Synaptic Mitochondria Are More Susceptible to Ca2+Overload than Nonsynaptic Mitochondria*

Mitochondria in nerve terminals are subjected to extensive Ca2+fluxes and high energy demands, but the extent to which the synaptic mitochondria buffer Ca2+ is unclear. In this study, we identified a difference in the Ca2+ clearance ability of nonsynaptic versus synaptic mitochondrial populations enriched from rat cerebral cortex. Mitochondria were isolated using Percoll discontinuous gradients in combination with high pressure nitrogen cell disruption. Mitochondria in the nonsynaptic fraction originate from neurons and other cell types including glia, whereas mitochondria enriched from a synaptosomal fraction are predominantly neuronal and presynaptic in origin. There were no differences in respiration or initial Ca2+ loads between nonsynaptic and synaptic mitochondrial populations. Following both bolus and infusion Ca2+ addition, nonsynaptic mitochondria were able to accumulate significantly more exogenously added Ca 2+ than the synaptic mitochondria before undergoing mitochondrial permeability transition, observed as a loss in mitochondrial membrane potential and decreased Ca2+ uptake. The limited ability of synaptic mitochondria to accumulate Ca2+ could result from several factors including a primary function of ATP production to support the high energy demand of presynaptic terminals, their relative isolation in comparison with the threads or clusters of mitochondria found in the soma of neurons and glia, or the older age and increased exposure to oxidative damage of synaptic versus nonsynaptic mitochondria. By more readily undergoing permeability transition, synaptic mitochondria may initiate neuron death in response to insults that elevate synaptic levels of intracellular Ca2+, consistent with the early degeneration of distal axon segments in neurodegenerative disorders.

Mitochondria in nerve terminals are subjected to extensive Ca 2؉ fluxes and high energy demands, but the extent to which the synaptic mitochondria buffer Ca 2؉ is unclear. In this study, we identified a difference in the Ca 2؉ clearance ability of nonsynaptic versus synaptic mitochondrial populations enriched from rat cerebral cortex. Mitochondria were isolated using Percoll discontinuous gradients in combination with high pressure nitrogen cell disruption. Mitochondria in the nonsynaptic fraction originate from neurons and other cell types including glia, whereas mitochondria enriched from a synaptosomal fraction are predominantly neuronal and presynaptic in origin. There were no differences in respiration or initial Ca 2؉ loads between nonsynaptic and synaptic mitochondrial populations. Following both bolus and infusion Ca 2؉ addition, nonsynaptic mitochondria were able to accumulate significantly more exogenously added Ca 2؉ than the synaptic mitochondria before undergoing mitochondrial permeability transition, observed as a loss in mitochondrial membrane potential and decreased Ca 2؉ uptake. The limited ability of synaptic mitochondria to accumulate Ca 2؉ could result from several factors including a primary function of ATP production to support the high energy demand of presynaptic terminals, their relative isolation in comparison with the threads or clusters of mitochondria found in the soma of neurons and glia, or the older age and increased exposure to oxidative damage of synaptic versus nonsynaptic mitochondria. By more readily undergoing permeability transition, synaptic mitochondria may initiate neuron death in response to insults that elevate synaptic levels of intracellular Ca 2؉ , consistent with the early degeneration of distal axon segments in neurodegenerative disorders.
Mitochondria are important regulators of cellular Ca 2ϩ homeostasis, producers of ATP via oxidative phosphorylation, and regulators of cell death pathways (for reviews see Refs. 1 and 2). Mitochondria assist in maintaining Ca 2ϩ homeostasis by sequestering and releasing Ca 2ϩ (2)(3)(4). Normal Ca 2ϩ cycling occurs by the movement of Ca 2ϩ into mitochondria via the Ca 2ϩ uniporter and slow efflux via the Na ϩ /Ca 2ϩ antiporter or by Na ϩindependent mechanisms (1,3). Isolated mitochondria in the presence of phosphate take up Ca 2ϩ to a fixed capacity, in a membrane potential (⌬⌿ m )dependent fashion (5)(6)(7). When the mitochondria become overloaded with Ca 2ϩ , they undergo the cataclysmic mitochondrial permeability transition (mPT) 3 via formation of a nonselective pore that allows solutes of 1500 daltons or smaller to pass through the usually impermeable inner mitochondrial membrane with a resultant rupture of the outer mitochondrial membrane caused by osmotic swelling (2, 8 -12).
Previous studies have demonstrated substantial mitochondrial heterogeneity that exists among organs and within the CNS. Nonsynaptic brain mitochondria are more resistant to Ca 2ϩ -induced opening of mPT, assessed by mitochondrial swelling, when compared with liver mitochondria (13)(14)(15). Within the CNS, there are regional differences in mitochondrial populations with regard to Ca 2ϩ -induced mPT threshold and reactive oxygen species (ROS) production (16 -18). There is also regional and cellular heterogeneity in the composition, morphology, and trafficking of mitochondria (19).
Synaptic mitochondria exist in an environment where they are exposed to extensive Ca 2ϩ influx. Although synaptic mitochondria are important in Ca 2ϩ clearance, it is uncertain whether this is predominantly by direct Ca 2ϩ removal or by providing ATP for the plasma membrane Na ϩ /Ca 2ϩ exchanger (20). Presynaptic mitochondria are typically punctuate and isolated, in contrast to the mitochondrial threads and clusters found in other regions of neurons and astrocytes (21,22). Synaptic mitochondria are synthesized in the cell body of neurons and then transported down the axon or dendrite (23)(24)(25). Within the central nervous system, mitochondria have an apparent half-life of approximately one month (26). As a result of transport, synaptic mitochondria may be "older" than mitochondria in the soma of neurons and glial cells and may exhibit greater cumulative damage from oxidative stress. As mitochondria age, they become more heterogeneous and on average become more depolarized (27). Because of their different function, morphology, or age, presynaptic mitochondria may handle Ca 2ϩ differently than mitochondria in other regions of neurons and other cell types.
Previous studies have compared isolated nonsynaptic and synaptic mitochondria with regard to metabolism and lipid composition (19, 28 -35). Distinctions in Ca 2ϩ handling between isolated synaptic and nonsynaptic mitochondria have not been examined previously. In studies using Ficoll density gradients, mitochondria were isolated from nonsynaptic, "light" synaptic, and "heavy" synaptic fractions. The heavy synaptic fractions exhibited a greater protein/lipid ratio, greater lipid peroxidation, and lower levels and activities of respiratory enzymes and were thought to reflect old mitochondria (34,35). In contrast, light synaptic and nonsynaptic mitochondria were largely similar in terms of * This work was supported by National Institutes of Health, United States Public Health enzyme, activities, lipid content, and lipid peroxidation. Percoll density gradients result in less contamination of the mitochondrial and synaptosomal fractions, and mitochondria isolated from synaptic and nonsynaptic populations exhibit similar enrichment, enzyme activity, and respiratory activity (32,36). The purpose of the present study was to compare the ability of well coupled isolated synaptic versus nonsynaptic brain mitochondria to accumulate exogenously added Ca 2ϩ .
Mitochondrial Isolation-All experimental protocols involving animals were approved by the University of Kentucky Animal Use and Care Committee. Male Sprague-Dawley (SD) rats (250 -300 g, 3 months of age) were used in all studies with the exception of the studies comparing the findings with those obtained in 3-month-old Fisher 344 (F344) rats. All of the animals were obtained from Harlan (Indianapolis, IN). As previously described (37), following carbon dioxide asphyxiation, the rats were decapitated, and the brains were rapidly removed. The cortices were dissected out and placed in a glass Dounce homogenizer containing five times the volume of isolation buffer (215 mM mannitol, 75 mM sucrose, 0.1% BSA, 20 mM HEPES, 1 mM EGTA, pH adjusted to 7.2 with KOH). The tissue was homogenized, and an equal volume of 30% Percoll in isolation buffer was added (ϳ4 ml). The resultant homogenate was layered on a discontinuous Percoll gradient with the bottom layer containing 40% Percoll solution in isolation buffer, followed by a 24% Percoll solution, and finally the sample in a 15% Percoll solution. The density gradients were spun in a Sorvall RC-5C plus superspeed refrigerated centrifuge (Asheville, NC) in a fixed angle SE-12 rotor at 30,400 ϫ g for 10 min. Usage of two Percoll density gradients for cortical regions from each animal improved the resolution of nonsynaptic mitochondria on the Percoll density gradient.
Following centrifugation, band 2 (synaptosomes) and band 3 (nonsynaptic mitochondria) (36) were separately removed from the density gradient. Each fraction was placed in separate tubes, and 10 ml of isolation buffer was added. The samples were washed by centrifugation at 16,700 ϫ g for 15 min. The supernatant was discarded, and the loose pellet was resuspended in the 1 ml of isolation buffer. A nitrogen cell disruption bomb (model 4639; Parr Instrument Company, Moline, IL) cooled to 4°C was used to burst the synaptosomes within this fraction (37,38). Both the nonsynaptic mitochondria and synaptosomes were placed in the nitrogen disruption bomb for 10 min at 1000 p.s.i. Previously, we demonstrated that the nitrogen disruption method does not impair mitochondrial function (37).
The nonsynaptic mitochondrial and the nitrogen-disrupted synaptosomal mitochondrial fractions were placed in separate 15-ml conical tubes. An equal volume of 30% Percoll was added to each sample, and discontinuous Percoll density gradient centrifugation was performed as described above. Band 3 was obtained from each of the gradients, and 10 ml of isolation buffer without EGTA (215 mM mannitol, 75 mM sucrose, 0.1% BSA, 20 mM HEPES, pH is adjusted to 7.2 with KOH) was added. The fractions were centrifuged at 16,700 ϫ g for 15 min and subsequently at 11,000 ϫ g for 10 min. The resultant pellet was resuspended in 1 ml of isolation buffer without EGTA and centrifuged at 10,000 ϫ g for 10 min. The final mitochondrial pellet was resuspended in isolation buffer without EGTA to yield a protein concentration of ϳ10 mg protein/ml and stored on ice. Protein concentration was determined using the bicinchoninic acid protein assay (Pierce).
Respiration Measurements-The respiratory activity of isolated mitochondria was measured using a Clark-type oxygen electrode (Hansatech Instruments, Norfolk, UK) as previously described (39). Approximately 100 g protein/ml of isolated nonsynaptic or synaptic mitochondria were suspended in a sealed, constantly stirred, and thermostatically controlled chamber at 37°C in KCl respiration buffer (125 mM KCl, 0.1% BSA, 20 mM HEPES, 2 mM MgCl 2 , 2.5 mM KH 2 PO 4 , pH 7.2). The rate of oxygen consumption was calculated based on the slope of the response of isolated mitochondria to the successive administration of oxidative substrates (5 mM pyruvate and 2.5 mM malate): 150 M ADP added twice in 1-min intervals; 1 M oligomycin; 1 M FCCP; and finally 1 mM succinate (40). The respiratory control ratio was determined by dividing the rate of oxygen consumption/min for state III (in the presence of ADP, second addition) by state IV (in the absence of ADP and presence of oligomycin). Only isolated mitochondrial preparations with an respiratory control ratio of over 5 were used in the study. The states of mitochondrial respiration described by Chance and Williams (41) were also calculated (nmol of oxygen consumed/mg of protein) in KCl respiration buffer.
Fluorescent Spectrofluorophotometer Assays-Fractions enriched in nonsynaptic and synaptic mitochondria (50 g protein/ml) were placed in 2 ml of KCl respiration buffer in a constantly stirred, temperature-controlled cuvette at 37°C with 100 nM CaG5N; excitation, 506 nm; emission, 532 nm; and 100 nM TMRE; excitation, 550 nm; emission, 575 nm; in the Shimadzu RF-5301PC spectrofluorophotometer (Kyoto, Japan). CaG5N was used to monitor extramitochondrial Ca 2ϩ , and TMRE was used to simultaneously monitor changes in ⌬⌿ m . Each time scan began with a base-line reading followed by a 5 mM pyruvate and 2.5 mM malate addition at 1 min, then 150 M ADP at 2 min, and then 1 M oligomycin at 3 min. At 5 min, Ca 2ϩ was added by a gradual delivery via an KD Scientific model 310 series infusion syringe pump (Holliston, MA) (5, 42) (160 nmol of Ca 2ϩ /mg of protein/min) or through bolus additions (1000 nmol of Ca 2ϩ /mg of protein or 50 M Ca 2ϩ ) until the mitochondria were no longer able to buffer the added Ca 2ϩ . The chemical uncoupler CCCP was added toward the end of each run. The traces presented are representative of at least three separate, independent experiments.
The spectrofluorophotometer traces were quantified by calculating the average base-line CaG5N fluorescence readings 1 min prior to the beginning of the Ca 2ϩ infusion or before the first bolus addition using the Shimadzu Hyper RF software and Microsoft Excel. The time point at which the CaG5N signal was 150% above the average base-line reading was considered to be the point at which the mitochondria were overloaded and no longer capable of removing Ca 2ϩ from the media. Mitochondrial Ca 2ϩ uptake capacity was calculated as the amount of Ca 2ϩ added or infused (nmol/mg) prior to the point at which the CaG5N signal was 150% above the average base-line reading.
Reactive Oxygen Species Production-Mitochondrial ROS production was measured using 25 M DCF (485 nm, 530 nm) in the Biotek Synergy HT plate reader as previously described (17,43,44). Isolated mitochondria (25 g of protein/ml) were added to 100 l of KCl respiration buffer with 5 mM pyruvate and 2.5 mM malate as oxidative substrates at 37°C. ROS production was calculated as the maximum DCF fluorescence following 15 min of incubation, expressed in arbitrary fluorescence units. Mitochondrial ROS production in the presence of oligomycin (to induce maximal ROS production) or FCCP (to induce minimal ROS production) was also determined to ensure that our measurements were within the range of the ROS indicator.
Western Blotting-Isolated nonsynaptic and synaptic mitochondria in isolation buffer plus a protease inhibitor mixture (Complete Mini; Roche Applied Science) were centrifuged at 10,000 ϫ g for 10 min. The resultant mitochondrial pellet was resuspended in 100 l of isolation buffer plus protease inhibitors with 0.01% Triton X-100, sonicated for 20 s, and centrifuged at 10,000 ϫ g for 10 min. The supernatant was used for Western blots. Sample buffer was added to the samples based on relative protein concentrations determined from the bicinchoninic acid protein assay, and all of the lanes were loaded with the same amount of protein (5 g/lane).
The samples were separated by SDS-PAGE using 10 or 12.5% Trisacrylamide/bis gels, along with Bio-Rad low range molecular weight markers. Following SDS-PAGE, the polypeptides were transferred electrophoretically onto 0.2 M nitrocellulose membranes. The membranes were incubated at room temperature for 1 h in 5% nonfat milk in 50 mM Tris-saline containing 0.05% Tween 20 at pH 7.5 (TTBS). The blots were incubated overnight in the primary antibody in TTBS at 22°C. The primary antibodies used in study included monoclonal cytochrome oxidase subunit IV (COXIV) at 1:20,000 (Molecular Probes); monoclonal post-synaptic density 95 protein (PSD-95) at 1:20,000 (BD Biosciences, San Jose, CA); and polyclonal voltage-dependent anion channel (VDAC) at 1:10,000 (Affinity Bioreagents, Golden, CO). After overnight incubation in primary antibody, the membranes were rinsed three times in TTBS and incubated in secondary antibody for 1 h either in horseradish peroxidase-conjugated goat anti-mouse IgG (1:3000) for COXIV and PSD-95 or in horseradish peroxidase-conjugated goat anti-rabbit IgG (1:3000) for VDAC. The blots were rinsed thoroughly in TTBS and were briefly incubated in the Pierce SuperSignal Pico chemiluminescent substrate. Finally, the blots were visualized using a Kodak Image Station 2000R and the Kodak Molecular Imaging software.
Statistics-Statistical analyses were performed using either an unpaired t test or a one-way analysis of variance ( p Ͻ 0.05) with Scheffé's post hoc analysis when appropriate. The results are expressed as the group means (Ϯ S.E.) from at least three independent experiments, and group size is indicated for each experiment in the figure legends.

RESULTS
Isolation of Synaptic Mitochondria-To isolate well coupled synaptic mitochondria, the isolation procedure utilized two separate centrifugations on discontinuous Percoll gradients and a nitrogen decompression technique (Fig. 1). Both the synaptic and nonsynaptic mitochondria underwent the nitrogen disruption and two runs through discontinuous Percoll density gradients. The nitrogen cell disruption avoids the damage to mitochondria caused by detergent-based disruption methods (45) and has been demonstrated to yield well coupled mitochondria (37). The average mitochondrial yields from cortical tissue pooled from two rats were 946 g for the nonsynaptic fraction and 453 g for the synaptic fraction. Western blots were performed using antibodies to probe for the outer mitochondrial membrane protein, VDAC; the inner mitochondrial membrane protein, COXIV; and the synaptosomal protein, PSD-95 (Fig. 2). The nonsynaptic mitochondrial fraction had strong immunoreactivity for the mitochondrial membrane proteins  COXIV and VDAC but not for PSD-95. The synaptosomes were positive for PSD-95 marker after the first Percoll centrifugation and after the nitrogen disruption. Following the second Percoll gradient, the synaptic mitochondrial-enriched fraction demonstrated strong immunoreactivity for both mitochondrial markers, but PSD-95 was not detected.
Similar Bioenergetics in Nonsynaptic and Synaptic Mitochondria-Following mitochondrial isolation, the rate of oxygen consumption in presence of pyruvate and malate as the oxidative substrates was measured (Fig.  3). Similar respiration rates were observed in the nonsynaptic and synaptic mitochondria (Fig. 3A). There was no significant difference in oxygen consumption in any of the different classical states of respiration (Fig. 3B) or in the respiratory control ratio (Fig. 3C). Together, the results demonstrate that both populations of isolated mitochondria were well coupled and bioenergetically active following the isolation procedure.
Increased Ca 2ϩ Accumulation in Nonsynaptic versus Synaptic Mitochondria in Two Rat Strains-Isolated nonsynaptic and synaptic mitochondria were incubated with an extramitochondrial, low affinity Ca 2ϩ fluorescent dye, CaG5N, and a ⌬⌿ m fluorescent indicator, TMRE, and placed in a constantly stirred, temperature-controlled cuvette at 37°C in a spectrofluorophotometer. All of the functional assays on the mitochondria were done in a KCl-based respiration buffer containing magnesium, inorganic phosphates, and BSA. The indicator TMRE was used in "quench" mode such that at high mitochondrial ⌬⌿ m fluorescence is lower than at lower ⌬⌿ m , because of dye stacking within the matrix (46). After obtaining a base-line reading, oxidative substrates (5 mM pyruvate and 2.5 mM malate) were added, allowing the mitochondria to generate a high ⌬⌿ m indicated by the sharp downward deflection of the TMRE trace (Fig. 4C). One min later, 150 M ADP was added, which caused the mitochondria to depolarize and use their ⌬⌿ m to phosphorylate the added ADP to ATP. Afterward, the ATP synthase inhibitor oligomycin (1 m) was added, and the high ⌬⌿ m was reestablished. These early additions served as internal controls in each experiment to ensure that both the nonsynaptic and the synaptic mitochondrial preparations were well coupled and bioenergetically competent for subsequent experiments.
Next, the ability of isolated nonsynaptic and synaptic mitochondria to buffer Ca 2ϩ was investigated using two approaches in two rat strains. First, Ca 2ϩ was infused at 160 nmol of Ca 2ϩ /mg of protein/min using an infusion pump (Fig. 4) into the cuvette. The infusion was terminated after the mitochondria were no longer able to accumulate the added Ca 2ϩ as demonstrated by an increase in CaG5N signal. Two minutes later, CCCP was added causing the total collapse of any remaining ⌬⌿ m and a total release of Ca 2ϩ within the mitochondria. To determine whether the results obtained were specific to the SD rats (Fig. 4, A and B), an outbred line of rats, experiments were also performed using an inbred line, F344 rats (Fig. 4, C  and D). In both rat strains, the nonsynaptic mitochondrial populations were able to buffer significantly more of the infused Ca 2ϩ than the synaptic mitochondria. Also, in both strains, collapse of the ⌬⌿ m preceded the impairment of Ca 2ϩ accumulation.
The second approach we utilized was bolus additions of 1000 nmol of Ca 2ϩ /mg of protein (50 M) to isolated mitochondria populations to simulate pathologic loads of Ca 2ϩ (Fig. 5). Following the first additions of Ca 2ϩ , the CaG5N signal increased transiently. This was accompanied by a loss in ⌬⌿ m, which was utilized to drive the uptake of Ca 2ϩ via the electrogenic Ca 2ϩ uniporter. Both the CaG5N and TMRE fluorescence then returned toward base line as the mitochondria were able to take up the extramitochondrial Ca 2ϩ (Fig. 5). As with a constant infusion, the synaptic mitochondrial populations were unable to buffer as much Ca 2ϩ before undergoing mPT as nonsynpatic mitochondria. The kinetics of Ca 2ϩ uptake in the two populations were similar with the first bolus addition, but with the second bolus addition the rate of Ca 2ϩ uptake was decreased in synaptic versus nonsynaptic mitochondria, and this decrease was exacerbated with subsequent additions of Ca 2ϩ . The loss of ⌬⌿ m was caused by Ca 2ϩ , because the FIGURE 3. Differences in the mitochondrial bioenergetics were not observed between nonsynaptic and synaptic mitochondria. A, representative traces from nonsynaptic (darker traces) and synaptic (lighter trace) mitochondrial oxygen consumption measurements in the presence of oxidative substrates (pyruvate and malate), ADP, oligomycin, FCCP, and succinate. B, there were no significant differences in any of the states of respiration. State III is particularly high because of the well coupled mitochondria produced through this isolation procedure. C, the respiratory control ratios (RCR), which is state III respiration divided by state IV respiration, were similar for both populations and were well above the acceptable range of 5. The results are the means Ϯ S.E. from seven independent experiments.
addition of EGTA, the Ca 2ϩ chelator, was able to partially rescue the ⌬⌿ m (Fig. 5B).
Overall, there was a significant increase in the amount of Ca 2ϩ (nmol of Ca 2ϩ /mg of protein) taken up prior to mPT by nonsynaptic mitochondrial populations as compared with synaptic mitochondrial populations isolated from SD and F344 brain tissue (Fig. 6B). When the mPT inhibitor cyclosporin A (CsA) was added, there was an increase in the amount of Ca 2ϩ accumulated prior to mPT by both nonsynaptic and synaptic mitochondria (Fig. 6, A and C). However, the influence of CsA on Ca 2ϩ accumulation was significant (p Ͻ 0.005) in nonsynaptic mitochondrial fractions but not in synaptic mitochondria. In the presence of CsA, Ca 2ϩ accumulation prior to mPT by nonsynaptic mitochondria remained greater than in synaptic mitochondria (p Ͻ 0.0001) (Fig. 6C). The greater uptake capacity of nonsynaptic mitochondria was also evident using the Ca 2ϩ bolus paradigm (p Ͻ 0.004) (Fig. 6D).
To investigate the possibility that the initial Ca 2ϩ loading within the mitochondria was different between the nonsynaptic and synaptic mitochondria after the isolation technique, CCCP, a chemical uncoupler, was added at 7 min after the addition of substrates to induce efflux of the stored Ca 2ϩ (Fig. 7). There was no difference in the level of the CaG5N signal after CCCP addition, indicating no substantial differences in the amount of Ca 2ϩ contained with the matrix after isolation. This argues against differential Ca 2ϩ loading in synaptic versus nonsynaptic mitochondria as a cause of the discrepancy in Ca 2ϩ uptake observed in the two mitochondrial populations.
Ca 2ϩ uptake evaluated using both the infusion method and the bolus additions was dependent upon the electrogenic Ca 2ϩ uniporter, as demonstrated by the addition of 600 nM RuRed, which totally prevented Ca 2ϩ uptake (Fig. 8). In the presence of RuRed, there was no loss of ⌬⌿ m following the addition of Ca 2ϩ .
Similar ROS Production between the Nonsynaptic and Synaptic Mitochondria-Another mechanism that might underlie the different abilities of nonsynaptic and synaptic mitochondria to accumulate Ca 2ϩ is greater ROS production and oxidative damage in the synaptic population. There was no significant difference in basal ROS production, measured using DCF, in isolated nonsynaptic and synaptic mitochondria (Fig. 9).

DISCUSSION
The results of this study demonstrate that nonsynaptic mitochondria isolated from rat cortex can accumulate significantly more Ca 2ϩ than FIGURE 4. Synaptic mitochondria are more susceptible to Ca 2؉ overload than nonsynaptic mitochondria following a slow infusion of Ca 2؉ . Isolated nonsynaptic or synaptic mitochondria were placed in a constantly stirred, temperature-controlled cuvette inside a spectrofluorophotometer. CaG5N (A and C) and TMRE (B and D) were monitored simultaneously. Each sample was given malate and pyruvate as oxidative substrates, which caused an increase in ⌬⌿ m marked by a downward deflection (B). ADP caused a loss of some ⌬⌿ m as ADP is phosphorylated into ATP. Next oligomycin, the ATP synthase inhibitor, was added, and the mitochondria were at maximal ⌬⌿ m . Ca 2ϩ infusion began at 5 min (160 nmol of Ca 2ϩ /mg of protein) causing a small, initial increase in CaG5N fluorescence until the mitochondria were able to accumulate the added Ca 2ϩ (A and C). The nonsynaptic mitochondria (darker trace) were able to buffer more of the infused Ca 2ϩ as compared with the synaptic mitochondria (lighter trace). By comparing the results obtained to the Ca 2ϩ signal observed in the absence of added mitochondria (buffer, lightest trace), the modest Ca 2ϩ accumulation by the synaptic mitochondria is evident. Similar results were observed in isolated nonsynaptic and synaptic mitochondria from SD (A and B) or F344 (C and D) rats. There was also a loss of ⌬⌿ m , as indicated by an increase in the TMRE fluorescence, earlier by the synaptic mitochondria from both strains, SD (C) and F344 (D). The y axis of the graphs is expressed in CaG5N or TMRE fluorescent arbitrary units (AU). These are representative traces from nonsynaptic and synaptic mitochondria from the same preparation and this experiment was repeated in eight (SD) and five (F344) independent experiments. the synaptic mitochondria before undergoing mPT. This difference in Ca 2ϩ accumulation was observed in two strains of rats utilizing both bolus Ca 2ϩ additions and gradual Ca 2ϩ infusion paradigms. CsA, an mPT inhibitor, increased the amount of Ca 2ϩ that could be accumu-lated in nonsynaptic mitochondria but did not have a significant effect in synaptic mitochondria. The uptake of Ca 2ϩ in both mitochondrial populations was dependent on the Ca 2ϩ uniporter and completely inhibited by the addition of RuRed. The differences were not due to initial Ca 2ϩ loads in the isolated mitochondria or the production of mitochondrial ROS. There were no differences between nonsynaptic and synaptic mitochondria in the rate of oxygen consumption, measured during the classical states of respiration, indicating that the electron transport chain was well coupled to oxidative phosphorylation allowing for the maintenance of normal ⌬⌿ m in both populations of mitochondria.
Previous methods for enriching synaptic mitochondria utilized either Ficoll (19,29,40,44) or Percoll (36) discontinuous density gradients and detergents such as digitonin to disrupt synaptosomal membranes. Yields of synaptic mitochondria were low in the previous studies, possibly because of the isolation protocol involving long centrifugation cycles (29) and damaging detergent techniques to release the synaptic mitochondria from synaptosomes (18). In the present study, cortical rat brain mitochondria were isolated from nonsynaptic and synaptic sources using centrifugation on successive discontinuous Percoll gradients and high pressure nitrogen cell disruption. The nonsynaptic and synaptic mitochondria underwent similar manipulations during their enrichment; both populations were exposed to the nitrogen disruption, and both underwent centrifugation through two discontinuous Percoll gradients. Use of the nitrogen disruption method (37) avoided the deleterious effects of detergents on mitochondrial proteins (18).
Synaptic mitochondria are largely presynaptic, because the homogenization procedure pinches off presynaptic terminals at the neck of the axon, and postsynaptic mitochondria are not retained in the synaptosomes (47,48). The mitochondria within the synaptic fraction are derived from both interneurons and projection neurons from subcortical regions. Within the presynaptic bouton, mitochondria serve a number of functions. In addition to ATP synthesis and Ca 2ϩ buffering, they contribute to neurotransmitter synthesis and catabolism. The bioenergetic demands in the presynaptic terminal are high, with ATP being required for endo-and exocytosis, in addition to maintenance of ion homeostasis. Presynaptic mitochondria are constantly exposed to high Ca 2ϩ transients associated with transmitter release (49).
Nonsynaptic mitochondria originate from both neurons and nonneuronal cells. In rat parietal cortex, glial cell density is approximately double the neuronal density, and mitochondria occupy a smaller percentage of the cytosol in glia as compared with neurons (50). Using a similar method to that used in the present study to isolate nonsynaptic mitochondria, Kristian and colleagues (51) found faint peripheral benzodiazepine receptor immunoreactivity, a marker of astrocytic mitochondria (52), in the nonsynaptic mitochondrial fraction in contrast to strong immunoreactivity in mitochondria isolated from cultured astrocytes. The weak peripheral benzodiazepine receptor immunoreactivity is not due to decreased expression in adult rats (52), indicating that the nonsynaptic mitochondrial fraction may be predominantly neuronal. At present, it is not possible to isolate mitochondria from neurons versus glia in the adult CNS. Although mitochondria can be isolated Similar results were obtained in SD rats (not shown). In B-D, the y axis indicates the amount of Ca 2ϩ added to the mitochondrial suspension. Using the 50% above base-line threshold calculation described under "Materials and Methods" and illustrated in A to quantify the amount of Ca 2ϩ added prior to the mitochondria undergoing permeability transition, the SD and F344 infusion data were analyzed by using an unpaired t test (B). There were no significant differences between the nonsynaptic SD (n ϭ 10) and the nonsynaptic F344 (n ϭ 5) or between the synaptic SD (n ϭ 8) and the synaptic F344 (n ϭ 5). Therefore all of the traces from the two strains were combined for subsequent analysis (C). The data from all of the strains were analyzed using a one-way analysis of variance with a Scheffé 's post-hoc analysis. The nonsynaptic mitochondria (n ϭ 15) were exposed to more Ca 2ϩ than the synaptic mitochondria (n ϭ 13) prior to undergoing mPT (p Ͻ 0.0001). The addition of CsA enabled nonsynaptic mitochondria (n ϭ 6) to withstand greater accumulated more Ca 2ϩ than the nonsynaptic alone (p Ͻ 0.0036) or the synaptic alone (p Ͻ 0.0001) or the synaptic with CsA (n ϭ 5) (p Ͻ 0.0001). The synaptic mitochondria with CsA were not significantly different from the synaptic mitochondria alone. D, finally, nonsynaptic (n ϭ 5) and synaptic (n ϭ 5) mitochondrial traces using the bolus addition paradigm were quantitated using an unpaired t test. The nonsynaptic mitochondria accumulated significantly more bolus additions of Ca 2ϩ than the synaptic mitochondria. Each group within all of these analyses was represented by at least five independent experiments. from primary cultures of neurons and astrocytes (51), the yields are low, and the Ca 2ϩ handling abilities of the two mitochondrial populations have not been evaluated.
The differences in mitochondrial Ca 2ϩ handling were evident following both bolus Ca 2ϩ additions and continuous Ca 2ϩ infusion to the isolated synaptic and nonsynaptic mitochondria. With both methods, a loss of ⌬⌿ m preceded or accompanied the impairment in mitochondrial Ca 2ϩ accumulation, similar to results previously obtained in brain and liver mitochondria (5). Using the bolus additions paradigm, there are high bioenergetic consequences because of immediate changes on the proton gradient (5). In addition, the various calcium phosphate salts (mono-, di-, and tri-calcium orthophosphate and hydroxyapatite) produced in response to the bolus addition may differ from those resulting from a slow accumulation of Ca 2ϩ (5). Use of a gradual infusion technique allows for mitochondria to slowly accumulate the Ca 2ϩ with minimal bioenergetic consequences and thereby allows a more accurate estimate of the Ca 2ϩ buffering capacity of the mitochondria (5,53). However, even with slow infusion, the amounts of Ca 2ϩ added are much greater than physiological free Ca 2ϩ cytosolic levels in presynaptic terminals, which are ϳ100 nM at rest and estimated to reach 8.5 M following sustained depolarization in the mouse calyx of Held, a large glutamatergic CNS terminal (54). The mechanism underlying the greater amount of Ca 2ϩ accumulation following bolus Ca 2ϩ addition as compared with gradual infusion is uncertain. Differences in Ca 2ϩ uptake may reflect increased recovery times in experiments using the bolus additions, whereas mitochondria are continuously exposed to elevated Ca 2ϩ in the infusion paradigm. It is interesting to note that the uptake dynamics following the first bolus addition was identical between the two populations, but that uptake was progressively decreased in the synaptic population with successive bolus Ca 2ϩ additions, suggesting that a subpopulation of the synaptic mitochondria undergoes permeability transition with each Ca 2ϩ bolus. This is currently being investigated in our laboratories. Alternatively, formation of various forms of Ca 2ϩ phos-phate (5, 55) could contribute to differences in Ca 2ϩ accumulation ability between the nonsynaptic and synaptic mitochondria in the two paradigms.
The difference in Ca 2ϩ handling between isolated synaptic and nonsynaptic CNS mitochondria has not been examined previously, although there is substantial evidence for differences in Ca 2ϩ buffering and mPT induction among various neural and non-neural mitochondrial populations. Isolated brain mitochondria are more resistant to mPT induction as compared with liver mitochondria (14, 15, 56 -58). However, some of this resistance has been attributed to the use of strong detergents in the preparations and may not reflect differences in the mitochondria themselves (45). Across brain regions, variability to undergo Ca 2ϩ -induced dysfunction has been observed (16 -18). This may be due, at least in part, to differing levels of cyclophilin D, a peptidyl-prolyl cis-trans isomerase that facilitates Ca 2ϩ -induced permeabilization of the inner mitochondrial membrane and mPT pore formation (59,60). CsA antagonizes mPT by binding to cylophilin D (61,62). We previously observed a difference in the response to CsA in isolated nonsynaptic cortical versus spinal cord mitochondria consistent with increased cyclophilin D mRNA expression in spinal cord (43). The cyclophilin D content of synaptic and nonsynaptic mitochondria remains to be determined. However, the response of the two mitochondrial populations to CsA suggests that alternate or additional mechanisms are likely responsible for the difference in Ca 2ϩ handling. Oxidation of the adenosine nucleotide translocator, a major component of the mPT pore, can also facilitate mPT (43,63). Mitochondria in the presynaptic terminal must be transported from the cell body. As they age, mitochondria undergo oxidative damage and become depolarized (27). Mitochondria with a high ⌬⌿ m are transported toward the presynaptic terminal via fast axonal transport, whereas mitochondria with low potential are transported retrogradely to the soma to be degraded (64). Thus, a possible explanation for the greater propensity of synaptic mitochodria to undergo Ca 2ϩ -induced mPT is their increased age and oxidation, although this remains to be determined experimentally.
The impaired Ca 2ϩ uptake capacity of synaptic mitochondria may also reflect their function and morphology. The extent of Ca 2ϩ buffering by neuronal, presynaptic mitochondria has been a subject of debate (65). Synaptic mitochondria may dampen large elevations in Ca 2ϩ associated with repetitive synaptic activity (66 -70). However, there is also evidence that Ca 2ϩ release from mitochondria can potentiate transmitter release (71). Within the large calyx of Held brainstem terminal, mitochondria are associated with a specialized adherens complex implicated in exocytosis and endocytosis (72). Similar structures are evident in other CNS presynaptic terminals (73). Presynaptic mitochondria are typically punctuate and isolated, in contrast to the mitochondrial threads found in neuronal dendrites and the mitochondrial clusters observed in neuronal soma (21,22). Within primary cortical astrocytes, mitochondria are predominantly observed as threads, although isolated mitochondria can be found in the periphery (22). Dynamin-related protein 1 is required for mitochondrial fission (74,75) but also reduces mitochondrial Ca 2ϩ retention capacity (76). In the Drosophila drp1 mutant, the loss of mitochondrial fission results in a near absence of synaptic mitochondria (77). This does not influence basal synaptic properties but impairs the ability to mobilize reserve pool vesicles, and this is partially rescued by the addition of exogenous ATP. Together, the above results suggest that the primary function of synaptic mitochondria to maintain high ratios of ATP:ADP that can be utilized for exocytosis and endocytosis. The limited ability of synaptic mitochondria to buffer large Ca 2ϩ loads may result from their punctuate morphology, in FIGURE 7. Addition of the uncoupler CCCP demonstrates similar Ca 2؉ levels within the mitochondrial matrix of synaptic and nonsynaptic mitochondria. CCCP was added at 7 min to cause release of Ca 2ϩ into the buffer, detected as an increase in CaG5N fluorescence (A) and complete loss of the ⌬⌿ m , assessed using TMRE (B). These representative traces are from the same mitochondrial preparation, and this experiment was repeated in two independent experiments.
contrast to the tubular mitochondrial morphology evident in other regions of neurons and glia.
The reduced Ca 2ϩ buffering capacity of synaptic mitochondria and their increased propensity to undergo mPT may also have pathological consequences. Degeneration of distal axon segments is one of the initial events observed in several neurodegenerative disorders including amyotrophic lateral sclerosis, spinocerebellar disorders, intoxications, and AIDS (78). Moreover, synapse elimination occurs early in the progression of Alzheimer disease (79,80) and is also implicated in schizophrenia (81).
It is clear that synapse and neurite degeneration can occur independently of cell death (82), and although the mechanisms are unclear, mitochondria are strongly implicated (83). In cultured hippocampal neurons exposed to glutamate, synapse loss precedes cell death (84). An age-or disease-related decline in mitochondrial Ca 2ϩ buffering capacity (17,85) may further increase the vulnerability of presynaptic terminals to mPT and subsequent release of death-related proteins (2,80,86) and thereby contribute to the pathogenesis of late onset neurodegenerative disorders. In summary, the results of the present study demonstrate that mitochondria isolated from rat cortical synaptosomes have an increased propensity to undergo mPT in response to added Ca 2ϩ as compared with mitochondria isolated from a nonsynaptosomal fraction. This difference is not the result of differences in mitochondrial bioenergetics or initial Ca 2ϩ load but may reflect the largely neuronal origin of synaptic mitochondria versus the mixed cellular origin of nonsynaptic mitochondria or the different functions of synaptic versus nonsynaptic mitochondria. It is also possible that the isolated nature of synaptic mitochondria in comparison with the threads and clusters found in other cellular locations, alterations in cyclophilin D content, or the greater age and cumulative oxidative damage to synaptic mitochondria may underlie the differences in Ca 2ϩ accumulation. Although the precise mechanisms underlying the Ca 2ϩ handling differences remain to be determined, the results provide an explanation for the vulnerability of distal axon segments and presynaptic terminals in several neurodegenerative disorders and following neuronal insult.