Excitotoxic Injury to Mitochondria Isolated from Cultured Neurons*

Neuronal death in response to excitotoxic levels of glutamate is dependent upon mitochondrial Ca2+ accumulation and is associated with a drop in ATP levels and a loss in ionic homeostasis. Yet the mapping of temporal events in mitochondria subsequent to Ca2+ sequestration is incomplete. By isolating mitochondria from primary cultures, we discovered that glutamate treatment of cortical neurons for 10 min caused 44% inhibition of ADP-stimulated respiration, whereas the maximal rate of electron transport (uncoupler-stimulated respiration) was inhibited by ∼10%. The Ca2+ load in mitochondria from glutamate-treated neurons was estimated to be 167 ± 19 nmol/mg protein. The glutamate-induced Ca2+ load was less than the maximal Ca2+ uptake capacity of the mitochondria determined in vitro (363 ± 35 nmol/mg protein). Comparatively, mitochondria isolated from cerebellar granule cells demonstrated a higher Ca2+ uptake capacity (686 ± 71 nmol/mg protein) than the cortical mitochondria, and the glutamate-induced load of Ca2+ was a smaller percentage of the maximal Ca2+ uptake capacity. Thus, this study indicated that Ca2+-induced impairment of mitochondrial ATP production is an early event in the excitotoxic cascade that may contribute to decreased cellular ATP and loss of ionic homeostasis that precede commitment to neuronal death.

Glutamate excitotoxicity underlies neuronal loss in ischemic and traumatic brain injury (1) and also likely contributes to dysfunction in chronic forms of neurodegeneration. In particular, the evidence for involvement of excitotoxicity in Huntington disease has been significantly strengthened by recent studies (for a review see Ref. 2). It is well established that glutamate-induced neuronal death depends on entry of extracellular Ca 2ϩ as a result of activation of the NMDA 1 subtype of glutamate receptors (3,4). The description of ensuing events in various types of cultured neurons is extensive (5) and includes an initial transient increase in cytoplasmic Ca 2ϩ , followed by a loss in ionic homeostasis (also termed delayed Ca 2ϩ deregulation) (4). The initial increase in cytoplasmic Ca 2ϩ is associated with mitochondrial Ca 2ϩ loading and slight mitochondrial depolarization followed by profound depolarization concurrent with the loss of ionic homeostasis. Other early events may include the release of apoptogenic cytochrome c and generation of reactive oxygen species and nitric oxide (5)(6)(7). As well, excitotoxic death in cultured neurons is preceded by a significant decline in cellular ATP (8,9).
Recent imaging studies of cultured neurons have demonstrated that mitochondrial Ca 2ϩ loading is in large part responsible for induction of cell death following exposure to glutamate (9 -12), yet the nature of the Ca 2ϩ -induced mitochondrial injury remains controversial. An alternative approach to the elucidation of these mechanisms has been the study of mitochondria isolated from the brains of rodents exposed to excitotoxic stimuli. An inherent problem in such studies with brain mitochondria lies in the heterogeneity of these preparations that contain the organelles from both neuronal and glial cells (13,14). Furthermore, frequently used isolation methods employing gradient centrifugation enrich the preparations in mitochondria from cell bodies, yet the excitotoxic events occur mostly within nerve terminals that are lost in the preparation as synaptosomal fractions. The use of digitonin to avoid the loss of synaptosomal mitochondria (15,16) has been questioned recently (17). Therefore, we have taken the approach of isolating mitochondria from relatively pure neuronal cultures exposed to glutamate to allow these problems to be overcome.
We investigated the early bioenergetic consequences of exposure to glutamate by isolating mitochondria from primary neuronal cultures exposed to glutamate for 10 min, prior to a commitment to neuronal death. By doing so, we avoid studying potential secondary effects of the cell death process that may affect mitochondrial function. Estimation of the extent of glutamate-induced Ca 2ϩ loading enabled us to test the effects of Ca 2ϩ at concentrations that mimic the glutamate stimulus. Additionally, Ca 2ϩ uptake capacities and susceptibility of mitochondria to the permeability transition were assessed. Our novel finding, which has not been revealed by other current approaches, is that mitochondria of neurons exposed to high levels of glutamate sustain significant Ca 2ϩ -induced injury to oxidative phosphorylation, and this injury occurs prior to any commitment to cell death. 1 min in 0.1% trypsin in a Ca 2ϩ /Mg 2ϩ -free phosphate-buffered saline solution supplemented with glucose (1.5 mM), after which trypsin was inactivated by addition of soybean trypsin inhibitor (0.1 mg/ml). The mixture was transferred into Hibernate E medium containing 20 units/ml DNase (Promega) in 0.2ϫ reaction buffer (Promega), and the cells were centrifuged at 200 ϫ g for 1.5 min. The supernatant was quickly aspirated, and the cells were resuspended in 10 ml of Neurobasal (E) medium (Invitrogen) plus glutamate (0.4 g/ml), 0.5 mM L-glutamine, penicillin, and streptomycin (100 units/ml and 100 g/ml, respectively), 1ϫ B27 supplement, and 5 mM sodium pyruvate. Once in suspension, the cells were diluted into 30 ml of the same medium without pyruvate (initial plating medium), and the number of viable cells was determined by trypan blue exclusion. Cells were plated on poly-D-lysine-coated 10-cm plates at a concentration of 750,000 cells per ml in a volume of 10 ml per plate and kept at 37°C in a 5% CO 2 incubator. For measurements of cellular ATP and cytoplasmic Ca 2ϩ , cells were plated on Biocoat poly-D-lysine-coated black 96-well plates (BD Biosciences) at 50,000 cells per well. After 4 -6 days in vitro, the initial plating medium was diluted with an equal volume of maintenance medium of the same composition lacking glutamate and L-glutamine and supplemented with 1% GlutaMAX-1 (Invitrogen). Cultures were fed every 3-4 days with fresh medium. All experiments were performed with cultures that were 13-15 days in vitro. These cultures were 91-95% neuronal, as estimated by immunocytochemical staining according to the manufacturer's protocols with anti-neuronal nuclei (Chemicon, mAB377) or anti-neurofilament 200 kDa (Calbiochem, IF06L).
Measurement of Viability of Cortical Neurons Treated with Glutamate-Cell viability was measured in neurons cultured in 96-well plates using a cytotoxicity detection kit (lactate dehydrogenase, LDH) according to the manufacturer's recommendations (Roche Applied Science). Data are expressed as percent cell death based upon measurement of LDH activity that was cell-associated versus that which was released to the medium. Cells were treated in HEPES-buffered salt solution (HBSS, containing 137 mM NaCl, 5 mM KCl, 10 mM NaHCO 3 , 20 mM HEPES, 5.5 mM glucose, 0.6 mM KH 2 PO 4 , 1.4 mM CaCl 2 , 0.9 mM MgSO 4 ) with 100 M glutamate plus 10 M glycine with or without pretreatment with 10 M of the NMDA receptor antagonist MK-801. Control cultures were exposed to HBSS without glutamate or glycine. Following a 10-min incubation at 37°C, an equal volume of maintenance medium was added (supplemented with glutamate plus glycine and/or MK-801 to maintain constant concentration of these effectors). The cells were then placed back in the incubator for 24 h prior to measurement of viability. Alternatively, at the addition of the maintenance medium, 2ϫ (20 M) MK-801 was included to block NMDA receptor activity.
Measurement of Neuronal Ca 2ϩ -Cytoplasmic Ca 2ϩ was measured in neurons cultured in the 96-well plates using magfluo-4, a cell permeant low affinity Ca 2ϩ dye (Molecular Probes). A stock solution of magfluo-4 (1 mg/ml) was prepared on the day of the experiment in Me 2 SO and then diluted in HBSS to a final concentration of 4 M. The culture medium was carefully aspirated, and cells were loaded with the dye (100 l of a 4 M solution per well) for 25 min in the incubator. The loading buffer was then replaced with dye-free HBSS, and the plates were assayed in a fluorescent plate reader (FLIPR, Molecular Devices) at 488 (excitation) and 525 nm (emission).
Measurement of Neuronal ATP-The level of neuronal ATP was determined by using a CellTiter-Glo TM luminescent assay kit (Promega). Cells cultured in 96-well plates were treated with glutamate for varying lengths of time (from 10 to 40 min). Control cells were exposed to HBSS without glutamate and glycine for 10 min. At the end of the treatment, CellTiter-Glo TM reagent was added, and the plates were placed on a shaker for 5 min. In some experiments, the glutamatecontaining medium was aspirated, and cells were incubated in Ca 2ϩfree HBSS (containing 100 M EGTA) for 3-6 min prior to the addition of CellTiter-Glo TM reagent. Luminescence was measured in a PolarStar plate reader (BMG). ATP levels are expressed as fmol/cell based upon an ATP standard curve, and data are corrected for background luminescent signal. Each set of data was collected from multiple replicate wells (n ϭ 12) and plotted as mean Ϯ S.E.
Preparation of Cerebellar Granule Cells-Granule cells were prepared as described previously (18) from 7-day-old postnatal Wistar rats. Cells were plated on poly-D-lysine-coated 10-cm plates at a concentration of 1 ϫ 10 6 cells per ml in a volume of ϳ9 ml per plate. Cells were cultured in minimal essential medium containing Earle's salts (Invitrogen) plus 10% (v/v) fetal calf serum (HyClone), 25 mM KCl, 30 mM glucose, 2 mM glutamine, penicillin, and streptomycin (100 units/ml and 100 g/ml, respectively). After 24 h, 10 mM cytosine arabinose was added to inhibit non-neuronal cell proliferation. Cells were maintained at 37°C in a 5% CO 2 incubator and were used after 7-8 days in vitro.
Isolation of Mitochondria from Control and Glutamate-treated Neuronal Cultures-A method for isolation of functional mitochondria from primary neuronal cultures has been described previously (19). We used a modified protocol for rapid preparation of neuronal mitochondria suitable for assessment of bioenergetic parameters and the Ca 2ϩ content. The culture medium was aspirated, and cells (cortical neurons or cerebellar granule cells) were exposed to 100 M glutamate and 10 M glycine in HBSS. Control cells were exposed to HBSS without glutamate and glycine. After glutamate addition, cells were returned to the incubator for 10 min. Next, the plates were placed on ice and washed twice with ice-cold Ca 2ϩ -free and Mg 2ϩ -free phosphate-buffered saline solution (Invitrogen) supplemented with 2 mM EGTA. Cells were rapidly scraped in 1 ml (per plate) of ice-cold mitochondrial isolation buffer containing 210 mM mannitol, 70 mM sucrose, 10 mM HEPES-KOH, pH 7.4, 2 mM EGTA, and 0.1% fatty acid-free bovine serum albumin and were then homogenized using a Dounce homogenizer (10 passes with a loose pestle and 10 passes with a tight pestle). Cell homogenates were centrifuged for 10 min at 1060 ϫ g at 4°C, and the supernatant was collected. The pellet was resuspended in the isolation buffer and centrifuged at 1060 ϫ g for 5 min. The first and second supernatants were pooled together and centrifuged at 14,600 ϫ g for 10 min at 4°C. The mitochondrial pellet was resuspended in the isolation buffer lacking EGTA and albumin and centrifuged at 14,600 ϫ g for 10 min. All isolation steps were performed on ice. The protein concentration was measured using the BCA protein assay kit (Pierce). Typically, 10 -15 plates of cells were used for one mitochondrial preparation. Note that mitochondria were isolated in the absence of digitonin that is commonly used for disruption of the cholesterol-rich synaptosomal membranes to release mitochondria entrapped in synaptosomes (15). Our mitochondrial preparations from cultured neurons appear to be relatively free of synaptosomes as evidenced by the lack of an effect of digitonin on the maximal rate of respiration in the presence of succinate and rotenone (data not shown). Succinate is a mitochondrial substrate that is poorly permeable to the plasma membrane.
Measurements of Mitochondrial Respiration, Membrane Potential, and Ca 2ϩ in Isolated Neuronal Mitochondria-Simultaneous measurements of mitochondrial respiration (O 2 consumption), Ca 2ϩ fluxes, membrane potential, and optical density of the mitochondrial suspension were conducted in a custom-constructed thermostatically controlled chamber (B. Krasnikov, Burke Medical Research Institute, NY). Membrane potential (⌬⌿) was monitored by distribution of tetraphenylphosphonium (2 M) with a tetraphenylphosphonium-selective electrode (20). Extramitochondrial Ca 2ϩ was monitored with a Selectophore I membrane (Fluka)-based Ca 2ϩ -selective electrode. A Clark-type electrode (Diamond General) was used to measure oxygen consumption. In addition, measurements of optical density at 660 nm by means of a light-emitting diode and photodetector permitted changes in mitochondrial volume to be followed. This parameter was also used to confirm consistency of protein concentrations between experimental runs. Digital data acquisition using a DynaRes 16 board and WorkBench software (Strawberry Tree) permitted simultaneous recording of all four parameters as well as plots of their first derivatives in real time. The first derivatives were used for quantification of respiration and Ca 2ϩ uptake rates.
For respiration measurements, mitochondria were incubated at 25°C in a basal saline medium (125 mM KCl, 5 mM HEPES/KOH, pH 7.4, and 2 mM phosphate) supplemented with 2 mM MgCl 2 and either complex I-linked substrates (a mixture of 5 mM glutamate and 5 mM malate) or a complex II substrate (5 mM succinate in the presence of 2 M rotenone). Protein concentrations were 1-1.5 mg/ml for cortical mitochondria and 0.7-1.3 mg/ml for cerebellar mitochondria. State 3 (phosphorylating) respiration was initiated by addition of 200 M ADP, after which state 4 (resting) respiration was induced by addition of 100 M atractyloside, an inhibitor of the ADP/ATP antiporter. Finally, to measure uncoupler-stimulated respiration (state 3u), sequential additions of the protonophore carbonyl cyanide p-(trifluoromethoxy)phenylhydrazone (FCCP) (20 -50 nM) were made until the respiration rate reached maximum. The acceptor control ratio was determined as the ratio of state 3 to state 4 respiration. The experiments were completed within 2-3 h after the isolation procedure.
Mitochondrial Ca 2ϩ content was determined in the same basal medium (above) supplemented with glutamate and malate. Protein concentrations were in the same range as for the respiration measurements. After 2-3 min of incubation to allow for uptake of contaminating Ca 2ϩ from the medium, mitochondrial Ca 2ϩ was released into the medium by addition of FCCP (0.5 M) followed 5 min later by the addition of the pore-forming peptide alamethicin (40 g/ml) to ensure complete Ca 2ϩ release. In some experiments, A23187 (1.2 M), a Ca 2ϩ ionophore, was also used to release sequestered Ca 2ϩ . The amount of Ca 2ϩ released from mitochondria was calculated based upon calibration of the Ca 2ϩ electrode performed at the end of each experimental day by additions of known concentrations of CaCl 2 to the medium. EGTA (1 mM) was then added to chelate free Ca 2ϩ in the medium. According to the Nernst equation, the voltage changes on the electrode are linearly related to the logarithm of Ca 2ϩ concentration values in the medium. The concentration of contaminating Ca 2ϩ in the medium was also experimentally determined. It was defined as an independent variable in the regression analysis shown in Equation 1, where a and b are regression parameters expressed in relative units, and [Ca 2ϩ ] 0 is the concentration of contaminating Ca 2ϩ . R 2 values for all calibration curves were no less than 0.995. The best fit parameters were used to calculate Ca 2ϩ concentrations from the voltage changes in the experimental runs. The mitochondrial Ca 2ϩ load was determined as the difference in Ca 2ϩ concentration after addition of the Ca 2ϩ -releasing agent (FCCP or alamethicin) minus the contaminating Ca 2ϩ in the medium and was normalized to mitochondrial protein concentration.
Mitochondrial Ca 2ϩ uptake capacity was measured in the same medium as that used for mitochondrial Ca 2ϩ content determination. Energized mitochondria were pulsed with successive additions of Ca 2ϩ (30 -50 M) every 2-3 min. The increasing Ca 2ϩ load caused a successive decline in the membrane potential and Ca 2ϩ uptake rates. Maximal Ca 2ϩ uptake capacity was defined as an amount of Ca 2ϩ (per mg of protein) required to decrease the Ca 2ϩ uptake rate by Ͼ90%. Note that because of the duration of the experimental run (25-40 min), most of these measurements were performed in an open chamber to avoid anoxia.
Data are expressed as mean Ϯ S.E., and n indicates the number of independent experiments. Each mitochondrial preparation was obtained from a separate neuronal harvest (or neuronal preparation).
Statistical Analysis-One-way analysis of variance followed by post hoc Tukey test was performed using SigmaStat. Where differences are noted, analysis of variance detected significant variance at p Ͻ 0.001, and pairwise comparisons are indicated in the figure legends.
Materials-Unless indicated otherwise, all reagents were obtained from Sigma.

Measurements of Cytoplasmic Ca 2ϩ in Cortical Cultures-
Mitochondrial Ca 2ϩ accumulation has been well established in different excitotoxicity models (21)(22)(23)(24)). An intact cell study utilizing a dye with low affinity to Ca 2ϩ has demonstrated that a significant fraction of accumulated Ca 2ϩ is retained within the mitochondria during brief glutamate exposures until FCCP is added to induce Ca 2ϩ efflux (25). Based on this approach, we performed measurements of cytoplasmic Ca 2ϩ in situ with our cortical cultures to demonstrate that the conditions of glutamate treatment used in our model produces results consistent with literature data. Fig. 1 depicts results of a representative experiment in which cortical neurons were challenged with various concentrations of glutamate (in the presence of 10 M glycine) for 10 min followed by chelation of extracellular Ca 2ϩ with EGTA and subsequent addition of FCCP. The responses to glutamate at concentrations between 30 and 300 M were similar, and the pattern is typified by an initial peak in cytoplasmic Ca 2ϩ that is followed relatively rapidly in these cultures by a more sustained and profound rise in cytoplasmic Ca 2ϩ . The response to 100 M glutamate appeared to be maximal with regard to cytoplasmic Ca 2ϩ and FCCP-induced release of Ca 2ϩ , as noted previously by others (25); therefore, we chose to probe in greater depth the bioenergetic changes occurring at this 10-min time point of glutamate exposure.
Effect of Glutamate Treatment on Cellular ATP-As shown in Fig. 2, 10 min of glutamate exposure results in a significant drop in the ATP level ( Fig. 2A). Longer exposures to glutamate (up to 40 min) did not further decrease the content of ATP.
Partial ATP depletion could be a direct consequence of acute activation of ATP-requiring efforts by the cell to re-establish ionic homeostasis during glutamate stimulation, and/or the ATP loss could result from impairment of cellular ATP production. To test whether ATP could be readily recovered if the glutamate and Ca 2ϩ challenge were halted, cells were exposed to glutamate for 10 min followed by incubation of the cells in a glutamate-and Ca 2ϩ -free buffer. Under these conditions, cytoplasmic Ca 2ϩ fairly rapidly returns to a steady state level (see Fig. 1 described above); however, ATP did not recover within this time frame. This result suggests that the compromise in cellular ATP levels reflects a compromise in the ability of the cell to produce ATP at a normal rate.
Effect of Glutamate Treatment on Neuronal Viability-We next tested the effects of glutamate exposure on the viability of the cortical cultures. By using LDH release as a method of assessing plasma membrane permeability, we found that 24 h of continuous exposure of these cultures to 100 M glutamate resulted in a significant loss in viability that was inhibited by 10 M of the NMDA receptor antagonist, MK-801 (Fig. 3). Even though ionic homeostasis appears to be lost and mitochondrial Ca 2ϩ loading is maximal after 10 min of glutamate exposure (Fig. 1), if this treatment is halted by addition of MK-801 for the ensuing 24 h, the injury is no longer sufficient to induce LDH release substantially above control levels (Fig. 3), at least at this time point. Therefore, cellular events leading up to the 10 min time point may be critical to initiation of the cell death pathway, but these events have not resulted in commitment of the cells to acute cell death.
Quantitative Estimation of Ca 2ϩ in Mitochondria from Control and Glutamate-treated Neurons-The ability of mitochondria in glutamate-treated cells to release accumulated Ca 2ϩ in response to FCCP argues against pervasive induction of a permeability transition in the mitochondrial population, as this would prevent retention of mitochondrial Ca 2ϩ . As shown on significant rate 5 min after the uncoupler addition (Fig. 4A, trace c; note that the y scale is logarithmic). Therefore, we used a pore-forming peptide alamethicin to quickly attain complete release (Fig. 4A, trace c). Quantitative estimation of mitochondrial Ca 2ϩ (performed as described under "Experimental Procedures") revealed that the total Ca 2ϩ released in response to FCCP and alamethicin reached 167 Ϯ 19 nmol/mg protein (n ϭ 8). In the 5-min time frame, FCCP-released about half of the total Ca 2ϩ (86 Ϯ 14 nmol/mg protein, n ϭ 8). More importantly, the amount of Ca 2ϩ released from mitochondria isolated from untreated cells was very low (1.9 Ϯ 0.8 nmol/mg protein, n ϭ 5; Fig. 4A, trace b). Note that the amount of contaminating Ca 2ϩ taken up into mitochondria from the incubation medium (2.5 Ϯ 1.6 nmol/mg protein, n ϭ 13) was subtracted from the amount of Ca 2ϩ released into the medium. We do not exclude the possibility that a fraction of damaged mitochondria could be lost during preparation, but there was no apparent difference in the yields of mitochondria isolated from control and glutamate-treated cells. As well, we cannot exclude the possibility that a subfraction of mitochondria may have undergone the permeability transition and released their accumulated Ca 2ϩ prior to isolation. Both of these possibilities would result in an underestimation of the quantities of retained Ca 2ϩ . Although it is possible that mitochondrial Ca 2ϩ could be lost during isolation, the standard conditions for preparation of mitochondria, such as low temperature, the presence of EGTA, and the absence of Na ϩ in the isolation buffers, minimize changes in mitochondrial Ca 2ϩ during the isolation procedure. This statement is supported by direct Ca 2ϩ measurements in Ca 2ϩloaded brain mitochondria after prolonged incubation under conditions that mimic our isolation procedure (27).
Because alamethicin is nonselective, and endoplasmic reticulum is a persistent contaminant of mitochondrial prepara- tions, we performed additional experiments to confirm that the released Ca 2ϩ was primarily of mitochondrial origin. Pretreatment of intact cortical neurons for 15 min prior to glutamate addition with 10 M cyclopiazonic acid, an endoplasmic reticulum Ca 2ϩ -ATPase inhibitor, did not affect Ca 2ϩ release in response to FCCP and alamethicin (Fig. 4B, traces a versus b). This observation suggests that the Ca 2ϩ released by alamethicin is primarily mitochondrial.
Determination of Ca 2ϩ Uptake Capacity in Mitochondria from Cortical Neurons-As discussed above, the results of Fig.  1 imply that the mitochondrial Ca 2ϩ load is saturated under our standard conditions (100 M glutamate and 10 min of treatment). At a low level of glutamate, the mitochondrial Ca 2ϩ load appears to be limited by Ca 2ϩ availability as indicated by a near basal steady state level of cytosolic Ca 2ϩ (Fig. 1, trace e). In contrast, one might presume that under the standard conditions, the mitochondrial Ca 2ϩ load represents the Ca 2ϩ uptake capacity of the mitochondria. To determine whether the uptake capacity of the mitochondria has indeed been saturated by glutamate exposure, we quantitatively determined the maximal Ca 2ϩ uptake capacity of mitochondria isolated from control cells that were not exposed to glutamate. Pulses of Ca 2ϩ were added to mitochondria from control neurons until virtually complete inhibition of Ca 2ϩ uptake was achieved. Repeated Ca 2ϩ additions caused gradual decreases in the membrane potential and Ca 2ϩ uptake rates (Fig. 5) to a point at which Ca 2ϩ -induced respiratory inhibition prevented further establishment of membrane potential (respiratory tracing not shown). The maximal Ca 2ϩ uptake capacity in cortical mitochondria was 363 Ϯ 35 nmol/mg protein (n ϭ 7), a value that is significantly higher than the glutamate-induced Ca 2ϩ load (167 Ϯ 19 nmol/mg protein). Most interestingly, even at the maximal Ca 2ϩ concentrations, mitochondria retained all the accumulated Ca 2ϩ and did not swell (Fig. 5). The addition of the pore-forming peptide alamethicin permitted detection of swelling under these conditions. These observations rule out induction of the permeability transition by Ca 2ϩ in this experiment. It should be noted that the measurements were performed in the absence of added Mg 2ϩ and adenine nucleotides, which are known to inhibit the classical permeability transition (28,29), and to enhance significantly the Ca 2ϩ uptake capacity of brain mitochondria (16,30,31). Therefore, the mitochondrial Ca 2ϩ uptake capacity may be even higher in intact neurons, as the cytoplasm contains millimolar concentrations of Mg 2ϩ and adenine nucleotides. However, factors other than the intrinsic mitochondrial Ca 2ϩ uptake capacity may limit Ca 2ϩ accumulation in mitochondria during acute glutamate treatment (see "Discussion").
Effect of Glutamate Treatment on Mitochondrial Respiration and Membrane Potential-Mitochondria isolated from untreated cortical neurons preserved functional integrity as evidenced by the ability to establish membrane potential (Fig. 6A) and by an acceptor control ratio of 4.8 Ϯ 0.3 (n ϭ 16) in the presence of complex I-linked substrates (Fig. 6B). Exposure of the cells to 100 M glutamate for 10 min did not compromise the ability of subsequently isolated mitochondria to generate membrane potential (Fig. 6A). During the isolation procedure, mitochondrial Ca 2ϩ transport processes that were taking place following glutamate exposure are inhibited by the low temperature and by permeabilization in medium containing EGTA. Therefore, these conditions are unlike those in the intact cell in which mitochondrial Ca 2ϩ cycling is ongoing. Our observation is in agreement with evidence that mitochondrial depolarization during acute glutamate treatment is reversible upon glutamate removal (5).
Most importantly, glutamate treatment markedly reduced (on average, by 44%) ADP-stimulated respiration (state 3) and decreased the amplitude of ADP-induced depolarization (Fig. 6,  A and B). Because state 3 respiration is dependent upon both electron transport and ADP phosphorylation, the observed inhibition could result from injury to either of the two processes. Uncoupler-stimulated respiration (state 3u) is an indicator of the maximal rate of electron transport. As shown in Fig. 6, A and B, state 3u respiration was reduced by ϳ10%. Under resting conditions, respiration rates (state 4) remained relatively unchanged. The acceptor control ratio in mitochondria from glutamate-treated cells dropped to 2.7 Ϯ 0.2 (n ϭ 15). Therefore, the most significant effect of glutamate treatment on mitochondria from cortical neurons was inhibition of ADP phosphorylation, whereas the effect on the maximal rate of electron transport was less pronounced, and there was no evidence of significant respiratory uncoupling.
Deleterious effects of elevated Ca 2ϩ on neural mitochondria, such as induction of the permeability transition (16,(31)(32)(33) or inhibition of respiration (34 -37), have been described in the literature. However, the pathophysiological relevance of the amounts of Ca 2ϩ used in many studies remains questionable. We examined the effects of adding exogenous Ca 2ϩ to control mitochondria at a concentration close to the quantity we measured for glutamate-induced Ca 2ϩ loading (Fig. 6B). In these experiments, mitochondria from untreated cortical neurons were exposed to a bolus of 150 M Ca 2ϩ (100 -125 nmol/mg). After sequestration of the added Ca 2ϩ was complete, respira-  (1) through Ca (n) was followed by addition of the pore-forming agent alamethicin (Alm). Note that alamethicin, but not Ca 2ϩ , induces hallmarks of permeability transition, swelling and release of previously retained Ca 2ϩ . Other experimental details are specified under "Experimental Procedures." tion was measured under the same conditions as in Fig. 6A. As shown on Fig. 6B, which summarizes the data on respiration from multiple experiments, the patterns of respiratory changes induced by 150 M Ca 2ϩ and glutamate treatment were simi-lar, although the added Ca 2ϩ caused a more dramatic inhibition of state 3 respiration. The data demonstrate that in this model Ca 2ϩ per se mimics the mitochondrial bioenergetic effects of acute glutamate treatment.
Because the extent of Ca 2ϩ -mediated injury may depend on the type of respiratory substrate (31,33,35,36,38,39), we also assessed the effects of glutamate treatment in the presence of the complex II substrate succinate. Mitochondria energized with succinate showed a similar pattern of inhibition of respiration as in the presence of complex I-linked substrates (Fig. 6,  C versus A). In the experiment shown in Fig. 6C, succinatesupported state 3 and state 3u respiration were inhibited by 47 and 19%, respectively.
Reversibility of Glutamate-induced Inhibition of ADP Phosphorylation-Next, we addressed the question of whether glutamate-induced partial inhibition of state 3 respiration can be reversed upon removal of the accumulated Ca 2ϩ . Addition of the Ca 2ϩ ionophore A23187 to mitochondria from glutamatetreated cells induced rapid release of the endogenous Ca 2ϩ that was subsequently chelated by EGTA (Fig. 7A). Ruthenium Red, the inhibitor of the Ca 2ϩ uniporter, was present in these experiments to prevent Ca 2ϩ re-uptake during A23187-induced Ca 2ϩ release. Respiration measurements showed that Ca 2ϩ depletion partially restored glutamate-inhibited state 3 respiration (Fig. 7B). This observation is consistent with previous studies with Ca 2ϩ -challenged heart (40) and brain (34) mitochondria, demonstrating that Ca 2ϩ -induced inhibition of oxidative phosphorylation can be at least partially relieved by treatment with A23187 and EGTA or after Ca 2ϩ release via the Ca 2ϩ /Na ϩ exchanger.

Measurements of Respiration, Ca 2ϩ Content, and Ca 2ϩ Uptake Capacity in Mitochondria from Cerebellar Granule Cells-
Little is known about the differences in mitochondria from different neuronal types, although such information is highly desirable in an effort to understand further the varying susceptibility of different neurons to injury in acute and chronic forms of neurodegeneration. Isolated mitochondria from different brain regions have been studied with regard to susceptibility to the permeability transition (41), susceptibility to dysfunction induced by traumatic brain injury (42), ischemic injury (27,43), and various toxicants (44) with the caveat that these are studies of mixed populations of mitochondria from varied cell types. The current approach allows more appropriate comparison of bioenergetic function between enriched neuronal cultures. A series of experiments was performed to compare the glutamate-induced changes in mitochondria from cortical neurons with those in mitochondria isolated from cerebellar granule cells. Under control conditions, the rates of respiration of mitochondria from the two neuronal cell types were comparable ( Fig. 6B and Fig. 8).
With regard to glutamate-induced mitochondrial injury, Atlante et al. (45) have reported progressive inhibition of succinate-supported state 3 respiration measured in homogenates prepared from cerebellar granule cell cultures at different times after glutamate exposure. In agreement with this study, glutamate treatment for 10 min induced 20 -30% inhibition of both state 3 and state 3u respiration in the presence of complex I-linked substrates in cerebellar mitochondria (Fig. 8). ADPstimulated respiration was therefore less inhibited than in cortical mitochondria (Fig. 8 versus Fig. 6). The observation of partial inhibition of ADP-stimulated respiration does not contradict a previous conclusion that cerebellar mitochondria in situ remain capable of producing ATP throughout excitotoxic glutamate exposure (Refs. 5 and 46; see "Discussion").
The most significant difference between cortical and cerebellar mitochondria was observed in the Ca 2ϩ uptake capacity. The maximal Ca 2ϩ uptake capacity for cerebellar mitochondria was 686 Ϯ 71 nmol/mg protein (n ϭ 4), almost twice as high as for cortical mitochondria. Despite the higher Ca 2ϩ uptake capacity, glutamate-induced increases in the Ca 2ϩ content were similar to those in cortical mitochondria. The Ca 2ϩ pool released by FCCP within 5 min in cerebellar mitochondria was 96 Ϯ 34 nmol/mg protein (n ϭ 7). The total Ca 2ϩ released in response to FCCP plus alamethicin was 173 Ϯ 64 nmol/mg protein (n ϭ 8). These data suggest that in cerebellar granule cells, similar to cortical neurons, glutamate-induced Ca 2ϩ loading does not saturate the mitochondrial Ca 2ϩ uptake capacity, but the extent of loading represents a smaller proportion of the maximal uptake capacity. As with cortical neuronal mitochondria, the alamethicin-releasable pool of Ca 2ϩ in mitochondria from untreated cells was a small fraction of the pool released following glutamate treatment (2.0 Ϯ 1.4% (n ϭ 4)). DISCUSSION In our cortical neuronal cultures, we identified impairment of ADP phosphorylation as the most dramatic consequence of glutamate exposure to mitochondrial bioenergetics, whereas inhibition of maximal respiratory chain activity was quite modest. The inhibition of oxidative phosphorylation occurred prior to commitment to cell death and therefore did not result from sequelae of the cell death process. The effect of excitotoxin exposure on oxidative phosphorylation was Ca 2ϩ -dependent as we found that it could be mimicked by additions in vitro of the relevant concentrations of Ca 2ϩ . Moreover, glutamate-induced inhibition of state 3 respiration could be reversed by rapid depletion of the accumulated Ca 2ϩ using ionophore. We found that the cellular ATP level was compromised within the first minutes of glutamate exposure ( Fig. 2A), and it was not readily recovered at 6 min after glutamate/Ca 2ϩ removal (Fig. 2B). The lack of recovery at this time point likely reflects the protracted elevation of matrix Ca 2ϩ following an excitotoxic stimulus because of the relatively slow kinetics of the mitochondrial Ca 2ϩ efflux pathway and continued Ca 2ϩ cycling (23,25,47,48). We propose that the early impairment of ADP phosphorylation contributes to cellular ATP depletion, and we further suggest that this inhibition is likely key to both ensuing neuronal dysfunction and/or cell death.
Certainly, there are other potential modes by which mitochondrial Ca 2ϩ sequestration may trigger neuronal dysfunction and/or death. Mitochondrial Ca 2ϩ overload may result in respiratory inhibition and/or the permeability transition that prevents sufficient mitochondrial ATP production to meet the increased energy demand. Alternatively, Ca 2ϩ sequestration may result in increased oxidative stress that compromises multiple components of cell function including Ca 2ϩ extrusion (5). Nicholls and co-workers (9,11,49) have reported that the cause of delayed Ca 2ϩ deregulation, at least in cerebellar granule cells, is a mitochondrial event other than insufficient ATP production. It should be noted, however, that the approaches in these elegant studies, including fluorescence imaging of intact cells and in situ respiration of monolayers, are capable of detecting whether ADP phosphorylation can take place, but they are unable to assess whether significant but incomplete inhibition of oxidative phosphorylation has occurred. This limitation is inherent to studies of mitochondrial bioenergetics using intact cells because of the inability to control ADP and substrate provision to the mitochondria. The isolation technique employed here overcomes this limitation. Jekabsons and Nicholls (49) found that glutamate stimulated the respiration of intact cerebellar granule cells to 40 -50% of their maximal electron transport chain capacity. We would suggest that this increase, although stimulated by ATP turnover, is not meeting ATP demand in a manner consistent with continued normal cell function and viability. There are potentially important differences, however, between the Ca 2ϩ responses of our cortical cultures and the cerebellar granule cells studied by Jekabsons and Nicholls (49). Foremost, Ca 2ϩ deregulation is much more delayed in a majority of the granule cells compared with our cortical neurons (Refs. 9, 11, and 49 versus Fig. 1). The inhibition of oxidative phosphorylation, the persistent decline in ATP, and the rapidity of the loss in ionic homeostasis in our cortical neurons are most consistent with death involving insufficient ATP generation. In fact, the pattern of cytoplasmic Ca 2ϩ responses to glutamate in cortical neurons appears sim- ilar to those in cerebellar granule cells when both respiration and mitochondrial ATP synthesis have been chemically blocked (11). As a result, we propose that mechanisms of injury are common in different neurons, but the events most critical to loss in ionic homeostasis and cell death may vary in different cultures and may depend on multiple factors, including the glycolytic capacity to produce ATP, reactive oxygen speciesdetoxifying systems, and the proportion of the mitochondria of the cells that sequesters excessive loads of Ca 2ϩ .
More importantly, the fact that our cortical cultures do not undergo cell death if the stimulus has been halted at 10 min (Fig. 3), despite the fact that they have undergone Ca 2ϩ deregulation (Fig. 1), clearly implies that additional event(s) critical to the commitment to cell death occur after this profound increase in cytoplasmic Ca 2ϩ . In fact, others have found that although antioxidants have no effect on the onset of delayed Ca 2ϩ deregulation, they protect against cell death measured 24 h later (50), suggesting that an oxidative event critical to death occurs after the loss of ionic homeostasis. Furthermore, our data suggest that inhibition of oxidative phosphorylation may be a consequence of glutamate exposure that, if brief, may result in a compromise in normal neuronal function but not necessarily result in neuronal death. This may have profound implications for the chronic dysfunction of neurons in diseases such as Huntington disease, in which expansion of the huntingtin protein may lead to NMDA receptor activation, exaggerated cytoplasmic Ca 2ϩ signaling via inositol 1,4,5-trisphosphate-mediated mechanisms, and thus enhanced mitochondrial Ca 2ϩ loading (2,51).
Inhibition of respiration and/or ADP phosphorylation by elevated Ca 2ϩ is not a unique feature of neuronal mitochondria. A similar injury was observed in earlier studies with mitochondria from tumor (38,52) and transformed neural cells, 2 brain (34,35), and heart (40). The precise mechanisms underlying this injury remain to be elucidated. The maximal rate of respiration (FCCP-stimulated respiration) is mildly decreased by glutamate treatment in both cortical and cerebellar mitochondria in our hands. Inhibition of maximal respiration in response to glutamate has been observed in cerebellar granule cells (45,49) and cortical neurons (53), although Atlante et al. (45) and Almeida et al. (53) did not correct for losses in cell viability and measured respiration at later time points. The effect that we observe is independent of the type of respiratory substrate (complex I versus complex II substrates) and therefore could be explained by a partial loss of cytochrome c (33,54,55). ADP phosphorylation (state 3 respiration) was significantly more inhibited than FCCP-stimulated respiration (state 3u respiration) in our cortical cultures, indicating that components other than the respiratory chain are affected, e.g. the ATP synthase and/or availability of adenine nucleotides and phosphate. As one potential mechanism, Ca 2ϩ has been proposed to bind the adenine nucleotides in the mitochondrial matrix causing inhibition of state 3 respiration and ATP synthase activity (34,56).
This study estimates that the glutamate-induced mitochondrial Ca 2ϩ loading for both cortical and cerebellar neurons is ϳ170 nmol/mg protein. This amount was determined by releasing the cation into the medium with the pore-forming agent alamethicin. We found that alamethicin was more effective than FCCP at rapidly releasing Ca 2ϩ , likely because of the fact that the bulk of Ca 2ϩ in mitochondria is stored in the form of the poorly soluble Ca 2ϩ -phosphate complex (16,(57)(58)(59)(60). The rate of solubilization of this complex may be slower in response to uncoupler as opposed to when alamethicin-induced pores allow dilution of the matrix contents. As noted earlier, one of the shortcomings of our current approach is that mitochondrial Ca 2ϩ can potentially be lost during the isolation. Although there is evidence against this (27), it must be stated that the actual mitochondrial Ca 2ϩ load may be even higher than our estimates. A second reason for this statement is that nonmitochondrial protein typically contaminates mitochondrial preparations. As judged by Western blot analysis (data not shown), the degree of cytosolic and membranous contaminations of our cortical preparation was comparable with that of brain mitochondria isolated according to a published protocol (15), and therefore, the Ca 2ϩ loads represent reasonable estimates for comparison to isolated mitochondria models.
The glutamate-induced mitochondrial Ca 2ϩ load is significantly below the maximal Ca 2ϩ uptake capacity that we measured in vitro. There are a number of potential explanations for this observation, including that the quantity of available phosphate (the necessary counter-ion in Ca 2ϩ transport) could limit glutamate-induced mitochondrial Ca 2ϩ uptake (61), that the activity of the mitochondrial Na ϩ /Ca 2ϩ exchanger is enhanced in response to cytosolic Na ϩ (23,48), or that there is loss of loaded Ca 2ϩ because of the permeability transition or the isolation procedure. Furthermore, mitochondrial Ca 2ϩ uptake may be altered by glutamate-induced generation of reactive oxygen species or nitric oxide (for reviews see Refs. 7 and 21). An equally tenable explanation for the apparent "intermediate" Ca 2ϩ loading in situ is that only a subpopulation of mitochondria may be sequestering Ca 2ϩ in intact neurons, for which there is strong evidence (62,63). Mitochondria within cells display functional and morphological heterogeneity that, among other factors, reflects their proximity to Ca 2ϩ sources (63)(64)(65). One can hypothesize that heterogeneous Ca 2ϩ loading causes more dramatic respiratory injury to a subpopulation of mitochondria that are in close proximity to NMDA receptors. Inhibition of ATP production to a critical subset of mitochondria may produce a localized inability to re-establish ionic homeostasis and membrane potential at the plasma membrane that ultimately perpetuates a collapse of ionic homeostasis.
The difference between the glutamate-induced Ca 2ϩ load and the maximal Ca 2ϩ uptake capacity is more dramatic in mitochondria from cerebellar granule cells. The higher Ca 2ϩ uptake capacity of cerebellar mitochondria (as compared with cortical mitochondria) correlates with the relative resistance of this brain region in excitotoxic and ischemia/reperfusion models (66,67). Mitochondria isolated from cerebellar tissue are less susceptible to a Ca 2ϩ -induced injury than mitochondria isolated from cortex and hippocampus (41) potentially because of higher adenine nucleotide content (38,41). Therefore, these data lead us to suggest that the proportional load of mitochondrial Ca 2ϩ resulting from glutamate exposure compared with the maximal Ca 2ϩ uptake capacity may determine susceptibility to excitotoxic neuronal death.
Our data do not support that glutamate exposure induces pervasive stimulation of the mitochondrial permeability transition in cortical neurons or cerebellar granule cells. If the permeability pore was open at the 10-min time point that we began our isolation, the EGTA in the buffer would sequester the matrix Ca 2ϩ (68). We observe substantial alamethicininduced Ca 2ϩ release from the mitochondria, and therefore a significant portion of mitochondria was impermeable to EGTA at the time of isolation. However, we cannot eliminate the possibility that a subpopulation of mitochondria has undergone the permeability transition, which would lead to an underestimation of the glutamate-induced mitochondrial Ca 2ϩ load. However, if the permeability transition occurs in a subpopulation of mitochondria during a 10-min glutamate exposure, this event is insufficient to induce significant neuronal death (Fig.  3). Neuronal mitochondria were also resistant to a permeability transition when they were loaded with Ca 2ϩ in vitro. At Ca 2ϩ concentrations sufficient to saturate Ca 2ϩ uptake capacity, no permeability increases were detected (Fig. 5). Although others have found indications of increased mitochondrial permeability in intact neurons (24, 69 -74), the lack of consistent protective effects of permeability transition inhibitors in excitotoxicity models (11, 70, 74 -76) renders the literature data controversial. Recently, it has been suggested that the permeability transition pore in brain mitochondria has altered sensitivity to conventional pore inhibitors (77). Regardless, the extent of permeability pore contribution to excitotoxic neuronal death requires further investigation.
In conclusion, these results have revealed that glutamate exposure of cortical neurons results in early inhibition of oxidative phosphorylation, in the absence of profound respiratory inhibition, or changes that might be the downstream result of commitment to cell death. Although the approach taken here has caveats, it revealed a type of Ca 2ϩ -induced injury that fluorescence imaging or whole cell respiration experiments are unable to detect. We propose that Ca 2ϩ -mediated inhibition of oxidative phosphorylation may have pathological relevance to acute as well as chronic excitotoxic injury.