Mitochondrial Uncoupling Protein-4 Regulates Calcium Homeostasis and Sensitivity to Store Depletion-induced Apoptosis in Neural Cells*

An increase in the cytoplasmic-free Ca2+ concentration mediates cellular responses to environmental signals that influence a range of processes, including gene expression, motility, secretion of hormones and neurotransmitters, changes in energy metabolism, and apoptosis. Mitochondria play important roles in cellular Ca2+ homeostasis and signaling, but the roles of specific mitochondrial proteins in these processes are unknown. Uncoupling proteins (UCPs) are a family of proteins located in the inner mitochondrial membrane that can dissociate oxidative phosphorylation from respiration, thereby promoting heat production and decreasing oxyradical production. Here we show that UCP4, a neuronal UCP, influences store-operated Ca2+ entry, a process in which depletion of endoplasmic reticulum Ca2+ stores triggers Ca2+ influx through plasma membrane “store-operated” channels. PC12 neural cells expressing human UCP4 exhibit reduced Ca2+ entry in response to thapsigargin-induced endoplasmic reticulum Ca2+ store depletion. The elevations of cytoplasmic and intramitochondrial Ca2+ concentrations and mitochondrial oxidative stress induced by thapsigargin were attenuated in cells expressing UCP4. The stabilization of Ca2+ homeostasis and preservation of mitochondrial function by UCP4 was correlated with reduced mitochondrial reactive oxygen species generation, oxidative stress, and Gadd153 up-regulation and increased resistance of the cells to death. Reduced Ca2+-dependent cytosolic phospholipase A2 activation and oxidative metabolism of arachidonic acid also contributed to the stabilization of mitochondrial function in cells expressing human UCP4. These findings demonstrate that UCP4 can regulate cellular Ca2+ homeostasis, suggesting that UCPs may play roles in modulating Ca2+ signaling in physiological and pathological conditions.

First identified and studied in brown fat cells, where they can dissociate oxidative phosphorylation from respiration and thereby promote heat production, mitochondrial uncoupling proteins (UCPs) 3 are a family of proton carrier proteins located in the inner mitochondrial membrane (1). Five mammalian UCPs, each with distinct tissue distributions, have been identified: UCP1 is expressed primarily in brown fat cells, UCP2 is widely expressed, including in neurons and glial cells, UCP3 is expressed principally in skeletal muscle cells, and UCP4 and UCP5 (also called BMCP1) are expressed in the nervous system (1). The functions of UCPs in neurons have not been established, but, based on their functions in other cell types, they are presumed to modulate the production of ATP and reactive oxygen species (ROS). One recent study showed that UCP5 is present in neurons throughout the brain and that overexpression of UCP5 in cultured neural cell line results in higher state 4 respiration and reduced mitochondrial membrane potential and ROS levels (2). Another study showed that levels of UCP2 are higher in the immature brain and that high levels of UCP2 are associated with increased resistance of neurons to seizures (3), excitotoxicity (4), and ischemia (5). UCP2 expression is induced by ischemic and oxidative damage (3,5) suggesting that it may protect neurons by functionally uncoupling mitochondria, thereby reducing ROS production. Although the brain-specific UCP4 was discovered more than 5 years ago (6), little is known about its function in neurons.
The intracellular messenger Ca 2ϩ regulates many crucial processes, including motility, secretion, gene expression, as well as many of the signaling cascades that drive proliferation, differentiation, and various metabolic reactions. Another major role for Ca 2ϩ is in cell death in physiological settings or during injury or diseases (7,8). The mechanisms that control cellular Ca 2ϩ dynamics are complex, with ion channels and pumps in the plasma membrane and endoplasmic reticulum (ER) playing major roles in regulating both rapid and long-term changes in the cytoplasmic free Ca 2ϩ concentration ([Ca 2ϩ ] i ). When Ca 2ϩ is released from the ER Ca 2ϩ store in response to signals that activate receptors coupled to inositol trisphosphate (IP 3 ) production, the emptying of the ER Ca 2ϩ stores triggers Ca 2ϩ influx through La 3ϩ -sensitive/nifedipine-insensitive Ca 2ϩconducting channels in the plasma membrane, a process that has been termed store-operated Ca 2ϩ entry (SOCE) (9). In addition to its function in refilling depleted IP 3 -sensitive stores, SOCE is important for fine-tuning the transient Ca 2ϩ spikes (10) and for inducing Ca 2ϩ dependent gene expression (11). Though most extensively studied in non-excitable cells, SOCE has also been described in excitable cells. Despite intensive investigations, neither the proteins that comprise these storeoperated Ca 2ϩ channels (SOCC), nor the molecular mechanisms underlying activation and gating of these Ca 2ϩ channels, have been established. Molecular cloning and functional expression suggest that SOCE is mediated by proteins homologous to the Drosophila transient receptor potential (TRP), a Ca 2ϩ -permeable non-selective cation channel. Mammalian homologues of TRPs are one of the largest groups of cation channels with diverse functions, biophysical properties, and modes of activation (12). SOCE activation might involve docking or fusion of SOCC containing secretory vesicles with plasma membrane, direct conformational coupling of IP 3 receptors with SOCCs, or the generation and release of a diffusible messenger from the ER upon depletion (13). Recent work has identified the ER membrane-spanning STIM1 (stromal interaction molecule) as the putative ER luminal Ca 2ϩ sensor that couples store depletion to SOCE activation (14). STIM1 functionally interacts with CRACM1 (also known as Orai1), a plasma membrane component required for SOCE activity suggesting that the latter may represent the SOCC itself or a regulatory subunit of the SOCC complex (15).
Recent studies have provided evidence that mitochondria can rapidly sequester Ca 2ϩ following its influx through SOCC and may modulate SOCE by acting as a local buffer for Ca 2ϩ (16 -19). Mitochondria are known to play key roles in the regulation of cellular Ca 2ϩ dynamics in physiological settings such as excitation-contraction coupling and neurotransmitter release, as well as in pathological states. Because mitochondria play major roles in energy metabolism and production of ROS, perturbations of mitochondrial ATP and ROS production can affect cellular Ca 2ϩ homeostasis, and vice versa (20,21). Changes in mitochondrial transmembrane potential can affect the ability of the mitochondria to take up Ca 2ϩ , which is important for the regulation of intramitochondrial oxidative metabolism (22). Although UCPs have been shown to modulate mitochondrial membrane potential, and ROS and ATP production, little is known about their contribution to neuronal Ca 2ϩ homeostasis. To elucidate whether UCPs may function in Ca 2ϩ homeostasis, we measured mitochondrial Ca 2ϩ sequestration following IP 3 -dependent and independent agonist-induced store depletion and the SOCE pathway in PC12 cells expressing or lacking human UCP4 (hUCP4). These data demonstrate that hUCP4 regulates mitochondrial Ca 2ϩ sequestration and Ca 2ϩ entry but not release from intracellular stores. Furthermore, expression of hUCP4 decreases the magnitude of sustained elevation in [Ca 2ϩ ] i after cellular Ca 2ϩ depletion, inhibits mitochondrial Ca 2ϩ overload and oxidative stress, and prevents cell death.

EXPERIMENTAL PROCEDURES
Cell Cultures and Transfection-Clones of PC12 cells stably overexpressing full-length hUCP4 were generated using methods similar to those described previously (23). The recombinant plasmid encoding full-length hUCP4 (pcDNA-hUCP4) was kindly provided by G. Pan (Genentech). PC12 cells were subcultured into 6-well plates coated with polyethyleneimine and grown to 80% confluency in culture medium consisting of Dulbecco's modified Eagle's medium supplemented with 10% horse and 5% fetal bovine serum. Transfection was performed using 1 g of purified plasmid DNA and 10 l of Lipofectamine reagent (Invitrogen). Stable transfectants were obtained after selection for growth in the presence of Geneticin (500 mg/liter). Several clones of cells expressing moderate levels of UCP4 protein were selected and maintained in the presence of Geneticin (200 mg/liter). For experiments, PC12 cells were plated onto glass coverslips and used between 18 and 48 h after plating. Transient transfection with the expression plasmid encoding the mitochondrially targeted fluorescent marker DsRed (mtDsRed) cloned into pcDNA3 (Invitrogen), was performed using Lipofectamine.
Experimental Treatments-Undifferentiated PC12 cells were treated with thapsigargin (TG), N-acetyl-D-sphingosine (C 2ceramide), C 2 -dihydroceramide, and aristolochic acid (Aris, Alexis Biochemicals). All reagents except when noted were from Sigma. When Me 2 SO or ethanol was used as the solvent, their final concentration did not exceed 0.1%. At the end of each treatment, cultures were processed for immunoblotting, and for evaluating the extent of cell death.
Targeting 21-mer siRNA duplexes with a two-base overhang corresponding to the Gadd153 coding regions nucleotides 270 -291 (siRNA Gadd153 , GenBank TM accession number NM_024134) were synthesized using the Silencer siRNA Construction Kit (Ambion, Austin, TX). Briefly, sense primers and corresponding antisense primers were mixed, heated to 95°C for 5 min, and then allowed to cool to 25°C at a rate of 1°C/min as previously described (23). Non-silencing control siRNA (siR-NA CTRL ) duplexes were synthesized using scrambled sequences. A GenBank TM search revealed no other known genes exhibiting sequence homology to these selected target sequences. To test the silencing effect of siRNAs on the induction of Gadd153, normal PC12 cells were transfected with the targeting or control siRNA duplexes (100 nM) for 14 h prior to exposure to TG. The effects of siRNA on vulnerability to TGinduced death were evaluated after transfection of siRNAs.
Immunoblot Analysis-Methods for protein quantification, electrophoretic separation, and transfer to nitrocellulose membranes were described previously (23). Membranes were incubated in blocking solution (5% milk in Tween Tris-buffered saline) overnight at 4°C followed by a 1-h incubation in primary antibody diluted in blocking solution at room temperature. Membranes were then incubated for 1 h in secondary antibody conjugated to horseradish peroxidase (Vector Laboratories), and bands were visualized using a chemiluminescence detection kit (ECL, Amersham Biosciences). The primary antibodies included rabbit polyclonal antibodies against hUCP4 (Alpha Diagnostic), SERCA2 (Dr. J. Lytton, University of Calgary), calreticulin (Sigma), TRPCs 1, 4, and 5 (Alomone Laboratories), and STIM1 (Dr. J. Roos, TorreyPines Therapeutics); mouse monoclonal antibodies against actin (Sigma), HSP60 (Signal Transduction), cytochrome c oxidase 1 (Molecular Probes), Bcl-2 (StressGen), Grp78 (StressGen), KDEL (StressGen), and Gadd153 (Santa Cruz Biotechnology). Membranes were stripped and re-probed with the actin antibody to verify and normalize protein loading (50 g of total protein, unless stated otherwise). To determine the presence of the mitochondria in subcellular fractions, membranes were re-probed with the cytochrome c oxidase 1 antibody.
Quantification of Cell Survival-Cell viability was assessed by the trypan blue exclusion method as described previously (23).
Confocal Laser Microscopy-Mitochondria organization was visualized in cells transiently transfected with mitochondrially targeted fluorescent marker DsRed as described previously (23). Cells were illuminated 24 h post-transfection at 575 nm, and emission was monitored at 528 -633 nm.
Measurement of Mitochondrial Membrane Potential and ROS Formation-Mitochondrial membrane potential was determined in cultured cells using cell permeant probe tetramethylrhodamine (TMRE, Molecular Probes) or MitoTracker Red CMXRos (Molecular Probes) as described previously (23). Accumulation of TMRE in mitochondria is dependent on the polarization state of the mitochondrial membrane. Each culture dish was incubated with 50 nM TMRE for 20 min at 37°C and then rinsed with Locke's buffer (in mM: 154 NaCl, 5.6 KCl, 2.3 CaCl 2 , 1.0 MgCl 2 , 3.6 NaHCO 3 , 5 HEPES, and 10 D-glucose, pH 7.2). Cells were imaged using a confocal laser scanning microscope coupled to an inverted microscope; cells were located under bright-field optics and then scanned once with the laser (488 nm excitation and 529 nm emission). Data were expressed as the ratio of the change in fluorescence (dF/F o ) after each treatment.
To assess formation of cellular ROS, cells were incubated for 30 min in the presence of the fluorescent probe dihydrorhodamine 123 (10 M DHR, Molecular Probes) or 2,7-dichlorodihydrofluorescein diacetate (10 M DCF, Molecular Probes). DHR localizes mainly in the mitochondria and is oxidized by several ROS, most notably hydrogen peroxide and peroxynitrite, to the positively charged rhodamine 123 derivative. The mitochondrial pixel intensity for 100 individual cells in five different fields per culture in triplicate was averaged, and background fluorescence was subtracted.
Measurement of Cytoplasmic and Intra-mitochondrial Free Ca 2ϩ Concentrations-Fluorescence ratio imaging of the Ca 2ϩ indicator dye fura-2 acetylmethyl ester (fura-2/AM, Molecular Probes) was used to quantify free intracellular Ca 2ϩ concentration ([Ca 2ϩ ] i ) using methods detailed previously (23). Cells were incubated for 15 min in the presence of 2 M fura-2/AM at 37°C, washed twice with Locke's buffer, and allowed to incubate for 10 min before imaging. Cells were imaged using the Zeiss Attofluor system with a 40ϫ oil immersion objective lens. The ratio of the fluorescence emission using two different excitation wavelengths (340 and 380 mm) was used to determine [Ca 2ϩ ] i . In some experiments Ca 2ϩ was omitted from the Locke's buffer without (Ca 2ϩ -deficient buffer) or with 100 M EGTA (Ca 2ϩ -free buffer). Where indicated, the Locke's buffer was supplemented with an appropriate amount of CaCl 2 . SOCE was induced by stimulation with TG or bradykinin (BK). SKF-96365 (Tocris) was used to block Ca 2ϩ entry, and the ionophore 4-bromo-A23187 (Calbiochem) was added to induce rapid extrusion of Ca 2ϩ from intracellular stores out of the cell. Ru360 (Calbiochem), a membrane-permeable analogue of ruthenium red, was used to prevent mitochondrial Ca 2ϩ uptake.
Cells treated with 2 M carbonyl cyanide m-chlorophenylhydrazone (CCCP, Sigma) were preincubated with 1 M oligomycin to inhibit mitochondrial ATPase and thereby to protect the cells from ATP depletion under conditions when membrane potential was collapsed by CCCP. Mitochondrial free Ca 2ϩ levels were quantified in cells loaded with the cell-permeant acetoxymethyl ester (AM) form of the fluorescence probe dihydrorhodamine-2 (Rhod-2, Molecular Probes). Cells were incubated with 2 M Rhod-2/AM for 45 min at 37°C and were then washed with Locke's buffer. Fluorescence images of randomly chosen fields with 30 -40 cells in each field were acquired (excitation, 543 nm; emission, 576 nm) using a confocal microscope. The laser beam intensity and photodetector sensitivity were held constant across cultures to allow quantitative comparisons of relative fluorescence intensity of cells between treatment groups. The illumination intensity was kept to a minimum to prevent artifactual mitochondrial Ca 2ϩ loading. Control experiments were performed using CCCP with oligomycin. Fluorescence (F) was quantified over the whole surface of each cell body (nucleus excluded) in the selected fields using software provided by the manufacturer (Zeiss), and the mean intensity per cell was then calculated. Values were expressed as the ratio of the change in fluorescence (dF/F o ) after each time point after treatment.
Loading and Cleavage of Bis-BODIPY-FL-C11-PC-Preparation of bis-BODIPY-FL-C11-PC (a synthetic phospholipid containing fluorescein in position 2 that is dequenched upon cleavage) in liposomes was performed according to the manufacturer's specification (Molecular Probes). PC12 cells were loaded with bis-BODIPY-FL-C11-PC for 30 min, rinsed twice and incubated in Dulbecco's modified Eagle's medium for another 30 min prior to exposure to TG. Fluorescence images were acquired with the confocal imaging system. The data values are corrected for background fluorescence.
Statistical Comparisons-Comparisons between cell types and treatment groups were made using analysis of variance with Scheffe post-hoc test for pairwise comparisons.

hUCP4 Localizes to Mitochondria and Alters Mitochondrial
Membrane Potential-Expression of hUCP4 in stably transfected PC12 cells was verified by RT-PCR and by immunoblot analysis using a UCP4 antibody. hUCP4 mRNA was detected in cells stably transfected with vector containing hUCP4 but not with the empty vector (Fig. 1A, left panel). A 32-kDa immunoreactive band, the predicted size of UCP4, was present in whole cell lysates and in the mitochondrial fraction of hUCP4 but not vectortransfected (VT) cells (Fig. 1A, right panel). A weak immunoreactive band with a slightly higher molecular weight was present in VT cells (indicated by the asterisk). Because we have been unable to detect endogenous rat UCP4 mRNA in PC12 cells (data not shown), the immunoreactive band is likely UCP2, which is known to be expressed in PC12 cells (24). Expression of hUCP4 in PC12 cells did not cause changes in the level of UCP2 mRNA (Fig. 1A, left panel).
To assess the effects of hUCP4 expression on mitochondrial membrane potential, we loaded cells with low concentrations of the fluorescence probe TMRE. Fluorescence images were acquired using a confocal microscope and quantified (25). hUCP4 cells exhibited reduced TMRE fluorescence compared with control cells demonstrating that expression of hUCP4 reduces steady-state mitochondrial membrane potential (Fig. 1B), which was sensitive to complete dissipation of the hydrogen ion gradient induce by the cell-permeant protonophore CCCP. Addition of oligomycin did not depolarize nor restore the difference in mitochondrial membrane potential suggesting that the mitochondrial membrane potential in hUCP4 cells was not maintained by the mitochondrial ATP synthase functioning in reverse mode and consuming ATP produced by glycolysis. This decrease in mitochondrial membrane potential led to a slight but significant increase in oxygen consumption suggesting the functional uncoupling activity is due to the expressed UCP4 in PC12 (26). Furthermore, we found that intracellular ATP content was maintained suggesting no disruption in cellular energetics in hUCP4-expressing cells (26). The staining pattern of the mitochondrial targeted fluorescence marker DsREd was similar in vector-transfected and hUCP4-expressing cells (Fig.  1C), indicating that hUCP4 expression did not grossly affect mitochondrial distribution (Fig. 1C). Levels of cytochrome oxidase (cytochrome c oxidase 1), a mitochondrial protein, remain unchanged ( Fig. 1D), indicating that hUCP4 did not induce mitochondrial proliferation as has been reported in brain cells of UCP2 transgenic mice (24).
hUCP4 Expression Reduces Mitochondrial Ca 2ϩ Sequestration-To evaluate how hUCP4 affects mitochondrial Ca 2ϩ homeostasis, we first measured the cytoplasmic free Ca 2ϩ concentrations in VT and hUCP4-transfected cells following stimulation with the IP 3 agonist bradykinin in the absence of extracellular Ca 2ϩ . The resting [Ca 2ϩ ] i level was similar in VT and hUCP4-expressing cells (Fig. 2, A and B). Compared with VT cells, the transient bradykinin-induced [Ca 2ϩ ] i increase in Ca 2ϩ -free buffer was higher in hUCP4 cells but did not reach statistically significance (237 Ϯ 21 and 261 Ϯ 24 nM, respectively, Fig. 2, A and B). Addition of CCCP to collapse the mitochondrial membrane potential caused a transient rise in [Ca 2ϩ ] i that was significantly lower in hUCP4-transfected cells compared with VT cells (208 Ϯ 21 and 272 Ϯ 25 nM, respectively, Fig. 2, A and B). Similar results were obtained when cells were treated with TG (an inhibitor of the sarco(endo)plasmic reticulum Ca 2ϩ ATPases) instead of BK (data not shown). Next, we determined whether hUCP4 expression affects ER Ca 2ϩ store content or release. In the presence of CCCP, the BK-induced [Ca 2ϩ ] i transient was comparable in hUCP4 and VT cells (241 Ϯ 20 and 231 Ϯ 21 nM, respectively, Fig. 2, C and D), suggesting that hUCP4 does not affect the kinetics of agonistinduced Ca 2ϩ release. The CCCP-induced [Ca 2ϩ ] i transient was similar in VT and hUCP4 cells (121 Ϯ 9 and 132 Ϯ 11 nM, respectively) indicating that hUCP4 expression did not signifi-cantly alter the steady-state [Ca 2ϩ ] m . We also exposed cells preincubated with or without CCCP to the Ca 2ϩ ionophore 4-bromo-A23187 (1 M) in Ca 2ϩ -deficient buffer to assess whether hUCP4 expression may alter the ER Ca 2ϩ store. The ionophore caused rapid extrusion of Ca 2ϩ from intracellular stores out of the cell. In the absence of CCCP, the ionophore-releasable Ca 2ϩ pool was significantly lower in hUCP4-transfected cells compared with VT cells (Fig. 2E, upper panel; 602 Ϯ 34 nM and 824 Ϯ 23 nM, respectively). On the other hand, the elevation of [Ca 2ϩ ] i induced by the ionophore when mitochondrial Ca 2ϩ uptake was inhibited was not different in hUCP4-transfected cells compared with VT cells (Fig. 2E, lower panel; 592 Ϯ 34 nM and 625 Ϯ 23 nM, respectively). These data suggest that hUCP4 expression has little or no effect on the ER Ca 2ϩ store content and release and that mitochondria in cells expressing hUCP4 have a reduced capacity to store Ca 2ϩ that has been released from the ER.
Evidence That hUCP4 Attenuates the Store-operated Ca 2ϩ Entry Pathway-Depletion of Ca 2ϩ stores by either IP 3 -dependent or IP 3 -independent pathways results in activation of SOCE (9). Recent studies indicate that mitochondrial Ca 2ϩ buffering is essential for SOCE and that inhibition of mitochondrial Ca 2ϩ uptake, either by mitochondrial depolarization or by inhibition of the mitochondrial Ca 2ϩ uniporter, impairs SOCE (16 -19). Because the data above suggested that UCP4 may affect mitochondrial Ca 2ϩ regulation, we investigated the mechanism by which UCP4 affects the SOCE pathway. To fully activate SOCE, we emptied intracellular stores with 1 M TG in the absence of extracellular Ca 2ϩ for 10 min. The slow and transient increase in [Ca 2ϩ ] i evoked by TG due to the leak of Ca 2ϩ from the stores was not significantly different in VT and hUCP4-transfected cells (Fig. 3A). Subsequent exposure of cultures to a range of external Ca 2ϩ concentrations ([Ca 2ϩ ] e ) resulted in a substantial concentration-dependent increase in [Ca 2ϩ ] i , which did not return to the basal level but, instead, decreased slowly to a steady-state level of elevated [Ca 2ϩ ] i . This plateau [Ca 2ϩ ] i reflects the balance between Ca 2ϩ influx into the cytosol through SOCE, which is gradually reduced by a feedback inhibition by Ca 2ϩ , and Ca 2ϩ efflux from the cells driven mainly by plasma membrane Ca 2ϩ ATPase. Expression of hUCP4 decreased both the peak amplitude and steady-state level of [Ca 2ϩ ] i during the store refilling period when the [Ca 2ϩ ] e was increased to 1.0 mM or greater (Fig. 3B). Following addition of 2 mM Ca 2ϩ , the average rise in [Ca 2ϩ ] i level was 762 Ϯ 56 and 553 Ϯ 43 nM in VT and hUCP4-transfected cells, respectively (Fig. 3B). A [Ca 2ϩ ] e of 2 mM, which represents the physiological level of extracellular calcium, was chosen for all subsequent experiments to better appreciate differences in Ca 2ϩ signals.
It is unlikely that hUCP4 interfered with the intensity of the coupling signal transmitted from ER to plasma membrane SOCC, because the initial rate of [Ca 2ϩ ] i change was similar in both VT and hUCP4-expressing cells (Fig. 3A). To determine whether hUCP4 may interfere with TG-induced depletion of intracellular Ca 2ϩ stores, we assessed the ability of BK to release Ca 2ϩ following exposure of cells to TG. Addition of 100 M BK failed to elicit further increase in [Ca 2ϩ ] i in either VT or hUCP4-transfected cells (data not shown) suggesting that stores were indeed fully depleted and that the reduction of SOCE in hUCP4 cells could not be attributed to incomplete store depletion. Preincubation of cells with SKF-96365, a widely employed inhibitor of SOCE, strongly inhibited TG-induced Ca 2ϩ entry (Fig. 3C), suggesting that the rise in [Ca 2ϩ ] i is mainly attributed to SOCE. To assess whether hUCP4 may enhance Ca 2ϩ extrusion mechanisms, we added EGTA during store refilling to abruptly inhibit Ca 2ϩ influx. The rapid decay of the steady state [Ca 2ϩ ] i and hence the initial rate of Ca 2ϩ clearance, which mostly reflects Ca 2ϩ extrusion rate was comparable in VT and hUCP4transfected cells (data not shown). Taken together, these data support the argument that expression of hUCP4 diminishes mitochondrial Ca 2ϩ sequestration and SOCE rather than enhancing Ca 2ϩ efflux across plasma membrane.
hUCP4 Does Not Alter Expression of Proteins in the SOCE Pathway-We next determined whether overexpression of hUCP4 affects levels of proteins involved in the SOCE pathway. Steady-state levels of SERCA2 and the ER luminal Ca 2ϩ binding/storage protein calreticulin, two ER proteins that regulate ER Ca 2ϩ store refilling and, indirectly, SOCE (27) were not significantly different in hUCP4-transfected and VT cells (Fig. 3D). Levels of Bcl-2, a protein that has been shown to play a role in ER Ca 2ϩ homeostasis (28,29), were also similar in VT and hUCP4 cells. Because the SOCE response was reduced in hUCP4 cells, we determined whether expression of TRP-related proteins was down-regulated in hUCP4 cells. Previous studies showed that induction of differentiation and chronic hypoxia increased expression of several TRPC proteins (a subfamily of TRPs), which was accompanied by an increase in SOCE activity (30). PC12 cells express most of the six TRPC (TRP 1-6) members (31). Despite their similarity in amino acid sequence, members of the TRPC subfamily have different modes of activation (12). Some TRPC proteins form Ca 2ϩ -permeable cation channels that do not appear to be storeoperated; they do not respond to treatment with TG, the most selective pharmacological tool for activating SOCE. Because store-depletion-insensitive TRPCs may heteromultimerize with store-depletion-sensitive TRPCs to form functional SOCC (12), we analyzed the expression of TRPs 1, 4, 5, and 6 in VT and hUCP4-transfected cells. TRPC4 mRNA was not detected in VT or hUCP4 cells (Fig. 3E). Steady-state levels of TRPCs 1, 5, and 6 mRNA measured by RT-PCR were not affected by hUCP4 expression (Fig. 3E); similarly, levels of TRPC proteins 1 and 5 were not different in VT and hUCP4 cells. Moreover, the protein level of STIM1, the putative ER calcium sensor (14,15), was not different in VT and hUCP4 cells. Thus, it is unlikely that altered expression of TRPCs or STIM1 contributed to the observed effect of hUCP4 in reducing SOCE.
hUCP4 Regulates SOCE at the Level of Mitochondrial Ca 2ϩ Sequestration-To determine whether mitochondrial Ca 2ϩ uptake is essential for maintenance of SOCE, we induced com- plete store depletion with TG in Ca 2ϩ -free medium and assessed the amount of releasable Ca 2ϩ from mitochondria ([Ca 2ϩ ] m ) in TG-treated cells during re-application of 2 mM Ca 2ϩ . Consistent with the results above (Fig. 3, A and B), the magnitude of the [Ca 2ϩ ] i rise and the average plateau Ca 2ϩ were significantly lower in hUCP4-transfected cells during Ca 2ϩ re-application (Fig. 4A). To terminate loading, we applied Ca 2ϩ -free buffer for another 150 s. Subsequent addition of 4-bromo-A23187 released the Ca 2ϩ sequestered in the TG-insensitive mitochondrial store, and the amount of released Ca 2ϩ was significantly lower in hUCP4-transfected cells than in VT cells (918 Ϯ 36 versus 618 Ϯ 24 nM, Fig. 4A) indicating that mitochondria of hUCP4-transfected cells take up less Ca 2ϩ during store refilling.
Next, we determined whether acute exposure to CCCP in the presence of oligomycin could mimic the inhibitory action of hUCP4 on the SOCE pathway. When CCCP was applied to TG-treated cells 1 min prior to and during the store-refilling period, TG-stimulated SOCE was largely blocked in both VT and hUCP4-transfected cells (388 Ϯ 34 nM and 340 Ϯ 28 nM, respectively, Fig. 4B). Subsequent application of 4-bromo-A23187 in Ca 2ϩ -deficient buffer released little Ca 2ϩ from the TG-insensitive stores in VT and hUCP4-transfected cells. As expected, CCCP applied to TG-treated cells during the store-refilling period caused release of Ca 2ϩ sequestered in the TG-insensitive stores, and the magnitude of this release was significantly lower in hUCP4-transfected cells compared with VT cells (719 Ϯ 46 and 1012 Ϯ 56 nM, respectively, Fig. 4C). To verify that the TG-insensitive Ca 2ϩ stores were mitochondria, we loaded cells with Rhod-2/AM, a Ca 2ϩ -sensitive fluorescent indicator commonly used to monitor changes in [Ca 2ϩ ] m (32). The magnitude of the increase in Rhod-2 fluorescence signal was significantly greater in VT cells than in hUCP4 cells following the addition of TG in the absence or presence of extracellular Ca 2ϩ . CCCP, which prevented uptake of Rhod-2 into the organelle, inhibited the increase in the fluorescence signal in both cell types, further confirming that the TG-insensitive store is mitochondria (Fig.  4D). To further demonstrate that Ca 2ϩ uptake by mitochondria is required to sustain SOCE, we used Ru360, a membrane-permeable analogue of ruthenium red that prevents mitochondrial Ca 2ϩ uptake by directly inhibiting the Ca 2ϩ uniporter without interfering with the mitochondrial membrane potential (33). Treatment with 10 M Ru360 significantly reduced SOCE in VT cells to an extent comparable to that in hUCP cells (258 Ϯ 19 and 247 Ϯ 13 nM, Fig. 4E). As expected, the amount of Ca 2ϩ released from the TG-insensitive stores was reduced in Ru360-treated VT and hUCP4 cells upon application of 4-bromo-A23187. Collectively, our data suggest that expression of hUCP4 attenuates the ability of the mitochondria to sequester cytosolic Ca 2ϩ and to sustain SOCE (16 -19).
hUCP4 Attenuates Store Depletion-induced Cell Death-To determine whether cells expressing hUCP4 are more resistant to death induced by TG, we measured viabilities of VT and hUCP4 cells exposed to TG for various time periods. Compared with VT cells, hUCP4 cells were significantly more resistant to death induced by TG. When exposed to TG for 24 h in medium containing 2 mM Ca 2ϩ , 28% of the hUCP4-expressing cells were dead compared with 50% of VT cells (Fig. 5A). Decreasing the [Ca 2ϩ ] e to 0.5 mM protected VT cells against TG-induced death, effectively eliminating the protective effect of hUCP4 (Fig. 5A). These findings suggested that hUCP4 protects cells by reducing SOCE. Consistent with this conclusion we found that the L-type Ca 2ϩ channel blocker nimodipine (1 M) had no significant effect on TG-induced death (data not shown). Because exposure of PC12 cells to SKF-96365 or Ca 2ϩ -free medium resulted in cell death within 6 -12 h (data not shown), we were unable to determine their effects on TG-induced cell death.
We found that basal and TG-induced levels of the ER chaperone GRP78, an indicator and key component of the ER stress response (34), were similar in VT and hUCP4-expressing cells (Fig. 5, B and D), indicating that hUCP4 did not protect against TG-induced death by facilitating induction of this ER chaperone. In contrast, protein levels of the growth-arrest DNA damage-inducible protein 153 (Gadd153), a ubiquitously expressed member of the CAAT/enhancer-binding protein family of transcription factors, increased more rapidly in VT compared with hUCP4 cells (Fig. 5, B and D). Because Gadd153 is highly induced in response to an elevation of [Ca 2ϩ ] i (35), its greater increase following TG treatment in VT cells compared with hUCP4 cells is likely due to the greater sustained elevation in Ca 2ϩ as the result of SOCE. Reducing [Ca 2ϩ ] e to 0.5 mM resulted in not only an attenuation but also a normalization of the TG-induced increase in Gadd153 protein levels in VT and hUCP4-expressing cells (Fig. 5, C and D). Thus, expression of hUCP4 attenuates cell death by suppressing SOCE and the upregulation of Gadd153.
hUCP4 Protects against Cell Death Induced by Ceramide-Ceramide is generated from membrane sphingomyelin in response to activation of various cytokine receptors and by exposure of cells to oxidative stress; ceramide can induce cell death by activating the SOCE pathway (36). We therefore determined how expression of hUCP4 expression might affect the vulnerability of PC12 cells to ceramide-induced death. VT and hUCP4 PC12 cells were exposed to increasing concentrations of the cell-permeable ceramide analogue C 2 -ceramide, and cell viability was determined 12 and 24 h later. Cells expressing hUCP4 were significantly more resistant to C 2 -ceramide-induced death compared with VT cells (Fig. 6A). The biologically inactive stereoisomer C 2 -dihydroceramide (40 M) had no significant effect on the survival of VT or hUCP4 cells during a 24-h exposure period. To determine whether the protective action of hUCP4 was mediated by its inhibitory role on the SOCE pathway, we measured Ca 2ϩ entry in VT and hUCP4 cells after exposure to C 2 -ceramide (Fig. 6B). C 2 -ceramide caused a slow rise in [Ca 2ϩ ] i that reached a long-lasting plateau in cells maintained in Ca 2ϩ -containing medium; the magnitude of this Ca 2ϩ plateau was significantly lower in hUCP4 cells compared with VT cells. In the absence of extracellular Ca 2ϩ , the rise in [Ca 2ϩ ] i was slower and only transient in both VT and hUCP4 cells (Fig. 6B), suggesting that Ca 2ϩ entry was required to maintain the Ca 2ϩ plateau in cells treated with C 2 -ceramide. To determine whether SOCE activity contributed to the increase in [Ca 2ϩ ] m in cells treated with C 2 -ceramide, we added the Ca 2ϩ ionophore 4-bromo-A23187 to cells 10 min after treatment with C 2 -ceramide. The rapid [Ca 2ϩ ] i rise upon addition of the Ca 2ϩ ionophore was markedly reduced in hUCP4-expressing cells (645 Ϯ 34 nM) compared with VT cells (1068 Ϯ 52 nM, p Ͻ 0.01) (Fig. 6C). The relative increase in the resting [Ca 2ϩ ] i was also much lower in C 2 -ceramide-treated hUCP4-expressing cells compared with VT cells. Reducing [Ca 2ϩ ] e to 0.5 mM partially protected VT cells from death induced by C 2 -ceramide (data not shown). Immunoblot analysis showed that expression of hUCP4 resulted in an attenuation of the ceramide-induced increase in Gadd153 protein levels compared with VT cells (Fig. 6D).
hUCP4 Expression Reduces Oxidative Stress and Preserves Mitochondrial Function-Mitochondria are a major source of intracellular ROS production, and sustained Ca 2ϩ entry and mitochondrial Ca 2ϩ uptake can increase mitochondrial oxidative metabolism and generation of ROS. Mitochondrial ROS are key inducers of mitochondrial transition pore opening, a process pivotal for cell death in many physiological and pathological settings. To determine whether hUCP4 modulates mitochondrial oxidative stress, we employed DHR, a cell-permeant mitochondrial dye, to assess formation of mitochondrial ROS in live cells. Exposure of VT cells to TG caused a rapid increase in DHR fluorescence within 2 h, followed by a sustained peak between 6 and 12 h (Fig. 7A). The increase in ROS production was significantly smaller in hUCP4 cells demonstrating that expression of hUCP4 suppresses mitochondrial ROS formation. The increase in DCF fluorescence, which measured cellular peroxide levels, was also reduced in hUCP4 cells at each time point (Fig. 7A). We also monitored mitochondrial membrane potential during an 18-h exposure to TG (Fig. 7B). The basal membrane potential was greater in VT cells compared with hUCP4 cells. After exposure to TG, the membrane potential in VT cells rose to a peak level at 6 h of exposure and then decreased to a level below the baseline at 18 h. In contrast, the membrane potential in hUCP4 cells rose modestly through 6 h and was maintained at this level through 18 h (Fig. 7B). A similar pattern of increases in DHR and TMRE intensities above the level of vehicle-treated cells was obtained in VT and hUCP4 cells exposed to C 2 -ceramide (data not shown).
Although mitochondria are prime targets of Ca 2ϩ -mediated apoptosis, SOCE could also contribute to Ca 2ϩ -dependent events in the cytosol that may indirectly influence mitochondrial function. Recent studies indicate that the extent of Ca 2ϩ entry through the SOCE pathway determines the activity of Ca 2ϩ -dependent cytosolic phospholipase A2 (cPLA2) (37). Upon activation and translocation to intracellular membranes, cPLA2 generates and releases arachidonic acid from membrane phospholipids (38). To determine whether expression of hUCP4 reduces the generation of arachidonic acid, we loaded cultured cells with the synthetic phospholipid bis-BODIPY-FL-C11-PC, which undergoes dequenching upon hydrolysis by PLA2 to release the labeled phospholipid within the cell. Exposure of bis-BODIPY-FL-C11-PC-loaded cells to TG triggered rapid phospholipid hydrolysis as indicated by an increased fluorescence signal (Fig.  7C). The magnitude of the TG-induced lipid hydrolysis was significantly lower, by ϳ30%, in cells expressing hUCP4 compared with VT cells (Fig. 7D). Aristolochic acid, a specific inhibitor of cPLA2, dramatically reduced the fluorescent signal associated with phospholipid release (Fig. 7 D), consistent with the activation of cPLA2 by TG. Incubation of cells in medium lacking Ca 2ϩ also largely inhibited TGinduced phospholipid hydrolysis (data not shown). Because arachidonic acid itself, and metabolites generated in the cyclooxygenase or lipoxygenase pathways, can contribute to oxidative stress (39), we determined whether inhibition of cPLA2 activation blocked ROS production. Pretreatment of cells with aristolochic acid partially inhibited TG-induced ROS production (data not shown), suggesting that cPLA2 and its associated metabolic pathways contribute to oxidative stress.
Because oxidative stress can contribute to Gadd153 up-regulation (40 -42) and because the magnitude of induction of Gadd153 protein levels was significantly greater in VT compared with hUCP4 cells (Fig. 5, B and D), we determined whether suppression of Gadd153 function by RNA interference normalized the differential vulnerability of the clones to TGinduced death. The TG-induced increase in Gadd153 protein was attenuated in VT and hUCP4 cells over a course of 14 h in the presence of 100 nM Gadd153 but not scrambled siRNA (Fig.  8A). Under these conditions, ER stress-induced expression of two other ER stress proteins, Grp78 and -94, were unaffected by the siRNA treatment. Suppression of endogenous Gadd153 expression partially rescued VT cells from TG-induced death (Fig. 8B). In addition, knockdown of Gadd153 function was also more effective compared with aristolochic acid in decreasing the vulnerability of VT cells to TG-induced death (Fig. 8B) suggesting that Gadd153 is a critical mediator of store depletioninduced oxidative stress and death.

DISCUSSION
Mitochondria play an important role in intracellular Ca 2ϩ dynamics and signaling by sensing microdomains of both Ca 2ϩ release from ER stores and Ca 2ϩ influx across the plasma membrane (16 -19). It has been shown that within these subcellular microdomains Ca 2ϩ is efficiently taken up by mitochondria. The ability of mitochondria to buffer Ca 2ϩ locally is essential for gating Ca 2ϩ -sensitive plasma membrane channels. Our data suggest that hUCP4 regulates mitochondrial Ca 2ϩ homeostasis and sensitivity of neural cells to store depletion-induced death by virtue of its ability to alter the membrane potential-dependent Ca 2ϩ buffering capacity of the mitochondria. The effects of hUCP4 on SOCE and cell death were associated with decreased levels of ROS. PC12 cells expressed little or no endogenous UCP4, the level of UCP2 expression was low and unaffected by expression of hUCP4, and hUCP4 had no detectable effects on mitochondrial shape or subcellular distribution. The latter findings suggest that the effects of hUCP4 on store depletioninduced Ca 2ϩ entry and cell survival were not secondary to changes in mitochondrial mass.
The key event in the SOCE pathway is intracellular Ca 2ϩ store depletion that triggers Ca 2ϩ entry upon activation of SOCC in the plasma membrane. The magnitude of the SOCE response is related to the degree to which stores are depleted (9). When exposed to a concentration of TG (1 M) that induced complete store emptying, PC12 cells expressing hUCP4 exhibited reduced SOCE compared with control VT cells. It is therefore unlikely that the presence of hUCP4 in mitochondria accelerates refilling of the ER Ca 2ϩ stores, thereby reducing SOCE. Additional data showed that the decreased SOCE in hUCP4-expressing cells was not attributed to reduced ER Ca 2ϩ store content or increased Ca 2ϩ extrusion. The steady-state mitochondrial Ca 2ϩ levels were not significantly altered in hUCP4 cells at resting state (Fig. 2, C and D) suggesting that a moderate dissipation of mitochondrial membrane potential has no significant impact on the trans-mitochondrial Ca 2ϩ flux responsible for ER refilling (43). Several studies have demonstrated that functional mitochondria are essential for maintenance of SOCE in different cell types (16 -19), and dissipation of the mitochondrial membrane potential, which impairs that ability of mitochondria to buffer cytosolic Ca 2ϩ , abolishes Ca 2ϩ influx through the SOCE pathway after the depletion of intracellular Ca 2ϩ stores. The exact mechanism of SOCE modulation by mitochondria is unknown. In addition to buffering Ca 2ϩ , mitochondria may release one or more diffusible factors that regulate SOCE activity (13)(14)(15)19). A recent study demonstrated that mitochondrial localization is also an important determinant of SOCE, because relocalization of mitochondria from the subplasma membrane region to perinuclear areas in HeLa cells overexpressing dynamitin caused a significant reduction in SOCE without impacting the ER Ca 2ϩ store content (44). Expression of hUCP4 had no detectable effect on mitochondrial numbers or subcellular localization, suggesting that the reduced mitochondrial Ca 2ϩ buffering in hUCP4 cells could not be ascribed to a reduction or altered distribution of mitochondria. Furthermore, protein levels of STIM1 (14,15) and TRPCs, some of which are involved in SOCE (12), were comparable in VT and hUCP4-expressing cells. We therefore propose that the activity of hUCP4 decreases the ability of mitochondria to sequester Ca 2ϩ entering the cell, resulting in a fast feedback inhibition of SOCC (16 -19). In support for this notion, we found that mitochondrial Ca 2ϩ uptake is reduced in hUCP4 cells during agonist-induced ER Ca 2ϩ release and Ca 2ϩ entry by the SOCE pathway. Similar reductions were seen in hUCP4 cells after store depletion induced by TG and C 2 -ceramide. Ca 2ϩ uptake is robust in mitochondria situated close to the membranes, which provides a mechanism for rapid sequestration of Ca 2ϩ by mitochondria near plasma membrane and ER Ca 2ϩ channels. However, massive Ca 2ϩ uptake by mitochondria, which can occur during excitotoxic injury and ischemia, can trigger cell death. Recent studies showed that high levels of UCP2 are associated with increased resistance of neurons to excitotoxicity (4), suggesting that a decrease in neuronal mitochondrial membrane potential can suppress Ca 2ϩ overload during excitotoxic insults. Consistent with this, Mattiasson et al. (5) demonstrated that transgenic mice overexpressing human UCP2 are more resistant to ischemic brain damage compared with nontransgenic control animals, and overexpression of UCP2 in cultured neurons protects them against oxygen/glucose deprivation-induced death. The latter study also showed that mild mitochondrial uncoupling by treatment with 2,4-dinitrophenol prevents the release of pro-apoptotic factors during oxygen-glucose deprivation and reduces neuronal death.
We found that expression of hUCP4 protects against Ca 2ϩ store depletion-induced death following exposure to TG and C 2 -ceramide. Studies in other cell systems showed that TG toxicity results primarily from mitochondrial Ca 2ϩ overload and secondarily from ER stress associated with Ca 2ϩ store depletion (45). Expression of hUCP4 blocked mitochondrial ROS formation by reducing the initial hyperpolarization and the subsequent depolarization of the inner mitochondrial membrane potential. Hyperpolarization has been attributed to rapid mitochondrial Ca 2ϩ uptake during exposure of neurons to glutamate and oxygen-glucose deprivation (46). With mitochondrial membrane hyperpolarization and extrusions of H ϩ ions from the mitochondrial matrix, the cytochromes within the electron transport chain becomes more reduced, which favors generation of ROS (47). The early hyperpolarization coincides with peak generation of ROS (Fig. 7, A and B). ROS are key inducers of mitochondrial permeability transition pore opening, and this process can trigger release of apoptogenic factors and activation of caspases (48). In addition to the mitochondrially derived ROS, expression of hUCP4 also blocked cPLA2 activation and subsequent oxidative metabolism of phospholipid-derived arachidonic acid via the cyclooxygenase or lipoxygenase pathways, which leads to O 2 . or OH ⅐ production (39). Recent investigations have demonstrated that members of the UCP family can prevent mitochondrial ROS formation, oxidative stress, and the terminal loss of the mitochondrial membrane potential.
Using gene knock-out models and in vitro systems, several groups demonstrated that UCP2 decreases cellular oxidative damage through a reduction in mitochondrial ROS production (49,50). The generation of ROS has been clearly established as a contributor to several neurodegenerative diseases and may underlie cellular changes seen in aging. Interestingly, targeted expression of UCP2 to the mitochondria of adult neurons can extend the life span in the fly by decreasing mitochondrial ROS production and attenuation of oxidative damage (51). We found that cells expressing hUCP4 exhibited reduced mitochondrial Ca 2ϩ uptake and ROS formation compared with control cells after store depletion with TG. Cells expressing hUCP4 also exhibited resistance to the toxicity of ceramide, a trigger of store depletion and activation of the SOCE pathway (32), which has been implicated in the degeneration of neurons in AD and related neurodegenerative disorders (52). Excessive production of ceramide may kill cells by triggering the mitochondrial-dependent pathway of apoptosis. The ability of hUCP4 to protect cells against ceramide-induced death therefore suggests an important role for this UCP in pathological conditions that involve ceramide-mediated cell death.
Grp78 is an ER chaperone that plays a critical role in maintaining cell viability under conditions of ER stress (34,53). However, the decreased susceptibility of cells expressing hUCP4 to store depletion-induced death was not due to an effect on expression of Grp78, because the magnitude of induction of Grp78 was similar in VT and hUCP4 cells. The size of the releasable ER Ca 2ϩ pool was also similar in VT and hUCP4 cells, ruling out the possibility that this parameter played a pivotal role in determining vulnerability to store depletion-induced death (36). On the other hand, we found that the rapid and sustained increase in Gadd153 protein level in response to ER store depletion was attenuated in hUCP4 cells. Gadd153 is a transcription factor that forms heterodimers with members of the CAAT/enhancer-binding protein family and is thought to be responsible for the induction of genes required for cell cycle arrest and apoptosis (54). The level of Gadd153 is very low under basal conditions, but its levels increase greatly in response to DNA damage and ER stress (54 -56). Previous studies have shown that TG and ionophores induce transcriptional activation of Gadd153 by elevating [Ca 2ϩ ] i (35), and induction of Gadd153 may be required for release of cytochrome c from mitochondria and caspase activation (57). Gadd153 expression is also induced by oxidative stress (40 -42), and the age-related increase in Gadd153 expression is correlated with heightened sensitivity to oxidant injury (58). Increased Gadd153 expression also mediates dopaminergic death following intrastriatal injection of 6-hydroxydopamine (59), a hydroxylated dopamine analogue that is readily oxidized causing oxidant injury. It remains to be determined whether the neuroprotective action of UCP2 in a mouse model of Parkinson disease (60) involves the suppression of Gadd153 induced by 1-methyl-4-phenylpyridinium (61). Because store depletion produces oxidative stress, which in turn induces Gadd153 expression, and because store depletion-induced death was significantly reduced by knockdown of Gadd153 expression, our data suggest that hUCP4 may contribute to the cytoprotective effect by reducing the up-regulation of oxidative stress-inducible and pro-apoptotic genes.
Collectively, our data provide evidence that hUCP4 can regulate Ca 2ϩ homeostasis and signaling, apparently by reducing mitochondrial membrane potential and Ca 2ϩ sequestration, the latter process playing an important role in SOCE. hUCP4 expression did not affect Bcl-2 expression but did reduce store depletion-induced oxidative stress and Gadd153 expression. Because ER stress, whether it results from protein misfolding and aggregation or from ischemic injury, inevitably leads to Ca 2ϩ store depletion (62), enhancing neuronal UCP function may prove to be a valuable protective strategy to counteract mitochondrial Ca 2ϩ overload and loss of mitochondrial function that precedes neurodegeneration. Future studies will determine if abnormalities in UCP4 expression or function contribute to the dysfunction and degeneration of neurons in acute and chronic neurodegenerative disorders.