Essential Role of Mitochondrial Ca2+ Uniporter in the Generation of Mitochondrial pH Gradient and Metabolism-Secretion Coupling in Insulin-releasing Cells*

Background: Mitochondrial Ca2+ uptake affects energy metabolism and insulin secretion. Results: Knockdown of mitochondrial Ca2+ uniporter decreases respiratory chain activity and mitochondrial pH gradient generation. Conclusion: Mitochondrial Ca2+ uptake via uniporter is essential for oxidative phosphorylation and metabolism-secretion coupling. Significance: The present study identifies mechanisms of action and bioenergetic consequences of mitochondrial Ca2+ transporters in insulin-releasing cells. In pancreatic β-cells, ATP acts as a signaling molecule initiating plasma membrane electrical activity linked to Ca2+ influx, which triggers insulin exocytosis. The mitochondrial Ca2+ uniporter (MCU) mediates Ca2+ uptake into the organelle, where energy metabolism is further stimulated for sustained second phase insulin secretion. Here, we have studied the contribution of the MCU to the regulation of oxidative phosphorylation and metabolism-secretion coupling in intact and permeabilized clonal β-cells as well as rat pancreatic islets. Knockdown of MCU with siRNA transfection blunted matrix Ca2+ rises, decreased nutrient-stimulated ATP production as well as insulin secretion. Furthermore, MCU knockdown lowered the expression of respiratory chain complexes, mitochondrial metabolic activity, and oxygen consumption. The pH gradient formed across the inner mitochondrial membrane following nutrient stimulation was markedly lowered in MCU-silenced cells. In contrast, nutrient-induced hyperpolarization of the electrical gradient was not altered. In permeabilized cells, knockdown of MCU ablated matrix acidification in response to extramitochondrial Ca2+. Suppression of the putative Ca2+/H+ antiporter leucine zipper-EF hand-containing transmembrane protein 1 (LETM1) also abolished Ca2+-induced matrix acidification. These results demonstrate that MCU-mediated Ca2+ uptake is essential to establish a nutrient-induced mitochondrial pH gradient which is critical for sustained ATP synthesis and metabolism-secretion coupling in insulin-releasing cells.

Pancreatic ␤-cells maintain blood glucose homeostasis by adapting insulin secretion to the changes in circulating nutrients. A major signaling molecule in this metabolism-secretion coupling linking nutrient metabolism to insulin secretion is cytosolic ATP most of which is synthesized from oxidative phosphorylation. Mitochondrial ATP synthesis is driven by the electrical (⌬⌿ mito , membrane potential) and chemical (⌬pH mito ) gradients across the mitochondrial inner membrane. These gradients are established as a result of electron transport and the associated export of protons mediated by the respiratory chain. Reducing equivalents in mitochondrial matrix are mainly produced by the tricarboxylic acid (TCA) 2 cycle and mitochondrial metabolite shuttles. Thus, the metabolic status of the ␤-cell mitochondria critically controls ATP synthesis and insulin secretory activity (1). Accumulating evidence suggests that defective mitochondrial function results in impaired glucose-stimulated insulin secretion (GSIS) and may contribute to the development of type 2 diabetes (2)(3)(4)(5).
In addition to [Ca 2ϩ ] mito , the matrix pH has been identified as a regulator of mitochondrial energy metabolism in ␤-cells. In contrast to other cell types, pancreatic ␤-cells have acidic pH mito under resting conditions. Nutrient stimulation causes matrix alkalinization without any marked cytosolic pH change (29). Preventing the resulting nutrient-induced increase of the ⌬pH mito changes using ionophores abrogated proton-coupled mitochondrial ion/metabolite transport, ATP synthesis, and GSIS regardless of elevated ⌬⌿ mito (29 -31). Therefore, pathogenic conditions causing a reduction of ⌬pH mito may seriously deteriorate ATP generation and insulin secretion in pancreatic ␤-cells.
Several recent reports demonstrate the functional role of MCU in pancreatic ␤-cells (26,32). MCU mediates glucosestimulated [Ca 2ϩ ] mito rise and second phase ATP/ADP increase (26). Knockdown of either MCU or MICU1 diminishes insulin secretion associated with defects in mitochondrial Ca 2ϩ uptake (32). Mice lacking MCU show a significant reduction of [Ca 2ϩ ] mito and Ca 2ϩ -stimulated oxygen consumption in muscle mitochondria, without changes in the basal respiration in embryonic fibroblasts (33). It remains unclear, however, how reduced MCU activity attenuates mitochondrial signal generation in pancreatic ␤-cell metabolism-secretion coupling. In this study, we observed that reduced mitochondrial Ca 2ϩ uptake following silencing of MCU significantly attenuated respiratory chain activity and ⌬pH mito increase in permeabilized as well as in intact insulin-secreting cells. These defects lead to impaired ATP synthesis and insulin secretion, demonstrating the crucial role of mitochondrial Ca 2ϩ uptake for the establishment of the ⌬pH mito in metabolism-secretion coupling. We also provide evidence for a novel role of the putative Ca 2ϩ /H ϩ antiporter leucine zipper-EF hand-containing transmembrane protein 1 (LETM1) as a Ca 2ϩ efflux route in insulin secreting cells, the role of which is altered in the absence of MCU.
siRNA Transfection-Cells were transfected with non-targeting or target-specific small interfering RNA (siRNA) using DharmaFECT1 (Dharmacon, Thermo Fisher Scientific Inc.). The target-specific siRNAs for rat MCU and LETM1 were purchased from Dharmacon, which is composed of siRNAs for four different targets of each gene (SMARTpool, Dharmacon).
Measurement of Mitochondrial Matrix Ca 2ϩ -To measure mitochondrial matrix Ca 2ϩ level, we used a mitochondria-targeted ratio-pericam (RPmit) plasmid, generously provided by Prof. Roger Tsien (UC San Diego). Cells were transfected with siRNA, and 24 h after, transfected with RPmit using X-tremeGENE (Roche Diagnostics GmbH, Mannheim, Germany). Fluorescence imaging of Ca 2ϩ was performed by using an inverted microscope (IX-81, Olympus, Tokyo, Japan) with an array laser confocal spinning disk (CSU10, Yokogawa Electric Corporation, Tokyo, Japan) and a cooled charge-coupled device (CCD) camera (Cascade 512B, Photometrics, Tucson, AZ). Intact or permeabilized cells on the confocal microscope were perfused with KRB solution or intracellular buffer, respectively, and fluorescence images (435 nm excitation and 535 nm emission) were acquired every 10 s and analyzed using Metafluor 6.3 software (Universal Imaging, Molecular Devices).
Measurement of Cytosolic ATP and Insulin-Cells plated onto 48 well-plates (2 ϫ 10 5 cells/well) were permeabilized with ␣-toxin and incubated for 5 or 15 min in an intracellular buffer containing ADP (10 M) with or without succinate (3 mM). To determine mitochondrial ATP release, the supernatant was harvested after incubation, and ATP level was measured by using the microplate reader (Synergy TM 2, BioTek Instruments Inc., Winooski, VT) with a bioluminescence assay kit (HS II, Roche Diagnostics, Mannheim, Germany).
For static insulin secretion measurement, cells in a 804Gcoated 24 well-plate (1.5 ϫ 10 5 cells/well) were transfected with siRNA and grown for 72 h. After deprivation of glucose for 1 h, cells were preincubated for 30 min with a KRB solution containing 2.8 mM glucose and 0.1% BSA. Then, cells were washed and incubated for 30 min with 2.8 mM or 16.7 mM glucose-containing KRB solution. Supernatant was collected for estimation of insulin release. Cellular insulin contents were determined in acid-ethanol extracts. Insulin levels were measured by using an insulin ELISA kit (Shibayagi Co., Gunma, Japan).
Mitochondrial Enzyme Activity and MTT Assay-For cytochrome c oxidase (COX) activity measurement, INS-1E cells were permeabilized by freeze-thaw cycle three times and mixed with isotonic solution (10 mM KH 2 PO 4 , 250 mM sucrose, 0.1% BSA, pH 6.5) with detergent (laurylmaltoside, 2.5 mM). Traces were started with the addition of reduced cytochrome c (25 nM; with a tiny amount of sodium hydrosulfite) and the enzymatic activity of COX was estimated by measuring absorbance at 550 nm continuously with a spectrophotometer (Amersham Biosciences, GE Healthcare Biosciences, Pittsburgh, PA). COX activity was expressed as moles of oxidized cytochrome c per min. Citrate synthase activity was measured with citrate synthase assay kit (Sigma) based on the manufacturer's instructions.
Measurement of Oxygen Consumption-Cellular oxygen consumption rate (OCR) was determined by Extracellular Flux Analyzer (XF-24, Seahorse Bioscience, North Billerica, MA). Cells (2 ϫ 10 4 cells/well) seeded on 24-well plates (Seahorse Bioscience) were transfected with siRNA and cultured for 72 h. On the experiment day, cells were incubated for 1 h at 37°C with KRB solution containing 2.8 mM glucose prior to 20 min of basal OCR measurement. Then, glucose (16.7 mM), oligomycin (3 g/ml), FCCP (3 M), and antimycin A (3 M) were added consecutively and the changes in OCR analyzed.
Measurement of Mitochondrial Membrane Potential-To measure the mitochondrial membrane potential (⌬⌿ mito ), cells seeded onto black-walled 96-well plates (5 ϫ 10 4 cells/well) were loaded with JC-1 (500 nM, Invitrogen) for 30 min and then permeabilized with ␣-toxin. The ratio of red (540 nm excitation and 590 nm emission) over green (490 nm excitation and 540 nm emission) fluorescence intensity was monitored from permeabilized cells in the presence of intracellular buffer containing JC-1 (500 nM) using a multi-well fluorescence reader (Flex-Station, Molecular Devices) (35).
As an alternative method to measure the mitochondrial membrane potential, cells seeded on coverslips were loaded with 5 nM TMRM for 20 min, and perfused with KRB solution containing TMRM (5 nM) on the inverted microscope. Fluorescence imaging with 514 nm excitation and 560 nm emission were recorded with the array laser confocal spinning disk microscopic system and analyzed by using Metamorph 6.1 software.
Statistical Analysis-Values are presented as mean Ϯ S.E. and N is the number of independent experiments. p values were obtained by Student's t test or one-way ANOVA and Ͻ 0.05 was considered to be significant.

Effects of MCU Knockdown on Mitochondrial Ca 2ϩ Uptake-
To understand the role of mitochondrial Ca 2ϩ transport in metabolism-secretion coupling, we transfected non-targeting siRNA (siControl) or siRNA selectively targeted to MCU (siMCU) in INS-1E cells, and assessed the effect of silencing after 72 h using quantitative real-time PCR and Western blotting. Application of siMCU efficiently reduced the transcript levels of MCU (73.8 Ϯ 5.3% reduction, Fig. 1A) compared with siControl-treated cells. Western blot analysis also revealed a strong siRNA-induced reduction of the MCU protein by 82.3 Ϯ 2.3% (Fig. 1, B and C).
To examine the impact of MCU knockdown on mitochondrial Ca 2ϩ uptake, we determined the effect of extramitochon-  (Fig. 1D). The [Ca 2ϩ ] mito declined slowly after returning to 10 nM [Ca 2ϩ ] o . Addition of 500 nM [Ca 2ϩ ] o resulted in more rapid and marked increase in [Ca 2ϩ ] mito , which slowly decreased again after cessation of the stimulus. In MCU-silenced cells, the responses to both Ca 2ϩ concentrations were markedly reduced by 61.8% and 58.2%, respectively (Fig. 1E). These results demonstrate that MCU contributes to mitochondrial Ca 2ϩ uptake at physiological Ca 2ϩ concentrations.

Effect of MCU Knockdown on ATP Synthesis and Insulin
Secretion-It is well known that [Ca 2ϩ ] mito amplifies metabolism-secretion coupling in ␤-cells and reduction of [Ca 2ϩ ] mito inhibits GSIS (6,7). To further investigate the role of [Ca 2ϩ ] mito in energy metabolism we studied succinate-dependent ATP synthesis in permeabilized cells. In the absence of substrate, silencing of MCU did not affect basal ATP formation (Fig. 1F). However, time-dependent ATP synthesis stimulated by succinate was markedly lowered in siMCU-treated cells. Consistent with the role of ATP as a signaling molecule for insulin exocytosis, GSIS in intact INS-1E cells was also dramatically decreased in MCU knockdown cells (76.3% inhibition, Fig. 1G). The role of MCU in metabolism-secretion coupling in pancreatic ␤-cells, was assessed following siMCU transfection of isolated rat pancreatic islets. Successful silencing of MCU in dispersed islet cells (Fig. 1H), resulted in significantly attenuated GSIS (Fig. 1I). Our data emphasize the importance of MCU-dependent mitochondrial Ca 2ϩ uptake in metabolism-secretion coupling of pancreatic ␤-cells.
Effects of MCU Knockdown on Mitochondrial Respiratory Function-Mitochondrial ATP synthesis by the F 1 F 0 -ATPase (complex V) is driven by the proton electrochemical gradient across the inner mitochondrial membrane, which is generated by the proton pumping activity of the respiratory complexes I, III, and IV. Protein expression of selected subunits of complex I, III, IV, and V was examined using Western blot analysis. MCU knockdown markedly reduced complex I (NDUFA9, nuclear DNA-en-  FEBRUARY 13, 2015 • VOLUME 290 • NUMBER 7 coded, Ϫ47.9%), complex III (UQCRC2, nuclear DNA-encoded, Ϫ47.5%), complex IV (subunit I, mitochondrial DNA-encoded, Ϫ67.1%), and complex V (ATP5A, nuclear DNA-encoded, Ϫ29%) (Fig. 2, A and B). Consistently, the enzyme activity of Complex IV was reduced by 16.6% after MCU knockdown (Fig. 2, D and E).

Role of MCU for Mitochondrial pH Gradient
Other mitochondrial functions such as citrate synthase activity (TCA cycle) or the TOM20 protein expression (subunit of mitochondrial protein import) were not altered when lowering MCU (Fig. 2, F and G) (Fig. 2, A and B). These results demonstrate that lowering mitochondrial calcium uptake selectively affect the expression of respiratory chain complexes.
Knockdown of MCU causes neither cell loss nor alteration in total soluble proteins (data not shown). Given these findings, we used the MTT assay as a read-out of mitochondrial reductive activity (36). Using an identical number of cells, silencing of MCU lowered the ability of cells to reduce MTT by 15% (Fig.  2C).
Finally, we investigated the effect of MCU knockdown on OCR. Under basal conditions (2.8 mM glucose), the OCR was modestly reduced in MCU knockdown cells (19.3% inhibition; Fig. 3, A and B). The OCR induced by high glucose (16.7 mM) was strongly impaired following MCU knockdown (35.0% inhibition; Fig. 3, A and C), suggesting that the activation of mitochondrial respiration is highly dependent on MCU activity. Taken together, in insulin-secreting cells, mitochondrial Ca 2ϩ uptake via MCU is necessary for mitochondrial functions at multiple levels from nutrient oxidation to ATP synthesis.

Nigericin-induced Mitochondrial Hyperpolarization Was
Lowered in MCU-silenced Cells-The mitochondrial electrical gradient (⌬⌿ mito ) is the main driving force for ATP synthesis as well as Ca 2ϩ transport through the MCU. MCU knockdown in turn may alter the ⌬⌿ mito . We therefore measured the ratio of fluorescence intensities (red/green) after loading with JC-1, which reflects the ⌬⌿ mito (35). The hyperpolarizing response to succinate in ␣-toxin permeabilized control and MCU knockdown cells, was not significantly different (Fig. 4A). Glucoseinduced hyperpolarization in intact cells was also not different between the two groups, which was measured by using the fluorescence probe TMRM in a non-quenching redistribution mode to measure ⌬⌿ mito (Fig. 4B) (on the figure it says JC-1 not TMRM). We also confirmed these findings by using JC-1 dye (data not shown), showing that silencing of MCU does not significantly alter ⌬⌿ mito in insulin-secreting cells.
The K ϩ /H ϩ electroneutral ionophore nigericin, dissipates the ⌬pH mito across the inner mitochondrial membrane. This results in a compensatory elevation of ⌬⌿ mito in order to maintain the total proton motive force (30). Therefore, hyperpolarization of ⌬⌿ mito by nigericin is proportional to the pH gradient prior to the addition of the ionophore. Interestingly, hyperpolarization by nigericin was markedly decreased (53.0% inhibition) following knockdown of MCU in permeabilized INS-1E cells (Fig. 4, C and D). This result suggests that MCU-silenced cells have defects in the establishment of a ⌬pH mito gradient but not ⌬⌿ mito , in response to nutrients. Knockdown of MCU Impaired Nutrient-generated pH Gradient-The ability of ␤-cells to elevate their ⌬pH mito following glucose stimulation is important for mitochondrial energy metabolism and thereby metabolism-secretion coupling (1). To directly assess the effect of MCU knockdown on pH mito , we expressed the mitochondria-targeted pH-sensitive protein mtAlpHi in siMCUtreated INS-1E cells. As shown in Fig. 5, A and C, succinate-induced alkalinization of pH mito in MCU-silenced cells was blunted compared with control cells (44.7% reduction). In intact MCU-silenced cells, glucose-induced matrix alkalinization was also strongly decreased (52.4% inhibition; Fig. 5, B and E). These results show that suppression of MCU-dependent mitochondrial Ca 2ϩ uptake also prevents the establishment of the nutrient-generated ⌬pH mito .
Addition of extramitochondrial Ca 2ϩ to succinate-stimulated mitochondria resulted in mitochondrial matrix acidification in permeabilized control cells. Interestingly, Ca 2ϩ -induced matrix acidification was not observed in MCU knockdown cells, indicating that this acidification is a secondary consequence of MCU-mediated Ca 2ϩ uptake (Fig. 5, A and D). Ca 2ϩ loading via MCU may be followed by Ca 2ϩ efflux in exchange for H ϩ in energized mitochondria causing the observed net acidification.
Acute Blocking of MCU Did Not Affect Metabolism-secretion Coupling-Gene silencing with siMCU transfection reduces the protein expression of MCU slowly over a time-course of several days. In order to understand the acute effects of blocking mitochondrial calcium import, we performed insulin measurement with the specific MCU blocker, Ru360. Pretreatment with Ru360 did not affect glucose-stimulated insulin secretion (Fig. 6A). MCU is a selective Ca 2ϩ channel mediating inward current and this Ca 2ϩ influx through MCU depolarizes the ⌬⌿ mito . We measured the effect of Ru360 on ⌬⌿ mito in intact cell during high glucose stimulation. As shown in Fig. 6, B and C, Ru360 further hyperpolarized the ⌬⌿ mito by blocking MCUmediated inward depolarizing currents. These results also confirm the effectiveness of Ru360 in intact cells. Taken together, long term reduction of mitochondrial Ca 2ϩ uptake leads to down-regulation of mitochondrial bioenergetics and metabolism-secretion coupling, which is not reproduced by acute pharmacological blocking of MCU.
LETM1 Participated as a Mitochondrial Ca 2ϩ -H ϩ Antiporter in INS-1E Cells-We hypothesized that the Ca 2ϩ /H ϩ antiporter LETM1 may be required for the observed matrix acidification triggered by Ca 2ϩ (23). To elucidate the role of LETM1 on Ca 2ϩ -coupled pH regulation, the change in pH mito upon extramitochondrial addition of Ca 2ϩ was measured in control or LETM1-silenced INS-1E cells. After 72 h of siRNA treatment, the knockdown effect was evaluated by using quantitative real-time PCR and Western blotting. Application of siLETM1 efficiently reduced the transcript (78.0 Ϯ 5.7% reduc-  . Knockdown of MCU did not affect mitochondrial membrane potential but decreased the hyperpolarizing response induced by dissipating the pH gradient with ionophore. A, mitochondrial membrane potential (⌿ mito ) in ␣-toxin-permeabilized INS-1E cells was measured in the presence of JC-1, a potential-sensitive fluorescence probe (500 nM). Increased JC-1 ratio (red/ green) reflects hyperpolarization of the ⌿ mito . B, intact cells were loaded and perfused with another potential-sensitive fluorescence dye, TMRM (5 nM), and the ⌿ mito was measured in a nonquenching redistribution mode. Hyperpolarization of the ⌿ mito by succinate (3 mM) in permeabilized cells (n ϭ 15) (A) and by high glucose (16.7 mM) in intact cells (n ϭ 6) (B) were compared between control and MCU-silenced cells. In ␣-toxin-permeabilized cells, nigericin (500 nM)-induced hyperpolarization of the ⌿ mito as compensation for the collapsed mitochondrial pH gradient (⌬pH mito ) was compared between control and MCU-silenced cells (n ϭ 6) (C and D). * denotes p Ͻ 0.05. FEBRUARY 13, 2015 • VOLUME 290 • NUMBER 7 tion, Fig. 7A) and LETM1 protein (78.9 Ϯ 0.9% reduction, Fig. 7, B and C) compared with siControl-treated cells. In ␣-toxinpermeabilized LETM1 knockdown cells, succinate-induced matrix alkalinization was not significantly altered (Fig. 7, D and  E). In contrast, Ca 2ϩ (500 nM)-elicited pH mito acidification was abolished in LETM1 knockdown cells, similar to our results in MCU knockdown cells (Fig. 7, D and F). Moreover, the increase in [Ca 2ϩ ] mito by extramitochondrial Ca 2ϩ was enhanced in LETM1-silenced cells (2.1-fold increase at 120 nM Ca 2ϩ , 1.4fold at 500 nM Ca 2ϩ ), likely caused by impaired Ca 2ϩ export (Fig. 7, G and H). Consistent with these findings the [Ca 2ϩ ] mito rise following K ϩ -induced Ca 2ϩ influx in intact cells was also enhanced in LETM1 knockdown cells (Fig. 7, I-L). Very similar results were obtained using the two different mitochondrial calcium probes ratiopericam (Fig. 7, I and J) and Rhod-2 ( Fig. 7, K and L). These results strongly suggest that LETM1 mediates at least one important component of Ca 2ϩ efflux in insulin secreting cells. In MCU knockdown cells, extramitochondrial Ca 2ϩ is taken up inefficiently, and therefore LETM1-mediated Ca 2ϩ efflux is strongly reduced. As a consequence the acidifying response of pH mito to extramitochondrial Ca 2ϩ is abrogated.

Role of MCU for Mitochondrial pH Gradient
We further investigated the functional consequences of LETM1 silencing on mitochondrial bioenergetics and metabolism-secretion coupling. Even though the stimulus-induced mitochondrial Ca 2ϩ response was augmented in LETM1 knockdown cells, glucose-induced insulin secretion and hyperpolarization of the ⌬⌿ mito were attenuated (Fig. 8, A and B). Reduction of LETM1 expression also lowered protein levels of subunits of respiratory chain complexes., These changes at the  level of the electron transport chain likely explain defective insulin secretion in LETM1-silenced cells (Fig. 8, C and D).

DISCUSSION
Increases in cytosolic Ca 2ϩ stimulate numerous energy consuming processes, including muscle contraction and neu-rotransmitter release. In particular, elevated mitochondrial matrix Ca 2ϩ ([Ca 2ϩ ] mito ) is a key stimulator of energy provision. The rise in [Ca 2ϩ ] mito activates TCA cycle dehydrogenases and ATP synthase leading to accelerated mitochondrial ATP production (6). For the proper coupling between energy demand and supply, propagation of Ca 2ϩ waves from the cytosol to the mitochondrial matrix through Ca 2ϩ transporters is necessary. MCU has been suggested to be the main communicating channel linking cytosolic and mitochondrial Ca 2ϩ signaling driven by the mitochondrial electrical gradient. Since the discovery of the molecular identity of MCU by two groups (11,12), several investigations on the role of MCU have been conducted. These studies consistently showed that MCU mediates the main mitochondrial Ca 2ϩ uptake route in HeLa cells (15), cardiac myocytes (37), neuronal cells (38), and pancreatic ␤-cells (26,32). In the latter cell type, the regulatory role of matrix Ca 2ϩ in mitochondrial ATP synthesis is therefore not limited to the provision of cellular energy, but also plays a key role as a signal in metabolism-secretion coupling.
In this study, we aimed to better understand the role of MCU as a regulator of mitochondrial metabolism and bioenergetics in pancreatic islet cells and clonal ␤-cells. Our findings demonstrate that silencing of MCU in insulin-releasing cells 1) decreases mitochondrial Ca 2ϩ uptake, 2) down-regulates electron transport chain proteins and enzyme activities, 3) reduces glucose-stimulated oxygen consumption, 4) impairs the generation of nutrient-stimulated ⌬pH mito , 5) does not affect ⌬⌿ mito , 6) reduces nutrient-stimulated ATP generation, and 7) impairs glucose-stimulated insulin secretion. We provide strong evidence that mitochondrial Ca 2ϩ uptake through MCU is a prerequisite for the establishment of the ⌬pH mito and activation of  5). B and C, reduced protein levels of LETM1 were validated by Western blots using primary antibodies against LETM1 and its densitometric analysis 72 h after transfection with siControl or siLETM1 (n ϭ 3). ␤-Actin was used as the reference control. INS-1E cells were transfected with non-targeting siRNA (siControl) or siRNA against LETM1 (siLETM1) and infected with an adenovirus carrying mtAlpHi or transfected with a ratiometric-pericam plasmid 24 h after siRNA transfection. D-F, changes in mitochondrial matrix pH (pH mito ) in response to succinate (3 mM) or extramitochondrial Ca 2ϩ (500 nM) were measured after 48 h of further incubation and compared between control (clear bar) and LETM1 knockdown cells (gray bar) (n ϭ 14 -21). G and H, increases in mitochondrial matrix Ca 2ϩ ([Ca 2ϩ ] mito ) by the addition of extramitochondrial Ca 2ϩ (120 nM and 500 nM) in ␣-toxin-permeabilized cells were compared between two groups (n ϭ 5-12). The changes in [Ca 2ϩ ] mito by the application of high K ϩ (30 mM) in intact cells were measured using ratiopericam probe (n ϭ 5) (I and J) or Rhod-2 dye (K and L). ** and *** denote p Ͻ 0.01 and Ͻ 0.001, respectively.  FEBRUARY 13, 2015 • VOLUME 290 • NUMBER 7 mitochondrial energy metabolism. The critical role of the ⌬pH mito in mitochondrial ATP synthesis and insulin secretion has been demonstrated previously (29,30). It is inferred, therefore, that the reduced [Ca 2ϩ ] mito together with impaired ⌬pH mito generation are the main reasons for defective metabolism-secretion coupling in MCU-silenced cells.

Role of MCU for Mitochondrial pH Gradient
MCU is an inwardly rectifying, highly Ca 2ϩ -selective ion channel driven by the negative ⌿ mito generated by the respiratory chain (39). Because of high ⌬⌿ mito (assumed as 180mV), energized mitochondria have the ability to capture cytosolic Ca 2ϩ over a wide range of concentrations (40,41). We observed that knockdown of MCU significantly diminished mitochondrial Ca 2ϩ influx at relatively low concentrations of extramitochondrial Ca 2ϩ (Ͻ 500 nM). These data imply that the MCU is the main mediator of mitochondrial Ca 2ϩ uptake from the cytosol under physiological conditions despite the negative regulation by other subunits of this complex such as MICU1/2 (18). Independent of the Ca 2ϩ source, either release from the ER or influx from the extracellular space, MCU works as the main Ca 2ϩ transport route into mitochondria (42). In MCUsilenced insulinoma cells, [Ca 2ϩ ] mito rises following either ER Ca 2ϩ release or high K ϩ -induced Ca 2ϩ influx are reduced (32).
We found a close functional connection between mitochondrial Ca 2ϩ uptake and ⌬pH mito regulation, not previously observed. Mitochondria in pancreatic ␤-cells are of relatively high volume density, facilitating nutrient metabolism and signal generation (35). Indeed, detection of plasma glucose levels is strictly dependent on mitochondrial oxidative phosphorylation in the ␤-cells (2). In this context, matrix alkalinization by glucose is a distinctive characteristic of ␤-cell mitochondria (29). In contrast, glycolytic cells such as HeLa cells or HepG2 cells have high resting pH mito and do not respond to nutrient stimulation (29,43). Our previous studies in insulin-secreting cells showed that short term attenuation of [Ca 2ϩ ] mito rises in an extracellular Ca 2ϩ free condition did not affect nutrient-stimulated alkalinization of pH mito. Conversely, a [Ca 2ϩ ] i transient caused by tolbutamide, a K ATP channel blocker, also did not affect ⌬pH mito (29). In the present study, however, continuous suppression of calcium uptake after knockdown of MCU had profound effects on matrix pH and oxidative phosphorylation. As strong evidence for impaired mitochondrial metabolism following MCU knockdown, we observed that protein levels and function of electron transport chain complexes, mitochondrial enzyme activities, and oxygen consumption rate were all reduced. Our findings are consistent with an earlier study which showed that effective buffering of matrix Ca 2ϩ lowered NAD(P)H levels, oxygen consumption, and ATP synthesis in hormone secreting cells (7). We propose that persistent inhibition of [Ca 2ϩ ] mito rises perturbs Krebs cycle and electron transport chain activities, which in turn causes defective ⌬pH mito generation and ATP synthesis, leading to impaired nutrientstimulated insulin secretion.
Nigericin-induced hyperpolarization of ⌿ mito reflects the preexisting ⌬pH mito . Therefore, the reduced hyperpolarizing response in MCU-silenced cells suggests reduction of the preexisting ⌬pH mito (Fig. 4, C and D). This effect on ⌬pH mito was confirmed using the mitochondrial pH-sensitive probe mtAlpHi (Fig. 5). On the other hand, the mitochondrial electri-cal gradient, the main component of the proton motive force, was not affected by MCU silencing (Fig. 4, A and B). This finding is similar to those observed in other cell types (11,12). It is not clear why there is a selective defect in ⌬pH mito generation without alterations in ⌬⌿ mito . Ca 2ϩ influx through MCU uses the electrical gradient as a driving force therefore rapid Ca 2ϩ influx elicits depolarization of ⌬⌿ mito (40,41). In MCU-silenced cells, mitochondrial Ca 2ϩ inward currents are reduced contributing to the preservation of ⌬⌿ mito . Another mechanism contributing to the maintenance of ⌬⌿ mito could be lower activity of the mitochondrial Na ϩ /Ca 2ϩ exchanger (NCLX) because the amplitude of [Ca 2ϩ ] mito is decreased in MCU-silenced cells. Ca 2ϩ efflux through NCLX leads to depolarization of ⌬⌿ mito because of its electrogenic property (24). Taken together, we can infer that reduced Ca 2ϩ influx and efflux in MCU-silenced cells prevent depolarization of ⌿ mito , which may countervail the effect of attenuated respiratory chain activity.
An interesting finding in this study is the acidification of pH mito by extramitochondrial Ca 2ϩ addition to energized mitochondria of permeabilized INS-1E cells. This pH mito acidification was dependent on mitochondrial Ca 2ϩ uptake via MCU. Mitochondrial Ca 2ϩ transport is coupled with H ϩ through a Ca 2ϩ /H ϩ exchanger, the molecular identity of which is not clear. In a genome-wide RNA interference screen Jiang et al. identified LETM1 to mediate this exchange in mitochondria (23). To investigate the role of LETM1 on Ca 2ϩ /H ϩ -coupled transport in insulin-secreting cells, we tested whether there is an alteration in pH mito acidification after knockdown of LETM1. Similar to the response in MCU knockdown cells, the pH mito in LETM1 knockdown cells was not acidified, instead slight alkalinization by extramitochondrial Ca 2ϩ was observed (Fig. 6C). Furthermore, the [Ca 2ϩ ] mito rises by extramitochondrial Ca 2ϩ addition were increased in LETM1-silenced cells (Fig. 6F). Our findings demonstrate that LETM1 mediates Ca 2ϩ efflux from mitochondria of insulin-secreting cells working in parallel with NCLX. A recent publication shows that purified human LETM1 mediates electroneutral 2 H ϩ /1 Ca 2ϩ antiport when reconstituted in artificial liposomes (44). Thus, in intact cells, Ca 2ϩ efflux via LETM1 is preferred, which is driven by the H ϩ gradient across the inner mitochondrial membrane (45). We suggest that Ca 2ϩ influx through MCU is coupled to LETM1-mediated Ca 2ϩ efflux with proton uptake favored by high [Ca 2ϩ ] mito and alkaline pH mito . In MCU-silenced cells, Ca 2ϩ efflux via LETM1 was prevented or even reversed to Ca 2ϩ influx due to reduced Ca 2ϩ and pH gradients, explaining the disappearance of pH mito acidification.
Why LETM1 knockdown has negative effects on the expression of subunits of the respiratory chain, glucose-stimulated ⌿ mito hyperpolarization and insulin secretion is not clear. A recent publication presented evidence that LETM1 haplo-insufficiency (ϩ/Ϫ) increases mitochondrial superoxide levels which is responsible for mitochondrial dysfunction (46). We propose that oxidative stress in LETM1-silenced insulin-releasing cells may negatively affect mitochondrial bioenergetics, respiratory activity, and metabolism-secretion coupling.
Our findings herein demonstrate that MCU-mediated Ca 2ϩ uptake is essential for the respiratory chain activity and the generation of ⌬pH mito in insulin-releasing cells. This chemical gradient (⌬pH mito ) is critically required for the substrate transport into the mitochondrial matrix, including pyruvate and inorganic phosphate (31) and ATP synthesis (29). Therefore, evidence in the present study suggests the bioenergetic mechanism to explain the defective metabolism-secretion coupling by MCU knockdown. In addition, mitochondrial Ca 2ϩ uptake regulates cytosolic Ca 2ϩ signaling and contributes to prevent local Ca 2ϩ overload in the cytosol (37). On the contrary, accumulation of Ca 2ϩ in the mitochondrial matrix induces permeability transition (PT) pore opening and apoptosis. To maintain Ca 2ϩ homeostasis, there is an interactive operation of mitochondrial transporters involved in Ca 2ϩ influx and efflux pathways. Further research focusing on the comprehensive understanding of mitochondrial Ca 2ϩ transporters may lead to the identification of novel therapeutic targets to improve mitochondrial energy metabolism and to prevent cytotoxicity. This may be especially relevant in insulin-releasing cells, where mitochondrial Ca 2ϩ transport plays a key role in metabolismsecretion coupling, dysfunction of which leads to the development of type 2 diabetes.