14.8 ml of Percoll Buffer; 40% Percoll in Percoll Buffer was prepared with 8 ml of Percoll and 12 ml of Percoll Buffer. Percoll Buffer contained 320 mM sucrose, 1 mM EGTA, 10 mM Hepes, pH 7.4.
Synaptic mitochondria were isolated from synaptosomes by the nitrogen cavitation method using a nitrogen cell disruption bomb, model 4639 (Parr Instrument Co., Moline, IL), cooled on ice (42) with some modifications. Briefly, the synaptosomes obtained during preparation of nonsynaptic mitochondria were transferred to a cooled beaker and placed into the nitrogen bomb on ice under 1,100 p.s.i. for 13 min. Then synaptosomes were layered on top of the discontinuous Percoll gradient (24%/ 40%) and centrifuged at 15,500 rpm (41,100 ϫ g) for 28 min in a ultracentrifuge Optima L100K, bucket rotor SW41Ti. The mitochondrial fraction in the interface between Percoll layers was transferred into a fresh tube; diluted 1:5 with medium containing 395 mM sucrose, 0.1 mM EGTA, 10 mM Hepes, pH 7.4; and centrifuged for 20 min at 15,500 rpm (41,100 ϫ g) in the Beckman ultracentrifuge Optima L100K, bucket rotor SW41Ti. The pellet was resuspended in 0.15 ml of the latter medium and kept on ice.
Liver, heart, and skeletal muscle mitochondria were isolated and purified in the same way as nonsynaptic brain mitochondria, but without Percoll gradient purification. Liver tissue from one mouse per strain was homogenized using 30 ml of Potter-Elvehjem homogenizers, and then mitochondria were isolated using differential centrifugation as described for nonsynaptic mitochondria before Percoll gradient purification. Hearts from three mice per strain as well as skeletal muscle (quadriceps of one hind limb) from one mouse per strain were disintegrated using Tissue Master 125 grinder (OMNI International, Kennesaw, GA). Then tissues were homogenized using 30-ml Potter-Elvehjem homogenizers, and mitochondria were isolated using differential centrifugation as described for nonsynaptic mitochondria without Percoll gradient purification. Mitochondrial protein concentration was measured by the Bradford method (44) with BSA as a standard.

Mitochondrial Ca 2؉ uptake measurements
Mitochondrial Ca 2ϩ uptake was measured with a miniature Ca 2ϩ -selective electrode in a 0.3-ml chamber at 37°C under continuous stirring. A decrease in Ca 2ϩ concentration in the incubation medium indicated mitochondrial Ca 2ϩ uptake. Here and in other experiments with isolated mitochondria, the standard incubation medium contained 125 mM KCl, 0.5 mM MgCl 2 , 3 mM KH 2 PO 4 , 10 mM Hepes, pH 7.4, 10 M EGTA and was supplemented with 3 mM pyruvate plus 1 mM malate. In addition, in some experiments, the incubation medium was supplemented with 0.1 mM ADP and 1 M oligomycin as described previously (30) and with 0.1% BSA (free from fatty acids, MP Biochemicals, Santa Ana, CA) as indicated. Ca 2ϩ was delivered to mitochondria as 10, 20, or 100 M CaCl 2 pulses.

Ca 2؉ uptake and PTP in MCU-KO brain mitochondria
Ca 2ϩ uptake rates were quantified as nmol of Ca 2ϩ /min/mg of mitochondrial protein.

Mitochondrial respiration
Mitochondrial respiration was assessed in a 0.4-ml continuously stirred chamber containing the standard incubation medium, which was supplemented either with 3 mM pyruvate plus 1 mM malate or 3 mM succinate plus 3 mM glutamate. In experiments with succinate, glutamate was used to prevent inhibition of succinate dehydrogenase by oxaloacetate due to elimination of oxaloacetate in the reaction catalyzed by aspartate aminotransferase (43,62). The chamber was maintained at 37°C and was equipped with a Clark-type oxygen electrode and a tightly sealed lid. The slope of the oxygen consumption trace corresponded to the respiratory rate.

Mitochondrial swelling and membrane potential
Mitochondrial swelling was evaluated in a 0.3-ml continuously stirred chamber at 37°C by following changes in light scattering of mitochondrial suspension at 525 nm with an incident light beam at 180°relative to the photodetector. The incubation medium for light scattering measurements contained 215 mM mannitol, 70 mM sucrose, 0.5 mM MgCl 2 , 3 mM KH 2 PO 4 , 10 mM Hepes, pH 7.4, 10 M EGTA, 3 mM pyruvate, 1 mM malate. A decrease in light scattering of mitochondrial suspension indicated mitochondrial swelling. Maximal mitochondrial swelling was induced by alamethicin (30 g/ml), which was considered as 100% swelling. Ca 2ϩ -induced swelling was calculated as a percentage of maximal, alamethicin-induced swelling (63). Mitochondrial membrane potential was assessed simultaneously with mitochondrial swelling with a TPP ϩ electrode by monitoring TPP ϩ distribution between the incubation medium and mitochondria (64). A decrease in external TPP ϩ concentration corresponds to mitochondrial polarization, whereas an increase of TPP ϩ concentration in the incubation medium corresponds to depolarization.

Mitochondrial ADP and ATP content
ADP and ATP levels were determined using a luciferin/luciferase-based ATP bioluminescent somatic cell assay kit (Sigma) and a GloMax 20/20 luminometer (Promega, Madison, WI). Mitochondria were incubated for 10 min at 37°C in the standard incubation medium supplemented with 3 mM pyruvate plus 1 mM malate, and then ADP and ATP were measured. Measurements were made in 4% perchloric acid extracts neutralized by KOH following the kit manufacturer's suggestions. ADP was converted to ATP using pyruvate kinase in the presence of phosphoenolpyruvate (67).

Statistics
Data are shown as mean Ϯ S.D. of the indicated number of separate experiments. Statistical analysis of the experimental results consisted of unpaired t test or one-way analysis of variance followed by Bonferroni's post hoc test if applicable (GraphPad Prism version 4.0, GraphPad Software Inc., La Jolla, CA). Every experiment was performed using several different preparations of isolated mitochondria.
Author contributions-J. H. conducted most of the experiments, analyzed the results, and contributed to writing the paper. T. B. maintained the MCU-KO breeding colony, conducted genotyping and immunoblotting experiments, and analyzed the results. J. E. R. and Z. L. maintained the MCU-KO colony and conducted genotyping. Y. M. U. analyzed the results and contributed to writing the paper. N. B. conceived the idea for the project, analyzed the results, and wrote the paper. All authors reviewed the results and approved the final version of the manuscript.