Address correspondence and reprint requests to Giovanni Manfredi, Weill Graduate School of Medical Sciences of Cornell University, 525 E. 68th St., A-505, New York, NY 10021, USA. E-mail: firstname.lastname@example.org
Mitochondrial respiratory chain dysfunction, impaired intracellular Ca2+ homeostasis and activation of the mitochondrial apoptotic pathway are pathological hallmarks in animal and cellular models of familial amyotrophic lateral sclerosis associated with Cu/Zn-superoxide dismutase mutations. Although intracellular Ca2+ homeostasis is thought to be intimately associated with mitochondrial functions, the temporal and causal correlation between mitochondrial Ca2+ uptake dysfunction and motor neuron death in familial amyotrophic lateral sclerosis remains to be established. We investigated mitochondrial Ca2+ handling in isolated brain, spinal cord and liver of mutant Cu/Zn-superoxide dismutase transgenic mice at different disease stages. In G93A mutant transgenic mice, we found a significant decrease in mitochondrial Ca2+ loading capacity in brain and spinal cord, as compared with age-matched controls, very early on in the course of the disease, long before the onset of motor weakness and massive neuronal death. Ca2+ loading capacity was not significantly changed in liver G93A mitochondria. We also confirmed Ca2+ capacity impairment in spinal cord mitochondria from a different line of mice expressing G85R mutant Cu/Zn-superoxide dismutase. In excitable cells, such as motor neurons, mitochondria play an important role in handling rapid cytosolic Ca2+ transients. Thus, mitochondrial dysfunction and Ca2+-mediated excitotoxicity are likely to be interconnected mechanisms that contribute to neuronal degeneration in familial amyotrophic lateral sclerosis.
Amyotrophic lateral sclerosis (ALS) is a devastating, neurodegenerative disease selectively affecting upper and lower motor neurons. Pathologically, ALS is characterized by extensive loss of motor neurons in the cortex, in the spinal cord and in the brain stem, atrophy of ventral roots, cortico-spinal tract degeneration, neuronal inclusions of aberrant neurofilament proteins, and reactive astrocytosis (Hirano 1991).
Approximately 10% of all ALS cases are familial (fALS), about 20% of which are associated with autosomal dominant mutations in the Cu/Zn-superoxide dismutase gene, SOD1 (Rosen et al. 1993). More than 100 different mutations in the SOD1 gene have been found in fALS patients (http://alsod1.iop.kcl.ac.uk/reports/mutations).
Since many of the fALS-associated SOD1 mutants have normal dismutase activity (Borchelt et al. 1994), SOD1 mutations are thought to result in a toxic gain of function, and a growing body of evidence suggests that mitochondria may be involved in mutant SOD1 toxicity (Xu et al. 2004).
A fundamental question that remains to be answered is whether mitochondrial dysfunction correlates with the progression of SOD1 fALS. To address this question, we have studied mitochondrial Ca2+ capacity in brain and spinal cord of SOD1 transgenic mice at different disease stages. Alterations in Ca2+ homeostasis have been widely implicated in the pathogenesis of several neurodegenerative disorders, including ALS (Harman 1996; Mattson 2000). In excitable cells, such as motor neurons, mitochondria play an important role in handling rapid cytosolic Ca2+ transients (Nicholls et al. 2003; Wang et al. 2003). Therefore, the impairment of this crucial mitochondrial function may contribute significantly to the pathogenesis of ALS.
To study the effects of mutant SOD1 on mitochondrial Ca2+ handling, we investigated Ca2+ capacity in mitochondria isolated from the brain and spinal cord of transgenic mice expressing mutant hSOD1 at different disease stages. We show that impaired mitochondrial Ca2+ handling arises very early on in the course of the disease in G93A mutant hSOD1 mice. This impairment is consistent with an increased susceptibility of hSOD1 mutant mitochondria to undergo permeability transition in response to Ca2+ loads. We find that in young, pre-symptomatic, G93A mice, mitochondrial Ca2+ capacity impairment precedes the onset of overt mitochondrial respiratory dysfunction. Furthermore, in G85R hSOD1 mutant mice, Ca2+ capacity impairment occurs in the absence of oxidative phosphorylation dysfunction. This suggests that Ca2+ capacity impairment may occur upstream in the cascade of pathogenic events leading to mitochondrial dysfunction in fALS.
Materials and methods
Transgenic mice expressing wild-type (Wt, N1029), G93A [high expressors (Gurney et al. 1994)] and G85R mutant hSOD1 (Bruijn et al. 1997), as well as non-transgenic littermates, were obtained from the Jackson Laboratory (Bar Harbor, ME, USA) and bred at the Weill Medical College of Cornell University animal facility. Both male and female mice expressing G93A mutant hSOD1 were studied at various disease stages and compared with non-transgenic age- and gender-matched littermates. In most experiments, a similar number of male and female pairs was used. In these mice, the symptomatic stage, characterized by muscle atrophy and weakness and gait abnormalities, is reached at approximately 3 months-of-age. The end stage, with severe weight loss and almost complete paralysis, occurs at approximately 4 months-of-age. Mice expressing G85R hSOD1 were studied at the symptomatic disease stage (i.e. at 10–12 moths-of-age). Mice expressing Wt hSOD1 (N1029) do not show any sign of paralysis. N1029 mice were killed at 45 and at 140 day-of-age.
Mitochondria isolation and purification
All procedures were approved by the Animal Care and Use Committee of the Weill Medical College of Cornell University. Non-synaptic brain and spinal cord mitochondria were isolated by a Percoll (Sigma, St. Louis, MO, USA) gradient purification method as described previously (Sims 1990). All procedures were conducted at 4°C. Mice were killed by decapitation, then the forebrain was excised and homogenized in a mannitol-sucrose isolation medium (225 mm mannitol, 75 mm ultrapure sucrose, 5 mm HEPES, pH 7.4) containing 0.1 mg/mL fatty acid-free bovine serum albumin (BSA) and 1 mm EGTA. Homogenates were centrifuged for 6 min at 2400 g in a Beckmann L8-80 ultracentrifuge (Beckmann Coulter, Fullerton, CA, USA) using a 50 TI rotor. The supernatant fluid was collected and centrifuged for 14 min at 13 000 g. The resulting pellet was re-suspended in isolation medium containing 15% Percoll and 1 mm EGTA, and layered on top of a discontinuous Percoll gradient (23/40%). The gradient was centrifuged for 22 min at 38 000 g, and the 23/40 interphase layer containing the mitochondria was collected. Mitochondria were washed three times with isolation medium without BSA and centrifuged for 14 min at 13 000 g. The pellet containing purified mitochondria was resuspended in 100 µL isolation buffer without EGTA and BSA.
Spinal cord mitochondria were isolated following the same procedures as for brain, except that in order to obtain sufficient material for the assays, four spinal cords were pooled for each preparation. To collect the spinal cords, mice were killed by decapitation, then cords were extruded from the spinal column by inserting a 16-gauge needle connected to a syringe filled with homogenization buffer into the lumbar thecal sac, and squirting out the entire spinal cord through the rostral foramen.
Liver mitochondria were isolated by tissue homogenization in mannitol-sucrose buffer, as described earlier, followed by centrifugation for 10 min at 596 g in a Beckmann L8-80 ultracentrifuge using a 50 TI rotor. The supernatant fluid was filtered through a gauze and centrifuged for 11 min at 4236 g. Both centrifugation steps were repeated a second time in the same buffer, but without BSA. The mitochondria-enriched pellet was washed in mannitol-sucrose buffer without BSA and EGTA, and finally re-suspended in the same buffer.
Mitochondrial protein concentration was measured with a protein assay kit (Bio-Rad, Hercules, CA, USA), using BSA as a standard.
Measurements of mitochondrial Ca2+ loading capacity
Mitochondrial Ca2+ capacity was estimated with the fluorescent ratiometric dye Fura 6F (Molecular Probes, Eugene, OR, USA). Incubation medium was composed of 125 mm KCl, 20 mm HEPES (pH 7.2), 2 mm KH2PO4, 2 mm MgCl2, 5 mm succinate, 1 µm rotenone and 0.2 mm ADP, with 1 µg/mL oligomycin and 0.3 µm Fura 6. Fura 6 does not enter into mitochondria, and its fluorescence reflects the extra-mitochondrial [Ca2+] (Andreyev and Fiskum 1999). The ratio between the fluorescence of ion-bound and ion-free indicator decreases as mitochondria take up Ca2+. The mitochondrial F1F0 ATPase inhibitor, oligomycin, was added to the incubation buffer to maximize the Ca2+ capacity of mitochondria. It was shown earlier that ADP increases the Ca2+-loading threshold for permeability transition pore (PTP) opening (reviewed in Zoratti and Szabo 1995), and that a combination of oligomycin and ADP greatly increases the Ca2+ capacity of brain mitochondria (Panov et al. 2004) and mitochondria from other tissues (reviewed in Zoratti and Szabo 1995). Mitochondria (65 µg/mL for brain and spinal cord and 125 µg/mL for liver) were added to 1 mL incubation medium placed in a thermostated (37°C) fluorimeter cuvette (VWR, Westchester, PA, USA) equipped with a magnetic stirrer. Bolus additions of CaCl2 were made to the mitochondrial suspension in 20 nanomole increments every 200 s, and changes in Fura 6 fluorescence were recorded at 340/380 nm excitation, 510 nm emission with an F-4500 fluorescence spectrophotometer (Hitachi, Tokyo, Japan). To calibrate Fura 6 fluorescence response to Ca2+ additions, mitochondria isolated from a non-transgenic control mouse were incubated in the same incubation medium supplemented with 1 mm KCN, 1 µm ruthenium red, and with succinate omitted. Changes in Fura 6 fluorescence ratios were recorded and plotted against the added Ca2+ amount. The curve obtained was fitted with a linear function, and the slope coefficient was used to convert Fura 6 fluorescence ratio into [Ca2+] in the medium. To calculate the amount of Ca2+ taken up after each bolus, we subtracted [Ca2+] at the end of the downward deflection, as estimated by the calibration curve, from the peak [Ca2+] determined by the bolus (20 µm), and normalized the resulting [Ca2+] difference by the mitochondrial protein. Ca2+ capacity results from the sum of Ca2+ taken up after all additions.
Brain and spinal cord mitochondria, isolated as described above, were fixed in 0.1 m Na-cacodylate buffer (containing glutaraldehyde 2.5%, paraformaldehyde 4% and picric acid 0.02%) for 60 min and then post-fixed in 1% OsO4−1.5% K-ferricyanide (aqueous) for 60 min. Samples were then dehydrated through a graded ethanol series, infiltrated, and embedded in Spurr's resin (Electron Microscopy Sciences, Fort Washington, PA, USA). Sections were cut at 55–60 nm using a Diatome diamond knife (Diatome, Fort Washington, PA, USA) on an RMC MT-7000 Ultramicrotome (RMC Instruments, Tucson, AZ, USA). Sections were contrasted with uranyl acetate and lead citrate, then viewed on a JSM 100 CX-II electron microscope (JEOL, USA, Inc., Peabody, MA, USA) operated at 80 kV. Images were recorded on Kodak 4489 Electron Image film (Eastman Kodak Company, Rochester, NY, USA) then digitized on an Epson Expression 1600 Pro scanner (Seiko-Epson Corporation, Suwa, Japan) at 900 dpi for publication.
Measurement of mitochondrial membrane potential (ΔΨm), O2 consumption, and ATP synthesis
Mitochondrial ΔΨm was estimated using the fluorescence of safranine O (3 µm for brain and spinal cord and 6 µm for liver) with excitation and emission wavelengths of 495 nm and 586 nm, respectively (Starkov et al. 2002). The rate of O2 consumption was measured in spinal cord mitochondria using a Clarke type electrode (Hansatech Inc., Cambridge, UK) as previously described (Mattiazzi et al. 2002), in the presence of either 5 mm succinate plus 1 µm rotenone or 7 mm pyruvate plus 1 mm malate. State 3 respiration was stimulated with 0.2 mm ADP. State 4 respiration was induced by addition of 1.6 µm carboxyatractylate. Mitochondrial ATP synthesis was measured in spinal cord mitochondria of G85R mutant hSOD1 mice using a luciferase/luciferin-based approach as previously described (Manfredi et al. 2002).
Measurement of cytochrome c release
Total mitochondrial cytochrome c content and the amount of cytochrome c released from mitochondria into the incubation buffer, with and without addition of Ca2+, were measured using an ELISA kit specific for mouse cytochrome c (R & D System, Minneapolis, MN, USA), according to the manufacturers' instructions. The assay was quantified by measuring absorbance at 450 nm in a microtiter plate reader (HTS 7000 Plus, Perkin Elmer, Boston, MA, USA).
All reagents were from Sigma (St Louis, MO, USA) unless otherwise stated.
For statistical comparison between two groups of data, such as those from the mitochondrial Ca2+ capacity and ΔΨm in G93A and non-transgenic control mice, since F-tests analyses revealed a homogeneity of variances between groups, p-values of the differences between wild-type and mutant mice were calculated by unpaired two-tailed Student's t-test. A level of confidence of p < 0.05 was adopted. Differences among different age groups within individual mouse lines were estimated using one-way anova.
Impairment of Ca2+ loading capacity in G93A brain mitochondria
A representative example of a mitochondrial Ca2+ uptake measurement is shown in Fig. 1. In isolated brain mitochondria, after each bolus, there is a rapid uptake of Ca2+, represented by a downward deflection of the Fura 6 fluorescence trace. When the Ca2+ capacity is exceeded, mitochondria fail to take up Ca2+ further, as shown by the lack of the downward deflection. The mitochondrial Ca2+ capacity of brain from a G93A mouse (Fig. 1a, G93A trace) is saturated after a smaller number of additions, compared with a littermate control (Fig. 1a, non-Tg trace). When cyclosporin A (CsA, 0.5 µg/mL), a classic inhibitor of the mitochondrial permeability transition pore (PTP), is added, G93A mitochondria (Fig. 1a, G93A-CsA trace) recover the Ca2+ capacity similar to that of non-transgenic mitochondria (Fig. 1a, non-Tg-CsA trace). This suggests that Ca2+ capacity impairment in G93A mitochondria is likely due to a lower threshold for Ca2+-induced PTP activation. In a Wt hSOD1 brain (Fig. 1b, N1029 trace), mitochondrial Ca2+ capacity does not differ from the non-transgenic littermate (Fig. 1b, non-Tg N1029 trace).
To confirm that the loss of Ca2+ capacity in G93A brain mitochondria results from Ca2+-induced damage rather than pre-existing structural damage, we compared mitochondrial morphology in G93A and non-transgenic brain mitochondria, before and after Ca2+ challenge. Non-transgenic or G93A mitochondria (120 days old) were subjected to the same number of Ca2+ additions, corresponding to the maximal amount of Ca2+ taken up by non-transgenic mitochondria. Mitochondria were imaged by electron microscopy. In the absence of Ca2+ challenge, both non-transgenic and G93A mitochondria have a normal appearance (Figs 1c and e). After Ca2+ challenge, a proportion of non-transgenic mitochondria acquires the typical swollen appearance resulting from membrane permeability transition, with distended outer membranes and fragmented cristae (Kristian et al. 2002), whereas some mitochondria preserve a normal morphology (Fig. 1d). After Ca2+ challenge, all of the G93A mitochondria appear either swollen or shrunken and condensed (Fig. 1f). The latter are reminiscent of de-energized mitochondria (Candipan and Sjostrand 1984).
Ca2+ loading capacity in G93A brain mitochondria was investigated at various disease stages. We studied pre-symptomatic (35 and 65 days-of-age), symptomatic (100 days-of-age) and end-stage (125 days-of-age) mice. In each experiment, age- and gender-matched non-transgenic littermates were used as controls. Ca2+ capacity in isolated brain mitochondria of 45 days and 140-day-old mice expressing Wt hSOD1 was also tested.
The average absolute brain mitochondria Ca2+ capacity in control mice increases with age, from 1131 ± 187 (standard error of the mean) nmoles/mg protein at 35 days-of-age to 2031 ± 647 nmoles/mg protein at 125 days (Fig. 2a). The age-related increase in mitochondrial Ca2+ capacity is statistically significant in both G93A transgenic and non-transgenic littermate groups (p < 0.03 and p < 0.003, respectively, by one-way anova). Despite the age-related increase in Ca2+ capacity, there is a significant reduction of mitochondrial Ca2+ capacity in G93A mice relative to non-transgenic littermates, which starts at 65 days-of-age and persists until disease end-stage (125 days; Fig. 2b). At all ages, the addition of CsA increases Ca2+ capacity and abolishes the differences between transgenic and control mice (data not shown). Ca2+ capacity in N1029 transgenic mice, both at 45 and 140 days-of-age, is unchanged relative to littermate controls (Fig. 2b). The lack of a defect in N1029 mice as compared with their non-transgenic controls indicates that the abnormal Ca2+ capacity in G93A mitochondria is not simply the result of hSOD1 over-expression, but a specific effect of G93A mutant hSOD1.
Impairment of mitochondrial Ca2+ capacity in G93A spinal cord
Since, in hSOD1 mutant mice, motor neuron degeneration affects the spinal cord more aggressively than the brain, we investigated the Ca2+ capacity of mitochondria isolated from spinal cord of G93A mice and non-transgenic controls. We studied spinal cord mitochondria at two different time points prior to the onset of muscle weakness, at 35 and 68 days-of-age. There is a marked reduction in Ca2+ capacity in G93A relative to age- and gender-matched, non-transgenic mitochondria at both ages. Ca2+ capacity in 45-day-old Wt hSOD1 spinal cord mitochondria is unchanged compared with littermate non-transgenic controls (Figs 3e and f).
Ca2+-induced morphological alterations of G93A spinal cord mitochondria do not correlate with increased cytochrome c release
We compared the morphological changes and cytochrome c release in 68-day-old G93A and non-transgenic spinal cord mitochondria challenged with Ca2+. For comparison, cytochrome c release of untreated mitochondria was studied by incubating them in the same buffer conditions, but without addition of Ca2+.
In the absence of Ca2+ challenge, both non-transgenic and G93A mitochondria preparations contained mostly organelles with intact appearance, while a fraction of mitochondria appear to be damaged, presumably by the isolation procedure (Figs 3a and c). However, the G93A preparation appears to contain a somewhat higher proportion of shrunken and condensed organelles (Fig. 3c). After Ca2+ challenge, the majority of non-transgenic mitochondria are swollen because of membrane permeability transition (Fig. 3b), whereas most of the G93A mitochondria appear condensed (Fig. 3d).
Approximately 10% of total mitochondrial cytochrome c is released in the buffer in the absence of Ca2+(Table 1), most likely due to mechanical damage. Upon Ca2+ challenge, the release of cytochrome c increases to only 20% of total cytochrome c. These results are consistent with earlier observations that Ca2+-induced PTP activation does not result in a massive cytochrome c release from neural mitochondria (Andreyev and Fiskum 1999). Under all conditions, there is no statistically significant difference in the amount of cytochrome c released from G93A and non-transgenic mitochondria (Table 1).
Table 1. Cytochrome c release from spinal cord mitochondria from 68-day-old mice. Cytochrome c was measured by ELISA in mitochondria re-suspended in Ca2+ assay buffer (Total) in the supernatant fluid after mitochondria sedimentation prior to Ca2+ challenge (Supernatant), and in the supernatant fluid after Ca2+ challenge and mitochondrial sedimentation (Supernatant + Ca2+). Each experiment was conducted on mitochondria isolated from a pool of four spinal cords from 68-day-old mice. In both G93A and non-Tg samples, Ca2+ boluses matched the number of additions corresponding to the maximal amount of Ca2+ taken up by non-transgenic mitochondria. There are no statistically significant differences between G93A and non-Tg mitochondria. Values are nmoles cytochrome c/mg protein ± SEM; number of pools of four spinal cords studied = 3
Supernatant + Ca2+
5710 ± 509
557 ± 36
1033 ± 120
4928 ± 449
517 ± 100
892 ± 96
Ca2+-induced loss of ΔΨm in mitochondria of G93A mice
We examined changes of mitochondrial membrane potential (ΔΨm) in response to Ca2+ load in G93A brain mitochondria at 35, 65 and 125 days-of-age. As shown by a representative trace in Fig. 4(a), after each Ca2+ addition, a drop in ΔΨm (de-polarization) is followed by a partial re-polarization. There is no statistically significant difference between G93A and non-transgenic mitochondria in ΔΨm prior to Ca2+ addition (ΔΨm-ini; data not shown). However, we observed that after each Ca2+ addition, mitochondrial re-polarization is less efficient in G93A than in non-transgenic mitochondria. In order to quantify this difference, experiments were conducted in parallel in one non-transgenic and one transgenic sample. ΔΨm sensitivity to Ca2+ challenge is measured as the ability of mitochondria to re-polarize after challenge with either low or high Ca2+ loads. Low loads were 40 nmol Ca2+ in all experiments; high Ca2+ loads were set for both non-Tg and transgenic mitochondria at the penultimate Ca2+ addition preceding the complete de-polarization of transgenic mitochondria (i.e. 120 nmoles Ca2+ in the example in Fig. 4a). In brain, G93A mitochondria exhibit a statistically significant reduction in ΔΨm in response to high Ca2+ loads starting at 65 days-of-age (Fig. 4b upper panel).
In spinal cord mitochondria, we did not find a significant difference in the decrease of ΔΨm in response to high Ca2+ loads between G93A and control mitochondria (Fig. 4c lower panel). However, in brain and spinal cords of younger G93A mice (65 and 35 days-of-age, respectively), mitochondria were significantly more sensitive to low Ca2+ loads (i.e. 40 nmoles; Figs 4b and c, upper panels).
Mitochondrial oxygen consumption is unchanged in spinal cord from pre-symptomatic G93A mice
We measured complex I and complex II driven oxygen consumption in the presence of malate plus pyruvate and succinate plus rotenone, respectively. State 3 respiration was inhibited by the addition of carboxyatractylate, a specific inhibitor of the adenine nucleotide translocator. Although previously it had been shown that, under certain experimental conditions, G93A spinal cord has impaired mitochondrial respiratory chain function (Mattiazzi et al. 2002; Kirkinezos et al. 2005), under our experimental conditions, spinal cord mitochondria from pre-symptomatic G93A mice do not display significant differences in respiration as compared with non-transgenic controls (Fig. 4d). These data suggest that the reduction of Ca2+ capacity does not result from impaired respiration per se, at least in the early stages of the disease.
Ca2+ capacity and ΔΨm sensitivity to Ca2+ challenge is unchanged in G93A liver mitochondria
To determine whether mitochondrial Ca2+ capacity impairment is a tissue-specific trait or a more generalized phenomenon that also affects non-neural tissues, we studied mitochondria isolated from liver of G93A mice. Ca2+ capacity in transgenic mice is not decreased compared with non-transgenic littermates, both at 67 and 120 days-of-age (Fig. 5a). Mitochondrial ΔΨm sensitivity to Ca2+ challenge, calculated as described earlier for brain and spinal cord mitochondria, is unchanged in liver from 120-day-old G93A transgenic mice, as compared with non-transgenic littermates (Fig. 5b).
Mitochondrial Ca2+ capacity and sensitivity of ΔΨm to Ca2+ challenge in G85R spinal cord
To assess whether a reduction in mitochondrial Ca2+ loading capacity is common to other murine models of fALS, we studied spinal cord mitochondria from symptomatic G85R mice at 10–12 months-of-age. We found that Ca2+ loading capacity is variable both in transgenic and non-transgenic groups, possibly because of an aging effect. However, there is a statistically significant reduction in G85R spinal cord mitochondria. The Ca2+ loading capacity in G85R spinal cords is 73 ± 7% of non-transgenic controls (p < 0.03). We did not find significant differences in ΔΨm sensitivity to Ca2+ challenge, calculated either after low or high Ca2+ loads between G85R mice and controls (Table 2). Unlike symptomatic G93A mice (Mattiazzi et al. 2002), in fully symptomatic G85R spinal cord mitochondria, there is no reduction in ATP synthesis and oxygen consumption (Table 2). These data confirm that a decrease in mitochondrial Ca2+ loading capacity is a common trait of mutant hSOD1 mice, and that it does not appear to correlate with a reduction in respiratory chain function.
Table 2. Oxidative phosphorylation and Ca2+ capacity in spinal cord mitochondria of 10–12-month- old G85R mice. Oxygen consumption and ATP synthesis were measured on isolated mitochondria using malate and glutamate as substrates, as previously described (Manfredi et al. 2002). Values are averages ± SD. Ca2+ capacity and ΔΨm were measured, as described in Methods. Values are averages ± SEM. G85R mice have a statistically significant decrease in Ca2+ capacity in spinal cord mitochondria, whereas differences in respiration, ATP synthesis and ΔΨm sensitivity to Ca2+ challenge were not significant
Respiration (nmol O2/min/mg)
ATP synthesis (nmol ATP/min/mg)
Ca2+ capacity (% of non-Tg)
ΔΨm 40 nmol (% of ΔΨm-ini)
ΔΨm Ca2+ (% of ΔΨm-ini)
557 ± 70 (n = 6)
189 ± 51 (n = 6)
100 (n = 4)
79.5 ± 12.1 (n = 4)
59.8 ± 19.9 (n = 4)
456 ± 80 (n = 5)
181 ± 54 (n = 5)
73 ± 7 (n = 4) (p < 0.03)
74.3 ± 11.6 (n = 4)
45.7 ± 14.7 (n = 4)
Although much is known about the genetic defects underlying SOD1-related fALS, the molecular and biochemical mechanisms that cause the ‘toxic’ effect of mutant SOD1 are still far from being understood. One of the major puzzles that remains to be solved is that of the tissue specificity of the degenerative process. Both in fALS patients and in transgenic animals, mutant SOD1 is ubiquitously expressed and its effect is not limited to motor neurons, since transgenic mice expressing mutant SOD1 exclusively in neurons or in glia do not develop disease (Gong et al. 2000; Lino et al. 2002), and both neurons and glia must express the mutant protein for the disease to take place (Clement et al. 2003). Yet, only certain types of neuronal cells degenerate.
Several lines of evidence indicate that mitochondria may be implicated in the selectivity of SOD1 toxicity. First, although the majority of SOD1 is localized in the cytosol, a portion of the protein is localized in mitochondria (Jaarsma et al. 2001; Okado-Matsumoto and Fridovich 2001; Sturtz et al. 2001; Higgins et al. 2002; Field et al. 2003). Mutant SOD1 is more abundant in mitochondria from nervous tissue than in other non-affected ones, such as liver or heart (Mattiazzi et al. 2002; Liu et al. 2004; Vijayvergiya et al. 2005). Furthermore, mutant hSOD1 has been shown to accumulate in the mitochondria of spinal cord from mutant hSOD1 transgenic mice (Liu et al. 2004) where it interacts with the anti-apoptotic protein Bcl-2 (Pasinelli et al. 2004). Second, mutant SOD1 forms aberrant aggregates in the mitochondria of the neural tissue and not in other, non-affected tissues of SOD1 transgenic mice (Mattiazzi et al. 2002; Liu et al. 2004; Vijayvergiya et al. 2005). Third, since Ca2+-mediated glutamate excitotoxicity is one of the potential mechanisms for the tissue specificity of SOD1 toxicity (Rothstein 1996), the impairment of Ca2+ handling in SOD1 mutant mitochondria may provide a link between mitochondrial dysfunction and glutamate excitotoxicity. Because mitochondria are a major short-term buffering system of intracellular Ca2+ (Nicholls et al. 2003; Wang et al. 2003), an impairment of mitochondrial Ca2+ handling may precipitate excitotoxic events in excitable cells that undergo high Ca2+ influx in response to glutamate (White and Reynolds 1996). Elevation of cytosolic Ca2+ levels in neurons may compromise mitochondrial integrity and functions in a number of ways, for example, by inducing enhanced production of free radicals from mitochondria (Reynolds and Hastings 1995). There is compelling evidence that free radicals play a major role in neuronal death (Beal 2004), although neither the source of these radicals nor the precise nature of the connection between Ca2+ mobilization and free radical production has been conclusively defined. It has been suggested that aberrant copper chemistry by mutant SOD1 causes oxidative damage and apoptosis in neuronal cells (Estevez et al. 1999), and that antioxidant agents extend survival in mutant hSOD1 transgenic mice (Jung et al. 2001; Wu et al. 2003).
Disturbed intracellular Ca2+ homeostasis has been previously reported in cultured primary motor neurons from G93A transgenic mice (Kruman et al. 1999; Kim et al. 2002). However, these studies did not address the role of mitochondria in Ca2+ handling. Here, we demonstrate impaired Ca2+ capacity in isolated mitochondria from brain and spinal cord of mutant hSOD1 mice (Figs 1–3). Liver, a non-affected tissue, does not display mitochondrial Ca2+ capacity impairment in G93A mice, even at late disease stages (Fig. 5). In spinal cord mitochondria of G93A mice, Ca2+ capacity is impaired at as early as 35 days-of-age (Fig. 3). At this stage, these mice do not manifest symptoms of paralysis or pathological signs of motor neuron degeneration. These results strongly suggest that an early impairment in a critical mitochondrial function, such as Ca2+ handling, may precede the onset of neurodegeneration in vivo.
In pre-symptomatic G93A mice, brain and spinal cord mitochondria do not show defects of oxidative phosphorylation (Mattiazzi et al. 2002). However, in these mitochondria, we observed a more pronounced depolarization induced by low Ca2+ loads compared with non-transgenic controls. This difference was not detected in mitochondria from older, symptomatic G93A or G85R mice (Figs 4b and c, Table 2). This observation suggests that mitochondria from younger mutant hSOD1 mice may be more prone to damage induced by intense workload. For high Ca2+ loads, we observed increased mitochondrial depolarization in G93A brain, but not in spinal cord (Figs 4b and c). There may be different explanations for this observation. First, it may reflect inherent differences in bioenergetics between brain and spinal cord mitochondria (Sullivan et al. 2004). This is confirmed by the observation that in both G93A and control spinal cord, the Ca2+-induced ΔΨm decrease was more pronounced than in brain (Figs 4b and c). It is also consistent with the observation that in both non-transgenic and G93A mice, Ca2+ capacity increases with age (Fig. 2a) in brain, but not in spinal cord (Fig. 3e). Second, it may also reflect the progressively different cellular composition in the degenerating tissue. It is possible that mitochondrial bioenergetics in SOD1 mutant astrocytes may be less affected than in neurons. Because, in mutant hSOD1 mice, astrocytosis is much more pronounced in spinal cord than in brain (Gurney et al. 1994), and because, in our experiments, mitochondria are isolated from the total cellular pool, in the later stages of the disease a shift of the cell population towards increased astrocytes may mask the bioenergetic defect in the spinal cord. Third, the ΔΨm decrease with high Ca2+ loads in the brain of G93A mice may reflect a compensatory mechanism. Such a mechanism may limit the amount of Ca2+ taken up by mitochondria, thereby preserving their integrity.
In vivo, reduced mitochondrial Ca2+ capacity may result in enhanced susceptibility to Ca2+-mediated mitochondrial PTP opening, especially under conditions of metabolic stress (Scanlon and Reynolds 1998). PTP opening is facilitated by high Ca2+ load, ΔΨm decrease and production of free radicals (Zoratti and Szabo 1995), and is prevented by CsA. We found that, in SOD1 mutant mitochondria, CsA attenuates the Ca2+ capacity impairment (Fig. 1a). This observation agrees with the hypothesis that PTP plays a role in the pathogenesis of fALS (Karlsson et al. 2004; Kirkinezos et al. 2004).
Interestingly, we also found that upon Ca2+ challenge, many mutant spinal cord mitochondria undergo morphological changes that are suggestive of energy loss (Candipan and Sjostrand 1984) rather than PTP-induced swelling (Fig. 3). A similar, although less pronounced, phenomenon is observed also in G93A brain mitochondria (Fig. 1). It is possible that the Ca2+ challenge de-energizes a portion of spinal cord G93A mitochondria before they can accumulate enough Ca2+ to reach the PTP opening threshold. It is also possible that Ca2+ may activate a ‘low conductance’ permeability pore, which does not result in mitochondrial swelling but is CsA sensitive (Novgorodov and Gudz 1996). This scenario appears to correlate well with the Ca2+-induced defect of re-polarization at ‘low Ca2+ loads’, in G93A spinal cord and brain mitochondria.
Although the mechanisms leading to cell death in ALS are still unclear, several lines of evidence suggest that the mitochondrial apoptotic pathway may play a role. In spinal motor neurons of G93A transgenic mice, cytochrome c is released from mitochondria, leading to caspase 9 activation (Guegan et al. 2001). Both inhibition of caspase activation (Li et al. 2000) and the overexpression of the mitochondrial anti-apoptotic protein Bcl-2 (Kostic et al. 1997) slow motor neuron degeneration and extend the survival of SOD1 mutant mice. However, other observations argue against a significant contribution of apoptosis to the pathogenesis of SOD1 fALS. SOD1 mice genetically lacking caspase 11 (Kang et al. 2003), an upstream regulator of the executioner caspases 1 and 3, showed no improvement of disease phenotype. Furthermore, morphological and biochemical markers of apoptotic cell death are difficult to detect, both in ALS patients and in transgenic mice (Migheli et al. 1999).
In vitro, mitochondrial PTP opening results in swelling and damage of organelles (Zoratti and Szabo 1995) that may result in cytochrome c release. However, we found no difference in the amount of cytochrome c released from G93A spinal cord mitochondria, as compared with controls, upon Ca2+ challenge (Table 1). The absence of correlation between PTP opening and cytochrome c release in mitochondria from neural tissue was reported earlier (Andreyev and Fiskum 1999).
It was shown that in symptomatic G93A spinal cord mitochondria, most respiratory chain complexes are partially suppressed (Jung et al. 2002; Mattiazzi et al. 2002). Here, we show that in G93A mice, decreased mitochondrial Ca2+ capacity precedes the respiratory chain defects (Fig. 4d) and, in G85R mice, it occurs in the absence of overt respiratory chain defects (Table 2). These results suggest that impaired mitochondrial Ca2+ capacity in mutant hSOD1 mice occurs early on in the progression of the disease, affects neural tissue preferentially and is not necessarily associated with respiratory chain defects.
Excitable cells, such as motor neurons, are exposed to repetitive Ca2+ challenges in vivo. It is possible that over time, these Ca2+ challenges in mutant hSOD1 neurons with abnormal mitochondrial Ca2+ handling may contribute to accelerating cell damage and, eventually, cell death.
We thank Leona Cohen-Gould (Weill Medical College of Cornell University microscopy facility) for preparation and imaging of electron micrographs. This work was supported by Telethon Italia ONLUS (MD) Foundation, the Robert Packard ALS Research Center ‘The New York Community Trust’ (GM), and NIH/NINDS grant P01 NS011766-27 (GM).