Essential Role of Aralar in the Transduction of Small Ca+ Signals to Neuronal Mitochondria*

Aralar, the neuronal Ca2+-binding mitochondrial aspartate-glutamate carrier, has Ca2+ binding domains facing the extramitochondrial space and functions in the malate-aspartate NADH shuttle (MAS). Here we showed that MAS activity in brain mitochondria is stimulated by extramitochondrial Ca2+ with an S0.5 of 324 nm. By employing primary neuronal cultures from control and aralar-deficient mice and NAD(P)H imaging with two-photon excitation microscopy, we showed that lactate utilization involves a substantial transfer of NAD(P)H to mitochondria in control but not aralar-deficient neurons, in agreement with the lack of MAS activity associated with aralar deficiency. The increase in mitochondrial NAD(P)H was greatly potentiated by large [Ca2+]i signals both in control and aralar-deficient neurons, showing that these large signals activate the Ca2+ uniporter and mitochondrial dehydrogenases but not MAS activity. On the other hand, small [Ca2+]i signals potentiate the increase in mitochondrial NAD(P)H only in control but not in aralar-deficient neurons. We concluded that neuronal MAS activity is selectively activated by small Ca2+ signals that fall below the activation range of the Ca2+ uniporter and plays an essential role in mitochondrial Ca2+ signaling.

Ca 2ϩ signaling in mitochondria is mainly achieved through the entry of Ca 2ϩ across the Ca 2ϩ uniporter, a highly selective Ca 2ϩ channel whose identity remains unknown (1,2). Because of its apparent low affinity for Ca 2ϩ , the fact that there is a substantial mitochondrial Ca 2ϩ uptake despite modest changes in average [Ca 2ϩ ] i is explained by the strategic location of mitochondria close to the endoplasmic reticulum Ca 2ϩ release channels, or plasma membrane Ca 2ϩ channels, and therefore to microdomains of very high [Ca 2ϩ ] i (3)(4)(5)(6).
In mitochondria, Ca 2ϩ activates pyruvate, isocitrate, and ␣-ketoglutarate dehydrogenases, resulting in an increase in the mitochondrial NADH/NAD ratio (7). The Ca 2ϩ uniporter-mitochondrial dehydroge-nases signaling pathway is very relevant for cell and neuronal function (8 -13). However, whether this is the only mechanism to transduce Ca 2ϩ signals to mitochondria has never been assessed. In fact, in studies where changes in NAD(P)H, [Ca 2ϩ ] i , and [Ca 2ϩ ] mit have been all measured in parallel, it was found that mitochondrial NAD(P)H levels did not closely match those of [Ca 2ϩ ] mit (12, 14 -17).
The identification of aralar (aralar1 (18,19)) as the brain isoform of the Ca 2ϩ -dependent aspartate-glutamate mitochondrial carrier (AGC) 5 SLC25A12 (20) opens up the possibility of an alternative way to transduce Ca 2ϩ signals to neuronal mitochondria. Indeed, the Ca 2ϩ binding domains of aralar face the mitochondrial intermembrane space, and increases in [Ca 2ϩ ] i result in the activation of aspartate-glutamate exchange, in the absence of calcium entry in mitochondria (20).
The AGCs are one of the transporters responsible for the malateaspartate NADH shuttle (MAS). Because of the electrogenic nature of Asp/Glu exchange (20,21), the AGC reaction is irreversible under physiological conditions and thus a potential site for regulation. The first aim of this work was to explore the potential of aralar as a brain AGC isoform to regulate MAS activity in brain at low Ca 2ϩ concentrations, below those required for the function of the mitochondrial Ca 2ϩ uniporter. MAS activity in brain mitochondria was found to have Ca 2ϩactivation properties adequate for this purpose.
The second aim of this work was to study the role of the aralar-MAS pathway in the supply of reducing equivalents to neuronal mitochondria. Aralar is expressed postnatally in rat and mouse brain, and it is located in neurons. Both MAS activity and aralar expression are acquired in parallel during neuronal maturation (22,23). Aralar is important for neuronal function as underscored by the finding that alterations in aralar gene and protein are associated with central nervous system diseases such as Mohr-Tranebjaerg syndrome, in which there is an impaired targeting of aralar to mitochondria (24), and autism (25). Aralar null mice also exhibit prominent motor coordination defects along with deficient myelination (26).
We have employed primary neuronal cultures from control and aralar-deficient mice (26) and two-photon excitation microscopy imaging of NAD(P)H to monitor the transfer of reducing equivalents from cytosol to mitochondria. We show that MAS is the main pathway to transfer reducing equivalents to neuronal mitochondria. High [Ca 2ϩ ] i signals activate the Ca 2ϩ uniporter-mitochondrial dehydrogenases signaling pathway, whereas small [Ca 2ϩ ] i signals selectively activate MAS activity in neurons. We conclude that the aralar-MAS pathway plays an essential role to transduce small [Ca 2ϩ ] i signals into neuronal mitochondrial energization.
Genotypes-Genotype was determined by PCR using genomic DNA obtained from tail or embryonic tissue samples (Nucleospin tissue kit, Macherey-Nagel). The following primers were used for genomic DNA amplification: sense primer mAra3Ј LTRF3 (5Ј-GTTCTCTA-GAAACTGCTGAGG-3Ј) that detects only mutated alleles, sense primer mAra int 13F1 (5Ј-GATGTGAGAACTCACCAGTGT-3Ј) that detects wild-type alleles, and antisense primer mAra int 13B (5Ј-AC-CACCACCAGCGTGTCAGC-3Ј) that detects both wild-type and mutant alleles (data not shown). PCR mixtures were preincubated at 94°C for 5 min, followed by 35 cycles of DNA amplification at 94°C for 60 s, 58°C for 60 s, and 72°C for 60 s; the process was finished with an incubation at 72°C for 5 min. Wild-type (271 bp) and mutant (406 bp) fragments were separated by electrophoresis on a 1.5% agarose gel.
Rat skeletal muscle mitochondria were obtained as described by Rolfe et al. (28) and suspended in MSK until used. Respiration rates were determined with a Clark-type oxygen electrode. Skeletal muscle mitochondria (0.1 to 0.2 mg protein) were added to 0.5 ml of MSK, and different Ca 2ϩ concentrations and the respiration rate in the presence of 5 mM glutamate ϩ 5 mM malate ϩ 0.5 mM ADP was recorded. The free Ca 2ϩ concentrations in each media was determined with fura-2 or Ca 2ϩ -Green as indicated below.
Reconstitution of MAS Activity in Brain Mitochondria-The reconstitution of MAS is described in Jalil et al. (26). Briefly, mitochondrial fractions (0.1 to 0.15 mg protein) were suspended in 3 ml of MSK and 100 M digitonin, and the shuttle was reconstituted in the presence of 4 units/ml glutamate-oxaloacetate transaminase, 6 units/ml malate dehydrogenase, 66 M NADH, 5 mM aspartate, 5 mM malate, 0.5 mM ADP, 200 nM ruthenium red, and appropriate CaCl 2 additions. The reaction was started by the addition of 5 mM glutamate, was determined from the decay in NADH fluorescence (excitation at 340 nm, emission at 465 nm), and was calibrated with appropriate NADH standards. The free Ca 2ϩ concentrations at each Ca 2ϩ addition were determined fluorimetrically with fura-2 (below 1 M free Ca 2ϩ ) and Ca 2ϩ -Green (above 1 M free Ca 2ϩ ). The concentrations of fura-2 (K d ϭ 224 nM; excitation, 340 and 380 nm; emission, 510 nm) and Ca 2ϩ -Green (K d ϭ 14 M; excitation, 506 nm; emission, 532 nm) were 5 and 0.1 M, respectively. The free Ca 2ϩ concentration was obtained by established procedures for ratiometric or nonratiometric probes (29).
To test the effect of intramitochondrial Asp on shuttle activity, mitochondria were preloaded with Asp following published procedures (21). To this end, digitonin-permeabilized mitochondrial fractions were first suspended (16 mg of protein/ml) in a glutamate-containing medium (100 mM potassium glutamate, 100 mM sucrose, 10 mM Tris, 0.1% BSA, pH 7.4) for 10 min at 0°C to allow for glutamate accumulation in mitochondria. Then the suspension was diluted 1:10 in 320 mM sucrose, 1 mM EDTA, 0.1% BSA, 10 mM Tris, pH 7.4, and centrifuged (10,000 ϫ g, 5 min). Glutamate-loaded mitochondria were washed with the same medium and resuspended (7 mg of protein/ml) in an Asp-loading medium (150 mM KCl, 10 mM MOPS, 0.1% BSA, pH 7.4). Asp loading was started by adding 3 mM oxaloacetate, and after 3 min at 28°C, the incubation was stopped (0°C, 5 min of centrifugation at 10,000 ϫ g). Asp-loaded mitochondria were washed with Asp-loading medium and resuspended in MSK to assay MAS activity.
Ca 2ϩ uptake in brain and muscle mitochondria preparations was measured in the presence of 0.1 M Ca 2ϩ -Green 5N (excitation 506 nm, emission 532 nm) in modified MSK (without EDTA and supplemented with 1 mM MgCl 2 ). Experiments were performed at 25°C under constant stirring, with ADP (0.2 mM), glutamate (5 mM), and malate (5 mM) as respiratory substrates. Measurements were started by the addition of mitochondria (to a final concentration of 0.5 mg/ml protein; not shown). After 4 min of preincubation, a first addition of Ca 2ϩ (50 M) was made, followed by sequential additions of 20 M Ca 2ϩ as indicated. Where appropriate, 200 nM ruthenium red was added.
Neuronal Cultures-Cortical neuronal cultures were prepared from 16-day-old mouse embryos, following the procedure used for rat embryos (23, 30) modified as follows. Mouse embryos were obtained from crosses between SVJ129/C57BL6 Aralar ϩ/Ϫ mice, and nonbrain tissue was used for determination of DNA genotype of every embryo. Because the number of Aralar ϩ/Ϫ embryos doubled that of Aralar ϩ/ϩ , Aralar ϩ/Ϫ mice were employed as controls, unless indicated otherwise. Cerebral cortices were removed free of meninges, cut into small pieces, and enzymatically dissociated in phosphate-buffered saline containing 1% BSA, 0.4 mg/ml papain, and 6 mM glucose and then mechanically dissociated, in the presence of DNase, by using glass pipettes of different pore size. Dissociated cells were collected by centrifugation (800 ϫ g, 5min) and seeded in medium containing 20% horse serum for 3 h. After this time, medium was completely replaced with serum-free B27-supplemented Neurobasal medium (31). The culture medium was partially replaced every 2nd day. Cells were plated at a density of 1 ϫ 10 5 cells/ cm 2 on poly-L-lysine and laminin-coated pretreated glass coverslips for the determination of [Ca 2ϩ ] i (30), or on glass coverslips sealed at the bottom of plastic wells (4-well LabTek cover slides, NUNC, Roskilde, Denmark), for two-photon microscopy. The different cell types in these neuronal cultures were characterized with specific antibodies as described earlier (30,32). Under the present culture conditions, neurons represented more than 90% of the total cell population. Cultures were used for experimentation between 9 and 11 days in vitro.
Two-photon Microscopy in Neuronal Cultures-Neuronal cultures, seeded on 4-well LabTek cover slides, were washed once and incubated in glucose-free HCSS (1 mM CaCl 2 , 120 mM NaCl, 0.8 mM MgCl 2 , 25 mM Hepes, 5.4 mM KCl, pH 7.4) for 1 h before experimentation. When Ca 2ϩ -free conditions were needed, cells were washed once and maintained with Ca 2ϩ -free HCSS in the presence of 100 M EGTA.
Two-photon excitation microscopy was performed by using Olympus IX70 inverted microscope with an ϫ60 Plan Apo NA 1.4 objective, coupled to a Bio-Rad Radiance 2000 MP confocal/multiphoton microscopy system. An infrared multiphoton laser (Coherent Mira 690 -1000 nm) provided excitation of intrinsic NADH fluorescence. The laser set-tings used for imaging were as follows: 4 milliwatts of power, with a neutral filter set to reduce the excitation light in 80%, and long wave at 735 nm with 150-fs pulses. Images were collected with a 390 -460 nm emission filter. Cells were maintained at 37°C with a temperature-controlled microscopic stage. Additions (0.1 ml) were made as a bolus to the cell chamber (initial volume, 0.5 ml).
Images (averages of three frames, 512 ϫ 512 pixels per frame, 0.4 m/pixel) were taken every 5 or 10 s for about 2-5 min. Image analysis was carried out with MetaMorph (Universal Imaging). Mitochondrial and cytosolic intensities were determined in individual cells or groups of cells following procedures used in islet beta cells (33,34). Because of their movement in and out of the focal plane, a particular mitochondria may not be imaged at all time points. Thus, the analysis was carried out only on those mitochondria that stood in the focal plane at each time point. Selected regions (with 4 -10 cells) or individual cells were outlined, and intensity thresholds were set that highlighted the bright areas shown to correspond to in-focus mitochondria (from Mitotracker labeling, see below). With these thresholds, which may underestimate the mitochondrial fluorescence but avoid contamination from the cytoplasmic compartment, the intensity of these bright areas was calculated. To estimate the cytoplasmic fluorescence, intensity thresholds were set in order to include the dim areas within cells and to exclude mitochondrial fluorescence. This probably entails an overestimation of the cytosol because of the dense neurite network where neurites and background are hard to tell apart.
Changes in NAD(P)H fluorescence were then calculated as mitochondrial/cytosolic (mit/cyt) NAD(P)H ratios or as normalized fluorescence values (cyt/cyt 0 , mit/mit 0 ). By using a more restrictive estimation of the cytosol, mit/cyt NAD(P)H ratios were increased, but the differences between genotypes or stimuli were maintained.
In each case, measurements to determine changes in NAD(P)H autofluorescence were performed in at least three independent experiments. Statistical significance was calculated for every experimental condition (genotype and/or stimuli applied) by using time course data sets of mit/cyt NAD(P)H fluorescence ratios and one-way analysis of variance (ANOVA). Thereafter, time courses for mit/cyt NAD(P)H fluorescence ratios obtained from different genotypes or stimuli were compared by a two-factor ANOVA test. The statistically significant results of the previous comparisons are indicated in the figure legends. In addition, data at the indicated time points were used to calculate the reducing equivalent transfer ratio (quotient between maximum and initial mit/cyt NAD(P)H ratios) and were analyzed for significance (unpaired two-tailed t test or Bonferroni's test).
To label mitochondria with MitoTracker, neurons were first incubated with lactate and malate to reduce mitochondrial NAD(P)H. After capturing the NAD(P)H fluorescence images, they were then incubated for 15 min with 100 nM MitoTracker Red CMXRos (Molecular Probes) on the microscope stage, where confocal images were captured using a He-Ne laser (543 nm excitation) and a 560 -600 nm emission filter (resolution, 0.16 m/pixel).
Pyruvate Determination-Pyruvate content was measured in the neuronal cultures exposed to the same experimental conditions used for two-photon microscopy. To measure pyruvate content, cells were washed once and incubated for 1 h in glucose-free HCSS and were then incubated for 3 min in glucose-and Ca 2ϩ -free HCSS (100 M EGTA) with 20 mM lactate and 5 mM malate. Cells were placed on ice, washed with cold phosphate-buffered saline, and scraped into 0.5 N perchloric acid. Deproteinized cellular pellets obtained in this way were lyophilized. Pyruvate content in these samples were spectrofluorimetrically measured by a standard enzymatic end points method (35).

Determination of [Ca 2ϩ ] i and [Ca 2ϩ ] mit in Neuronal
Cultures-It was determined that [Ca 2ϩ ] i neurons were grown on coverslips and loaded with fura-2AM by incubation in Ca 2ϩ -free HCSS with 5 M fura-2AM and 50 M pluronic acid (Molecular Probes), for 20 min at 37°C, and rinsed with HCSS, 1 mM CaCl 2 , for 30 min. The [Ca 2ϩ ] i was then determined by Ca 2ϩ imaging as described before (30).
To determine [Ca 2ϩ ] mit , cells were loaded with rhod-2 as described previously (15). Cortical neurons were incubated at 4°C for 15 min in HCSS containing 4 M rhod-2 AM. The dye-containing buffer was then replaced by culture medium and incubated overnight at 37°C, under 95% air and 7% CO 2 , to dissipate any cytosolic dye. The combination of cold loading, to decrease the activity of intracellular cytosolic esterases, and the cationic properties of rhod-2 allowed a preferential accumulation of the dye into mitochondria.
Rhod-2-loaded neurons were preincubated exactly under the same conditions used for two-photon microscopy, mounted on the stage of a confocal laser scanning (LSM510META) microscope (Axiovert 200, Zeiss), and imaged using a 40ϫ oil Plan-Neofluar objective (1.3 NA) and an additional digital magnification of 1.2ϫ. Images were obtained upon illumination with a helium-neon laser at 543 nm, and the emitted light was recorded with a 560 nm long-pass filter. Images (averages of 200 frames: 512 ϫ 512 pixels, 192 m ϫ 192 m) were acquired with a confocal slit aperture setting of 72 m, once every 5 s. Confocal images were analyzed with MetaMorph software (Universal Imaging). To study the variations of Ca 2ϩ concentrations in mitochondria in single neurons, individual cells were outlined manually, and the variations in fluorescence were expressed as F/F 0 , where F is the emitted fluorescence at any given time, and F 0 is the fluorescence of the same zone at time ϭ 0.

Ca 2ϩ Activation of the Malate-Aspartate NADH Shuttle in Brain
Mitochondria-MAS activity in brain mitochondria, measured as a decrease in NADH in the incubation medium, was dependent upon Glu addition, and under our assay conditions, it did not require Asp-preloading of mitochondria (Fig. 1, A and B). MAS activity was completely absent in brain mitochondrial fractions from Aralar Ϫ/Ϫ mice (26). In addition, respiration on glutamate plus malate was extremely reduced in skeletal muscle mitochondria with no compensatory increase in respiration with pyruvate plus malate (26). Similarly, the lack of citrin, the liver AGC isoform, is not accompanied by compensatory increases in the activity of the glycerol phosphate NADH shuttle (36).
To study Ca 2ϩ activation of MAS, shuttle activity was measured after the addition of calibrated Ca 2ϩ loads in the presence of 200 nM ruthenium red (RR) to block Ca 2ϩ uptake through the Ca 2ϩ uniporter. Under these conditions, any activation of the shuttle can be attributed to the effect of Ca 2ϩ on the external face of the inner mitochondrial membrane, where the Ca 2ϩ binding domains of aralar are located (20). Fig. 1, C and D, shows that MAS activity increased with Ca 2ϩ . The maximal activation was about 3.1-fold (from 26.7 Ϯ 5.28 to 85.18 Ϯ 10.4 nmol ϫ mg protein Ϫ1 ϫ min Ϫ1 , in the absence of Ca 2ϩ or at saturating Ca 2ϩ concentrations, respectively) with an S 0.5 of 324 Ϯ 114 nM. A similar increase in activity (2.55 Ϯ 0.57-fold) was observed when the assay was carried out with aspartate-preloaded mitochondria. The S 0.5 value for Ca 2ϩ activation is substantially lower than that of the Ca 2ϩ uniporter of mitochondria (apparent K m about 10 -20 M (2, 37)), suggesting that Ca 2ϩ activation of the shuttle may contribute to NADH reduction in mitochondria at Ca 2ϩ concentrations where the Ca 2ϩ uniporter is still inactive.

Two-photon Measurement of Lactate-induced Changes in NAD(P)H in Cultured Neurons-Primary cortical neuronal cultures from
Aralar ϩ/Ϫ mouse embryos used as control were equilibrated with glucose-free HCSS for 1 h and then switched to glucose-and Ca 2ϩ -free HCSS, 100 M EGTA, to start autofluorescence imaging. After acquiring 2 images (one every 10 s), 20 mM lactate ϩ 5 mM malate (LM) was added to the incubation chamber, and fluorescence was recorded for 2-5 min.
The total fluorescence of the neurons (average of 30 cells) increased markedly after LM addition (Fig. 2, A and B), and this increase was not observed when vehicle was added ( Fig. 2A), indicating that it was LMdependent. There were hardly any changes in total, mitochondrial, or cytosolic fluorescence in vehicle-exposed neurons ( Fig. 2A and data not shown), indicating that photobleaching in these experiments was very low.
At the subcellular level, autofluorescence was found to concentrate in bright punctate regions stained with Mitotracker (Fig. 2B) that were presumed to be mitochondria. The LM-induced increases in fluorescence of mitochondria, and the neighboring cytoplasm, are represented in Fig. 2, C-F. As observed, the addition of LM resulted in an increase in the two regions, with individual variation of the responses among cells (Fig. 2, C and E). This was expected because lactate entering the cells via neuronal monocarboxylate carriers (38) would immediately increase the cytosolic NADH/NAD level through the lactate dehydrogenase reaction. However, mitochondria had the greatest increase in NAD(P)H fluorescence (note the different scales). We have not quantified the actual increase in NAD(P)H in mitochondria or cytosol, because it may involve both free but mostly bound NAD(P)H, and the fluorescence of these two forms is very different, about 10-fold higher when bound (39,40).
In order to test that cytosolic and mitochondrial fluorescence signals were adequately separated in the imaging experiments, we studied the response to rotenone, which has selective effects on mitochondrial NAD(P)H. Fig. 2G shows that rotenone addition results in a very large increase in normalized mitochondrial fluorescence, with no changes in cytoplasmic fluorescence, indicating that contamination between mitochondrial and cytosolic signals is below detection in our current set up.
Lack of Increase in Mitochondrial NAD(P)H in Response to Lactate in Aralar-deficient Neurons-To study the role of aralar, the response to LM addition was studied in control and aralar-deficient neuronal cultures derived from littermate embryos in Ca 2ϩ -free media and in the presence of 1 mM Ca 2ϩ . Fig. 3A shows the changes in the mit/cyt NAD(P)H fluorescence ratio in response to LM (added at the arrow) in Ca 2ϩ -free media. The changes in normalized mitochondrial fluorescence ratio are shown in Fig. 3B. Control neurons showed varied responses to LM and reached a plateau at about 1 min after LM addition (Fig. 3A). In contrast, neurons from aralar-deficient mice showed a very modest rise in both the mit/cyt and normalized mitochondrial fluorescence ratio (Fig. 3, A and B). These changes were essentially the same in the presence of Ca 2ϩ . In fact, the reducing equivalent transfer (see 'Materials and Methods') was not statistically different in the presence of 1 mM Ca 2ϩ or in Ca 2ϩ -free media both in control (n ϭ 30 -180 neurons, p Ͼ 0.6) and aralar-deficient neurons (n ϭ 30 -180, p Ͼ 0.8).
[Ca 2ϩ ] i was not modified by LM addition either in the absence or presence of Ca 2ϩ (results not shown), indicating that MAS activity is activated solely by the rise in cytosolic NADH obtained upon lactate entry.
Lactate gives rise to pyruvate through lactate dehydrogenase. Therefore, the increase in mitochondrial NAD(P)H may be due to NADH shuttle activity and to the entry and oxidation of pyruvate in mitochondria. To determine the contribution of lactate-derived pyruvate oxidation in the increase in mitochondrial NAD(P)H, we have studied the responses to pyruvate plus malate in control and aralar-deficient neurons. Two and 10 mM (not shown) pyruvate (plus 5 mM malate) induced concentration-dependent increases in the mit/cyt NAD(P)H fluorescence ratio, which were similar in control and aralar-deficient neurons (Fig. 3C).
Because the responses to pyruvate and the internal pyruvate concentrations observed 2 min after LM addition (nmol/5 ϫ 10 6 cells) were the same in control (0.79 Ϯ 0.09) and aralar-deficient (1.08 Ϯ 0.18) neurons, the blunted response to LM caused by aralar deficiency is clearly due to a decreased transfer of reducing equivalents from cytosolic NADH to mitochondria. Although a small and nonsignificant increase in cytosolic NAD(P)H fluorescence has been observed in aralar-deficient neurons (results not shown), its contribution to the blunted response to LM is probably a minor one, as the changes in mitochondrial/cytosolic or normalized mitochondrial fluorescence ratios are the same (Fig. 3, A  and B).
Reducing equivalent transfer at 70 s after LM addition in Ca 2ϩ -free medium (quotient between mit/cyt NAD(P)H ratios at 70 s and initial values) was 1.76 Ϯ 0.03 in control and 1.3 Ϯ 0.01 in aralar-deficient neurons (mean Ϯ S.E. of 72-180 neurons, p Ͻ 0.0001; ANOVA followed by unpaired two-tailed t test). Similar results were obtained in the presence of 1 mM Ca 2ϩ (reducing equivalent transfer was 1.82 Ϯ 0.11 and 1.29 Ϯ 0.16, in control and aralar-deficient neurons, n ϭ 30 -100 neurons, p Ͻ 0.05, unpaired t test). The contribution to this increase of pyruvate (lactate-derived) oxidation can be estimated from the values in aralar-deficient neurons, and it represents about 40% of the increase, with MAS activity corresponding to the remaining 60%. It is unlikely that glycerol 3-phosphate shuttle activity contributes significantly to this process, because it is relatively small in neurons (41). It donates electrons to complex III, and any change in mitochondrial NAD(P)H level from that activity would depend on reverse electron transport. Thus, MAS activity is a major pathway for lactate-dependent reduction of mitochondrial NAD(P)H in control neurons.

Regulation of Mitochondrial NAD(P)H Response by Large Ca 2ϩ
Signals-By having established that lactate-induced increase in reducing equivalent transfer to mitochondria is mostly due to MAS activity, we sought to determine the role of [Ca 2ϩ ] i in the regulation of shuttle activity.
To this end, we have studied the increase in mitochondrial NAD(P)H in neurons in response to two types of Ca 2ϩ signals as follows: large [Ca 2ϩ ] i signals brought about by depolarization-dependent activation of voltage-operated Ca 2ϩ channels in the presence of millimolar external Ca 2ϩ concentrations, and small [Ca 2ϩ ] i signals, generated in a Ca 2ϩfree medium by Ca 2ϩ mobilization via activation of IP 3 receptors. In the first case, in addition to Ca 2ϩ -mediated MAS activation, it was expected that voltage-operated Ca 2ϩ channel-dependent Ca 2ϩ entry would result in Ca 2ϩ uptake by mitochondria and subsequent dehydrogenase activation (8). In the second case, the small Ca 2ϩ signals generated by agonist addition were expected to remain below the activation threshold of the mitochondrial Ca 2ϩ uniporter but still capable of activating the aralar-MAS pathway. Fig. 4A shows the increase in mitochondrial NAD(P)H obtained when LM was added together with 60 mM KCl and in the presence of 1 mM external Ca 2ϩ in control and aralar-deficient neurons. There was no difference between the increases in [Ca 2ϩ ] i and [Ca 2ϩ ] mit , obtained after LM ϩ KCl addition in control and aralar-deficient neurons, in agreement with the lack of effect of aralar overexpression on Ca 2ϩ homeostatic mechanisms reported earlier (42) (Fig. 4, B and C). The increase in the mit/cyt NAD(P)H fluorescence ratio obtained in the presence of KCl (Fig. 4, D and E) was significantly larger than that obtained by LM addition itself both in control and aralar-deficient neurons (p Յ 0.0001). Reducing equivalent transfer within 40 -70 s after LM or LM ϩ KCl addition was 1.89 Ϯ 0.09 and 3.35 Ϯ 0.35, respectively, in control neurons (n ϭ 30 -100, p Ͻ 0.0001) or 1.34 Ϯ 0.16 and 2.77 Ϯ 0.4 in aralardeficient neurons (n ϭ 30 -100 neurons, p Ͻ 0.0019). Furthermore, the time course of the response to LM ϩ KCl was the same in control and aralar-deficient neurons (p ϭ 0.397).
To verify that the lack of difference between wild-type and aralardeficient neurons was not due to NAD(P)H being maximally reduced in both cases, we compared the increases in mitochondria/cytosol NAD(P)H ratios obtained with LM plus high K ϩ with the NAD(P)H signals in mitochondria obtained in response to rotenone. By inhibiting electron flow through complex I, the addition of 2 M rotenone plus LM induced a progressive increase in reducing equivalent transfer to mitochondria, as observed previously (13). The quotient between final and initial mit/cyt NAD(P)H ratio obtained after 2 min incubation with rotenone was 8.32 Ϯ 0.62 (n ϭ 36 -90 neurons), i.e. substantially larger than those observed with LM plus high K ϩ .
The increase in mitochondrial NAD(P)H obtained with LM plus high K ϩ is expected to result from the additive effects of Ca 2ϩ activation of MAS activity and Ca 2ϩ activation of mitochondrial dehydrogenases. The lack of differences in mitochondrial NAD(P)H increase between wild-type and aralar-deficient neurons under conditions appropriate for Ca 2ϩ uniporter signaling indicates that MAS activity contributes very little, if at all, to this process, suggesting that MAS activity may be inhibited under conditions allowing mitochondrial Ca 2ϩ uptake.
By studying MAS activity in isolated mitochondria in the presence or absence of 200 nM RR (Fig. 4F), it was observed that Ca 2ϩ activation of MAS was abolished if no RR was present. At this RR concentration, Ca 2ϩ uptake in brain mitochondria was fully inhibited (Fig. 4G). These results indicate that Ca 2ϩ activation of MAS is blocked if the Ca 2ϩ uniporter is active and Ca 2ϩ is allowed to enter the mitochondria. We believe that this is because of the effect of mitochondrial Ca 2ϩ on the affinity for ␣-KG of ␣-KGDH. This Krebs cycle dehydrogenase and the ␣-KG-malate carrier (OGC) compete for the substrate ␣-KG (Fig. 1E), so that Ca 2ϩ activation of ␣-KGDH activity results in an increased affinity for ␣-KG and a decrease in ␣-KG efflux from mitochondria that would immediately oppose MAS activation, as shown by O'Donnell et al. (43,44). (45) showed that ATP activation of metabotropic P2Y receptors present in neonatal neurons in a Ca 2ϩ -free medium results in small [Ca 2ϩ ] i transients. To study the effect of small [Ca 2ϩ ] i signals on MAS activity, neurons incubated in Ca 2ϩ -free HCSS plus 100 M EGTA were exposed to LM together with 100 M ATP or different Ca 2ϩ -mobilizing agonists. In our neuronal cultures, ATP-induced [Ca 2ϩ ] i transients were much smaller than those obtained with high K ϩ , with departures of Յ100 nM from resting values that lasted 1 min at most (Fig. 5A); these transients were not accompanied by any detectable increase in [Ca 2ϩ ] mit in control or aralar-deficient neurons (Fig. 5B). Therefore, [Ca 2ϩ ] mit variations are below the detection limit of rhod-2 (dissociation constant (K d )) determined in situ in 1.3 M mitochondria (46), which is somewhat lower than that of fura-2 (K d 0.2 M). However, this small Ca 2ϩ signal resulted in a remarkable potentiation of LM-dependent increase in mitochondrial NAD(P)H fluorescence in control neurons (Fig. 5, C and D), which was notably absent in aralardeficient neurons (Fig. 5, C and E). In control neurons, reducing equivalent transfer at 80 s after LM ϩ ATP addition was 2.88 Ϯ 0.076, and derived from littermate embryos were preincubated in glucose-free HCSS, 1 mM CaCl 2 for 1 h at 37°C, switched to Ca 2ϩ -and glucose-free HCSS (100 M EGTA), and exposed to 20 mM lactate plus 5 mM malate (LM) (added at the arrow). The changes in mitochondrial/cytosolic NAD(P)H fluorescence were computed within 18 regions containing 4 -10 neurons/region. In controls, but not in aralar-deficient neurons, mit/cyt fluorescence ratios were greater than that at t ϭ 0 (p Ͻ 0.0001, one-way ANOVA; from 60 s after LM addition onward, p Ͻ 0.05, Bonferroni's test). Time course data sets of mit/cyt NAD(P)H ratios from control and aralar-deficient neurons were compared (p Ͻ 0.0001, two-factor factorial ANOVA). B, time course data sets of normalized mitochondrial (mit/mit 0 ) fluorescence ratios from control and aralar-deficient neurons were also analyzed (p Ͻ 0.0001, two-factor factorial ANOVA). C, aralar-deficient and control neuronal cultures derived from littermate embryos were preincubated as above and exposed to 2 mM pyruvate ϩ 5 mM malate (PM) (squares) (added at the arrow), and the changes in mitochondrial/cytosolic NAD(P)H fluorescence ratios were computed.  Because lactate-derived pyruvate contributes to NAD(P)H generation, we have tested whether the effects of ATP involve Ca 2ϩ -stimulated pyruvate metabolism rather than Ca 2ϩ -stimulated MAS activity. Fig. 5F shows that the small [Ca 2ϩ ] i signals triggered by ATP had no effect on the increase in mitochondrial NAD(P)H induced by pyruvate. Therefore, these results clearly showed that the neuronal aralar-MAS pathway is selectively activated by small Ca 2ϩ signals, below the threshold for Ca 2ϩ uniporter activation.

DISCUSSION
Our results show that brain mitochondria have a Ca 2ϩ -dependent malate-aspartate NADH shuttle activated by low extramitochondrial Ca 2ϩ concentrations, with an S 0.5 for Ca 2ϩ of 324 nM. In neurons, this property is exploited to transduce Ca 2ϩ signals to mitochondria via the aralar-MAS pathway, by transferring NAD(P)H reducing equivalents from the cytosol. Quite remarkably, we find that small [Ca 2ϩ ] i signals activate MAS activity under conditions where Ca 2ϩ uptake in mitochondria is not observed. Moreover, these same [Ca 2ϩ ] i signals do not activate the pyruvate-dependent reduction of NAD(P)H in mitochondria, a process dependent on the activity of the Ca 2ϩ uniporter. We have observed that synaptosomal mitochondria start taking up Ca 2ϩ at a global [Ca 2ϩ ] i of about 350 -400 nM (27). Clearly, Ca 2ϩ signaling through the aralar-MAS pathway in neuronal mitochondria, with an S 0.5 of 324 nM, would be only significant below these Ca 2ϩ concentrations, which is exactly what we found. Our results show for the first time that mitochondria respond to [Ca 2ϩ ] i signals with a substantial increase in NAD(P)H through a Ca 2ϩ uniporter-independent pathway.
The Ca 2ϩ uniporter-dehydrogenase signaling pathway results in a large activation of mitochondrial dehydrogenases (Ca 2ϩ activation of about 4 -5-fold for pyruvate dehydrogenase and ␣-KGDH) (47,48). However, Ca 2ϩ activation of MAS is smaller (3-fold activation in brain mitochondria). Moreover, in the mitochondrial matrix three dehydrogenases are activated by Ca 2ϩ , whereas only one (malate dehydrogenase) is activated by the aralar-MAS pathway. It was very surprising to find that the magnitude of the response via the aralar-MAS pathway is very large in neurons, in fact not far from that evoked by the activity of the Ca 2ϩ uniporter-dehydrogenases pathway.
Ca 2ϩ Activation of Malate-Aspartate NADH Shuttle in Neurons-Lactate is produced by astrocytes and taken up by neurons that use it as an oxidizable substrate, particularly during periods of high activity (49 -52). When neurons are supplied with lactate, cytosolic NAD(P)H levels increase, and reducing equivalents are immediately transferred to mitochondria via the aralar-MAS pathway. In the absence of aralar, mitochondrial NAD(P)H levels show a very small increase, which may be attributed to mitochondrial metabolism of lactate-derived pyruvate. Thus, during lactate utilization in resting conditions, most of the transfer of reducing equivalents to mitochondria (about 60%) is carried out by MAS activity.
Small [Ca 2ϩ ] i signals evoked by agonists that, by themselves, do not stimulate mitochondrial Ca 2ϩ uptake and dehydrogenase activity selec-tively activate the aralar-MAS pathway so that reducing equivalent transfer to mitochondria is now strongly potentiated. ATP and carbachol applied in Ca 2ϩ -free media behave this way in neurons and result in about 3-fold MAS activation (the difference between reducing equivalent transfer values in control and aralar-deficient neurons, at 50 -70 s after LM addition in the absence or presence of ATP). This activation is close to the maximal activation of MAS observed in isolated mitochondria (about 3-fold). However, the ATP-evoked [Ca 2ϩ ] i transients only reach 200 nM at best, clearly below the Ca 2ϩ concentrations required to fully activate MAS in brain mitochondria (about 1 M; Fig. 1). This implies a close proximity between mitochondrial aralar and the IP 3 receptors, the Ca 2ϩ release sites activated by ATP, so that localized Ca 2ϩ concentrations, higher than those of the bulk cytosol, could exist and result in full activation of MAS. However, these local Ca 2ϩ concentrations would still be unable to activate the Ca 2ϩ uniporter-dehydrogenase pathway.
Neurotransmitters (glutamate and acetylcholine) and/or neuromodulators (histamine, adenosine, and neurotrophin-3) acting on G-protein-coupled receptors or Trk receptors activate phospholipase C isoenzymes resulting in different dynamics of IP 3 production and diverse [Ca 2ϩ ] i transients (53)(54)(55)(56). This variability may arise from the engagement of different G-proteins and phospholipase C isoforms in a cell-specific manner. IP 3 -dependent Ca 2ϩ mobilization is critically involved in neuromodulation, as it controls excitability, transmitter release, and gene expression (53)(54)(55) in a cell-and receptor-specific way. Agonists that evoke IP 3 production may produce quite different [Ca 2ϩ ] i and [Ca 2ϩ ] mit transients, depending on the specific receptor being activated and agonist concentration (57,56), and some may produce small [Ca 2ϩ ] i transients that selectively activate the aralar-MAS pathway in neuronal mitochondria. The energization obtained by Ca 2ϩ activation of MAS can be utilized to produce more ATP in mitochondria, as found in studies with aralar-overexpressing cells (42). Indeed, preliminary data indicate that lactate supply during mild hypoglycemia increases ATP levels in neurons and even more so in the presence of nonhydrolyzable ATP agonists, but none of these effects were observed in aralar-deficient neurons. 6 Further consequences of this energization have yet to be explored, but it is tempting to suggest that it could prime mitochondria for subsequent energy-requiring processes such as metabolite transport, etc., that may be involved in the final response to the neuromodulator.
Activity-dependent changes in the NAD(P)/NAD(P)H ratio have been detected long ago in different cell types (58). In neurons and synaptosomes, high K ϩ depolarization leads to an initial transient drop in NAD(P)H that is followed by an increase in NAD(P)H (8). Similar changes are produced by neural activity (12,13,52,59,60). These changes have been attributed to physiological increases in [Ca 2ϩ ] i and Ca 2ϩ (and possibly ADP) entry in mitochondria leading to the following: 1) increased respiration and mitochondrial NADH oxidation, and 2) Ca 2ϩ activation of mitochondrial dehydrogenases resulting in NADH reduction, whereas Shuttleworth et al. (13) found that they were Ca 2ϩindependent. In our experimental conditions, the transient decrease in mitochondrial NAD(P)H was not observed, possibly because the imaging rate was too slow, and lactate was added together with high K ϩ . However, a rapid increase in mitochondrial NAD(P)H in response to high K ϩ stimulation was clearly shown, and this was not modified by aralar deficiency, in agreement with the notion that it involves the Ca 2ϩ uniporter signaling pathway. Indeed, our results show that MAS activity in neuronal mitochondria is superseded under conditions where high [Ca 2ϩ ] i signals drive Ca 2ϩ uniporter activity. This is probably because of competition for substrate between ␣-KGDH of Krebs cycle and the ␣-KG-malate transporter (OGC) of MAS. The two reactions compete for ␣-KG by virtue of their apparent K m values. The K m value for ␣-KG of the OGC on the matrix side of the carrier is 1.5 mM (61) and that of ␣-KGDH is around 0.2 mM (47). Activation by Ca 2ϩ of ␣-KGDH increases its affinity for ␣-KG (7,47), causing a reduction in OGC activity (43) and an inhibition of MAS activity (44). Although this effect is present within the whole time window of our experiments (about 1.5 min), it could disappear when the activation by Ca 2ϩ of ␣-KGDH comes to an end. Thus, it is possible that after the decay of [Ca 2ϩ ] mit transients, MAS activity could prolong the increase in mitochondrial NAD(P)H induced by high [Ca 2ϩ ] i signals and thus contribute to ATP synthesis and recovery of the resting state, conditions that rely on neuronal lactate utilization (62).
It should be noted that the increase in mitochondrial NAD(P)H obtained through the aralar-MAS pathway does not require the presence of Ca 2ϩ in the mitochondrial matrix, an event that contributes to cell death in a number of cell types (63), including neurons (64). It may also provide an alternative mechanism to supply mitochondrial NADH under oxidative stress conditions, where ␣-KGDH is selectively inhibited (65). Lactate, and not glucose, is the major neuronal energy substrate after an insult, and overexpression of a lactate transporter in neurons enhances neuronal resistance to excitotoxicity (66), suggesting that the aralar-MAS pathway might also enhance survival to glutamate excitotoxicity.