NCLX Protein, but Not LETM1, Mediates Mitochondrial Ca2+ Extrusion, Thereby Limiting Ca2+-induced NAD(P)H Production and Modulating Matrix Redox State*

Background: Whether mitochondrial Ca2+ extrusion is mediated by NCLX (mitochondrial sodium/calcium exchanger) or LETM1 (leucine zipper-EF-hand-containing transmembrane protein 1) and controls matrix redox state is unknown. Results: NCLX, but not LETM1, increases Ca2+ extrusion, limits NAD(P)H production, and promotes matrix oxidation. Conclusion: NCLX controls the duration of matrix Ca2+ elevations and their impact on redox signaling. Significance: NCLX is a potential target for the treatment of redox-dependent diseases. Mitochondria capture and subsequently release Ca2+ ions, thereby sensing and shaping cellular Ca2+ signals. The Ca2+ uniporter MCU mediates Ca2+ uptake, whereas NCLX (mitochondrial Na/Ca exchanger) and LETM1 (leucine zipper-EF-hand-containing transmembrane protein 1) were proposed to exchange Ca2+ against Na+ or H+, respectively. Here we study the role of these ion exchangers in mitochondrial Ca2+ extrusion and in Ca2+-metabolic coupling. Both NCLX and LETM1 proteins were expressed in HeLa cells mitochondria. The rate of mitochondrial Ca2+ efflux, measured with a genetically encoded indicator during agonist stimulations, increased with the amplitude of mitochondrial Ca2+ ([Ca2+]mt) elevations. NCLX overexpression enhanced the rates of Ca2+ efflux, whereas increasing LETM1 levels had no impact on Ca2+ extrusion. The fluorescence of the redox-sensitive probe roGFP increased during [Ca2+]mt elevations, indicating a net reduction of the matrix. This redox response was abolished by NCLX overexpression and restored by the Na+/Ca2+ exchanger inhibitor CGP37157. The [Ca2+]mt elevations were associated with increases in the autofluorescence of NAD(P)H, whose amplitude was strongly reduced by NCLX overexpression, an effect reverted by Na+/Ca2+ exchange inhibition. We conclude that NCLX, but not LETM1, mediates Ca2+ extrusion from mitochondria. By controlling the duration of matrix Ca2+ elevations, NCLX contributes to the regulation of NAD(P)H production and to the conversion of Ca2+ signals into redox changes.

hand domain-containing protein 1, mitochondrial; or leucine zipper-EF-hand-containing transmembrane protein 1) have been recently proposed to exchange Ca 2ϩ against Na ϩ or H ϩ , respectively. Functional analysis strongly suggests that NCLX is a mitochondrial Na ϩ /Ca 2ϩ exchanger because overexpression of this protein enhances mitochondrial Ca 2ϩ efflux, whereas its knockdown diminishes Ca 2ϩ extrusion. Furthermore, pharmacological inhibition of mitochondrial Ca 2ϩ efflux with the benzothiazepine derivative CGP37157 completely blocks NCLXdependent Ca 2ϩ export. LETM1 was proposed to be a high affinity mitochondrial Ca 2ϩ /H ϩ exchanger (33,34) able to drive both extrusion and uptake of Ca 2ϩ into energized mitochondria at submicromolar Ca 2ϩ concentrations. Previous studies, however, indicated that LETM1 mediates mitochondrial K ϩ /H ϩ exchange (35,36), and the contribution of LETM1 to mitochondrial Ca 2ϩ transport is not yet firmly established (37). One factor hindering studies of mitochondrial Ca 2ϩ extrusion is the large variability in the kinetics of mitochondrial Ca 2ϩ efflux between cells during physiological stimuli. Perfectly detailed protocols have been available from the 1970s, for the quantitative analysis of Ca 2ϩ efflux, and an elegant series of studies carried out by Carafoli and co-workers (38,39) on isolated mitochondria examined the pathway and mechanism of Ca 2ϩ release. Nevertheless, protocols enabling the quantitative analysis of mitochondrial Ca 2ϩ efflux in live cells, where the analysis of this process is complicated by cell-to-cell variability, are lacking.
The transient matrix Ca 2ϩ elevations have several effects on mitochondrial function. The energetic redox balance in particular is a primary target of the mitochondrial Ca 2ϩ homeostasis (40), with strong impact on metabolic regulation (41) and human health (42). Inside the organelle, Ca 2ϩ activates oxidative metabolism and respiration. In addition, [Ca 2ϩ ] mt elevations can have several and sometimes opposing effects on the redox balance. On the one hand, [Ca 2ϩ ] mt elevations activate Ca 2ϩ -dependent dehydrogenases, accelerating NADH production (25). As a result, the ratio of the redox couple NAD(P)H/ NAD(P) will increase (43)(44)(45)(46)(47). On the other hand, [Ca 2ϩ ] mt elevations accelerate respiration. This will increase the associated formation of reactive oxygen species (ROS) (48,49), with a net oxidizing effect in the matrix space.
Here we have studied the role of NCLX and LETM1 in the export of Ca 2ϩ from the mitochondrial matrix space. In order to properly describe mitochondrial Ca 2ϩ export kinetics, we have applied a biparametric single-cell analysis. This novel approach allowed us to determine the contribution of different Ca 2ϩ export systems in an amplitude-dependent manner. Furthermore, we have assessed the importance of Ca 2ϩ extrusion kinetics in the regulation of oxidative metabolism and in the control of the mitochondrial redox state.

EXPERIMENTAL PROCEDURES
Reagents-Histamine, dithiothreitol (DTT), H 2 O 2 , and rotenone were obtained from Sigma, and CGP37157 was from Calbiochem. Preparation of NCLX-encoding plasmid was described previously (32). The 4mtD3cpv construct (50) was provided by Drs. Amy Palmer and Roger Tsien (University of California, San Diego). The mitochondrial redox indicator roGFP1 (51) was provided by Dr. S. James Remington (University of Oregon). The LETM1-encoding plasmid (35) was provided by Dr. Luca Scorrano (University of Geneva). The mitochondrial pH sensor mitoSypHer was described previously (52).
Mitochondrial Ca 2ϩ Measurements-Experiments were performed in HEPES buffer containing 140 mM NaCl, 5 mM KCl, 1 mM MgCl 2 , 2 mM CaCl 2 , 20 mM Hepes, 10 mM glucose, pH 7.4, with NaOH at 37°C. Glass coverslips were inserted in a thermostatic chamber (Harvard Apparatus, Holliston, MA), and solutions were changed by hand. Cells were imaged on an Axiovert s100 TV using a ϫ40, 1.3 numeric aperture oil immersion objective (Carl Zeiss AG, Feldbach, Switzerland) and a cooled, 16-bit CCD back-illuminated frame transfer MicroMax camera (Roper Scientific, Trenton, NJ). [Ca 2ϩ ] mt was measured with the genetically encoded 4mtD3cpv sensor. Cells were excited at 430 nm through a 455DRLP dichroic and alternately imaged with 480AF30 and 535DF25 emission filters (Omega Optical). Images were acquired every 2 s. Fluorescence ratios were calculated in MetaFluor 6.3 (Universal Imaging) and analyzed in Excel (Microsoft) and GraphPad Prism 5 (GraphPad). [Ca 2ϩ ] mt was calculated in situ in semipermeabilized cells as described previously (55) from 4mtD3cpv ratios (R) using the following equation.
[Ca 2ϩ ] ϭ [KЈd n ϫ ͑R Ϫ R min ͒/͑R max Ϫ R͒] 1/n (Eq. 1) R min was obtained by treating the cells with 1 mM EGTA along with 10 M ionomycin, and R max was obtained by treating the cells with 10 M ionomycin and 10 mM Ca 2ϩ . The maximal Ca 2ϩ efflux rates were calculated by performing a first order derivative on the data obtained during the first minute of the decay phase of the Ca 2ϩ response.
Mitochondrial Matrix pH Measurements in Permeabilized Cells-Ratiometric measurements of the mitochondrial pH were performed on the same instrument as for [Ca 2ϩ ] mt measurements, using the mitochondrial targeted sensor mito-SypHer. Cells were alternately excited at 420 and 490 nm through a 505DCXR dichroic filter and imaged with a 535DF25 band pass filter (Omega Optical) as described previously (52). Images were acquired every 5 s. MitoSypHer-expressing HeLa cells were permeabilized on the microscope with a 1-min exposure to digitonin (100 M) in Ca 2ϩ -free intracellular buffer, containing 235 mM sucrose, 20 mM HEPES, 5 mM succinic acid, 1 mM EGTA, adjusted to pH 7.4 with N-methyl-D-glucamine. After digitonin washout, cells were kept in intracellular buffer for 10 min, before K ϩ -driven H ϩ extrusion was evoked by changing the intracellular solution with a K ϩ -gluconate solution containing 50 mM potassium gluconate, 135 mM sucrose, 20 mM HEPES, 5 mM succinic acid, 1 mm EGTA, adjusted to pH 7.4 with N-methyl-D-glucamine. The ratiometric 490/420 signals were normalized to the basal level (set to 1), and the amplitude of K ϩ -evoked ⌬pH was calculated.
Mitochondrial Redox State Measurements-Ratiometric measurements of the mitochondrial redox state were performed by using the same instrument as described for [Ca 2ϩ ] mt measurements, using the mitochondrially targeted, genetically encoded sensor roGFP1. Cells were excited at 410 and 480 nm, and emission was collected at 535 nm (535DF45, Omega Optical) through a 505DCXR (Omega Optical) dichroic mirror. Images were acquired every 2 s. The 480/410 fluorescence ratios were normalized to both the minimum of fluorescence (obtained after the addition of 1 mM H 2 O 2 ) and the maximum (obtained after the addition of 10 mM DTT). NAD(P)H Measurements-Cells were allowed to adhere to glass bottom dishes (MatTek, Ashland, MA) for 2 days, and experiments were performed in the Hepes buffer solution as described above. A laser-scanning confocal microscope (Leica TCS SP5 II MP, Manheim, Germany) with a HCX IRAPO L ϫ25/0.95 water objective was utilized to monitor NAD(P)H autofluorescence at 37°C (Life Imaging Services). Laser scans at 727 nm (IR Laser Chameleon ultra, Coherent) were used for two-photon NAD(P)H excitation. Each 512 ϫ 512-pixel image represents an average of 16 scans taken with a resonant scanner at 8000 Hz. NAD(P)H emission was collected at 445-495-nm wavelength every 20 s. The excitation power of the laser was set to avoid cellular photodamage. Data were processed with LAS AF software (Leica) and analyzed in Excel (Microsoft) and GraphPad Prism 5 (GraphPad). In order to normalize the NAD(P)H responses, each experiment was concluded by adding the complex I inhibitor rotenone, which results in the accumulation of NAD(P)H (autofluorescence close to maximal), followed by peroxide, where the autofluorescence is almost completely lost (minimal value). Fluorescence intensity data were normalized to both rotenone addition (equal to 1) and minimal value (hydrogen peroxide ϭ 0). Change in fluorescence (⌬NAD(P)H) were calculated as the difference between the effect of histamine after reaching a plateau and the baseline value.

Biparametric Analysis of Mitochondrial Ca 2ϩ Extrusion
Rates-To assess the kinetics of Ca 2ϩ efflux from mitochondria, we measured [Ca 2ϩ ] mt changes evoked by physiological agonists in HeLa cells with the mitochondrially targeted Ca 2ϩ sensor 4mtD3cpv. Application of the endoplasmic reticulum Ca 2ϩ -mobilizing agonist histamine rapidly increased [Ca 2ϩ ] mt to ϳ2.5 M (Fig. 1A), but the amplitude and kinetics of the elevations were highly variable, even between cells of the same clonal population recorded simultaneously (Fig. 1B). Importantly, the Ca 2ϩ efflux kinetics increased together with the amplitude of [Ca 2ϩ ] mt elevations, with large elevations followed by rapid Ca 2ϩ efflux and small elevations followed by a sustained plateau. This heterogeneity prompted us to perform a biparametric analysis and to measure both the maximal rates of Ca 2ϩ extrusion and the amplitude of the [Ca 2ϩ ] mt response (Fig. 1C). The efflux rates were then expressed as a function of the corresponding signal amplitude by aggregating the data over defined ranges of [Ca 2ϩ ] mt , which allowed us to model the [Ca 2ϩ ] mt -Ca 2ϩ efflux relationship by an exponential fit (Fig.  1D). This approach enabled us to include all of the cells recorded while preserving the complexity of the underlying biological process.
NCLX Levels but Not LETM1 Levels Modulate Matrix Ca 2ϩ Extrusion at High [Ca 2ϩ ] mt -We then assessed the contribution of NCLX and LETM1 to mitochondrial Ca 2ϩ extrusion. Consistent with their proposed roles as mitochondrial Ca 2ϩ / Na ϩ and Ca 2ϩ /H ϩ exchangers, both proteins were strongly enriched together with the outer mitochondrial membrane protein TOM20 in mitochondrial fractions from HeLa cells ( Fig. 2A). NCLX overexpression (32) did not alter the average amplitude of the [Ca 2ϩ ] mt elevations evoked by histamine (not shown) but accelerated the kinetics of mitochondrial Ca 2ϩ efflux (Fig. 2B). The biparametric analysis revealed that NCLX accelerated Ca 2ϩ efflux exclusively in cells undergoing large [Ca 2ϩ ] mt elevations (Fig. 3C). Separate analysis of cells exhibiting small (⌬R/R 0 Յ 0.3) and large (⌬R/R 0 Ͼ 0.3) elevations confirmed that NCLX enhanced mitochondrial Ca 2ϩ extrusion only in cells experiencing large [Ca 2ϩ ] mt elevations (Fig. 2D). CGP37157, an inhibitor of the mitochondrial Na ϩ /Ca 2ϩ exchanger, almost completely prevented Ca 2ϩ efflux, regardless of NCLX overexpression (Fig. 2, B-D). The inhibition was reversible (not shown) and particularly evident in cells undergoing large [Ca 2ϩ ] mt elevations (Fig. 2D). We then tested whether the proposed Ca 2ϩ /H ϩ exchanger LETM1 could affect Ca 2ϩ efflux, possibly over a range of [Ca 2ϩ ] mt distinct from NCLX. Overexpression of LETM1 (35) was validated by Western blot, and the function of the overexpressed protein was tested in permeabilized cells (Fig. 3, A and B). LETM1 was originally identified as a key element of mitochondrial volume homeostasis through regulation of K ϩ /H ϩ exchange (37). In our hands, overexpression of LETM1 increased K ϩ -driven matrix proton extrusion (Fig. 3B), consistent with the overexpression of a functional K ϩ /H ϩ exchanger (Fig. 3B). On the other hand, LETM1 did not alter Ca 2ϩ efflux rates, regardless of the amplitude of [Ca 2ϩ ] mt elevations (Fig. 3, C-E). These data demonstrate that NCLX but not LETM1 levels limit the rates of Ca 2ϩ extrusion from mitochondria, an effect most apparent during large [Ca 2ϩ ] mt elevations in HeLa cells that endogenously express both exchangers.
NCLX Levels Modulate the Mitochondrial Redox State during Stimulation-Through its impact on the duration of [Ca 2ϩ ] mt elevations, NCLX may affect intramitochondrial Ca 2ϩ -dependent processes and alter the mitochondrial redox status. To test this possibility, we measured the mitochondrial redox state using the genetically encoded redox-sensitive probe roGFP1 (51), which contains engineered surface cysteine groups positioned to reversibly form disulfide bonds. Expression of matrixtargeted roGFP1 labeled mitochondria (Fig. 4A), and the fluorescence signal increased during histamine application, indicating a more reduced state (Fig. 4B). To compare redox responses, we determined the full signal range of the probe in each experiment using peroxide to oxidize roGFP1 followed by reduction of the probe with dithiothreitol (Fig. 4B). NCLX overexpression did not affect the basal redox state prior to histamine addition (Fig. 4C) but completely prevented the histamine-induced redox changes (Fig. 4, B and D), an effect that was partially reverted by CGP37157 (Fig. 4, B and D). These results demonstrate that NCLX levels regulate the mitochondrial redox state, probably by altering the duration of [Ca 2ϩ ] mt elevations.
NCLX Levels Limit Histamine-induced Mitochondrial NAD(P)H Production-The observed net reduction of mitochondrial matrix redox state led us to speculate that NCLX, via its effect on matrix Ca 2ϩ , might regulate Ca 2ϩ -dependent matrix dehydrogenases. To test this possibility, we measured changes in NAD(P)H autofluorescence by two-photon microscopy. Histamine application increased the NAD(P)H autofluorescence of HeLa cells (Fig. 5A), indicating a net reduction of NAD(P) ϩ to NAD(P)H, confirming earlier studies (56,57). To compare NAD(P)H responses, we calibrated each recording by adding the complex I inhibitor rotenone to promote maximal accumulation of NAD(P)H, followed by peroxide to decrease the autofluorescence to the minimal value. NCLX overexpression did not affect the basal autofluorescence levels of HeLa cells (Fig. 5B) but severely blunted histamine-induced NAD(P)H formation, by 73%, an effect fully prevented by CGP37157 (Fig. 5C). We conclude that NCLX levels are critical for the [Ca 2ϩ ] mt -dependent regulation of the mitochondrial oxidative metabolism.

DISCUSSION
Mitochondria sense and shape Ca 2ϩ signals during cell stimulation (2-6, 58) via their ability to take up and subsequently release Ca 2ϩ ions. Ca 2ϩ sequestration in the mitochondrial matrix contributes to the buffering of cytosolic Ca 2ϩ elevations and serves as a signal that activates mitochondrial Ca 2ϩ -dependent processes (24,25). In contrast, prolonged accumulation of Ca 2ϩ in the matrix triggers mitochondria-induced cell death (28,31,58). Here, we assessed the contribution of the two ion exchangers NCLX and LETM1 in the kinetics of mitochondrial Ca 2ϩ extrusion and in the control of the matrix redox state.
The proteins that transport Ca 2ϩ across the mitochondrial inner membrane were recently identified, and most efforts are currently devoted to defining the mechanism of mitochondrial Ca 2ϩ uptake (14, 15, 21-23, 59, 60). However, the extrusion process is equally critical to achieve mitochondrial Ca 2ϩ -dependent signaling without triggering cell death. Here, we demonstrate that the rates of mitochondrial Ca 2ϩ extrusion are related to the [Ca 2ϩ ] mt amplitude, with high rates of Ca 2ϩ efflux following large elevations and very slow rates following small elevations. This amplitude-dependent control of mitochondrial Ca 2ϩ export probably serves to maintain [Ca 2ϩ ] mt in a range that is sufficient to activate mitochondrial metabolism (24 -26) without reaching the levels that could initiate apoptosis. Our observation that Ca 2ϩ extrusion is minimal at low [Ca 2ϩ ] mt and maximal when mitochondria experience Ca 2ϩ signals of large amplitude therefore suggests that the Ca 2ϩ extrusion system is tuned to avoid long lasting [Ca 2ϩ ] mt elevations that can trigger cell death by promoting Ca 2ϩ -dependent mitochondrial permeability transition pore opening (28,31). The nature of this [Ca 2ϩ ] mt -sensing mechanism is not known, but several studies suggest the existence of regulatory mechanisms controlling mitochondrial Ca 2ϩ export kinetics via direct or indirect interactions with the mitochondrial Na ϩ /Ca 2ϩ exchanger. The protein kinases PKC (61) and PINK1 (62) were reported to modulate the activity of this ion exchanger, and the stomatin-like protein SLP-2, which localizes to the inner mitochondrial membrane, was shown to inhibit mitochondrial Na ϩ /Ca 2ϩ exchange (63). Direct regulation of the exchanger by Ca 2ϩ cannot be excluded, but NCLX does not share the hallmark Ca 2ϩ regulatory site of plasma membrane Na ϩ /Ca 2ϩ exchange proteins (64).
The [Ca 2ϩ ] mt dependence of the Ca 2ϩ extrusion system, combined with the large cellular variability in the amplitude of [Ca 2ϩ ] mt responses, prompted us to express Ca 2ϩ efflux rates as a function of the [Ca 2ϩ ] mt amplitude to assess how molecular and pharmacological manipulations alter mitochondrial Ca 2ϩ efflux. Using this biparametric analysis, we found that NCLX levels are rate-limiting for mitochondrial Ca 2ϩ export during physiological stimulations, whereas LETM1 levels are inconsequential. This confirms earlier studies on NCLX (32) and casts doubt as to the Ca 2ϩ exchanger function of LETM1, which remains to be clarified (37). Although mitochondrial Ca 2ϩ extrusion is mainly mediated by exchangers (5,10,39), the permeability transition pore (27,29,30) may contribute to Ca 2ϩ release through fast and reversible opening of the channel (65). It would be interesting to test the contribution of the permeability transition pore under physiological conditions using the biparametric analysis presented here.
Given the strong impact of NCLX on mitochondrial Ca 2ϩ extrusion kinetics, we further investigated its role in the regulation of mitochondrial oxidative metabolism and redox state. A complex relationship exists between [Ca 2ϩ ] mt and matrix redox processes. Ca 2ϩ is able to affect the redox state by activating oxidative metabolism but also by influencing the formation of reactive oxygen species. A rise in [Ca 2ϩ ] mt stimulates several matrix dehydrogenases, which in the presence of substrate are able to increase the NAD(P)H/NAD(P) ratio and, as a consequence, cause a net reduction of the matrix redox state. At the same time as respiration is accelerated, more reactive oxygen species are formed, which should shift redox couples in the direction of oxidation (43)(44)(45)(46)(47)(48)(49). Following stimulation with histamine, we observed a net reduction of the redox-sensitive thiol groups of roGFP1 expressed in the mitochondria of HeLa cells (Fig. 4). Similarly, the histamine-induced [Ca 2ϩ ] mt rise also increased the NAD(P)H/NAD(P) ratio (Fig. 5). The net redox changes due to histamine-induced [Ca 2ϩ ] mt elevations appear to be dominated by the activation of Ca 2ϩ -dependent dehydrogenases. Such redox changes are known to have further downstream effects modulating electron transport, ATP-synthase (66), and matrix enzyme activities (67). Mitochondrial redox signaling has therefore been proposed to be crucial for the regulation of energy metabolism (41,68). Furthermore, mitochondrial activation and redox changes are linked to the production of ROS at concentrations that impact signaling functions, as shown for glucose-induced insulin secretion (69). Our results establish a direct link between NCLX-mediated mitochondrial Ca 2ϩ extrusion and matrix redox state. Redox-dependent processes are therefore sensitive to the regulation of NCLX during stimulus-induced [Ca 2ϩ ] mt elevations. Given that NCLX shortens the duration of the mitochondrial Ca 2ϩ transient (Fig. 2, B and C) without significantly lowering the amplitude, the effect on the mitochondrial redox state is surprisingly strong (Fig. 4). These results suggest that the fast uptake of Ca 2ϩ is not sufficient to modulate the mitochondrial redox state. Instead, [Ca 2ϩ ] mt elevations must last for a sufficient time to boost NAD(P)H production. This is consistent with previous studies showing that the metabolic decoding of cytosolic Ca 2ϩ elevations requires the integration of multiple repetitive elevations (56,57,70).
The inhibitor CGP37157 rescued all of the mitochondrial functions affected by NCLX overexpression, indicating that Na ϩ /Ca 2ϩ exchange activity accounts for the changes in oxidative metabolism and redox state. In the presence of the inhibitor, Ca 2ϩ extrusion was minimal regardless of NCLX overexpression, whereas  redox changes and NAD(P)H generation in NCLX-overexpressing cells were restored to control levels (Figs. 4 and 5). Based on the almost complete block of Ca 2ϩ extrusion, one could have expected a further reduction of the NAD(P)H/NAD(P) ratio in treated cells and a more reduced state in the matrix than control levels. The sustained [Ca 2ϩ ] mt elevation evoked by CGP37157, however, is expected to augment not only oxidative metabolism and respiration but also ROS formation, which would oxidize the matrix and decrease the NAD(P)H/NAD(P) ratio. The redox state of CGP37157-treated cells might therefore reflect the balance between accelerated NAD(P)H formation and increased ROS-dependent oxidation.
Our data demonstrate that NCLX plays a key role in cell physiology, providing mechanistic insight into the complex interrelations between Ca 2ϩ and redox signaling. These results also provide strong evidence for the importance of mitochondrial Ca 2ϩ export in the regulation of the mitochondrial oxidative metabolism. The present work identifies NCLX as a target for the modulation of redox-dependent processes.