Similar Patterns of Mitochondrial Vulnerability and Rescue Induced by Genetic Modification of α-Synuclein, Parkin, and DJ-1 in Caenorhabditis elegans*

How genetic and environmental factors interact in Parkinson disease is poorly understood. We have now compared the patterns of vulnerability and rescue of Caenorhabditis elegans with genetic modifications of three different genetic factors implicated in Parkinson disease (PD). We observed that expressing α-synuclein, deleting parkin (K08E3.7), or knocking down DJ-1 (B0432.2) or parkin produces similar patterns of pharmacological vulnerability and rescue. C. elegans lines with these genetic changes were more vulnerable than nontransgenic nematodes to mitochondrial complex I inhibitors, including rotenone, fenperoximate, pyridaben, or stigmatellin. In contrast, the genetic manipulations did not increase sensitivity to paraquat, sodium azide, divalent metal ions (Fe(II) or Cu(II)), or etoposide compared with the nontransgenic nematodes. Each of the PD-related lines was also partially rescued by the antioxidant probucol, the mitochondrial complex II activator, d-β-hydroxybutyrate, or the anti-apoptotic bile acid tauroursodeoxycholic acid. Complete protection in all lines was achieved by combining d-β-hydroxybutyrate with tauroursodeoxycholic acid but not with probucol. These results show that diverse PD-related genetic modifications disrupt the mitochondrial function in C. elegans, and they raise the possibility that mitochondrial disruption is a pathway shared in common by many types of familial PD.

The etiology of Parkinson disease has both genetic and environmental components (1). Epidemiological studies show that PD 3 is more common in rural areas, and increased rates of PD are associated with the use of agricultural toxins, such as pesticides and herbicides (2). Attention has focused on inhibitors of the mitochondrial electron transport chain because some of the agricultural toxins implicated in PD are complex I inhibitors (2). In addition, ingestion of complex I inhibitors causes syndromes related to PD. The complex I inhibitor 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine (MPTP) selectively kills dopaminergic neurons in many types of animals (1). Rotenone, another complex I inhibitor, also causes a PD-related syndrome in rats and causes multiple changes in the mitochondria of cultured neurons relevant to PD (3)(4)(5)(6). These factors implicate disruption of mitochondrial function, and particularly complex I inhibition, in the etiology of PD.
Many of the genes associated with familial cases of PD have been identified, but no clear consensus exists over whether the different disease-related proteins converge onto a common pathway. Mutation of ␣-synuclein (at A53T, A30P, or K46E) or duplication of ␣-synuclein is associated with familial parkinsonisms (7)(8)(9)(10). Loss of the putative ubiquitin ligase, parkin, causes autosomal recessive juvenile parkinsonism (11). Mutations in the genes coding for UCH-L1, DJ-1, and PINK1 are also all associated with autosomal recessive PD, and mutations in LRRK2 are associated with autosomal dominant PD (12)(13)(14)(15). ␣-Synuclein is a small ubiquitous protein that binds lipids and might regulate vesicular function, proteasomal activity, and/or signal transduction (16 -20). ␣-Synuclein has a tendency to form oligomers and fibrils, particularly after exposure to oxidative stress (21)(22)(23)(24). ␣-Synuclein is most abundant in neurons in the brain and is present in all mammals, but does not exist in Caenorhabditis elegans or Drosophila. Parkin is a putative E3 ubiquitin ligase that is neuroprotective in cell culture and transgenic Drosophila (25)(26)(27)(28). Overexpression of parkin protects against ␣-synuclein toxicity, which suggests that the two proteins affect intersecting biochemical pathways (27,28). DJ-1 is an oncogene that modulates oxidative stress possibly by affecting mitochondrial function (29 -31). The varied functions of these proteins obscures any potential convergent pathways that might underlie the pathophysiology of Parkinson disease.
In this study, we demonstrate that genetic modulation of ␣-synuclein, parkin, and DJ-1 all disrupt mitochondrial function in C. elegans. We also describe pharmacologic treatments that protect the mutant C. elegans strains against toxicity caused by the genetic mutations that support the mitochondrial convergence of the genetic pathways. These results point to disruption of mitochondrial function as a general cause of most, if not all, cases of PD.

MATERIALS AND METHODS
Generation of the ␣-Synuclein Transgenic Lines-Human wild type ␣-synuclein cDNA was inserted into the vector pPD95.75 downstream of a synaptobrevin promoter and expressed in Bristol N2 non-tg lines by co-injecting with a pDPSU006-GFP vector to produce extrachromosomal arrays. The pDPSU006-GFP vector was used as a marker for gene transmission. The A53T ␣-synuclein expression is driven by an unc-119 neuronal promoter. GFP driven by the dopamine transporter promoter was used as a marker to identify those nematodes expressing the extrachromosomal array. Expression of ␤-synuclein expression was also driven by an unc-119 neuronal promoter. Transgenes were integrated by exposing animals to 1,800 rads of ␥-irradiation, and animals were out-crossed five times.
Detection of the Wild Type and A53T ␣-Synuclein Transgene-PCR was performed to detect the transgene by using the following primers to detect the transgene: sense primer 5Ј-ATTAATTCATAGCC-3Ј and antisense primer 5Ј-CTGGGAGCAAAGATA-3Ј (Integrated DNA Technologies, Coralville, IA).
Generation of the K08E3. 7 Knockout-A deletion library of nematodes derived from 460,000 trimethlypsoralen/UV-mutagenized non-tg animals was used to generate the knockout. Previously described methods were used for library construction and screening (32). PCR with the primers listed below was used to verify a K08E3.7 deletion within the mutagenized animals. Animals verified as K08E3.7 knockouts were outcrossed five times to the wild type strain to produce a clean genetic background. The K08E3.7 knockout is a deletion of 1132 bp whose limits are GGACATTTCAAACTTTGAAT . . . AAAATAACTTCAG-GAGTGGT. The following primers were used to detect the deletion: sense primer 5Ј-ATACAAGATAGAAAAACAGGTC-3Ј and antisense primer 5Ј-GCAAAACGAAAAAAAACA-3Ј (Integrated DNA Technologies, Coralville, IA).
Other C. elegans Lines-The ced-3 knock-out strain used in this study was kindly given to us by Dr. Horvitz, Massachusetts Institute of Technology (Cambridge, MA). The BY200 lines was a generous gift from R. Blakely (Vanderbilt University). The tau P301L line was a generous gift from B. Kraemer (University of Washington). The Mito-GFP line was a generous gift from A. van der Bliek (UCLA).
Life Span Analysis-The life spans of the non-tg strains expressing wild type and A53T ␣-synuclein and the K08E3.7 KO strains were determined by using techniques described previously (33). Three NGM plates (2.5 g of bactopeptone (BD Biosciences), 3 g of NaCl, 20 g of agar, 1 mM CaCl 2 , 1 mM MgSO 4 , 5 mg of cholesterol, 25 mM KH 2 PO 4 (Sigma), 1 liter of double distilled H 2 O) seeded with OP50-1 Escherichia coli) were seeded with 40 nematodes each and were monitored throughout the life cycle for each strain.
Culturing of Nematodes and Isolation of Nematode Lysates-The Bristol N2 non-tg strain was used as the wild type strain. Growth and culture of nematodes were performed using standard techniques (34,35). Previously described methods were used for isolation. Mixed stage nematode lysates were prepared as described previously, except that the isolated pellet of nematodes was further washed with 5 ml of cold 1ϫ PBS ϩ protease inhibitor mixture (Sigma), spun down, and sonicated for lysis (36).
Immunoblot Analysis-Protein concentration of nematode lysates was determined using the BCA protein assay (Pierce). 30 g of protein were used for immunoblot assays. 2ϫ dithiothreitol protein loading buffer was added to each sample. The samples were then heated for 5 min at 90°C and run on 8 -16% SDS gradient polyacrylamide gels (Bio-Whittaker, Walkersville, MD). Proteins on the polyacrylamide gels were then transferred to polyvinylidene difluoride (Bio-Rad) overnight at 4°C at 0.1 A/gel in transfer buffer. The immunoblot was blocked in 5% milk in 100 mM Tris-buffered saline, 0.1% Tween 20 (TBST) and shaken for 1 h at room temperature. The blots were then incubated overnight at 4°C in primary antibody at the appropriate concentration in 5% bovine serum albumin in TBST. The blots were washed three times, 10 min each, and incubated for 3 h in secondary antibody (1:5000) (The Jackson Laboratory, West Grove, PA) in 5% milk in TBST at room temperature. Blots were washed three times and developed using a chemiluminescent reaction (PerkinElmer Life Sciences).
Chemical Treatments-For chemical treatments, 35-mm plates were poured with the indicated concentrations of chemicals, as described under "Results" diluted in NGM agar. The plates were seeded with bacteria medium containing rotenone, paraquat etoposide, or metal. 40 synchronized L4 nematodes of the appropriate strain were placed on each plate. Three plates were analyzed for each point; each experiment was repeated 3-6 times. Every other day the surviving nematodes were transferred onto a new plate with the same concentrations of rotenone, paraquat, etoposide, or metal. The number of surviving nematodes was counted on days 2 and 4. The number of live nematodes was determined by counting nematodes that were either moving or those that responded to light touch with a platinum wire. Toxicity was measured as a function of percent survival of the nematodes following treatment.
Treatment with Small Interfering RNA; Generation of the B0432.2 Knockdown-E. coli containing a B0432.2 genomic fragment cloned into L4440 was used to generate the C. elegans knockdown (37). The bacteria was grown for 12 h in LB ϩ 50 g/ml ampicillin (Sigma). This culture was seeded onto NGM plates containing 25 g/ml carbenicillin (Sigma) and 1 mM isopropyl ␤-D-thiogalactopyranoside (Research Products International Corp., Mt. Prospect, IL). B0432.2 RNAi was induced overnight at room temperature. The L3-L4 hermaphrodites were transferred onto the NGM plates. These nematodes were grown for 72 h at 15°C for the RNAi to take effect. Single adults were then plated onto new individual NGM plates with additives and RNAi containing bacteria. The progeny of these adults was used in subsequent assays.
Quantification of Knockdown-N2 worms were grown from hatching on HT115(DE3) bacteria containing L4440 B0432.2 (DJ-1) plasmid for knockout; worms were grown on freshly prepared bacterial plates containing 1 mM isopropyl ␤-D-thio-galactopyranoside for induction of B0432.2. For quantification of B0432.2 (DJ-1) RNA, adult worms (6 days old) were transferred to micro-tubes with 30 ml of PBS. Single worms were placed in tubes and dissolved in 300 ml of TRIzol (Invitrogen), and 12 mg of linear polyacrylamide (GenElute LPA, Sigma) was added to it as a carrier for RNA extraction. Aqueous phase with RNA was precipitated with isopropyl alcohol and washed once with 70% ethanol. The RNA pellet was dried at room temperature and dissolved in diethyl pyrocarbonate-treated water. The extracted RNA was immediately reverse-transcribed with a 3Ј DJ-1 primer (TGC CAC TGA CAA CAA CAC GA) for 1 h at 50°C. The reverse-transcribed products were PCRanalyzed with a Bio-Rad real time PCR. A 5Ј primer (CAA GTC ATC CAA GTG TTA AGG AGA AA) and a Taqman probe (TCG AGA AAG GAG GCT ACA AGT ACT CGG AGG) were added to the PCR mixture to monitor the threshold cycles for DJ-1 amplifications. PCR products were analyzed with a 3% agarose gel electrophoresis. Band intensities were measured by Labworks software (UVP Inc.) DNA sequencing was also performed to confirm the identity of the amplified band.
Thioflavine S Histochemistry-Worms were treated with 25 M rotenone for 2 days, harvested with M9 media, and fixed overnight in 4% paraformaldehyde at 4°C. After being washed three times with PBS buffer, the worms were incubated in a shaker at 37°C with TMT solution (1% Triton X-100, 5% ␤-mercaptoethanol, and 125 mM Tris, pH 7.6) for 3 h. The worms were then washed two times with a solution containing 100 mM Tris, pH 7.6, 1 mM CaCl 2 and were incubated in the same solution with 2 mg/ml collagenase for another 3 h at 37°C with vigorous agitation. After washing three times with PBS buffer, animals were put on a poly-D-lysine-coated slide in a 30-l volume and covered with a coverslip. The slides were kept at Ϫ80°C overnight. Coverslips were removed quickly at Ϫ80°C, and the slides were gradually hydrated with graded frozen methanol (100 -70%) followed by a 10-min wash with water. The slides were then incubated with 0.015% thioflavine S for 10 min and washed three times in 80% ethanol for 5 min. The slides were washed once with double distilled H 2 O. After 10 min of drying at room temperature, the slides were mounted with glass coverslips with thin layer of Fluoromount-G (Electron Microscopy Sciences, Hatfield, PA).
Dot Blot Analysis-The non-tg and A53T ␣-synuclein strains treated with rotenone were analyzed via dot blot for ␣-synuclein aggregation. Samples were vacuumed through a dot blot apparatus onto cellulose acetate membrane. Recombinant aggregated ␣-synuclein was used as a positive control. The cellulose acetate membrane was blocked and probed with the Syn303 antibody (1:500) (38) as described previously under "Immunoblot Analysis." Oxygen Consumption-5000 stage-synchronized adult worms were grown on large plates, collected in S-basal buffer, and washed free of bacteria. Changes in oxygen concentration of the worms was performed in 1 ml of S-basal buffer containing heat-killed E. coli in a closed chamber cuvette mixed with a mini-stirring bar, maintained at 22°C, and using a Clark-type electrode (Hansatech Instruments Ltd., Norfolk, UK). The worms were incubated in the chamber 5 min before measurement, and then oxygen consumption was measured for usually 1-3 min. The slope of the straight portion of the plot was used to derive the oxygen consumption rate. After the measurement, the worms were collected for protein quantification. Oxygen consumption rate for each population was normalized to its protein content.
Preparation of Mitochondria-The methods used were those described by Kayser et al. (39). Two grams of worms cleaned of E. coli were suspended in MSM-E buffer (220 mM mannitol, 70 mM sucrose, 5 mM MOPS, 2 mM EDTA, pH 7.4). The worms were ruptured with a Polytron, and then proteinase type XXVII (5 mg/g worms, Sigma) was added and incubated for 10 min. Next the slurry was homogenized in a Dounce. The slurry was mixed in 1 volume of MSM-E buffer containing 0.4% bovine serum albumin, and the homogenate was centrifuged (300 ϫ g, 10 min, 4°C). The mitochondrial pellet was resuspended in MSM-E and washed twice by centrifugation (7000 ϫ g, 10 min, 4°C). The final pellet was resuspended in 100 l of MSM-E, and total protein was determined by BCA assay.
NADH-CoQ Reductase Assay-The assay was performed as described by Kramer et al. (40). Mitochondria were mixed in freshly prepared assay buffer (0.25 mM potassium EDTA, 0.2% bovine serum albumin, 1 mM KCN, 25 mM potassium phosphate, pH 8.0, 100 M NADH at 30°C). After 3 min of preincubation, 0.02 mM decylubiquinone was added for the measurement of complex I-III. All assays contained potassium cyanide to inhibit complex IV (1.0 mM final concentration). Upon addition of substrate, samples were placed in a spectrophotometer (Beckman), and oxidation of NADH was measured at 340 nM for 15 min at 30°C. Rates of oxidation were calculated according to Equation 1, ͑⌬A/min͒͑dilution factor͒͑1/⑀͒͑1/1 cm͒ ⅐͑1000 g/mg͒/͑g/L protein͒ ϭ enzyme activity/mg protein (Eq. 1) where ⌬A/min is the change in absorbance; 1 cm is the cuvette path length, and ⑀ ϭ 6.81 is the extinction coefficient (in mM/liter cm).
Statistical Analysis-Statistical analysis on the raw numbers was done using an unpaired t test or analysis of variance followed by a Newman Keuls post-hoc analysis (Statview), when necessary. All experiments were performed in triplicate.

Characterization of the Wild Type and A53T ␣-Synuclein
Nematodes-The presence of wild type or A53T ␣-synuclein in the C. elegans transgenic lines was demonstrated by PCR and immunoblot analysis (Fig. 1, A-C). PCR of the transgenic C. elegans genomes yielded 450-bp bands that corresponded to full-length wild type and A53T ␣-synuclein cDNA (Fig. 1A, 1st and 2nd lanes). The plasmid used to generate the wild type ␣-synuclein transgenic lines was used as a control to verify the size of ␣-synuclein (Fig. 1A, 4th lane). No ␣-synuclein was detected through PCR analysis in the Bristol N2 nontransgenic (non-tg) strain, which is consistent with the absence of a homologue for ␣-synuclein in the C. elegans genome (Fig. 1A, 3rd lane). The presence of ␣-synuclein was validated by immunoprecipitating ␣-synuclein from the A53T ␣-synuclein transgenic or Bristol N2 non-tg lines. A 14-kDa band, characteristic of ␣-synuclein, was evident only in the A53T lines; no such band was present in the N2 line or in A53T ␣-synuclein lysates precipitated with preimmune serum (Fig. 1B).
The life spans of the C. elegans strains were examined as a simple test to determine whether expression of ␣-synuclein or lack of K08E3.7 affects the metabolism of the nematodes, as is seen for loss of the parkin homologue in Drosophila (58). The mean life span was similar among the non-tg, wild type, and A53T ␣-synuclein-expressing strains; however, the K08E3.7 KO strain had a 15.4% shorter life span that the non-tg strain ( Fig. 2A). The mean survival for the non-tg strain was 18.5 Ϯ 0.169 days, whereas that for the K08E3.7 KO strains was 15.65 Ϯ 0.211 days, p Ͻ 0.01 ( Fig. 2A). The transgenic and nontransgenic C. elegans strains were also examined visually at multiple life stages from larval development through adulthood. No defects were observed in any transgenic strains through development and into adulthood.
KO8E3.7/Parkin KO Reduces Ubiquitination in C. elegans-The K08E3.7 KO and non-tg nematodes were examined to explore the func-tion of K08E3.7 in the nematode. The results show that the K08E3.7 KO strain exhibits reduced levels of basal ubiquitination (Fig. 2, B-D). Basal ubiquitination in the K08E3.7 KO strain was 55 Ϯ 6.5% (p Ͻ 0.01) than of the non-tg strain (Fig. 2D); in comparison, basal ubiquitination was unchanged in the A53T ␣-synuclein line (data not shown). The reduction in high molecular weight ubiquitinated proteins in the K08E3.7 KO worm strain contrasts with results reported for parkin knock-out mice (41). C. elegans might lack the intrinsic ability to compensate for lack of parkin, which could lead to a result similar to that observed for acute knockdown of parkin in cell culture. Reduced ubiquitination in the K08E3.7 KO strain lends support to the hypothesis that parkin is involved in the ubiquitin proteasomal system.
Other functions examined in the K08E3.7 knock-out strain were unaffected. For instance, defecation and egg laying were not significantly changed (supplemental Fig. 1, A and B). Chemotaxis to isoamyl alcohol, trimethylthiazol, and pyrazin also were not different from the N2 strain, although chemotaxis to diacetyl was significantly decreased in the K08E3.7 knock-out line (supplemental Fig. 1C).

Survey of Toxicology
Transgenic Lines Expressing PD-related Genetic Factors Show Selectively Increased Vulnerability to Mitochondrial Complex I Inhibitors-Environmental factors are hypothesized to contribute to the pathophysiology of PD, but whether such factors interact with gene mutations associated with PD is poorly understood. To investigate the interaction of genes with environmental toxins, we examined the vulnerability of the A53T ␣-synuclein and K08E3.7 worms to a variety of toxins, including inhibitors of complex I-IV, redox active metals, and DNA repair inhibitors. For controls, we used the Bristol N2 nontransgenic worms, a transgenic worm line expressing GFP driven by a promoter for the dopamine transporter (BY200), a transgenic worm line expressing human ␤-synuclein driven by the neuron specific unc119 promoter, and a transgenic worm line expressing human P301L tau driven by the muscle-specific myo 3 promoter (42,43).
The first treatment was with rotenone, a known mitochondrial complex I inhibitor. Rotenone (25 or 50 M) induced a significant dose-dependent mortality in all the strains, which demonstrates that rotenone is toxic to the non-tg, K08E3.7 KO and ␣-synuclein-expressing transgenic strains (Fig. 3, A, dose response; B, time course). Because most of the worms were dead by 4 days, the life span analysis was stopped at 4 days. Most interestingly, the K08E3.7 KO, wild type, and A53T ␣-synucleinexpressing strains showed reduced viability at each concentration of rotenone and at both 2 and 4 days of treatment compared with the non-tg strain or BY200 line at the same dose (Fig. 3, A and B, only the non-tg is shown as the control, and see Fig. 4, E and F, for BY200 data). After 25 M rotenone treatment for 4 days, the K08E3.7, wild type, and A53T ␣-synuclein-expressing strains showed 74 Ϯ 1.48%, 60 Ϯ 1.2%, and 56 Ϯ 1.12% survival, respectively (p Ͻ 0.01). The A53T ␣-synucleinexpressing strain showed greater vulnerability to rotenone compared with the wild type ␣-synuclein-expressing strain with 50 M rotenone treatment, exhibiting 68.4% (p Ͻ 0.01) lower survival after 4 days of 50 M rotenone (Fig. 3B). Because the K08E3.7 KO represents just one line of worms, we also examined the effects of knockdown with RNAi to K08E3.7 or Sel-9 (the latter representing a gene that should not affect the response to rotenone and is a control for RNAi). Worms that were treated with RNAi for K08E3.7 showed 43 Ϯ 1.3% (p Ͻ 0.01) more toxicity than worms treated with RNAi for Sel-9, confirming the effects of K08E3.7 on rotenone toxicity.
To determine whether the increased vulnerability of the PD-related nematodes generalized to other complex I inhibitors, we tested other known mitochondrial complex I inhibitors. We examined the sensitivity to fenperoximate, pyridaben, stigmatellin, piericidin, papaverin, capsaicin, and 4-phenylpyridin, all of which are known to inhibit complex I (44). The following three of these compounds were toxic to C. elegans: fenperoximate, pyridaben, and stigmatellin. We observed that the PDrelated nematode lines showed increased vulnerability to each of these complex I inhibitors, similar to that observed with rotenone ( Fig. 3,  C-E). These data suggest that the PD-related C. elegans lines are generally more vulnerable to complex I inhibitors.
Next we tested whether the sensitivity of the PD-related C. elegans was restricted to complex I inhibition or extended to inhibition of the mitochondrial complexes II-IV. Each of the C. elegans lines was exposed to antimycin A, 3-nitroproprionic acid (3-NP) or sodium azide, which inhibit mitochondrial complexes II, III and IV, respectively. The non-tg and transgenic strains were equally vulnerable to 500 mM azide (Fig. 4A). We saw 50% survival for the non-tg and transgenic strains at 12.5 h following azide treatment. Antimycin A and 3-NP had no effect on the nematodes, suggesting that the worms do not take up the compounds (data not shown).
PD-related Genetic Factors Do Not Increase Vulnerability to Redox Active Metals, Paraquat, or Etoposide-To test whether PD-related genetic factors affect the vulnerability of nematodes to other toxins, each strain was exposed to different concentrations of CuCl 2 or FeCl 2 (Fig. 4B), paraquat (Fig. 4C), or etoposide (Fig. 4D). The synucleinexpressing or K08E3.7 KO lines did not exhibit higher sensitivity to the metals or paraquat compared with the non-tg strain, and the  DECEMBER 30, 2005 • VOLUME 280 • NUMBER 52 ␣-synuclein-expressing strains actually showed slightly reduced vulnerability to etoposide, a topoisomerase II inhibitor (Fig. 4D, 4-day treatment). The data suggest that ␣-synuclein can be mildly protective against etoposide toxicity when expressed in the transgenic strains. Loss of K08E3.7 was associated with a small increase in vulnerability to etoposide-induced toxicity compared with the non-tg strain (Fig. 4C). These results indicate that the PD-related genetic changes cause increases in vulnerability to toxins that are largely selective for inhibition of mitochondrial complex I in C. elegans, although we cannot rule out the possibility that the vulnerability might also include inhibition of complexes II or III, because the worms were not susceptible to 3-NP or antimycin A.

Parkinson Genes Increase Rotenone Toxicity in C. elegans
As further controls, we examined the toxicity profile of worm strains expressing either human P301L tau driven by a muscle-specific promoter (myo-3) or human ␤-synuclein driven by the unc119 neuronal specific promoter (43). Under basal conditions, the P301L worms moved slower and laid fewer eggs than the other lines, suggesting that they might be somewhat less healthy. The ␤-synuclein worms exhibited no gross abnormalities; they moved at normal speeds and laid normal numbers of eggs. The pattern of toxicity for the tau worm line differed distinctly from that of the PD-related lines. The P301L tau worms showed increased toxicity to copper (CuCl 2 , 0.75 mM) but no difference in vulnerability to rotenone (25 M), compared with Bristol N2 worms or BY200 worms (Fig. 4, E and F). In contrast, rotenone (25 M) did not induce any extra toxicity in the ␤-synuclein worms compared with the Bristol N2 worms (Fig.  4G). These data demonstrate that expression of a protein prone to aggregation, such as P301L tau, does not necessarily cause a selective vulnerability to complex I inhibitors.

Mechanism of Toxicity, Production of Free Radicals, Activation of Apoptosis, and Formation of Protein Aggregates
Rotenone Inhibits Respiration-The vulnerability to complex I inhibitors raises the possibility that the function of the electron transport chain might be reduced in worms with PD-related genetic changes. To . Vulnerability to other toxins unchanged in K08E3.7 KO, wild type, and A53T ␣-synuclein-expressing lines. A, kill curve over a 24-h treatment with 500 mM sodium azide. Neither the K08E3.7 KO, wild type (WT), nor A53T ␣-synuclein-expressing strains show increased sensitivity to sodium azide compared with the non-tg strain; ϩ, p Ͻ 0.01. B and C, no difference in sensitivity to Fe(II) (B), Cu(II) (B), or paraquat (C) compared with the non-tg strain. D, both wild type and A53T ␣-synuclein are slightly protective against etoposide treatment, whereas K08E3.7 KO strain is slightly vulnerable compared with the non-tg strain. E, the P301L tau C. elegans line shows increased vulnerability to copper compared with non-tg or BY200 worm lines after 6 days of exposure to 0.75 mM CuCl 2 . F, the P301L tau C. elegans line shows no difference in vulnerability to rotenone (25 M) compared with non-tg or BY200 worm lines; *, p Ͻ 0.01 compared with non-tg. G, the ␤-synuclein-expressing C. elegans line shows no difference in vulnerability to rotenone (25 M) compared with non-tg, whereas the A53T ␣-synuclein line shows increased toxicity; *, p Ͻ 0.001 compared with non-tg and ␤-synuclein (B-syn), n ϭ 8. H, the K08E3.7 KO and A53T ␣-synuclein show reduced oxygen consumption under basal conditions compared with the non-tg line. ϩ, p Ͻ 0.01 compared with untreated nematodes of the same strain. Rot, rotenone.
investigate this, we examined whether the doses of rotenone used inhibited oxygen consumption. C. elegans lines that were non-tg or expressing A53T␣-synuclein or lacking K08E3.7 were exposed to rotenone (1 M) for 1 h, and oxygen consumption was measured; the line expressing P301L tau driven by a muscle promoter was used as an additional control in this experiment. Lower doses and shorter exposures of rotenone were used based on titrations of the non-tg line to determine doses of rotenone that would yield detectable oxygen consumption signals over the course of the assay. The results indicated that the worms expressing A53T ␣-synuclein or lacking K08E3.7 showed lower levels of oxygen consumption under basal conditions or after treatment with rotenone compared with the nontransgenic line (Fig. 4H).
To verify the mechanism of action of the other complex I inhibitors, we also investigated whether pyridaben and phenylpyridin inhibited respiration in the worm. Bristol N2 nontransgenic C. elegans were incubated in rotenone (50 M, 1 h), pyridaben (25 M, 1 h), or phenylpyridin (10 mM, 1 h) after which respiration was measured. Both compounds significantly decreased respiration (supplemental Fig. 2). In contrast, etoposide (50 M, 1 h) did not significantly decrease respiration (supplemental Fig. 2). To verify further that rotenone was acting by inhibiting complex I, we isolated mitochondria from Bristol N2 nontransgenic C. elegans, and we monitored mitochondrial NADH-CoQ reductase activity using decylubiquinone as the electron acceptor (40,45). Under control conditions, C. elegans exhibited an activity of 44 units/mg, whereas after treatment with rotenone (50 M) no activity was detectable.

Altered Morphology of Dopaminergic Neurons and Aggregation of Synuclein during Rotenone Treatment in A53T ␣-Synuclein-expressing
Worms-To investigate whether rotenone was affecting neurons in C. elegans lines, we subjected C. elegans line BY200 (expressing GFP driven by the dopamine transporter promoter) to 25 M rotenone for 1-4 days. The morphology of the dopaminergic neurons was not changed by 2 days of rotenone treatment (Fig. 5, A and B). However, staining with thioflavine S revealed the presence of multiple thioflavine-positive inclusions that were present in 38.7 Ϯ 1.3% of the worms (Fig. 5C). In addition, after thioflavine staining, occasional worms showed processes that were strongly thioflavine-positive (Fig. 5D). We also examined mitochondrial morphology in response to 25 M rotenone by examining the response of a C. elegans line expressing a GFP with a mitochondrial promoter (provided by A. van der Bliek) (46). However, no changes in morphology were observed in response to rotenone, and we did not observe selective mitochondrial degeneration in any particular anatomic entity in the worm (supplemental Fig. 3).
Presence of Thioflavine-positive Aggregation of ␣-Synuclein in Response to Rotenone-To test whether the ␣-synuclein was aggregating, the non-tg and A53T ␣-synuclein lines were exposed to rotenone (25 M, 4 days), after which treated and untreated worms were homogenized in PBS. The insoluble material was captured with cellulose acetate membrane and probed with the Syn303 antibody that selectively recognizes aggregated ␣-synuclein (38). The Syn303 antibody reacted with lysates from the rotenone-treated A53T ␣-synuclein line but did not react with lysates from untreated worms or the rotenone-treated non-tg line (Fig. 5, E and F). The synuclein aggregates from either total lysates or a Sarkosyl insoluble fraction were not detectable following PAGE and immunoblotting (with or without prior immunoprecipitation), suggesting that the aggregates were not SDS-resistant; in addition, immunoprecipitated ␣-synuclein was not ubiquitinated (data not shown). Previous studies suggest that oxidation stimulates synuclein aggregation. To determine whether the aggregation was associated with increased protein oxidation, we performed oxyblots on the worm lysates (Fig. 5G). The results demonstrated a clear increase in the amount of oxidation in the lysates (Fig. 5G). These results indicate that aggregation accompanies rotenone treatment in the nematodes expressing A53T ␣-synuclein and is consistent with prior studies showing that ␣-synuclein aggregates in response to oxidative stresses (21,24,(47)(48)(49)(50).
Rotenone Induces Caspase Activation-Next we explored whether the rotenone activated apoptosis. Non-tg, A53T-␣-synuclein-expressing worms, or worms lacking K08E3.7 were treated with 25 M rotenone for 2 days, and levels of caspase activity were compared with untreated worms. Caspase activity showed a trend toward elevation in the non-tg worms (33 Ϯ 19%, p ϭ 0.88, n ϭ 4) but was significantly elevated in the worms expressing A53T ␣-synuclein (101 Ϯ 22%, p Ͻ 0.001, n ϭ 4); caspase activation was not observed in the K08E3.7 KO worm. This suggests that significant levels of apoptosis were induced by rotenone in the worms expressing A53T ␣-synuclein and identifies a difference in the types of cell death processes activated by rotenone among the worm strains.

Rescue of Mitochondrial Function Protects against Rotenone-induced
Toxicity-Rotenone is known to induce free radical generation, inhibition of complex I, and apoptosis. To test whether preventing these processes protects against rotenone toxicity, the non-tg and PD-related lines (A53T ␣-synuclein strain and KO8E3.7 KO strain) were co-treated with rotenone and putative protective compounds such as anti-oxidants, mitochondrial complex II activators, or anti-apoptotic compounds. Testing of the ␣-synuclein lines was limited to the A53T ␣-synuclein line to limit the complexity of the analysis.
Two known anti-oxidants, N-acetylcysteine and probucol, were tested for protection against rotenone-induced free radical generation (51). N-Acetylcysteine was unable to protect any of the strains against rotenone toxicity (data not shown). In contrast, probucol significantly reduced rotenone toxicity, yielding 12.13, 13.34, and 24.45% higher survival rates for the non-tg, K08E3.7 KO, and A53T ␣-synuclein lines than treatment with rotenone alone (Fig. 6A). However, probucol was unable to fully protect any of the nematode strains against rotenone-induced toxicity. The ability of probucol to partially protect against rotenoneinduced toxicity suggests that the mechanism of toxicity involves generation of free radicals, although direct measurement of free radical production would be required to prove this hypothesis.
To test a strategy of bypassing rotenone-mediated inhibition of mitochondrial complex I, we utilized D-␤-hydroxybutyrate (D␤HB), which increases cellular succinate, a substrate for complex II (52,53). Treatment with D␤HB protected the non-tg and PD-related lines against rotenone-induced toxicity in a dose-dependent manner, fully protecting the nontransgenic strain and partial protecting the K08E3.7 KO and the A53T ␣-synuclein lines (Fig. 6B). Further studies in C. elegans suggest that the D␤HB is acting by stimulating respiration, consistent with a mechanism of action via the electron transport chain. Treating C. elegans with rotenone (50 M, 1 h) blocked respiration, but concurrent treatment with 50 mM D␤HB partially restored respiration (supplemental Fig. 4). Full protection by D␤HB in the non-tg strain is consistent with results from prior studies indicating that rotenone acts via inhibition of mitochondrial complex I.
The cytoprotective bile acid tauroursodeoxycholic acid (TUDCA), which is a more soluble form of ursodeoxycholic acid (a medicine used clinically for treatment of certain cholestatic liver diseases), was examined next for protection against rotenone-induced toxicity. TUDCA is a putative anti-apoptotic compound that inhibits Bax activity in eukaryotes and might affect the apoptotic system in nematodes (54 -56). Treatment with TUDCA yielded dose-dependent protection against rotenone-induced toxicity with full protection of the non-tg line occurring at 10 mM TUDCA. Treatment with 10 mM TUDCA also increased survival of the K08E3.7 KO and A53T ␣-synuclein-expressing strains by 23.8 and 25%, respectively, but was unable to provide full protection (Fig. 6C). We also tested other putative cytoprotective agents, including radicicol (0.1-10 mM), resverotrol (0.5-10 mM), and taxol (25-2000 M); however, none of these agents protected against rotenone toxicity.
D␤HB Shows Additive Protection in Combination with TUDCA but Not Probucol-No dose of probucol, D␤HB, or TUDCA provided complete protection against rotenone-induced toxicity in the K08E3.7 KO or A53T ␣-synuclein-expressing transgenic strains (Fig. 6, A-C). To test whether protection by the different agents was additive, the nematode lines were treated with different combinations of probucol, D␤HB, or TUDCA. Probucol did not supplement the protection against rotenoneinduced toxicity provided by either D␤HB or TUDCA in the K08E3.7 KO and A53T ␣-synuclein-expressing lines (Fig. 6, D and E). However, co-treatment with D␤HB and TUDCA fully prevented against rotenone-induced toxicity in the non-tg and transgenic lines (Fig. 6F). This result suggests that a combined pharmacological approach based on activating mitochondrial complex II and inhibiting apoptosis in C. elegans can prevent the toxicity induced by rotenone in the presence of PD-related genetic changes, assuming that TUDCA acts in the worm by inhibiting apoptosis.
To test whether TUDCA inhibits apoptosis in C. elegans, we treated C. elegans lacking the executioner caspase ced-3 with 25 M rotenone Ϯ 10 mM TUDCA (Fig. 7A). The worms lacking ced-3 showed less vulnerability to rotenone treatment, suggesting that the apoptotic machinery contributes to rotenone-induced cell death (Fig. 7A). TUDCA did not provide any additional protection from rotenone-induced toxicity to the ced-3 knock-out nematodes, although it did partially protect the non-tg worms (Fig. 7A). The inability of TUDCA to add to the protection provided by ced-3 suggests that the mechanism of protection by TUDCA occurs at least in part through a pathway that involves the apoptotic apparatus.
Toxicity from Paraquat Is Partially Protected by Probucol but Not D␤HB-Treatment of the non-tg and transgenic strains with paraquat indicated that all the strains were equally vulnerable to paraquat treatment (Fig. 3B). Because the mechanism of paraquat toxicity is not fully understood, we sought to use the nematodes to explore the mechanism of toxicity. Co-treatment of non-tg nematodes with 25 M paraquat Ϯ 25 or 50 mM D␤HB did not attenuate paraquat toxicity (Fig. 7B). However, co-treatment with paraquat and probucol increased survival of all strains by 30% compared with paraquat treated alone (Fig. 7C). The partial protection provided by probucol indicates that free radical generation contributes to paraquat toxicity. The inability of D␤HB to protect against paraquat toxicity suggests that the toxicity in C. elegans is independent of complex I inhibition. A lack of full protection by probucol suggests either that paraquat has a secondary unknown mechanism of toxicity or that multiple free radical scavengers are required to minimize oxidation.
Knockdown of DJ-1 (B0432.2) Increases Vulnerability to Rotenone, Which Is Rescued by a Combination of D␤HB and TUDCA-The results described above show that C. elegans lines expressing ␣-synuclein (wild type or A53T) respond to complex I inhibition in much the same manner as C. elegans lines lacking parkin. To investigate whether this might be a trait that generalizes to other PDrelated genes, we examined the responses of nematodes following knockdown of the B0432.2, the C. elegans homologue of DJ-1. The non-tg Bristol N2 C. elegans line was grown on HT115(DE3) E. coli containing the L4440 vector coding for RNAi for B0432.2 (DJ-1). Two days later, the nematodes were exposed to 25 M rotenone, and after 4 days the viability of the nematodes was scored. Knockdown of B0432.2 (DJ-1) produced a 53% reduction in transcript level (Fig.  8A). The response of the nematodes resembled that of the other PD-related lines because the nematodes with B0432.2 (DJ-1) knockdown were observed to be significantly more sensitive to rotenone treatment than control nematodes (Fig. 8B); knockdown by a bacterial line carrying a control vector (Sel-9) had no effect on the vulnerability to rotenone (data not shown). Rescue by D␤HB and TUDCA was also examined. Nematodes with B0432.2 (DJ-1) knockdown were exposed to rotenone and either 25 mM D␤HB, 10 mM TUDCA, or both. Treatment with either D␤HB or TUDCA alone induced a partial rescue of the B0432.2 (DJ-1) knock-down nematodes, but treatment with D␤HB and TUDCA together elicited full protection (Fig. 8C). These results provide evidence that PD-related lines of C. FIGURE 6. Protection against rotenone-induced toxicity. A, co-treatment of the K08E3.7 KO and A53T ␣-synuclein-expressing strains with rotenone and probucol yielded partial protection of the non-tg, K08E3.7 KO, and A53T ␣-synuclein (syn) strains compared with untreated of the same strain; ϩ, p Ͻ 0.01. B and C, co-treatment of the non-tg, K08E3.7 KO, and A53T ␣-synuclein-expressing strains with rotenone and D␤HB (B) or TUDCA (C) fully protected the non-tg strain, but partially protected the K08E3.7 KO and A53T ␣-synuclein-expressing strains compared with untreated nematodes of the same strain; ϩ, p Ͻ 0.01. Treatment with probucol, D␤HB, or TUDCA was unable to completely overcome the enhanced rotenone-induced toxicity of the transgenic strains as compared with the non-tg strain in any treatment group; *, p Ͻ 0.01. D, both probucol and D␤HB provided partial protection against rotenone for K08E3.7 KO and A53T ␣-synuclein-expressing strains but did not show any additive benefit when combined together; ϩ, p Ͻ 0.01 comparing the treatments; *, p Ͻ 0.01 comparing the strains. E, probucol also did not show any additive benefit against rotenone toxicity when combined with TUDCA; ϩ, p Ͻ 0.01 comparing the treatments; *, p Ͻ 0.01 comparing the strains. F, combining D␤HB and TUDCA fully protected against rotenone-induced toxicity in all the strains compared with untreated of the same strain; ϩ, p Ͻ 0.01. elegans carrying three different PD-related genetic factors exhibit similar responses to complex I inhibition.

DISCUSSION
Studies implicate a diverse array of environmental and genetic factors in the pathophysiology of PD. Increasing lines of evidence suggest that mitochondrial toxicity represents a common target of each of these agents. The research described above investigated the toxicological profile of three different genetic models in C. elegans related to PD. We examined C. elegans lines overexpressing wild type or A53T human ␣-synuclein or lacking KO8E3.7, the homologue of parkin in C. elegans. Each line showed a selective increase in the vulnerability to inhibitors of the mitochondrial complex I, including rotenone, fenperoximate, pyridaben, or stigmatellin. Knockdown of B0432.2, the homologue of DJ-1 in C. elegans, also increased the vulnerability to complex I inhibition.
The sensitivity of C. elegans to rotenone is less than that of mammalian organisms, but lower sensitivity to toxins is commonly observed in worms and might reflect the thick protective cuticle present in worms. In contrast, the C. elegans lines did not show increased sensitivity to other types of toxins, including inhibition of complex IV, oxidative stress, or DNA damage. The convergence of the toxicology profiles on complex I prompted us to examine whether treatments that improve mitochondrial function might be applicable to nematode models related to PD. First we examined D␤HB, which increases succinate levels and provides a secondary source of substrate for complex II (53). Treating the nematodes with D␤HB fully protected the nontransgenic nema-todes against rotenone-mediated toxicity, and partially protected the PD-related lines against complex I inhibition. Combining D␤HB with TUDCA, which inhibits apoptosis, another mitochondrial function, fully protected all the lines against complex I inhibition. These results suggest that PD-related genetic changes interfere with mitochondrial function and disrupt complex I function. Whether they also affect the function of complexes II and/or III remains to be determined because the worms were not sensitive to 3-NP or antimycin A.
Each aspect of the genetic and toxic models studied above provide potentially important insights into the pathophysiology of human disease. The C. elegans line with the KO8E3.7/parkin deletion shows reduced levels of high molecular weight ubiquitin conjugates. This represents the first in vivo evidence that loss of a parkin homologue affects the ubiquitin proteasomal system. The studies reporting knockout of parkin in mice and knockout of parkin in Drosophila did not report any changes in ubiquitination, although levels of some mitochondrial proteins were noted to be changed by Palacino and co-workers (41,57,58). Once antibodies to the nematode homologues of parkin substrates are available, it will be of great interest to study the turnover of these proteins.
The pathophysiological processes associated with rotenone toxicity in C. elegans bear many similarities to those occurring in mammals and provide valuable insights into the pathophysiology of complex I inhibition. Prior studies have shown that compounds such as rotenone and MPTP inhibit complex I of the mitochondrial electron transport chain in many species (1). The ability of D␤HB to protect against rotenone- FIGURE 7. Analysis of the mechanisms of action of TUDCA and paraquat. A, treatment with TUDCA does not protect against rotenone-induced toxicity in the Ced-3 knock-out worm compared with the untreated group; ϩ, p Ͻ 0.01. The Ced-3 knock-out worms were less vulnerable to rotenone toxicity than the non-tg strain, but addition of TUDCA did not provide any additional protection; *, p Ͻ 0.01. B, treatment with D␤HB did not protect the non-tg, K08E3.7 KO, or A53T ␣-synuclein-expressing strain against paraquat-induced toxicity. ϩ, p Ͻ 0.01 compared with untreated of the same strain. C, treatment with probucol partially protects against paraquat induced toxicity in the non-tg, K08E3.7 KO, and A53T ␣-synuclein-expressing strains; #, p Ͻ 0.01 compared with paraquat treated with the same strain; ϩ, p Ͻ 0.01 compared with untreated of the same strain. mediated toxicity provides further evidence that rotenone acts by inhibiting complex I activity. The secondary effects of complex I inhibition also appear to be similar in mammals and C. elegans. Rotenone has been shown to cause oxidative damage in rats in addition to inhibiting electron transport, and our results indicate that rotenone causes oxidative damage in C. elegans (4). Oxyblots of lysates from rotenone-treated nematodes showed increased reactivity compared with untreated nematodes. Rotenone treatment causes the accumulation of ␣-synuclein aggregates in rats, possibly through a mechanism mediated by oxidative stress, and similar processes occur in C. elegans (3). Thioflavine-positive deposits and ␣-synuclein aggregation were apparent following rotenone treatment of the A53T C. elegans line. In addition, fragmentation of the GFP signal in the A53T ␣-synuclein worms following rotenone treatment suggests damage to dopaminergic neurons. The pathology that we observed complements a previous report by Lakso et al. (42) of ␣-synuclein-induced dopaminergic dysfunction in a transgenic synuclein worm line. The synuclein aggregates developed following rotenone treatment did not show SDS resistance, unlike that which occurs in humans, but this susceptibility to SDS might reflect the acute nature of the rotenone treatment paradigm used in this study. SDSresistant ␣-synuclein aggregates only appear in Drosophila after 3 weeks of aging, in rat after 6 weeks of rotenone treatment, and in transgenic mice after 6 -11 months of aging (42, 47, 50, 59 -61). These results suggest that the pathophysiology of rotenone toxicity in C. elegans shares important similarities with that occurring in mammals.
The specificity of rotenone and MPTP for complex I raises the possibility that treatments that act distal to complex I to enhance the function of the electron transport chain might be beneficial. D␤HB is a ketone that increases the levels of succinate, which is a substrate for complex II (52). By increasing substrate levels for complex II, D␤HB increases the activity of complex II and protects against a variety of toxic insults (52,53,62). Recent studies demonstrated that D␤HB partially protects against MPTP in cell culture and in vivo (52,53). However, humans with PD might have genetic changes that predispose to the disease, which could render them less responsive to D␤HB. The nematode models of PD have allowed us to explore how PD-related genetic changes alter the response to D␤HB. The studies confirm that D␤HB effectively protects the nontransgenic nematode against complex I inhibition. However, D␤HB was unable to fully protect against complex I inhibition in nematodes containing PD-related genetic changes. Although most patients with PD do not have mutations in parkin, synuclein, or DJ-1, patients with PD might have other genetic changes that affect the mitochondria in a similar manner and render them less responsive to D␤HB.
A limited screen of other known neuroprotective compounds identified two compounds, probucol and TUDCA, that partially protected against mitochondrial toxicity. Probucol was designed to lower cholesterol and has been used to treat hyperlipidemia in humans, but it is also a potent anti-oxidant (51). Protection by probucol is consistent with the suggestion by Sherer et al. (4) that rotenone induces free radical production and oxidative injury. However, probucol did not show protection that was additive with that of D␤HB, which suggests that D␤HB reduces oxidative stress in addition to increasing substrate availability for complex II. In contrast, TUDCA did show protection that was additive with that of D␤HB. TUDCA has been shown to inhibit apoptosis in mammals, and its cousin ursodeoxycholic acid is clinically approved for use in humans (55,56,63). TUDCA has not been studied in C. elegans previously, but our studies indicated that TUDCA does not protect C. elegans lacking ced 3, which suggest that TUDCA acts in part by inhibiting apoptosis. A corollary of these results is the likelihood that complex I inhibition induces apoptosis in the nematode as it does in mammalian cells; measurement of caspase activity supports this hypothesis. The complex I inhibitor MPTP induces apoptosis in vivo, and mice lacking the pro-apoptotic protein Bax are less vulnerable to MPTP (64,65). Post-mortem studies of patients who died from PD also show histochemical evidence of apoptosis (66,67). Together, these results suggest that combination therapy that stimulates complex II function and inhibits apoptosis enhances protection against complex I inhibition.