Mitochondrial electron transport chain complex dysfunction in a transgenic mouse model for amyotrophic lateral sclerosis

Authors


Address correspondence and reprint requests to Zuoshang Xu, Department of Biochemistry and Pharmacology, University of Massachusetts Medical School, 55 Lake Avenue North, Worcester, MA 01655, USA. E-mail: Zuoshang.Xu@umassmed.edu

Abstract

Amyotrophic lateral sclerosis is a fatal neurodegenerative disease that causes degeneration of motoneurons. Mutation of Cu,Zn superoxide dismutase (SOD1) is one cause for this disease. In mice, expression of mutant protein causes motoneuron degeneration and paralysis resembling the human disease. Morphological change, indicative of mitochondrial damage, occurs at early stages of the disease. To determine whether mitochondrial function changes during the course of disease progression, enzyme activities of mitochondrial electron transport chain in spinal cords from mice at different disease stages were measured using three different methods: spectrophotometric assay, in situ histochemical enzyme assay, and blue native gel electrophoresis combined with in-gel histochemical reaction. The enzyme activities were decreased in the spinal cord, particularly in the ventral horn, beginning at early disease stages. This decrease persisted throughout the course of disease progression. This decrease was not detected in the spinal cords of non-transgenic animals, of mice expressing the wild-type protein, and in cerebellum and dorsal horn of the spinal cords from mice expressing mutant protein. These results demonstrate a functional defect in mitochondria in the ventral horn region and support the view that mitochondrial damage plays a role in mutant SOD1-induced motoneuron degeneration pathway.

Abbreviations used
ALS

amyotrophic lateral sclerosis

BN

blue native

DAB

3,3′-diaminobenzidine

DB

decylubiqunone

DCIP

dichloroindophenol

DTNB

5,5′-dithiobis(2-nitrobenzoic acid)

ETC

electron transport chain

G93A

mice expressing mutant human SOD1 G93A

NBT

nitro blue tetrazolium

Para

paralysis stage

PAGE

polyacrylamide gel electrophoresis

PMSF

phenylmethylsulfonyl fluoride

PMW

premuscle weakness stage

PMS

phenazine methasulfate

RD

rapid declining stage

ROS

reactive oxidative species

SD

slow declining stage

SDH

succinate dehydrogenase

SOD1

Cu,Zn superoxide dismutase

WS

wild-type human SOD1 expressing mice

WT

wild-type mice

Amyotrophic lateral sclerosis (ALS) is caused by the selective degeneration of motoneurons in spinal cord, brain stem, and cerebral cortex. The degeneration of motoneurons leads to skeletal muscle atrophy, paralysis and death. About 10% of ALS cases are familial and the rest are sporadic (Rowland and Shneider 2001). Among the familial cases, 20% are caused by mutations of Cu,Zn superoxide dismutase (SOD1) (Rosen et al. 1993). It has been well demonstrated that mutant SOD1 causes motoneuron degeneration by a gain of toxic property (Xu 2000). This raises two important questions: what is this toxic property and what is the cellular target of this toxicity?

Definitive answers to these questions have not emerged but current evidence has pointed to several possibilities. With regard to the first question, the toxicity generated by the mutant SOD1 may involve enhanced capacities to produce reactive oxidative species (ROS) such as hydroxyl radicals and peroxynitrite (Wiedau-Pazos et al. 1996; Yim et al. 1997; Estevez et al. 1999). Alternatively, the mutant protein may have an enhanced propensity to form intracellular aggregates (Johnston et al. 2000), which in turn generate cellular toxicity similar to other neurodegenerative diseases (Cleveland and Rothstein 2001).

With regard to the second question, there are several possible targets for the mutant SOD1 toxicity. First, mutant SOD1 might damage neurofilaments and axonal transport machinery, leading to accumulation of neurofilaments in proximal axons and formation of giant neurofilament swellings. These changes could further cause blockage of axonal transport and axon degeneration (Julien 2001). There is also evidence that mutant SOD1 might damage the Golgi apparatus (Mourelatos et al. 1996). Strong evidence suggests that mitochondria are a major target for mutant SOD1 toxicity. In mice expressing mutant SOD1 G93A and G37R, mitochondrial swelling and vacuolation begin at early disease stages (Dal Canto and Gurney 1995; Wong et al. 1995) and peak at the onset of muscle weakness (Kong and Xu 1998). In cultured neuroblastoma cells or motoneurons, expression of mutant SOD1 causes mitochondrial damage and dysfunction (Carri et al. 1997; Kruman et al. 1999).

In the current work, we investigated the changes in mitochondrial electron transport chain (ETC) complex activities at different disease stages in mice expressing mutant SOD1 G93A (G93A mice). Mitochondrial ETC complexes are directly involved in the most important function of mitochondria, ATP production. ETC activities have been measured as a functional indicator of mitochondria. Several previous studies reported changes in mitochondrial ETC activities in ALS (Bowling et al. 1993; Browne et al. 1998; Swerdlow et al. 1998; Wiedemann et al. 1998; Borthwick et al. 1999). However, no systematic measurement of ETC activities has been carried out in the spinal cord along the course of disease progression. Therefore, how mitochondrial function changes in association with disease progression is not known. Using three different methods to measure three ETC (complexes I, II and IV) activities, here we show that mitochondrial ETC activities are decreased in the ventral horn prior to the disease onset and during the course of disease progression. These results support the view that mitochondrial damage plays a role in mutant SOD1-induced motoneuron degeneration pathway.

Materials and methods

Chemicals

Reagents for BCA protein assay were purchased from Pierce (Rockford, IL, USA). All other chemicals and enzymes were purchased from (Sigma, St Louis, MO, USA) except where stated otherwise.

Transgenic mice and categorization of disease stages

Mice transgenic for the human SOD1 mutant G93A [TgN(SOD1-G93A)1 GUR] and the human wild-type SOD1 [TgN(SOD1)2GUR] were purchased from the Jackson Laboratory and bred at the University of Massachusetts Medical School animal facility. Transgenic mice were identified using PCR according to Gurney et al. (1994). Disease stages of transgenic mice were determined by timing the mouse hanging onto a vertical rotating wire (Kong and Xu 1998). Based on the change in the length of time that the mice were capable of hanging onto the wire, the disease progression was categorized into four stages: a premuscle weakness stage (PMW), when muscle strength was maintained at the normal level; a rapid declining stage (RD), when hanging time declined sharply; a slow declining stage (SD), which followed the RD stage and lasted for 4–11 weeks; and a paralysis stage (Para) when paralysis of limbs occurred and mice could no longer hold onto the wire (Kong and Xu 1998). All animal procedures are approved by University of Massachusetts Medical School IACUC.

Sample preparation

Mitochondria-enriched samples were prepared as described by Schagger (1996) with minor modifications (Jung et al. 2000). Briefly, CNS tissues were homogenized in buffer 1 [440 mm sucrose, 20 mm 3-morpholino propanesulfonic acid (MOPS), 1 mm EDTA, 0.2 mm phenylmethylsulfonyl fluoride (PMSF), pH 7.2] at a ratio of 25 µL/mg tissue. Homogenized samples were centrifuged at 20 000 g for 20 min. Mitochondria-enriched pellets were dissolved in buffer 2 [1 m aminocaproic acid, 50 mm bis-Tris-HCl, 1 µg/mL pepstatin, 1 µg/mL leupeptin, 10 µL/mL PMSF, and 20 µmN-p-tosyl-L-phenylalanine chloromethyl ketone (TPCK), pH 7.0] at a ratio of 4 µL/mg tissue and freshly prepared 10% dodecyl maltoside was then added at 2 µL/mg tissue to dissolve membrane proteins. These homogenates were centrifuged at 100 000 g for 15 min. The supernatants were collected and used for ETC enzymatic assays. Protein concentrations in the supernatants were determined by the BCA assay. Protein contents of cytochrome a in the mitochondrial preparations were spectrophotometrically determined from the difference in absorbance between 603 and 630 nm after reduction with sodium ascorbate (extinction coefficient ε 16.5 /mm/cm) (Sekuzu et al. 1967).

Spectrophotometric enzyme assay

All spectrophotometric assays were performed by established methods using a Beckman DU-600 spectrophotometer, as described (Jung et al. 2000). In brief, complex I (NADH dehydrogenase) activity was measured by monitoring rotenone-sensitive NADH oxidation with Coenzyme Q1 (100 µm) at 340–380 nm (ε 5.5 /mm/cm) (Estornell et al. 1993). Complex II activity was measured by monitoring malonate-sensitive reduction of dichloroindophenol (DCIP) at 600–750 nm (ε 19.1 /mm/cm) when coupled to complex II-catalyzed reduction of decylubiqunone (DB) (Trounce et al. 1996). Complex IV (cytochrome c oxidase) activity was monitored by KCN-sensitive oxidation of reduced cytochrome c (ferrocytochrome c) at 550–540 nm (ε 19.0 /mm/cm). Citrate synthase activity was measured by monitoring the reduction of 5,5′-dithiobis(2-nitrobenzoic acid) (DTNB) at 412–360 nm (ε 13.6 /mm/cm), coupled to the reduction of CoA with oxaloacetate (Trounce et al. 1996).

Blue native gel electrophoresis and in-gel enzymatic histochemical reactions

The methods have been described in detail (Jung et al. 2000). The proteins were separated on a gradient gel (5–13% polyacrylamide) with a 4% polyacrylamide stacking gel. To assure the same conditions, eight gels were made simultaneously using Mini-Protein II Multi-gel casting chamber (Bio-Rad Laboratories, Hercules, CA, USA). Coomassie blue brilliant G-250 (5%) in 1 m aminocaproic acid was added to mitochondria-enriched samples at a ratio of 1 : 14. Thirty micrograms of each sample was loaded onto the gels and then electrophoresed overnight with 1.5 mA of constant current per gel at 4°C. One gel was stained with Coomassie blue for protein content measurement and the others were incubated histochemically for enzyme activity measurement.

To measure complex I activity, the gel was incubated in 0.1 m Tris–HCl, 0.1 mg/mL NADH and 1.0 mg/mL NBT (nitro blue tetrazolium), pH 7.4. As a measure for complex II activity, succinate dehydrogenase (SDH) activity was determined as follows: The gel was incubated in 50 mm phosphate buffer, 84 mm of succinic acid, 0.2 mm PMS (phenazine methasulfate), 2 mg/mL NBT, 4.5 mm EDTA and 2 mm KCN, pH 7.4. The incubation tray was wrapped with aluminum foil to shield PMS from light. To measure complex IV activity, the gel was incubated in 50 mm phosphate buffer, 1.0 mg/mL DAB (3,3′-diaminobenzidine), 24 units/mL catalase, 1 mg/mL cytochrome c and 75 mg/mL sucrose, pH of 7.4. Reactions were carried out at room temperature and stopped at various times by fixing the gel in 45% methanol and 10% acetic acid overnight. This removes excess Coomassie blue used in the gel-running buffer. The gels were immersed in 10% methanol and 10% acetic acid overnight, wrapped in gel drying membranes (Promega, Madison, WI, USA) and air dried for 48 h.

The dried gels were scanned using a Bio-Rad GS-700 imaging densitometer. Band intensity was measured using Scion Image (version beta 3b, 1998). Initial reaction velocities were measured from the slopes generated by plotting log[(ODmax − ODt)/ODmax] against the reaction times (t). ODmax was the maximum optical density (OD) obtained from the longest incubation; ODt was the OD of the bands at different incubation times. The initial reaction velocities were then normalized against the protein contents of the ETC complexes to obtain specific activities. The specific activity of complex I was obtained by dividing its initial velocity by its protein content. For complex II and IV, measurement of protein content on blue native (BN)–polyacrylamide gel electrophoresis (PAGE) was unreliable because complex II was often poorly stained by Coomassie blue and complex IV band often partially overlapped with an abundant, unidentified protein band on the gel. Therefore, the specific activities of SDH and complex IV were estimated from normalizing the initial velocities of these complexes against the average protein content of complex I, III and V of the same animal tissues. This data analysis appeared reasonable since ratios between the protein contents of these three complexes were not significantly different throughout different ages or at different disease stages (data not shown), suggesting that the proportions of different complexes were relatively constant during aging or at different disease stages.

Histochemical reactions on spinal cord sections

Detailed protocol and characterization of the method has been described (Jung et al. 2002). In brief, a 5-mm lumber spinal cord segment was embedded in OCT compound (Sakura Finetechnical Co., Tokyo, Japan) on dry ice and stored at −80°C until sectioning. Eight-micrometer sections were cut at −20°C using a cryostat (HM 500 OM, Microm, Heidelberg) and mounted on Superfrost Plus micro slides (VWR). To measure complex I activity, sections were incubated in 0.1 m Tris-HCl, 0.1 mg/mL NADH, 1.0 mg/mL NBT, 2 µg/mL antimycin A, 84 mm malonate and 2 mm KCN, pH 7.4 for 20 min. To test rotenone sensitivity of complex I reaction, sections were preincubated with 60 µm rotenone in the dark for 15 min and then incubated with the reaction solution as above in the presence of the same amount of rotenone. To measure SDH activity (indicative of complex II activity), sections were incubated in 50 mm phosphate buffer, 84 mm of succinic acid, 0.2 mm PMS, 2 mg/mL NBT, 4.5 mm EDTA, 60 µm rotenone, 2 µg/mL antimycin A and 2 mm KCN, pH 7.4 for 10 min. To measure complex IV activity, the gel was incubated in 50 mm phosphate buffer, 1.0 mg/mL DAB, 24 units/mL catalase, 1 mg/mL cytochrome c and 75 mg/mL sucrose, 60 µm rotenone and 84 mm malonate, 2 µg/mL antimycin A, pH of 7.4 for 20 min (Seligman et al. 1968). Histochemical reactions of SDH and complex IV were performed in the dark and all reactions were carried out at room temperature.

The prepared slides were examined using an Olympus IX-70 microscope and 16 bit digital images taken with a digital camera (Roper Scientific, model TE/CCD-1000PB) operated by MetaMorph image analysis software (Universal Imaging, West Chester, PA, USA). Images were taken with particular care to use uniform gray scales and below the level of saturation. The average optical intensity (per pixel) of the whole spinal cord, ventral and dorsal horns on these sections was quantified using the MetaMorph software on an IBM PC. When either ventral or dorsal horn was quantified, the areas were first drawn on the paralyzed sections because the areas of ventral horn are the smallest among all animals. Care was taken to exclude any white matter. The areas were then copied and pasted to the other sections. In this way we insure that the same sized areas were included for measurement. The optical intensity was converted to optical densities (OD) by the formula, OD = log10(Ibk/Im), where Ibk is background intensity; Im is measured intensities from different regions of the sections.

Data analysis

Results are presented as the mean ± SEM. Relative values in Figs 1, 5 and 6 were calculated using the value of non-transgenic 60-day-old (WT 60) mice as the 100% value. Data were analyzed by one-way anova. Multiple comparisons were followed with post-hoc Dunnett's t-test and Tukey HSD.

Figure 1.

Enzyme activities of electron transport chain (ETC) complex I, II and IV measured by the spectrophotometric assays. Enzyme activities are normalized against either protein amount measured using a BCA assay (gray bars) or citrate synthase activity (white bars) in the mitochondria-enriched fractions. WT, WS and G93A stand for wild-type mice, mice expressing wild-type human SOD1 and mice expressing mutant human SOD1 G93A, respectively. PMW, RD, SD and Par stand for premuscle weakness, rapid declining, slow declining and paralysis disease stages (Kong and Xu 1998). Relative enzyme activities in all spinal cords are calculated using the value of the 60-day-old WT as the reference (100%). The actual values of the 100% enzyme activities are 63.1 nmol/min/mg for complex I, 37.6 nmol/min/mg for complex II and 254.7 nmol/min/mg for complex IV (gray bars). Five spinal cords were measured for every age or disease stage.

Figure 5.

Quantification of enzyme activities of electron transport chain (ETC) complex I, SDH and complex IV from spinal cord sections as shown in Figs 2–4. Enzyme activities were measured based on optical densities (OD) in ventral (gray bars) and dorsal (white bars) horn areas (see Figs 2–4). Relative ODs were calculated by setting the value of the 60-day-old WT at 100% and presented as mean ± SEM. All ODs were statistically compared with the values of WT 100 and those that are significantly different are marked by *. OD values were also compared between ventral and dorsal horn areas, and those that are significantly different are marked by #. Statistical analysis was carried out by one-way anova followed by post-hoc Dunnett's t-test. *#p < 0.05, **##p < 0.01. Four animals from each age or disease stage and three sections from each animal were measured.

Figure 6.

Specific activities of complex I, II (by SDH activity measurement) and IV in the spinal cords. Proteins in mitochondria-enriched fractions were resolved by BN–PAGE. Protein content of electron transport chain (ETC) complexes was measured from Coomassie blue stained gels. Activities were measured from the gels after the histochemical reactions (Jung et al. 2000). Specific activity for complex I was calculated from division of complex I activity by complex I protein amount. Specific activities for complex II and IV were calculated from division of the activities by the average protein amount of complex I, III and V (for details, see Materials and methods). The relative ODs of WT and WS groups were calculated by setting the values of the 60-day-old WT at 100%. Relative ODs of the G93A group were calculated by setting the values of the WT sample within G93A group at 100%. Values are mean activities ± SEM. Statistical comparisons were made between WT 100 and other ages in the control group and between WS and each disease stage in the G93A group using one-way anova followed by post-hoc Dunnett's t-test, *p < 0.05, **p < 0.01. Eight animals were measured at each age or disease stage.

Results

No significant changes are detected in mitochondrial ETC activities during ALS progression by spectrophotometric assays

To evaluate mitochondrial function during ALS progression, ETC complex I and IV, and SDH activities were measured using the spectrophotometric assays in mitochondria-enriched fractions from spinal cords that were dissected at different disease stages. These measurements were normalized against the protein contents in the mitochondrial fraction (Fig. 1, gray bars) or citrate synthase activities (Fig. 1, white bars). The protein contents in the mitochondrial fraction were not changed significantly during aging in wild-type (WT), wild-type human SOD1 expressing (WS) and mutant human SOD1 G93A expressing (G93A) mice (Table 1), while a decreasing trend (not statistically significant) in citrate synthase activity during aging was apparent (Table 1). As a result, there was an apparent decreasing trend (not statistically significant) in the ETC activities in G93A mice, but not in those in WT and WS mice, if the activities were normalized against the protein content (Fig. 1 gray bars). This trend disappeared when the ETC activities were normalized against the citrate synthase activities (Fig. 1). To obtain a relatively specific measure of the protein content in the ETC complexes, we measured the content of cytochrome a (a component in complex IV) from a separate set of spinal cords. We did not find a significant change during aging and in disease progression (Table 1).

Table 1.  Protein content and citrate synthase activity in mitochondria preparations from spinal cord in G93A transgenic mice
  1. WT, wild-type mice; WS, mice expressing wild-type human SOD1; G93A, mice expressing mutant human SOD1 G93A; PMW. premuscle weakness stage; RD, rapid declining stage; SD, slow declining stage; Para, paralysis stage. Protein content in mitochondrial preparations, cytochrome a (cyto a) or citrate synthase (CS) activity show no significant changes statistically with disease stage and age. aData are mean ±SEM protein amount measured by BCA assay (mg/g spinal cord tissue). bData are mean ±SEM CS activity (nmol/min/mg protein in the mitochondrial prep). cData are mean ±SEM protein quantity of cytochrome a measured spectrophotometrically (nmol/g spinal mitochondrial preparation). Five animals were measured at each age or disease stage. Statistical analysis was carried out by anova followed by Dunnett's t-tests using non-transgenic 60-day-old mice (WT 60) as a control group and comparing other groups against it.

 Age (days)60100150200240 
 Proteina27.5 ± 2.427.6 ± 2.326.8 ± 2.324.9 ± 1.525.4 ± 1.8 
WTCS actb226 ± 23224 ± 28226 ± 26216 ± 14199 ± 22 
 Cyto ac386 ± 21427 ± 12416 ± 14411 ± 24432 ± 29 
 Proteina26.6 ± 1.927.4 ± 2.027.1 ± 2.625.4 ± 1.825.0 ± 2.3 
WSCS actb245 ± 19213 ± 18225 ± 16230 ± 13209 ± 20 
 Cyto ac370 ± 25394 ± 13480 ± 31382 ± 54453 ± 39 
 DiseasePMWPMWPMW   
 stage60 day100 day150 dayRDSDPara
 Proteina27.1 ± 2.626.2 ± 2.426.4 ± 2.126.2 ± 2.324.5 ± 2.626.7 ± 3.1
G93ACS actb246 ± 24229 ± 18231 ± 16220 ± 14212 ± 5.9201 ± 11
Cyto ac400 ± 21331 ± 36394 ± 5337 ± 37375 ± 56320 ± 38

Activities of mitochondrial ETC complexes are decreased selectively in the ventral horn of the spinal cord starting at early disease stages

ALS is known to affect motoneurons in the ventral horn most severely. Measuring ETC activities in spinal cord homogenate mixes the affected and unaffected regions. This may diminish the detectable changes of ETC activities in the vulnerable region. To examine ETC activities selectively in the ventral horn, in situ histochemical reactions using unfixed cryostat spinal cord sections were carried out. The densities of the reaction product for complex I and IV, and SDH on these sections were similar at young ages between G93A mice and WT mice (Figs 2, 3 and 4, two top left panels). But a decrease in the ventral horn could be detected in sections from 150-day-old G93A mice (just prior to disease onset) compared with the age-matched WT mice (Figs 2, 3 and 4, bottom left panel). As the disease progressed, this decrease in the ventral horn became more pronounced in G93A mice compared with the WT mice (Figs 2, 3 and 4, right panels).

Figure 2.

In situ histochemical reactions of complex I (NADH dehydrogenase) on spinal cord sections. Spinal cords from G93A mice at different disease stages were paired with age-matched wild-type mice and the activity of complex I was detected by the in situ histochemical reaction. Regions bounded by dotted lines in the top left panel illustrate dorsal and ventral horn areas from which optical densities were quantified (see Fig. 5). The same regions were quantified in all other spinal cord sections.

Figure 3.

In situ histochemical reactions of succinate dehydrogenase (SDH, indicative of complex II activity) on spinal cord sections. Spinal cords from G93A mice at different disease stages were paired with age-matched wild type mice and the activity of SDH was detected by the in situ histochemical reaction. Symbols are the same as those in Fig. 2.

Figure 4.

In situ histochemical reactions of complex IV (cytochrome c oxidase) on spinal cord sections. Spinal cords from G93A mice at different disease stages were paired with age-matched wild type mice and the activity of cytochrome c oxidase was detected by the in situ histochemical reaction. Symbols are the same as in Fig. 2.

To quantify these visual differences, ODs in the ventral horn (gray bars) and the dorsal horn (white bars) were measured. A progressive decrease of complex I and IV, and SDH activities began in the ventral horn prior to the onset of muscle weakness in G93A mice (Fig. 5). The decrease ranges from ∼20% in 150-day-old asymptomatic animals to ∼40% in paralyzed animals. The percentage decreases in complex I are probably underestimates as ∼20% of the activity in the ventral horn is rotenone-insensitive (Jung et al. 2002). Significant changes were not found in the dorsal horn of G93A mice, neither were they found in ventral and dorsal horns of the WT mice (Fig. 5).

Specific activities of ETC complexes are reduced in the spinal cord of G93A mice

To determine specific activities of ETC complexes based on the ETC complex protein quantity, mitochondrial ETC complexes from spinal cords were resolved using histochemical BN–PAGE (Jung et al. 2000). Protein quantities of complex I, III and V were measured from Coomassie blue stained BN–PAGE gels and were found not significantly changed at different disease stages of the G93A mice and not significantly different between G93A mice and age-matched WT and WS mice (data not shown). Activities of complex I and IV, and SDH were measured using histochemical reactions in BN–PAGE gels (see Materials and methods). A statistically significant decrease was observed in the specific activities of complex I from the earliest age (60 days) that we measured and from the older ages (Fig. 6). Specific enzyme activity of SDH was decreased from the 100-day-old and the older animals, although these decreases were not statistically significant (Fig. 6). A statistically significant decrease was observed in the specific activities of complex IV after the disease onset. However, a decreasing trend begins at the younger ages (PMW 60, 100, 150), although those decreases did not reach the level of statistical significance (Fig. 6). Similar decreases were not observed in mitochondrial fractions prepared from spinal cords of WT and WS animals at matching ages (Fig. 6) and from the cerebella of the same group of G93A animals (data not shown).

Discussion

Several groups have shown widespread morphological abnormalities of mitochondria prior to the disease onset in mice expressing G93A and G37R mutations (Dal Canto and Gurney 1995; Wong et al. 1995; Kong and Xu 1998). This work investigates functional changes in mitochondria along the course of disease progression. We measured mitochondrial ETC activities at different times before and after disease onset in G93A mice using three different but complementary methods: spectrophotometric assays on mitochondria-enriched fractions from the spinal cords (Jung et al. 2000), quantitative in situ histochemistry on spinal cord sections (Jung et al. 2002) and BN–PAGE combined with in-gel histochemistry (Jung et al. 2000). The first method is simple and quick, but only provides an average measure of the ETC activities in the whole spinal cord. Since ALS affects predominantly the motoneurons in the ventral horn, the reduction in the ETC activities may be masked. This is probably the reason why no significant changes in ETC activities were found using this method, although decreasing trends exist during disease progression (Fig. 1). This result is similar to the outcome of previous measurements of the ETC activities in spinal cord homogenates from humans and mice (Bowling et al. 1993; Browne et al. 1998; Wiedemann et al. 2002).

The second method, in situ histochemical measurement in spinal cord sections, measures regional specific changes in ETC activities, thus overcoming the shortcoming of the first method. Region-specific, disease-stage-dependent decreases in complex I and IV, and SDH activities are revealed by this method (Figs 2–5). The limitation of this method is that the results are influenced by both mitochondrial numbers and specific enzyme activities. This limitation is complemented by the third method, the histochemical BN–PAGE (Jung et al. 2000). Data obtained using this method reveal that the specific activity of complex I is reduced from the very early disease stages (Fig. 6). Unfortunately such a measurement could not be achieved for complex II and IV because of the difficulties in measuring protein quantity of these two complexes in gels. As an approximation, we estimated the specific activities of these two complexes by normalizing the activities to the average protein quantities of other measurable complexes (I, III and V). Since there is no evidence that the ratio among these complexes is changed during the progression of the disease, this approximation is perhaps reasonable. By this approximation, significant decreases are found in the specific activity of complex IV after the disease onset (Fig. 6). Decreases in both SDH and IV specific activities are also found in earlier stages, although they did not reach statistical significance.

Mitochondrial dysfunction has been observed in ALS and in cultured cell models before. Early work measuring mitochondrial ETC activities in the cortex of human and mouse ALS showed an increase in complex I activity (Bowling et al. 1993; Browne et al. 1998; Swerdlow et al. 1998), suggesting a compensatory reaction. Using cybrid technology, Swerdlow et al. reported a decrease in complex I, III and IV activities in platelets isolated from sporadic ALS patients (Bowling et al. 1993; Browne et al. 1998; Swerdlow et al. 1998). Similarly, a decrease in complex I activity was detected in muscle biopsy from sporadic ALS patients (Wiedemann et al. 1998; Vielhaber et al. 2000). Furthermore, a decrease in complex IV activity was detected in spinal cords of patients that had died of ALS (Fujita et al. 1996; Borthwick et al. 1999). In cultured neuroblastoma cells transfected with G93A mutant SOD1, Carri et al. (1997) observed a decrease in mitochondrial membrane potential. Similar findings were reported by Kruman et al. (1999) in cultured motoneurons isolated from mice expressing mutant SOD1 G93A.

To this background our results add two new points: First, by systematically measuring the ETC activities across the full range of ages and disease stages in this transgenic animal model, we demonstrate that mitochondrial dysfunction begins early in vivo, prior to disease onset. Second, this decrease is in the specific activities of the ETCs (especially in complex I). The quantity of mitochondrial proteins, as indicated by the levels of citric synthase, cytochrome a and three ETC complexes (I, III and V), are insignificantly changed during the course of disease progression, although a decreasing trend is noticeable at the end stage of the disease stage (Table 1). The specific activity decrease suggests that there is damage to these mitochondrial enzymes, resulting in inactive or less active enzymes. This differs from an early study, which showed that decreases in mitochondrial ETC activities are associated with loss of mitochondrial mass at the end stage of disease (Wiedemann et al. 2002). Although the decreases in the specific activities are apparently small (10–30%), it is dramatic considering that motoneurons are only a fraction of cells in the spinal cord.

How mutant SOD1 damages mitochondrial ETC enzymes is unclear and requires further investigation. Several studies have provided strong evidence, using rat liver and yeast, that SOD1 is present inside mitochondria (Okado-Matsumoto and Fridovich 2001; Sturtz et al. 2001), reaffirming the early observation (Weisiger and Fridovich 1973). Furthermore, mouse endogenous SOD1, as well as wild-type and mutant human SOD1 in transgenic mice are inside mitochondria in the spinal cord (Higgins et al. 2002). In addition, mutant SOD1 is associated with abnormal mitochondria in mouse CNS (Levine et al. 1999; Jaarsma et al. 2001). Therefore, mutant SOD1 is in the vicinity of mitochondrial ETC enzymes and could damage the ETC enzyme function directly. Our result indicates that such damage is induced at young ages, far ahead of the onset of the muscle weakness and significant motoneuron loss. Thus, in conjunction with the previous observations on mitochondrial morphology, our current results suggest that mitochondrial damage in motoneurons occurs early, and therefore, plays a causative role in motoneuron death. Consistent with this conclusion is the finding that dietary creatine supplement is efficacious in delaying the onset and prolonging survival of mice expressing mutant SOD1 (Klivenyi et al. 1999).

How might mitochondrial damage lead to motoneuron death? Mitochondrial damage results in energy deficiency, which in turn might lead to deficiencies in ATP-dependent ionic pumps and ionic imbalance in cells (Beal 1992). Mitochondrial dysfunction also increases the production of ROS, which in turn causes further damage to functional components in cells, including nucleic acids, biomembranes and proteins (Andreassen et al. 2000). Most relevant to ALS may be that mitochondrial dysfunction increases the sensitivity of motoneurons to excitotoxicity.

Excitotoxicity may play a role in motoneuron cell death in ALS (Rothstein 1996). Motoneurons express glutamate receptors and are susceptible to high concentrations of glutamate receptor agonists (Rothstein et al. 1993; Carriedo et al. 1996; Ikonomidou et al. 1996; Rothstein et al. 1996). In sporadic ALS patients, there is an increase in extracellular glutamate concentration (Rothstein et al. 1991). Recent experiments have also shown that expression of mutant SOD1 in cultured motoneurons enhances their sensitivity to excitatory amino acids (Baker et al. 1998; Kruman et al. 1999). The observed efficacy (although mild) on a mutant SOD1 transgenic mice using riluzole (Gurney et al. 1998), a compound thought to antagonize excitotoxicity, also suggests that excitotoxicity contributes to motoneuron death induced by expression of mutant SOD1.

Excitotoxic cell death is related to mitochondrial function. Mitochondrial dysfunction is known to sensitize neurons to glutamate toxicity (Ikonomidou and Turski 1996; Ikonomidou et al. 1996; Bittigau and Ikonomidou 1997). Furthermore, mitochondrial Ca2+ overload is required for glutamate-induced excitatory cell death in cortical neurons and spinal cord motoneurons (Stout et al. 1998; Carriedo et al. 2000). The double assaults by mitochondrial dysfunction and glutamate induced excitotoxicity may explain the selectivity of motoneuron death in ALS. Motoneurons display a higher AMPA receptor density on their surface than other neurons (Vandenberghe et al. 2000) and are particularly susceptible to mitochondrial calcium overload when exposed to AMPA (Carriedo et al. 2000). Not surprisingly, chronic partial disruption of mitochondrial function causes motoneuron death and this death can be inhibited by applying excitatory amino acid blockers (Kaal et al. 2000).

Mutant SOD1 may cause excitatory cell death of motoneurons in two ways. First, mutant SOD1 causes mitochondrial damage in motoneurons, as reviewed above. This may weaken the tolerance of motoneurons to glutamate. Second, mutant SOD1 is also capable of impairing the function of glial glutamate transporter GLT-1 (Canton et al. 1998; Pedersen et al. 1998; Trotti et al. 1999). Particularly striking is the recent finding that transgenic rats expressing mutant SOD1 G93A have a dramatic decrease in GLT-1 in the ventral horn of their spinal cords (Howland et al. 2002). This could increase the extracellular concentration of glutamate and exacerbate excitotoxicity.

Finally, mitochondria are known to play a critical role in apoptosis. Several important factors involved in the apoptosis pathway (e.g. cytochrome c, SMAC/DIABLO, APAF1 and bcl2 family proteins) either reside in mitochondria or are acting through mitochondria (Bratton and Cohen 2001). Consequently, mitochondrial damage could cause deregulation of the apoptotic pathway by releasing proapoptotic molecules from mitochondria. Indeed, apoptotic neurons are detected in mutant SOD1 transgenic mice (Martin 1999; Li et al. 2000). Many molecules involved in apoptosis are up-regulated and activated (Pasinelli et al. 2000; Almer et al. 2001). Anti-apoptotic treatments have shown efficacy in prolonging the survival of mutant SOD1 transgenic mice (Kostic et al. 1997; Li et al. 2000).

Taken together, in concert with previous observations on the widespread morphological abnormalities of mitochondria at early disease stages in mutant SOD1-induced motoneuron degeneration, our current effort demonstrates that functional deficiencies of mitochondria also occur in ventral horn of the spinal cord at early disease stages. This result supports the view that mitochondrial damage is an early step in the mutant SOD1-induced motoneuron degeneration pathway and indicates that mitochondria are a downstream target of mutant SOD1 toxicity.

Acknowledgements

We thank Ms Laura Fenton and Ellen Trang for technical assistance, Drs Kenneth Fletcher and Lang Lin for statistical analysis. This work is supported by NINDS (RO1 NS35750, NS41739) and ALS Association. CMJH is supported by NIH training grant (5-T32-NSO-7366).

Ancillary