Calcium Signaling in Brain Mitochondria

Ca2+ signaling in mitochondria has been mainly attributed to Ca2+ entry to the matrix through the Ca2+ uniporter and activation of mitochondrial matrix dehydrogenases. However, mitochondria can also sense increases in cytosolic Ca2+ through a mechanism that involves the aspartate-glutamate carriers, extramitochondrial Ca2+ activation of the NADH malate-aspartate shuttle (MAS). Both pathways are linked through the shared substrate α-ketoglutarate (αKG). Here we have studied the interplay between the two pathways under conditions of Ca2+ activation. We show that αKG becomes limiting when Ca2+ enters in brain or heart mitochondria, but not liver mitochondria, resulting in a drop in αKG efflux through the oxoglutarate carrier and in a drop in MAS activity. Inhibition of αKG efflux and MAS activity by matrix Ca2+ in brain mitochondria was fully reversible upon Ca2+ efflux. Because of their differences in cytosolic calcium concentration requirements, the MAS and Ca2+ uniporter-mitochondrial dehydrogenase pathways are probably sequentially activated during a Ca2+ transient, and the inhibition of MAS at the center of the transient may provide an explanation for part of the increase in lactate observed in the stimulated brain in vivo.

mate carriers (AGCs) and not the CaU (2)(3)(4)(5). Aralar (Slc25a12), also named aralar1, the AGC isoform with predominant expression in brain (6 -8), is a component of the NADH malateaspartate shuttle (MAS), which in brain is activated by extramitochondrial Ca 2ϩ (S 0.5 324 nM) (4). Immunocytochemistry and in situ hybridization data and mRNA levels in acutely isolated brain cells indicate that aralar is localized preferentially in neurons (8 -12). This is consistent with a higher MAS activity in neuronal than astrocyte cultures (8) and with aralar being one of the more enriched proteins during differentiation of P19 cells to a neuronal phenotype (13). It is also consistent with the higher levels of aralar in total than in synaptosome-free mitochondrial fractions (9).
Studies in cultured neurons, which have aralar as only AGC isoform, showed that small Ca 2ϩ signals that have limited access to mitochondria are able to activate the aralar-MAS pathway (4). However, large Ca 2ϩ signals that induce robust mitochondrial Ca 2ϩ transients fail to activate the pathway (4). This suggested that in neuronal mitochondria the aralar-MAS pathway is inhibited under conditions in which the CaU-mitDH pathway is activated. This surprising result indicates that Ca 2ϩ activation of the malate aspartate shuttle and tricarboxylic acid cycle activity are somehow mutually exclusive in neurons. We have now studied the interplay between the AGG-MAS and CaU-mitDH pathways in brain mitochondria under Ca 2ϩ -stimulation conditions. Our results show that the shared metabolite ␣KG controls the relationship between the second transporter of MAS, the oxoglutarate carrier (OGC, Slc25a11), and ␣KGDH in the Krebs cycle, by virtue of the effects of Ca 2ϩ on the kinetics of ␣KGDH. Interestingly, the inhibition of OGC and MAS is fully reversible, in parallel with Ca 2ϩ egress from mitochondria.
Behaviorally evoked brain activation results in an increased cerebral blood flow and increased glucose utilization, but paradoxically, this is not accompanied with an equivalent increase in oxygen utilization (14,15). As a consequence, the oxygen glucose index, which is close to 6 when glucose is fully oxidized in resting conditions, falls to about 5 (16). This is accompanied by an increase in brain lactate production (14, 16 -19). Our results suggest that MAS inhibition during Ca 2ϩ -induced Krebs cycle activation would drive pyruvate to lactate formation and may play a role in lactate formation during brain activation.

EXPERIMENTAL PROCEDURES
Animals and Materials-3-Month-old C57BL/6xSv129 mice were housed with a 12-h light cycle and fed ad libitum on standard chow. Animals were sacrificed by cervical dislocation, and the tissue of interest was quickly dissected and kept in ice-cold media for mitochondrial isolation, which was carried out at 4°C. All animal procedures were approved by European guidelines. All reagents were obtained from Sigma, except malate dehydrogenase (Roche Applied Science), Fura-2, Calcium-Green5N (Molecular Probes, Eugene, OR), and CGP-37157 (Tocris Biosciences, Ellisville, MO).
MAS Reconstitution and mitGPDH Measurement-MAS was reconstituted following published methods (24 -26), with modifications (4,23). Briefly, mitochondria (0.1-0.15 mg of brain and liver and 0.020 -0.030 mg of heart) were resuspended in 3 ml of MSK(with 100 M digitonin for brain preparations), and the shuttle was reconstituted in the presence of 4 units/ml aspartate aminotransferase, 6 units/ml malate dehydrogenase, 66 M NADH, 5 mM aspartate, 5 mM malate, 0.5 mM ADP. When appropriate, 200 nM Ruthenium Red (RR) and calibrated calcium concentrations were present. mitGPDH activity was measured in a similar way in MSK, with 0.5 mM ADP, 66 M NADH, and 2 units/ml glycerol-3-phosphate dehydrogenase (27). Once a base line is achieved, either 5 mM glutamate (MAS) or 5 mM glycerol 3-phosphate (mitGPDH) was added to trigger shuttle activity, coupled to the decrease in NADH fluorescence (excitation 340 nm, excitation 465 nm). All assays were performed at 37°C under constant stirring.
Measurement of OGC Activity-Transport of ␣-KG through the OGC was measured by a modification of the MAS reconstitution system. Appropriate amounts of mitochondrial protein of the tissue of interest were resuspended in 3 ml of MSK supplemented with 66 M NADH, 0.5 mM ADP, 3 mM (NH 4 ) 2 SO 4 , 5 mM glutamate, 10 units/ml GDH (and 100 M digitonin when brain mitochondria were assayed). 5 mM malate addition initiated ␣KG efflux from mitochondria, which was followed by a decrease in NADH fluorescence. Experiments were performed at 37°C under constant stirring. When indicated, 200 nM RR and calibrated calcium additions were made.
Free Calcium Calibration-The free calcium concentrations in the assays obtained after the different CaCl 2 additions were determined in the presence of Fura-2 (under 1 M, K d ϭ 224 nM; excitation, 340 and 380 nm; emission, 510 nm; concentration 5 M) or Calcium-Green5N (over 1 M, K d ϭ 14 M; excitation, 506 nm; emission, 532 nm; concentration 0.1 M). Free calcium concentration was calculated as established for ratiometric and nonratiometric probes (28,29).
Enzymatic Assays-Enzyme activities were assayed in either freshly isolated or freeze-thawed mitochondria from different tissues of 3-month-old C57BL/6xSv129 mice. Experiments were performed in a final volume of 200 l and 20 -50 g of protein in a FLUOstar optima (BMG-Labtech) plate fluorimeter at 30°C. Activities were calculated as the initial slope of NADH fluorescence changes per mg of protein. An NADH cal-ibration curve was performed daily by adding known amounts of NADH to medium. The following additions were made to the medium (MME: 50 mM KCl, 10 mM HEPES, pH 7.4; 0.2% Triton, and 10 M rotenone (32): 0.2 mM NAD ϩ for measurement of GDH; 0.2 mM NADH, 5 units/ml malate dehydrogenase, and 12.5 mM ␣KG for aspartate aminotransferase measurement; 0.2 mM NAD ϩ , 0.3 mM thiamine pyrophosphate, 10 M CaCl 2 , 0.2 mM MgCl 2 , 0.14 mM CoASH, and 0.5 mM ADP to assay ␣KGDH. After establishment of a base line, activity was triggered with the appropriate substrate as follows: 5 mM glutamate or aspartate and 12.5 mM ␣KG (GDH, aspartate aminotransferase, and ␣KGDH, respectively (30 -33).
Mitochondrial Membrane Potential-Mitochondrial membrane potential was estimated by quenching of tetramethylrhodamine methyl ester (TMRM, 549 nm excitation, 575 nm emission) fluorescence, as described previously (34). Briefly, after stabilization of TMRM (300 nM) fluorescence, 100 g of mitochondria were added to 2 ml of MSK containing 5 mM aspartate. Changes in membrane potential were followed qualitatively by the variations of the TMRM fluorescence after addition of substrates (glutamate/malate, 5 mM), ADP (0.5 mM), and calcium (5 M free calcium). When indicated, 50 mM Na ϩ replaced 50 mM K ϩ in MSK (MSKNa medium). A calcium unbuffered medium (MSK or MSKNa without EDTA) was also used.

Regulation of MAS and Glycerol-3-P Dehydrogenase
Activity by Extramitochondrial and Intramitochondrial Ca 2ϩ -MAS activity was shown to increase in response to extramitochondrial calcium in brain and heart mitochondria (3,4,36). Paradoxically, it became inhibited in a calcium-dependent manner when calcium was allowed to enter the mitochondria through the Ca 2ϩ uniporter (4). In brain mitochondria, the remaining activity drops to 79 Ϯ 4%, 58 Ϯ 7%, and 17 Ϯ 3% in the presence of 0, 0.3, and 5 M [Ca 2ϩ ] free , respectively (Fig. 1A). MAS inhibition by intramitochondrial Ca 2ϩ is not as large in heart mitochondria, with 77 Ϯ 6% residual activity at 5 M free calcium (Fig. 1B). In contrast, MAS activity in liver mitochondria is not affected by intramitochondrial Ca 2ϩ , as its activity is the same in the absence or presence of RR, although it is activated by extramitochondrial calcium to a smaller extent than in brain or heart (Fig. 1C) as reported earlier (36).
It is unlikely that the inhibitory effects of matrix Ca 2ϩ on MAS activity are because of Ca 2ϩ -induced permeability transition pore opening and loss of mitochondrial metabolites (particularly NAD ϩ ) as follows: first, because assays were con-ducted in the presence of 0.5 mM ADP, which is a known inhibitor of permeability transition pore opening (37); second, because inhibition of MAS by Ca 2ϩ matrix was not observed in liver mitochondria (Fig. 1C), which retains half the amount of Ca 2ϩ than brain mitochondria before undergoing permeability transition (38); and third, because inhibition of MAS by matrix Ca 2ϩ in brain and heart mitochondria is still observed in the presence of 5 M cyclosporin A (results not shown).
We also investigated the effect of extra-and intramitochondrial Ca 2ϩ on mitochondrial glycerol-3-phosphate dehydrogenase (mitGPDH), the mitochondrial component of the glycerol 3-phosphate shuttle. The activity of this enzyme is known to be activated by low concentrations of extramitochondrial calcium (Ϸ35-100 nM (39) through a decrease in the K m value for glycerol 3-phosphate of the mitGPDH (40,41). Unlike MAS, mitGPDH was completely Ca 2ϩ -dependent, with no detectable activity in the total absence of free Ca 2ϩ in brain and liver mitochondria (supplemental Fig. 1). The activity of mitGPDH in heart mitochondria was not detectable under the conditions of this study, in agreement with the data of Scholz et al. (42,43) who found a marked decrease of glycerol 3-phosphate shuttle activity in the heart with development. The enzyme requires Ca 2ϩ in the intermembrane space, as full activation is obtained in the presence of RR in all tissues. The kinetics of activation agrees with a low S 0.5 for activation, as the maximal activity was attained at 0.3 M free Ca 2ϩ (supplemental Fig. 1). On the other hand, Ca 2ϩ entry in the matrix (i.e. RR not present in the assay) did not affect mitGPDH activity in either brain or liver mitochondria (supplemental Fig. 1).
Therefore, the results indicate that in brain mitochondria Ca 2ϩ activation of MAS, but not mitGPDH, is prevented by strong Ca 2ϩ signals that reach mitochondrial matrix. This suggests that reducing equivalents transfer to mitochondria may still proceed through mitGPDH when Ca 2ϩ entry in mitochondria leads to an inhibition of MAS activity. This depends on whether a glycerol-P shuttle is actually present in brain, with its two enzymes within the same cell, as suggested to be the case from the astrocyte and neuronal transcriptome findings (12, 44 -46).
Effect of Extra-and Intramitochondrial Calcium on OGC Activity-The activity of the OGC was studied by measuring the efflux of ␣KG from mitochondria, in medium containing glutamate dehydrogenase (GDH). GDH will transform ␣KG into glutamate in the presence of NADH and ammonium, resulting in a decrease in NADH fluorescence, when malate is added to trigger ␣KG efflux from mitochondria (supplemental Fig. 2). The supplemental Fig. 2B shows that the decrease of NADH fluorescence was dependent on the presence of ammonium and was prevented by the OGC inhibitor phenylsuccinate (45,47).
To test the effects of both extra-and intramitochondrial calcium on OGC activity, ␣KG efflux was measured in the absence or presence of the calcium uniporter inhibitor RR, both in Ca 2ϩ -free medium or in the presence of 5 M free Ca 2ϩ . In the presence of 200 nM RR, there was no difference in ␣KG efflux at the two calcium concentrations in brain, heart and liver mitochondria (Table 1), indicating that the OGC is not activated by extramitochondrial Ca 2ϩ , in accordance to the lack of calciumbinding motifs in the OGC sequence (48).
Having shown that extramitochondrial Ca 2ϩ does not modify the OGC activity, the influence of intramitochondrial Ca 2ϩ was studied by carrying out the same experiments in the absence of RR, when the Ca 2ϩ uniporter is not inhibited ( Table  1). Under these conditions the addition of Ca 2ϩ results in a decrease of ␣KG efflux in mitochondria from brain and heart (to 37 Ϯ 5% and 78 Ϯ 10% residual activity, respectively (Table  1)), indicating that the OGC in these tissues is inhibited by intramitochondrial Ca 2ϩ . In liver, the absence of RR resulted in an inhibition of ␣KG efflux both in the absence or presence of Ca 2ϩ (Table 1), clearly a nonspecific effect, unrelated to Ca 2ϩ entry in mitochondria, which was not studied any further. Therefore, inhibition of ␣KG efflux by mitochondrial Ca 2ϩ may explain the inhibition of MAS in brain and heart mitochondria.

TABLE 1 Effect of intra-and extramitochondrial calcium on the activity of the OGC
␣KG efflux was measured in brain, heart, and liver mitochondria from 3-month-old C57BL/6xSv129 mice at 0 (i.e. under the Fura2 limit detection, Ͻ20 nM) and 5 M free calcium, in the presence (RR present)  Influence of Mitochondrial ␣KG Sources and Utilization Pathways on ␣KG Efflux Along the OGC-OGC and ␣KGDH have a common substrate, ␣KG. When mitochondrial Ca 2ϩ increases, it activates ␣KGDH, which causes a decrease in the K m value for its substrate, ␣KG, and a decrease in mitochondrial ␣KG levels (49,50). Our working hypothesis is that this leads to the decrease in OGC activity that we have shown above in brain and heart mitochondria. As OGC is a member of MAS, this may explain the decrease of MAS activity in brain and heart mitochondria when calcium is allowed to enter the organelle (see Fig. 1A and supplemental Fig. 2A).
We figured that a high OGC activity with respect to ␣KGDH would make the OGC relatively independent of ␣KGDH activity, and this could be a possible explanation for the lack of effect of intramitochondrial Ca 2ϩ on liver MAS and OGC activities. However, Table 2 shows that this is not the case. Measurements of ␣KGDH activity in mitochondria from different tissues showed that the ratio of ␣KG efflux to ␣KGDH activity is similar in liver and brain mitochondria.
A second possibility is that liver mitochondria OGC could utilize an additional source of ␣KG making it independent of ␣KGDH. The source of ␣KG in the OGC assays is external glutamate, which enters mitochondria through the AGC and is transaminated with oxaloacetate by mitochondrial aspartate aminotransferase (see supplemental Fig. 2A). The transaminase can be inhibited by 5 mM aminooxyacetate (AOAA) (45), which leads to a complete block of ␣KG efflux in brain and heart mitochondria (supplemental Fig. 3, A and B). However, this is not the case in liver mitochondria, where AOAA only decreased the efflux to 62 Ϯ 15% in the presence of RR and 5 M calcium, but it had no effect in the absence of RR and 5 M calcium (supplemental Fig. 3C).
The lack of effect of AOAA on ␣KG efflux in liver mitochondria points to an extra(s) source(s) of ␣KG in this tissue. This is probably glutamate dehydrogenase (GDH), which in liver has the highest activity among the tissues explored in this study ( Table 2).
Inhibition of OGC and MAS Activity in Brain Mitochondria Is Reversible-We have shown that Ca 2ϩ entry in heart and brain mitochondria prevents Ca 2ϩ activation of MAS activity through an inhibition of ␣KG efflux along the OGC because of the Ca 2ϩ activation of ␣KGDH, a situation that does not take place in liver. In these two types of mitochondria, the maximal activity of the malate-aspartate shuttle (limited by the AGC) is only slightly lower than that of the OGC (see Table 2), as has also been observed in rat brain mitochondria using an entirely different method (9). This makes MAS activity very sensitive to the decrease of the OGC activity brought about by matrix Ca 2ϩ . Because Ca 2ϩ activation of ␣KGDH should last as long as Ca 2ϩ remains in the mitochondrial matrix, we have studied whether net Ca 2ϩ efflux from mitochondria allows reactivation of MAS.
To this end, we have first allowed mitochondria to take up Ca 2ϩ , while carrying out MAS activity during 3 min, and then Ca 2ϩ uptake was stopped by addition of RR, and mitochondria were allowed to release the accumulated Ca 2ϩ along RR-insensitive Ca 2ϩ efflux pathways. The recovery of MAS and OGC activities was studied at the same time (Fig. 2). MAS activity was clearly inhibited after the incubation in the presence of Ca 2ϩ (compare slopes of traces a and b in Fig. 2A). As observed in Fig.  2A, RR addition results in a reactivation of MAS, which is almost complete 10 min after RR addition ( Fig. 2A, compare final slopes of traces a and c, and Fig. 2B). ␣KG efflux along the OGC was also re-activated upon Ca 2ϩ efflux, as expected, although its activity was not fully recovered within this time period (Fig. 2C, recovery after 10 min addition of RR).

Enhanced Recovery of OGC and MAS Activity in Brain Mitochondria by NCX-mediated Ca 2ϩ Efflux-
The previous experiments were performed in a sodium-free medium (MSK), in which calcium efflux proceeds mainly through the Na ϩ -independent Ca 2ϩ efflux pathway, mitochondrial Na ϩ -independent/ Ca 2ϩ exchanger (51-53), or a putative low conductance form of the permeability transition pore (54 -56). As the main mitochondrial calcium efflux pathway in brain is the Na ϩ /Ca 2ϩ exchanger, NCX (52, 57, 58), we have studied the recovery of MAS activity in a medium with sodium (MSK with 50 mM NaCl

TABLE 2 Enzyme and transporter activities involved in ␣KG production and utilization in mitochondria using glutamate and malate as substrates
The results are mean Ϯ S.E. of four independent experiments performed at least in duplicate. ␣KG production was computed as the sum of AGC-MAS and GDH, whereas ␣KG utilization was the sum of ␣KGDH and OGC. Differences (Bonferroni test) with liver (*, p Ͻ 0.05; **, p Ͻ 0.005) or heart (***, p Ͻ 0.005) are indicated.

Brain
Heart Liver replacing 50 mM KCl; MSKNa). Unexpectedly, we found that MAS activity was more inhibited by matrix Ca 2ϩ in a Na ϩcontaining than in a Na ϩ -free medium, and recovery of MAS activity after RR addition was not complete (supplemental Fig. 4). In contrast, inhibition of ␣KG efflux through the OGC by intramitochondrial calcium is not so prominent in MSKNa, and its recovery induced by RR addition is much larger (Fig. 3C). These results indicate that failure to recover MAS activity in Na ϩ medium is not because of a failure to recover OGC activity. On the contrary, OGC activity is not as inhibited in Na ϩcontaining as it is in Na ϩ -free media, probably because the free calcium levels in mitochondria (and ␣KGDH activity) are lower when NCX is active, and Ca 2ϩ efflux and deactivation of ␣KGDH are faster. A possible explanation for the paradoxical effects of Na ϩ in the recovery of MAS activity is based on the steep dependence of AGC on the proton electrochemical gradient. The transport of glutamate plus H ϩ against aspartate is electrogenic (3, 59 -61), and the mitochondrial membrane potential may be dissipated by Ca 2ϩ ϩ Na ϩ cycling, with Ca 2ϩ entry along the uniporter, and exit through the NCX coupled to efflux of Na ϩ back to the cytosol through Na ϩ /H ϩ exchange (62,63). This paradoxical effect of Na ϩ would not affect the OGC, which is electroneutral and thus independent upon ⌬p (64,65).
Indeed, we found that mitochondria were partially depolarized when incubated in the presence of 50 mM NaCl (MSKNa; supplemental Fig. 4B). However, depolarization was not Ca 2ϩ -dependent and consequently not due to Na ϩ and Ca 2ϩ cycling. It was possibly caused by Na ϩ influx through a channel in the mitochondrial membrane, which is inhibited by Mg 2ϩ (66), and consequently a complete depolarization was obtained in the presence of Mg 2ϩ (supplemental Fig. 4C). Under these conditions RR recovery of MAS activity was almost complete (supplemental Fig. 4A).
To verify the relationship between matrix Ca 2ϩ and Ca 2ϩ inhibition of MAS, we have next studied MAS recovery and Ca 2ϩ efflux upon RR addition in parallel. To this end, Ca 2ϩ fluxes were determined through the changes in extramitochondrial Ca 2ϩ levels measured with Calcium Green in a medium with no calcium buffers and with traces of Mg 2ϩ (i.e. MSK or MSKNa without EDTA). There were no differences in the mitochondrial membrane potential generated in the presence of glutamate ϩ malate when mitochondria were incubated in Ca 2ϩ -unbuffered media in the absence (MSK without EDTA) or presence (MSKNa without EDTA) of Na ϩ (compare traces in A1 and B1 in Fig. 3). After the addition of Ca 2ϩ (40 nmol/mg protein, giving 5 M free Ca 2ϩ ), most of it was taken up by energized mitochondria. However, the total amount of Ca 2ϩ taken up after glutamate addition was higher in the absence than in the presence of Na ϩ (about 41 Ϯ 1.6 versus 26.6 Ϯ 3.7 nmol/mg protein in MSK and MSKNa, respectively), as observed previously with rat brain mitochondria (58,67). Interestingly, MAS activity measured in these two conditions varied inversely with the accumulated Ca 2ϩ . It was strongly inhibited (about 50%) in the absence of Na ϩ (Fig. 3, A3  and A4), but much less so in the presence of Na ϩ (about 20%) (Fig. 3, B3 and B4).
Ca 2ϩ efflux triggered by RR addition was strongly increased in a Na ϩ medium (182 Ϯ 10 and 22.5 Ϯ 8 nmol/min/mg initial speed in MSKNa and MSK, respectively), and after 10 min the amount of Ca 2ϩ released was 24 or 66% of the calcium taken up in MSK and MSKNa, respectively (compare traces A2 and B2 in Fig. 3). This difference was abolished in the presence of the NCX inhibitor CGP-37157 (results not shown and see Ref. 35). Under these conditions MAS activity fully recovered control levels in Na ϩ media (Fig. 3B4), but recovery was not complete in the absence of Na ϩ (Fig. 3A4). The loss of the inhibitory effect of matrix Ca 2ϩ on MAS was clearly because of stimulated Ca 2ϩ efflux along the NCX because the effect of external Na ϩ was abolished in the presence of the NCX inhibitor (Fig. 3B4).

Tissue-specific Effects of Matrix
Ca 2ϩ on MAS Activity-Activation by Ca 2ϩ of mitochondrial metabolism is mainly due to Ca 2ϩ entry to the matrix through the calcium uniporter and activation of three matrix dehydrogenases (pyruvate dehydrogenase, ␣KGDH, and isocitrate dehydrogenase) (1,5,68). In addition, Ca 2ϩ binding to aralar on the external side of the inner mitochondrial membrane activates MAS (4,36) and explains the activation of respiration of brain mitochondria by extramitochondrial Ca 2ϩ when malate plus glutamate are used as substrates (69). In this study we report that the effect of matrix Ca 2ϩ on ␣KGDH results in an inhibition of MAS activity, at least in two tissues, brain and heart, but not in liver. We conclude that MAS inhibition is because of matrix Ca 2ϩ activation of ␣KGDH and consequent inhibition of OGC, as suggested for heart mitochondria (70,71). Both activities have similar K m value for ␣KG (1.5 and 2.1 mM for the OGC and ␣KGDH, respectively (49,72)). Upon entry through the uniporter, Ca 2ϩ stimulates ␣KGDH by reducing its K m value for ␣KG (from 2.1 to 0.16 mM, with an S 0.5 Ϸ1.2 M for calcium (49)). In this situation, OGC will be at a disadvantage, and thus ␣KG efflux will become impaired (Fig.  4C). This competition of the two pathways for a common substrate is consistent with the findings of O'Donnell et al. (71), who reported a 4-fold reduction of MAS activity in heart under high workload conditions.
In liver mitochondria we have failed to observe such an inhibition of MAS by matrix Ca 2ϩ (Fig. 1C), and a similar situation was found in beta cell mitochondria (73). As ␣KG efflux in liver mitochondria is only partially blocked by the transaminase inhibitor, this points to an alternative source of ␣KG different from transamination. This is probably GDH, which has a high activity in liver (Table 2), and betacell (74,75) mitochondria. In mitochondria utilizing malate plus glutamate as substrates, ␣KG production can be expressed as the sum of the glutamate transported through AGC and then transaminated (aspartate aminotransferase is highly active, and thus the limiting step in this pathway is the AGC), and that generated in the GDH activity. ␣KG can then either be transported out of mitochondria by the OGC or serve as substrate of ␣KGDH (␣KG consumption). If enough ␣KG is produced to comply with the activity of consuming pathways (␣KG production/consumption ratio Ͼ1 as in liver mitochondria), no inhibition of OGC will be observed, oth- . C, if the Ca 2ϩ transient is high enough, uptake through the Ca 2ϩ uniporter will take place and mitDH will be activated. Because of the competition between OGC and ␣KGDH, MAS activity is reduced even in the presence of elevated extramitochondrial Ca 2ϩ . D, when the transient decreases or if CaU is inhibited after a prolonged Ca 2ϩ elevation (see "Discussion"), net efflux will lower matrix Ca 2ϩ to basal levels, and MAS activity would recover. In the brain, this would take place faster because of the prominent presence of the more active NCX.
erwise ␣KG efflux (and MAS activity) will be inhibited upon calcium entry to the mitochondria and activation of ␣KGDH (␣KG production/consumption ratio Ͻ1, as in brain and heart mitochondria; see Table 2).
Physiological Role of Reversible Matrix Ca 2ϩ -induced MAS Inhibition-The fact that the activation by extramitochondrial Ca 2ϩ of the AGC-MAS pathway is abolished when matrix dehydrogenases are activated by intramitochondrial Ca 2ϩ seems paradoxical, especially because MAS is required to keep a stimulated production of glucose-derived pyruvate. However, it must be borne in mind that Ca 2ϩ signals are usually of a transient nature, and it is their frequency more than their amplitude that matters in terms of decoding and generation of an output. In every Ca 2ϩ spike that reaches mitochondria eliciting a mitochondrial Ca 2ϩ transient, the AGC-MAS pathway would be active at the beginning and end of the spike, but in the central part of it the CaU-mitDH pathway would take over (5). The prevalence of MAS at the beginning of a spike is based on the lower S 0.5 of the system (300 nM) compared with the apparent affinity for Ca 2ϩ of the CaU (1-20 M) and to the lag between cytosolic and mitochondrial calcium increases (76). With a limited Ca 2ϩ entry in mitochondria, the activation of the AGC-MAS pathway accounts for most of the increase in mitochondrial NADH production elicited by the Ca 2ϩ spike (4), and this may prime mitochondria in the face of an immediate rise in energy expenditure (Fig. 4B) (4,5). At the end of the spike, MAS re-activation could take place if basal levels of matrix Ca 2ϩ are regained before cytosolic ones are reached (Fig. 4D). Indeed, MAS inhibition by intramitochondrial calcium disappears when Ca 2ϩ efflux is enhanced over calcium uptake upon CaU inhibition. Furthermore, if the cytosolic Na ϩ concentration increases, as expected in depolarizing conditions, the more active NCX will reduce the extent of MAS inhibition at the center of the spike, and it will hasten the recovery of activity both of OGC and MAS activities at the end of the spike.
Our results suggest that the main NADH-producing pathway in mitochondria at the center of the spike is the Ca 2ϩactivated tricarboxylic acid cycle or Krebs cycle, and not MAS.
As MAS is the main NADH shuttle in brain, this may disrupt the cytosolic NAD ϩ /NADH ratio, directing some of the pyruvate to lactate and limiting pyruvate supply to mitochondria. Because of the transitory nature of the calcium spikes, this imbalance should be a temporal one, and the reactivation of MAS at the end of the spike and beyond the duration of mitochondrial Ca 2ϩ transient may be important to re-establish resting cytosolic NAD ϩ /NADH and pyruvate/lactate ratios. Interestingly, a new mechanism to limit the duration of the mitochondrial Ca 2ϩ elevation in response to a persistent elevation in cytosolic Ca 2ϩ , as would occur during repeated stimulation, has been described (77). According to this mechanism, CaU is inhibited by extramitochondrial calcium within a similar range of calcium concentrations as those leading to its activation (10 -20 M), but with a time constant for the uptake (Ϸ7.9 s at 10 M calcium) at about half that for the inhibition (Ϸ16 s at 10 M calcium) (77). Whether this mechanism is also present in brain mitochondria remains to be established.
Studies in heart have shown that activity-dependent inhibition of the OGC and MAS decreases the cytosolic levels of ␣KG and glutamate (71), and this will lead to a decrease in aspartate efflux from mitochondria and a fall in cytosolic aspartate. Brain aspartate levels are extremely dependent on the AGC activity, and they fall drastically in the aralar/AGC1 knock-out mouse (23). Although it is unlikely that total brain glutamate levels reflect MAS activity, as they are only marginally affected in the aralar knock-out mouse, 3 the changes in aspartate levels could give some clues to MAS activity. Interestingly, Table 3 shows that aspartate levels consistently decrease during brain stimulation, supporting the possibility that MAS is actually inhibited under these conditions. From the above considerations it is obvious that increases in the lactate/pyruvate ratio would be predicted to occur in the brain in vivo, whether as the consequence of sporadic transient increases in cytosolic NADH/ NAD ϩ or of trains of cytosolic NADH/NAD ϩ transients upon 3 B. Padro, L. Contreras, and J. Satrú stegui, unpublished results.

TABLE 3 Metabolite levels during brain stimulation conditions in vivo and in vitro
Data for glutamate (Glu), aspartate (Asp), and Lactate (Lact) content both at control (rest) and during brain work (stimulated) conditions from indicated Tables of the selected references are shown. ND, not determined in the original study; NMDA, N-methyl-D-aspartic acid. repeated depolarization, as one would expect during periods of brain activation. In agreement with this, a number of studies have shown that during task-dependent brain activation, there is an increase in brain glucose utilization larger than the increase in oxygen use and a simultaneous lactate production (14, 16 -19) (see Table 3). It has been proposed that most of the lactate produced upon brain activation arises in astrocytes, which increase the glycolytic rate to energize the uptake of glutamate along with Na ϩ in excitatory synapses (78,79). Indeed, it is likely that aerobic glycolysis (i.e. the production of lactate from glucose in the presence of oxygen) takes place preferentially in astrocytes because, compared with neurons, these cells have reduced MAS activity and reduced aralar/AGC1 levels (8,9,11), and freshly isolated astrocytes release more lactate than neurons (see Refs. 80, 81 and reviewed in Ref. 82). However, our results show that even neurons, which have a robust malate-aspartate shuttle, can provide an alternative source of lactate, which would specifically arise under stimulation conditions. Interestingly, a recent study of evoked glucose and oxygen metabolism in rat cerebellum in vivo has shown that increases in evoked lactate production required a time threshold to start and were not dependent on glutamate reuptake (83), a result that may point to neurons, and not only astrocytes, as lactate producers (84). It is conceivable that the prolonged stimulation required to increase lactate release could lead to repeated mitochondrial calcium transients causing persistent Ca 2ϩ activation of mitochondrial dehydrogenases, inhibition of the OGC and MAS, and lactate production.

System
Conclusions-In this work, we show that MAS activation by extramitochondrial calcium is abolished in the event of ␣KGDH activation by intramitochondrial calcium, in brain and heart mitochondria. This inhibition is based on the competition of both pathways for the shared substrate ␣KG. When matrix calcium concentrations regain a basal level, MAS activity is recovered, and this may extend the mitochondrial energization beyond the end of the cytosolic Ca 2ϩ signal. This inhibition of MAS activity may contribute, in part, to explain the "lactate paradox" observed upon brain stimulation in vivo, as it would lead to increased aerobic glycolysis and lactate formation, although enough oxygen is available and the Krebs cycle is activated.