Chronic systemic complex I inhibition induces a hypokinetic multisystem degeneration in rats


Address correspondence and reprint requests to G. U. Höglinger; INSERM U 289, Experimental Neurology and Therapeutics; Hôpital de la Salpêtrière; 47, Boulevard de l'Hôpital; 75651 Paris Cedex 13, France. E-mail:


In Parkinson's disease, nigral dopaminergic neurones degenerate, whereas post-synaptic striatal target neurones are spared. In some atypical parkinsonian syndromes, both nigral and striatal neurones degenerate. Reduced activity of complex I of the mitochondrial respiratory chain has been implicated in both conditions, but it remains unclear if this affects the whole organism or only the degenerating brain structures. We therefore investigated the differential vulnerability of various brain structures to generalized complex I inhibition. Male Lewis rats infused with rotenone, a lipophilic complex I inhibitor [2.5 mg/kg/day intraveneously (i.v.) for 28 days], were compared with vehicle-infused controls. They showed reduced locomotor activity and loss of striatal dopaminergic fibres (54%), nigral dopaminergic neurones (28.5%), striatal serotoninergic fibres (34%), striatal DARPP-32-positive projection neurones (26.5%), striatal cholinergic interneurones (22.1%), cholinergic neurones in the pedunculopontine tegmental nucleus (23.7%) and noradrenergic neurones in the locus ceruleus (26.4%). Silver impregnation revealed pronounced degeneration in basal ganglia and brain stem nuclei, whereas the hippocampus, cerebellum and cerebral cortex were less affected. These data suggest that a generalized mitochondrial failure may be implicated in atypical parkinsonian syndromes but do not support the hypothesis that a generalized complex I inhibition results in the rather selective nigral lesion observed in Parkinson's disease.

Abbreviations used



choline acetyltransferase


dopamine- and cAMP-regulated phosphoprotein


3,3-diaminobenzidine tetrahydrochloride




entopeduncular nucleus


fetal calf serum


frontal cortex


glial fibrillary acidic protein


external segment of the globus pallidus


granule layer of the dentate gyrus




myelin oligodendrocyte-glycoprotein


normal goat serum


3-nitropropionic acid


phosphate-buffered saline


Parkinson's disease


polyethyleneglycol 400


pedunculopontine tegmental nucleus


progressive supranuclear palsy


hippocampal pyramidal layer


serotonin transporter


substantia nigra pars compacta


subthalamic nucleus


tyrosine hydroxylase


ventral tegmental area

Parkinson's disease (PD) is a hypokinetic movement disorder characterized by degeneration of dopaminergic nigrostriatal projection neurones (Ehringer and Hornykiewicz 1960; Jellinger 2001). In most cases, there is additional degeneration of cholinergic, serotoninergic and noradrenalinergic neurones, but striatal neurones are not lost (Jellinger 2001; Hirsch et al. 2002). Atypical parkinsonian syndromes, such as progressive supranuclear palsy (PSP; Steele et al. 1964; Daniel et al. 1995), are distinct neuropathological entities, in which more widespread neurodegeneration occurs. This includes degeneration of both nigral and striatal neurones, a feature which distinguishes them from idiopathic PD.

It has been suggested that reduced activity of complex I of the mitochondrial respiratory chain in the substantia nigra (SN) may be involved in the pathophysiology of PD (Schapira et al. 1989, 1990; Mann et al. 1992). Some studies also described complex I inhibition in platelets and muscle of PD patients (Bindoff et al. 1989; Parker et al. 1989), suggesting that in PD the SN may be particularly vulnerable to a generalized complex I dysfunction affecting the whole body. Other studies, however, were either unable to confirm peripheral complex I alterations (Mann et al. 1992; DiDonato et al. 1993) or demonstrated that, even within the brain, the complex I deficiency is confined to the SN (Schapira et al. 1990).

In PSP, there is some evidence in favour of a generalized impairment of mitochondrial function (Albers and Beal 2000). Reduced activity of complex I has been found in PSP cybrids (Swerdlow et al. 2000), leading to significantly decreased O2 consumption and adenosine triphosphate (ATP) levels (Albers et al. 2001). In addition, epidemiological, clinical and neuropathological investigations have associated a PSP-like disease in Guadeloupe with regular intake of tropical plants containing potent lipophilic complex I inhibitors (Caparros-Lefebvre and Elbaz 1999; Caparros-Lefebvre et al. 2002). Thus, complex I dysfunction may be involved in both PSP and PD, but its distribution (generalized vs. localized in the SN) and relevance to the pathological events remains unclear.

The complex I inhibitors 1-methyl-4-phenylpyridine and 6-hydroxydopamine (Nicklas et al. 1985; Glinka and Youdim 1995), widely used as a means of experimentally inducing selective nigral degeneration as a model of PD, both accumulate selectively in dopaminergic neurones (Waddington 1980; Javitch et al. 1985). Thus, they cannot be used to test the hypothesis that the SN is particularly susceptible to a generalized inhibition of complex I. Rotenone, a specific inhibitor of mitochondrial complex I, is highly lipophilic, easily crosses biological membranes and does not selectively accumulate in nigral neurones. Nevertheless, chronic systemic administration of rotenone in rats [2–3 mg/kg/day intravenously (i.v.), 28–36 days] has been reported to result in a selective nigral degeneration (Betarbet et al. 2000). We used this model to characterize the differential vulnerability of various brain regions towards a generalized complex I inhibition and found non-dopaminergic lesions, particularly in the striatum, in all animals with nigral lesion, as seen in atypical parkinsonism but not in idiopathic PD.

Materials and methods

Animals and surgery

Animal work was carried out in accordance with the declaration of Helsinki and the guide for the care and use of laboratory animals adopted and promulgated by the National Institutes of Health. Male Lewis rats (CERJ, Le Genest St Isle, France), weighing 285–330 g at the time of operation, were housed in separate cages with free access to food and water under a 12-h/12-h light–dark cycle. Alzet osmotic mini pumps (2 ML4; IFFA CREDO, Arbresle, France) were filled with the test solutions and incubated in sterile 0.9% (wt/vol) NaCl for at least 4 h prior to implantation. Rats were anaesthetized by an intramuscular (i.m.) injection of 1.2 mL/kg ketamine : xylazine : H2O (2 : 1 : 1). Pumps were implanted under the skin on the back of the rats and connected by a subcutaneous cannula to the left inguinal region, where it was inserted into the femoral vein. Twenty-six rats were infused with rotenone (Sigma, Lyons, France) dissolved in equal volumes of dimethylsulphoxide (DMSO) and polyethyleneglycol 400 (PEG) to obtain a final delivery of 2.5 mg/kg/day calculated with respect to the average body weight (rotenone group). Six rats were infused with rotenone at a rate of 2.2 mg/kg/day for 28 days (low-dose rotenone group). Five rats were infused with DMSO : PEG (1 : 1) only (vehicle group).

Behavioural assessment

Spontaneous locomotor activity was monitored in cages that automatically recorded interruptions of two orthogonal light beams. The total number of beam interruptions was recorded for each rat for 1 h between 07.00 and 10.00 h. The average value recorded on 2 consecutive days was taken as a measure of the animal's activity prior to the operation and at 1, 2, 3 and 4 weeks thereafter.

Tissue preparation

After 28 days of infusion, the rats were deeply anaesthetized by an intraperitoneal (i.p.) injection of 30 mg/kg sodium pentobarbital and killed by transcardial perfusion. Brains were post-fixed in 4% (wt/vol) paraformaldehyde in 0.1 mol/L phosphate-buffered saline (PBS) for 24 h, cryoprotected in 10% (wt/vol) sucrose in 0.1 mol/L PBS for 24 h, frozen in isopentane at − 30°C for 2 min and stored at − 80°C. Correct intravenous placement of the cannula was verified after perfusion.


Brains were cut into 30-µm sections on a freezing microtome. Sections were collected in 15 regularly spaced series, and stored in 0.1 mol/L PBS containing 0.02% (wt/vol) sodium azide at 4°C. For immunohistochemistry, free-floating sections were incubated successively for 30 min with 0.1% H2O2 in 0.1 mol/L PBS to block endogenous peroxidase activity, for 30 min with 5% (vo/vol) normal goat serum (NGS) in 0.1 mol/L PBS to inhibit non-specific binding sites, and for 24 h at 4°C with the primary antibodies: anti-tyrosine hydroxylase (TH) rabbit polyclonal antibody to detect dopaminergic and noradrenergic neurones, Peel Freez (Abcyc, Paris, France), 1/500; antidopamine- and cAMP-regulated phosphoprotein (DARPP-32) mouse monoclonal antibody to detect a subpopulation of dopaminoceptive neurones, a gift from Dr HC Hemmings (Department of Anesthesiology, NY Presbyterian Hospital, Cornell University, NY, USA), 1/20000; anti-SERT (5-HT-transporter AB-1) rabbit polyclonal antibody to detect serotoninergic fibres, Oncogene (Cambridge, MA, USA), 1/3000; anticholine acetyltransferase (ChAT) rabbit monoclonal antibody to detect cholinergic neurones, Chemicon (Temecula, CA, USA), 1/3000; anti-cow glial fibrillary acidic protein (GFAP) rabbit polyclonal antibody to label astrocytes, Dako (Glostrup, Denmark), 1/1000; anti-RIP mouse monoclonal antibody to detect oligodendrocytes, a gift from Dr B Zalc (INSERM U 495, Hôpital de la Salpêtrière, Paris, France), 1/20; antimyelin oligodendrocyte-glycoprotein (MOG) mouse monoclonal antibody to detect myelin sheets, a gift from Dr Zalc, 1/10; anti-ED1 mouse monoclonal antibody to detect microglia, Serotec (Oxford, UK), 1/300; anti-alpha-synuclein rabbit polyclonal antibody, Chemicon, 1/2000; anti-ubiquitin rabbit polyclonal antibody, Dako, 1/750. All antibodies were diluted in 0.1 mol/L PBS with 2.5% NGS and 0.15% Triton X-100. Triton was omitted for detection of the membrane-bound DARPP-32. Sections were then incubated for 2 h at room temperature with the appropriate biotinylated secondary antibody (anti-rabbit or anti-mouse IgG; Vector Laboratories, Burlingame, CA, USA) in 0.1 mol/L PBS with 2.5% NGS and 0.15% Triton. The avidin–biotin method was used to amplify the signal (ABC Kit, Vector) and 3,3′-diaminobenzidine tetrachloride (DAB) was used to visualize bound antibodies. To exclude non-specific labeling, the primary antibodies were omitted.

Bodian silver impregnation, cresyl violet staining and NADPH diaphorase reaction

For Bodian silver impregnation, sections were mounted on double gelatin-coated glass slides, dehydrated in 100% ethanol for 1 h and placed vertically in 300 mL H2O on a layer of 15-g copper shavings and covered with 3-g proteinated silver for incubation at 56°C for 24 h in a water bath. After cooling to room temperature for 3 h, sections were rinsed rapidly in H2O, then incubated in a reducing solution (1% (wt/vol) hydroquinone, 5% (wt/vol) sodium sulphite in H2O) for 8 min. After three rinses in H2O, the slides were incubated in 1% (wt/vol) sodium tetrachloraurate III until they became gray, washed with H2O, incubated in 2% (v/v) oxalic acid until they became purple, washed again, fixed in 5% (wt/vol) sodium thiosulphate for 5 min and mounted in a permanent mounting medium.

For cresyl violet staining, sections were mounted on glass carrier slides, dehydrated for 1 h in 95% (v/v) ethanol, incubated for 1 min in 4% (v/v) cresyl violet and rinsed.

For NADPH diaphorase reaction, sections were incubated for 3 h at 37°C in 0.1 mol/L PBS containing 0.2% (wt/vol) NADPH (Sigma, St Louis, MO, USA), 0.09% (wt/vol) nitro blue tetrazolium (Sigma) and 12% (v/v) DMSO.

Image analysis

Images of the histological sections were acquired with a bright-field Nikon® Optiphot 2 microscope via a Sony® DXC-950P 3CCD colour video camera and analysed using Biocom® (Les Ulis, France) VisioScan version T4.18 image analysis software.

For semiquantitative evaluation of the distribution and quality of lesions on Bodian silver-impregnated slides, one investigator (GH) acquired images (objective × 50) from a pre-defined location in the following regions in all animals: frontal cortex (fCx), hippocampal pyramidal layer (Py) and the granule layer of the dentate gyrus (GrDG), caudate-putamen (striatum), external segment of the globus pallidus (GPe), entopeduncular nucleus (EP), subthalamic nucleus (STN), substantia nigra pars compacta (SNc), cerebellar (Cb) granule layer and Purkinje cells. Two other investigators (AP, KP) studied open-label Bodian silver-stained sections from the rotenone and vehicle groups in all of the brain regions and agreed on criteria for differentiating lesioned animals from controls. The images were then shown twice in random order to both investigators blind to the treatment of the rats, resulting in four analyses of each image. The investigators noted the presence or absence of the lesion criteria, scored from 0 (not lesioned) to +++ (severely lesioned), then classified the animal as lesioned or not. For statistical analysis, an image was classified as lesioned if the presence of a lesion was detected in at least two of the four analyses. All images were shown again in an open-label manner and the severity of the lesion in each anatomical structure was scored from 0 (not lesioned) to +++ (severely lesioned).

For quantification of cell loss, stereological cell counts were done on regularly spaced sections in the following regions: striatum [DARPP-32- and ChAT-immunoreactive (-ir), NADPH diaphorase-positive]; SNc, ventral tegmental area (VTA), locus ceruleus (TH-ir); pedunculopontine tegmental nucleus (PPN; ChAT-ir). Total cell numbers were estimated by integration along the rostrocaudal extent of the structures. In the striatum and SNc, the cross-sectional cell area of at least 15 neurones per animal homogeneously distributed across the structures was measured. For quantification of TH- and SERT-ir structures in the striatum, immunolabelled sections were exposed to bright-field illumination under controlled temperature, and the optic density was determined with the Biocom® VisioScan software package. GFAP-, RIP- and ED1-ir glial cells were counted on one representative section per animal in the striatum, 0.8–1.0 mm anterior to the rostral limit of the anterior commissure. At the same anatomical level, the TH-ir optic density and the number of DARPP-32-ir cells was compared between the centre of the striatum and its periphery, defined by a circle whose diameter covered half the striatum, to assess differential distribution of the lesion within the striatum. Ubiquitin and α-synuclein immunoreactivity was analysed qualitatively in the SN.

Primary rat mesencephalic cultures

Post-mitotic dopaminergic neurones from the ventral mesencephalon of embryonic day 15.5 Wistar rat embryos (CERJ) were dissociated mechanically, plated onto polyethylenimine-pre-coated 24-well culture plates, grown in N5 medium (Kawamoto and Barrett 1986) supplemented with 5 mm glucose, 5% horse serum and 2.5% fetal calf serum (FCS), as described previously (Michel and Agid 1996). After 3 days, FCS was reduced to 0.5% to limit astrocyte proliferation. Rotenone treatment was initiated at day 5 in N5 medium containing 1% horse serum and 0.1% FCS. Rotenone was diluted to 100 mm in DMSO and further pre-diluted in N5 medium prior to addition to the cultures to achieve a final concentration of 15 nm and 100 nm. After 2 h or 24 h, cultures were washed twice in PBS and fixed for 2 h in 2.5% glutaraldehyde in PBS (v/v).

Electron microscopy

For electron microscopy, brain tissue sections and cultured cells were post-fixed in 1% osmium tetroxide for 30 min and embedded in epon, as previously described (Muriel et al. 1999). Ultrathin sections were cut using an ultramicrotome (Ultracut E, Leica, France), contrasted with uranyl acetate and lead citrate and observed with an electron microscope JEOL 1200 EX at 80 kV.

Statistical analysis

A commercially available software package (Statistica 5.0, StatSoft, Tulsa, OK, USA) was used for statistical analysis. Results were expressed as the mean ± SEM. A p-value < 0.05 was assumed to be statistically significant. The experimental groups were compared with a two-sided t-test for independent samples, or two-way analysis of variance (anova; independent variables: time and treatment for activity scores; cell population and treatment for cell numbers and optical densities) followed by a post-hoc least significant difference (LSD) test, as appropriate. For the semiquantitative histological evaluation, the sensitivity and specificity of the detection of rotenone-induced lesions was calculated. The Mann–Whitney U-test was used to compare lesion severity in the rotenone and vehicle groups. Spearman's R and Pearson's χ2 were calculated to analyse interinvestigator and re-test stability for detection of rotenone-induced histological alterations.


Animals included in the analysis

Nine of the 26 rats infused with 2.5 mg/kg/day rotenone and two of the six rats infused with 2.2 mg/kg/day died spontaneously before the end of the 28-day infusion period. In eight rats infused with 2.5 mg/kg/day, the catheter had come out of the vein due to local tissue degeneration, as observed after perfusion. These rats were excluded from analysis. These complications were not observed in vehicle-infused rats. Finally, five vehicle-infused rats, nine 2.5 mg/kg/day rats and four 2.2 mg/kg/day rats met the defined experimental conditions and were included for analysis. The low-dose rotenone group is only referred to when explicitly mentioned in the following results.

In vivo analysis of the animals

Activity scores prior to the operation did not differ significantly between the vehicle group (227 ± 49) and the rotenone group (292 ± 38). The scores obtained at the end of the experimental period were not significantly different from the pre-operative values in the vehicle group (298 ± 56). The rotenone group, however, showed a significant reduction in spontaneous locomotor activity at the end of the treatment compared to pre-operative values (107 ± 31, p < 0.001, t-test for dependent samples). There was a significant effect of treatment (p < 0.001) on locomotor activity (two-way anova), but not of time (p = 0.48) or treatment–time interaction (p = 0.32; Fig. 1). No rigidity or tremor were observed. Postural instability and dystonic limb posturing were observed in some of the animals that died prematurely, 1–2 days prior to death.

Figure 1.

Spontaneous locomotor activity in vehicle-infused (□) and rotenone-infused (▴) rats prior to operation (pre-OP) and 1, 2, 3 and 4 weeks thereafter is presented as a percentage of the respective pre-operative group mean. There was a significant effect of treatment (two-way anova: p < 0.001), but not of time or time–treatment interaction.

Semiquantitative analysis of lesion distribution

Two investigators compared Bodian silver-impregnated sections from rotenone- and vehicle-infused rats. They identified the criteria ‘altered nuclear structure’, ‘increased nuclear size’, ‘reduced number of nuclei’ and ‘dystrophic fibres’ to differentiate rotenone- from vehicle-infused rats. Typical morphological changes are illustrated in Fig. 2. These alterations were scored from 0 (not present) to +++ (severe). The scores for these alterations were significantly higher (Mann–Whitney U-test) in rotenone-infused animals in many of the brain regions assessed (Table 1). Altered nuclear structure was consistently observed in rotenone-infused animals in all the brain regions analysed. Increased nuclear size was seen in all these regions except the cerebellar granule cell layer and the cerebral cortex. A reduced number of nuclei was observed in the basal ganglia nuclei examined and in the dentate gyrus. Dystrophic fibres were seen in the STN, the GPe and the cerebellum. Table 1 gives the sensitivity and specificity of the blind investigators' ability to identify rotenone-infused animals from a single image per animal. The interinvestigator correlation (Spearman R = 0.86, p < 0.001; χ2 102.0) and re-test stability (Spearman R = 0.91, p < 0.001; χ2 294.3) for the blind detection of rotenone-induced alterations were excellent. Even in the cerebral cortex, where histological alterations were mild, high sensitivity (100.0%) and specificity (83.3%) were obtained, demonstrating the consistent presence of characteristic histological alterations in rotenone-infused rats. Histological alterations – graded by agreement of both observers – were severe in the basal ganglia nuclei, moderate in the hippocampus and cerebellum and minor in the cerebral cortex (Table 1).

Figure 2.

Neurodegeneration visualized by cresyl violet and Bodian silver impregnation. Histological alterations in rotenone-infused rats (a′–n′) in comparison to vehicle-infused animals (a–n) as visualized by cresyl violet (first and second column) and Bodian silver impregnation (third and fourth column). SNc (a–b′), striatum (c–d′), GPe (e–f′), GrDG (g–h′), Cb (I–j′), fCx (k–l′). Arrowheads in (f′) and (j′) point to dystrophic fibres. Cytoplasmic vacuolization (arrowheads) of a GPe neurone (m,m′) and a dystrophic fibre in the Cb (n,n′; detail from j,j′) at higher magnification. Scale bars: 10 µm (a–l′), 10 µm (m,m′), 20 µm (n,n′).

Table 1.  Lesions visualized by Bodian silver impregnation
RegionLesion severityLesion detectionLesion quality
Sensitivity (%)Specificity (%)Altered nuclear structureIncreased nuclear sizeReduced number of nucleiDystrophic fibres
  1. Lesion severity: + (mildly lesioned) to +++ (severely lesioned). Lesion quality: p-values obtained by Mann–Whitney U-test, rotenone versus vehicle. n.s. not significant. Not all criteria were pertinent to rotenone-induced alterations in all regions and were therefore not assessed blindly (n.a.). SNc, substantia nigra pars compacta; GPe, external segment of the globus pallidus; STN, subthalamic nucleus; EP, entopeduncular nucleus; Py, hippocampal pyramidal layer; GrDG, granule layer of the dentate gyrus; Cb, cerebellum; fCx, frontal cortex.

Cb, Purkinje cells++85.766.70.0010.05n.s.0.001
Cb, granule layer++85.766.70.001n.a.n.a.0.01

Quantitative analysis of neuronal lesions

The number of TH-ir cells in the SNc was 28.0% lower in rotenone-infused rats than in vehicle-infused rats (p < 0.01; Fig. 6). The dopaminergic cell population in the VTA in the immediate vicinity of the SNc, however, was unchanged in rotenone-treated rats. The mean cross-sectional area of TH-ir cell bodies in the SNc was increased in rotenone-treated rats (369.5 ± 12.6 µm2; vehicle: 288.5 ± 2.5 µm2; t-test: p < 0.001). Morphologically, lesioned TH-ir SNc neurones were characterized by nuclear enlargement and cytoplasmic vacuolization (Fig. 3h). Degeneration of dopaminergic fibres resulted in a 54.9% decrease in the optical density of TH-ir structures in the striatum of rotenone-infused rats (p < 0.001; Fig. 3a–g and 6). All rotenone-infused rats that met the inclusion criteria (28-day survival, intact infusion system) had a nigral TH-ir cell number and striatal TH-ir optical density inferior to the lowest value observed in vehicle-infused animals, demonstrating a lesion of the nigrostriatal dopaminergic system.

Figure 6.

Quantitative analysis of immunohistochemistry. Immunohistochemical detection of TH, ChAT, DARPP-32 and SERT followed by stereological cell counts or determination of the optical density (OD) were performed in the VTA, striatum, PPN, locus ceruleus and SNc of rotenone-infused animals, and were expressed as percentage of the group mean for vehicle-infused rats. Two-way anova demonstrated significant effects of treatment (p < 0.001), neuronal population (p < 0.01), and treatment–population interaction (p < 0.01). Post-hoc LSD test: *p < 0.05, **p < 0.01, ***p < 0.001.

Figure 3.

Degeneration of nigrostriatal dopaminergic neurones visualized by TH-immunolabeling. Intact striatal dopaminergic innervation in a vehicle-infused animal (a), central loss in a rat surviving only 16 days of rotenone infusion (b) and global loss after 28 days of infusion (c). Dystrophic fibres on the periphery of the denervated centre of the striatum in (b) at higher magnification (d). Dystrophic fibre at high magnification (f). Absence of dystrophic fibres in vehicle-infused rats (e,g). Vacuolization in a nigral dopaminergic neurone (h), which is absent in vehicle-infused controls (i). Scale bars: 1 mm (a–c), 100 µm (d,e), 20 µm (f,g), 10 µm (h,i).

In the striatum, a 34.7% reduction in SERT-ir optical density was observed in rotenone-infused rats (p < 0.001; Figs 4a and 6). The number of striatal DARPP-32-ir projection neurones was reduced by 25.4% in rotenone-infused animals (p < 0.01; Figs 4c and 6) and their mean cross-sectional cell area increased (121.4 ± 5.2 µm2, vehicle: 83.4 ± 2.2 µm2, t-test: p < 0.001). Morphologically, there was pronounced nuclear swelling, but no cytoplasmic vacuolization (Fig. 4c). Striatal cholinergic interneurones were reduced in number by 22.1% (p < 0.05; Fig. 6) and cytoplasmic vacuolization was obvious (Fig. 4e), but their cross-sectional cell area remained unchanged (244.5 ± 10.2 µm2, vehicle: 251.5 ± 7.2 µm2). The number of NADPH diaphorase-positive neurones was not significantly reduced in the striatum (− 12.3%; Figs 4g and 6).

Figure 4.

Striatal neurodegeneration. Striatal damage is evidenced by a rarefaction of SERT-ir fibres (rotenone: a, control: b), swelling and rarefaction of DARPP-32-ir projection neurones (rotenone: c, control: d), and vacuolization and rarefaction of ChAT-ir interneurones (rotenone: e, control: f). NADPH diaphorase-positive neurones were relatively spared (rotenone: g, control: h). Scale bar: 10 µm.

In the animals that died prior to the end of the 28-day infusion period, the loss of TH-ir fibres was, if present, confined to the centre of the striatum (Fig. 3b). In animals that survived 28 days of infusion, however, the decrease of TH-ir optical density was no more pronounced in the centre (− 55.4 ± 5.6%) than on the periphery (− 54.9 ± 5.6%; Fig. 3c). Similarly, DARPP-32-ir cell loss was no more severe in the centre of the striatum (− 30.3 ± 7.9%) than on its periphery: (− 32.7 ± 6.9%), but was slightly more pronounced in the dorsal half of the striatum (− 32.4 ± 5.9%) than in the ventral half (− 26.9 ± 5.2%), although the difference was not statistically significant.

A significant cell loss in rotenone-infused animals was also observed in the noradrenalinergic neurones of the locus ceruleus (26.4%, p < 0.01; Figs 5a and 6) and the cholinergic neurones of the PPN (29.3%, p < 0.01; Figs 5e and 6). Cell loss was also evident in the cholinergic nucleus basalis of Meynert (not quantified; Fig. 5c).

Figure 5.

Other brain stem nuclei. Loss of TH-ir neurones in the locus ceruleus (rotenone: a, control: b), and of ChAT-ir neurones in the nucleus basalis of Meynert (rotenone: c, control: d) and PPN (rotenone: e, control: f). Scale bar: 100 µm.

To assess whether lower rotenone doses would induce a selective dopaminergic lesion, we infused four rats for 28 days with 2.2 mg/kg/day and found a 20.6 ± 5.1% loss of ChAT-ir striatal interneurones (p < 0.01) and a 22.6 ± 8.4% loss of DARPP-32-ir striatal projection neurones (p < 0.05), while there was no significant loss of striatal TH-ir fibres (optical density: − 15.7 ± 20.4%, p = 0.22).

Glial cells

Glial cells were quantified in the striatum 0.8–1.0 mm anterior to the anterior commissure, because both ChAT- and DARPP-32-immunoreactivity showed the most pronounced degree of neuronal lesion at this anatomical level within the striatum. In rotenone-infused animals there was no change in the number of RIP-ir oligodendrocytes (102.6 ± 6.0% of control) and a slight but not significant increase in GFAP-ir astrocytes (110.0 ± 7.9% of control) and ED-1-ir microglial cells (115.1 ± 7.1% of control). Morphologically, there were no apparent alterations of MOG-ir myelin sheets (Fig. 7a), RIP-ir oligodendrocytes (Fig. 7c), GFAP-ir astrocytes (Fig. 7e) and ED1-ir microglial cells (Fig. 7g).

Figure 7.

Striatal glial cells. MOG-ir myelin sheets (rotenone: a, control: b), RIP-ir oligodendrocytes (rotenone: c, control: d), GFAP-ir astrocytes (rotenone: e, control: f), and ED1-ir microglial cells (rotenone: g, control: h). Scale bars 20 µm (a,b), 10 µm (c–h).

Nigral cytoplasmic inclusions

In the SN of rotenone-infused animals, a small proportion of cells displayed focal cytoplasmic immunoreactivity for ubiquitin (Fig. 8a) and α-synuclein (Fig. 8c) suggesting the presence of protein aggregates. These aggregates were not observed in control animals.

Figure 8.

Nigral cytoplasmic inclusions. Ubiquitin-ir (rotenone: a, control: b) and α-synuclein-ir (rotenone: c, control: d) dense foci are found in a cytoplasmic localization in some cells in the substantia nigra of rotenone-infused animals (arrowheads), but not in control animals. Nuclei in (c) and (d) are counterstained with cresyl violet. Scale bar 10 µm.

Ultrastructural alterations induced by rotenone

Electron microscopic analysis of neurones in primary rat mesencephalic cultures demonstrated that exposure to a low concentration of rotenone (15 nm) for 24 h resulted in a moderate swelling of the endoplasmic reticulum, while mitochondria appeared morphologically intact (Fig. 9b). Exposure to a high concentration (100 nm) for 2 h resulted in swelling of both the endoplasmic reticulum and of mitochondria (Fig. 9c). Analysis of neurones in the SNc of rotenone-infused rats (2.5 mg/kg/day) demonstrated swelling of the endoplasmic reticulum, while mitochondria appeared morphologically intact (Fig. 9e).

Figure 9.

Ultrastructural alterations induced by rotenone treatment. In comparison to control conditions (a), neurones in primary rat mesencephalic cultures exposed to 15 nm rotenone for 24 h (b) showed morphologically intact mitochondria (arrowheads) but a moderate swelling of the endoplasmic reticulum (arrows). Exposure of cultures to 100 nm rotenone for 2 h (c) resulted in swelling of both the endoplasmic reticulum (arrows) and of mitochondria (arrowheads). In comparison to control rats (d), neurones in the SNc of rats infused with rotenone (2.5 mg/kg/day; e) showed swollen endoplasmic reticulum (arrows), but morphologically intact mitochondria (arrowheads). Scale bar 500 nm.


This study addressed the differential vulnerability of various regions in the rat brain to a chronic generalized inhibition of mitochondrial complex I by the pesticide rotenone. Based on a semiquantitative analysis of silver-impregnated sections and stereological cell counts in immunohistochemically stained sections, we demonstrated a pronounced degeneration of neurones in brain stem and basal ganglia nuclei, whereas alterations in the hippocampus, cerebellum and cerebral cortex were mild to moderate. These data provide the first experimental evidence that systemic exposure to a naturally occurring, lipophilic complex I inhibitor that is not preferentially accumulated in nigral neurones can induce a pattern of neurodegeneration resembling an atypical parkinsonian syndrome rather than PD.

Intravenous administration of rotenone (10–18 mg/kg/day, 7–9 days, Sprague–Dawley rats) was initially reported to result in striatal and pallidal lesions sparing the SN (Ferrante et al. 1997). Recently, Betarbet et al. (2000) reported that lower doses administered for extended periods of time (2–3 mg/kg/day, 28–36 days, Lewis rats) induced selective degeneration of nigral dopaminergic neurones. In the present study, we infused male Lewis rats at a dose of 2.5 mg/kg/day i.v. for 28 days. At this dose we found a lethality of 35%, which is in close agreement with the known LD50 of 5 mg/kg i.p. in rats (Windholz et al. 1976). In agreement with the findings of Betarbet et al. (2000), our experimental animals became hypokinetic, lost nigral neurones and showed intracellular focal immunoreactivity for ubiquitin and α-synuclein. Further studies are warranted to assess whether the formation of these inclusions shares a common pathophysiological basis with formation of Lewy bodies in PD. In addition, we found neuronal cytoplasmic vacuolization in rotenone-infused rats. Electron microscopy suggested that swelling of the endoplasmic reticulum may be the origin of these vacuoles, while mitochondria appeared morphologically intact. Consistently, electron microscopy of neurones in primary mesencephalic cultures confirmed that a low concentration of rotenone (15 nm) can induce a selective swelling of the endoplasmic reticulum, while a rather high concentration (100 nm) is necessary to induce swelling of both the endoplasmatic reticulum and of mitochondria. The pronounced loss of spontaneous locomotor activity in our experimental animals associated with a loss of only 54.9% of striatal dopaminergic fibres suggests that factors other than dopaminergic denervation may be implicated in the behavioural changes. Consistently with this, we also found degeneration of non-dopaminergic systems, which together may be responsible for the reduced locomotor activity. In the striatum, dopaminoceptive DARPP-32-ir projection neurones and ChAT-ir interneurones degenerated significantly, while NADPH diaphorase-positive neurones were preserved. The latter neurones have already been shown to be more resistant to chronic inhibition of mitochondrial complex II by 3-nitroproprionic acid than striatal projection neurones (Beal et al. 1993). Glial cells were also resistant to rotenone treatment as the number of striatal oligodendrocytes was unchanged and that of astrocytes and microglial cells slightly, but not significantly, increased after chronic rotenone exposure. The degeneration of multiple neuronal systems observed in the present study, however, is in marked contrast to the report by Betarbet et al. (2000).

Every effort was made in our study to obtain controlled experimental conditions. Pumps were implanted at two separate time points by two independent investigators in independent sets of animals with newly prepared test solutions. Rats from the rotenone and the vehicle group were implanted alternately. At the time of perfusion, the volume remaining in the pumps was quantified and rats with dislocated catheters were excluded in order to ensure that delivery of the test solution was correctly intravenous. For all staining procedures, rotenone and control tissues were processed in parallel in the same working solutions. The histological evaluation was performed in a blind manner. Thus, it is unlikely that the differences found in our study compared to that of Betarbet et al. (2000) were due to a bias in the experimentation and preparation of our histological material. We also tested whether the lack of nigral selectivity was due to the concentration of rotenone. Using a lower dose (2.2 mg/kg/day), however, we found a significant loss of intrinsic striatal neurones without a significant loss of dopaminergic fibres.

Several factors might explain the difference between the results obtained in this study and those of Betarbet et al. (2000). A different susceptibility of the strain of Lewis rats used in our study cannot be excluded. Furthermore, there were some methodological differences. Firstly, Betarbet et al. (2000) observed silver deposits in the striatum which they attributed to degenerating dopaminergic fibres, but they provided no evidence that nigrostriatal fibres rather than intrinsic striatal cells were labeled. It is unlikely, however, that in striatal regions completely lacking phenotypic markers of dopaminergic innervation (TH, DAT, VMAT2) there were enough dopaminergic fibres remaining to cause the strong silver-labelling seen in their image. It was therefore most likely related to intrinsic striatal neurones as in our study. Secondly, Betarbet et al. (2000) suggested that, in striatal regions completely lacking TH-immunoreactivity, GAD-immunofluorescence was not affected. This conclusion is not substantiated, however, by the images they show (Figs 4c and d in Betarbet et al. 2000), in which neuronal perikarya and neurites cannot be clearly distinguished. Thirdly, Betarbet et al. (2000) used acetylcholinesterase histochemistry to demonstrate the integrity of cholinergic striatal interneurones. Acetylcholinesterase is not specific to these neurones, however, as it is also expressed by cholinoceptive neurones and glial cells. With the specific marker ChAT, we found a significant loss of cholinergic interneurones. Finally, Betarbet et al. (2000) did not quantify their histological examinations, so that preservation of non-dopaminergic neurones cannot be deduced. The discrepancy between the results of the two studies may therefore be more apparent than real.

Does the lesion pattern observed in rotenone-infused animals bear similarities to a parkinsonian syndrome in humans? In most parkinsonian syndromes in humans, multiple neuronal systems degenerate. In PD, dopaminergic, noradrenergic, serotoninergic and cholinergic systems degenerate (Jellinger 2001; Hirsch et al. 2002). The degree and distribution of non-dopaminergic lesions varies considerably, however, and may account for the diversity of non-motor symptoms (Jellinger 1999). The loss of striatal neurones in rotenone-infused rats is not observed in PD, however. Nevertheless, these neurones do degenerate in atypical parkinsonian syndromes (Adams et al. 1961; Steele et al. 1964; Fearnley and Lees 1990; Daniel et al. 1995). In these disorders, which include PSP and multiple system atrophy, lesions are predominant in the basal ganglia and brain stem nuclei, moderate in the cerebellum and hippocampus, and mild, though clearly detectable, in the cerebral cortex (Adams et al. 1961; Steele et al. 1964; Fearnley and Lees 1990; Daniel et al. 1995). This is reminiscent of the observations in rotenone-infused rats. Neuronal vacuolization is another characteristic common to our model and PSP (Steele et al. 1964). Thus, rotenone-infused rats reproduce features of atypical human parkinsonism rather than of idiopathic PD.

How does the pattern of complex I inhibition in rotenone-infused rats relate to human parkinsonian syndromes? In PD brains, reduced complex I activity has been found in the SN, but not in nuclei unaffected by neurodegeneration (Schapira et al. 1989, 1990; Mann et al. 1992). This situation is mimicked experimentally by the complex I inhibitors 6-hydroxydopamine and 1-methyl-4-phenylpyridinium, which are taken up by dopaminergic neurones (Waddington 1980; Javitch et al. 1985) and consistently result in selective nigral lesions. Studies demonstrating complex I inhibition in peripheral tissue of PD patients (Bindoff et al. 1989; Parker et al. 1989) have suggested that the SN may be particularly vulnerable to generalized mitochondrial dysfunction, but this was subsequently contradicted (Mann et al. 1992; DiDonato et al. 1993). Because rotenone infusion induces a homogenous complex I inhibition throughout the brain (Betarbet et al. 2000), it may be used to test this hypothesis on an experimental basis. Our data show that nigral TH-ir cells were not selectively lost, nor even significantly more affected than other brain stem or basal ganglia nuclei. Therefore, the present data do not support the view that a generalized complex I inhibition is responsible for the rather selective nigral degeneration in PD. In PSP, there is some evidence to suggest a more widespread impairment of mitochondrial energy metabolism, including complex I inhibition, within the brain as well as in peripheral tissues (Albers and Beal 2000). Although other factors, such as a genetic predisposition (Albers and Augood 2001), appear to be crucially implicated in the aetiology of PSP, our study demonstrates that a chronic generalized complex I inhibition in rats is sufficient to produce the characteristic distribution of lesions.

The origins of the mitochondrial dysfunctions in PD and PSP remain obscure. However, several epidemiological studies suggest that exposure to pesticides, many of which are mitochondrial toxins, is a risk factor for PD (Lockwood 2000). Unfortunately, most epidemiological studies linking PD to pesticide exposure are complicated by the uncertainty of the clinical diagnosis (Lockwood 2000), particularly as atypical parkinsonism has also been reported in humans after exposure to pesticides or complex I inhibitors. First, development of a parkinsonian syndrome refractory to levodopa was reported after organophosphate exposure (Bhatt et al. 1999). Second, a mutation in the mitochondrial ND4 gene encoding a subunit of complex I was identified in a family with maternally inherited multisystem degeneration with prominent parkinsonism (Simon et al. 1999). Third, in Guadeloupe, an unusually high frequency of a PSP-like tauopathy was associated in a case–control study with regular consumption of Annona muricata plants (Caparros-Lefebvre and Elbaz 1999; Caparros-Lefebvre et al. 2002), which contain a variety of potent complex I inhibitors (Alali et al. 1999). In fact, we observed a lesion pattern closely resembling rotenone-induced lesions when we infused rats chronically with annonacine (P. Champy et al. in preparation), a lipophilic complex I inhibitor (Alali et al. 1999) that is the major acetogenin in Annona muricata. Thus, our data provide the first experimental evidence that exposure to a lipophilic complex I inhibitor can indeed trigger a widespread degeneration in basal ganglia and brain stem nuclei that bears some similarity to human atypical parkinsonian syndromes.

Interestingly, chronic systemic inhibition of mitochondrial complex II in rats by subcutaneous delivery of 3-nitropropionic acid (3-NP) via osmotic minipumps has been shown to result in a 30–35% neuronal loss in the striatum with relative sparing of NADPH diaphorase-positive interneurones and mild astrogliosis (Beal et al. 1993; Brouillet et al. 1993). This pattern of degeneration resembles that seen in our study after chronic rotenone infusion. However, in contrast to the lesion pattern in rotenone-infused animals, the nigrostriatal dopaminergic inputs have been reported to be relatively intact after 3-NP intoxication (Beal et al. 1993; Brouillet et al. 1993; Wüllner et al. 1994), although extrastriatal lesions have been described in the pallidum, the hippocampus and the substantia nigra pars reticulata when more acute intoxication protocols were used (Hamilton and Gould 1987; Beal et al. 1993; Guyot et al. 1997). From a behavioural standpoint, 3-NP treatment of rats also resulted in a persistent bradykinesia, which may be preceded by a transient phase of hyperkinesia under certain conditions (Borlongan et al. 1997; Guyot et al. 1997). On this basis, 3-NP intoxication has been proposed as a model for Huntington's disease (Beal et al. 1993; Brouillet et al. 1993). In contrast, rotenone infusion, which also targets the mitochondria, does not reproduce the pattern of degeneration seen in PD or Huntington's disease but rather that of a multisystem degeneration with a given selectivity for subpopulations of neurones. Identifying the factors involved in such a differential vulnerability of the neurones to mitochondrial toxins (Albin 2000) may help to better understand the pathological mechanisms of neurodegenerative diseases.

In conclusion, we did not find that the SN is more susceptible to generalized inhibition of complex I than other basal ganglia or brain stem nuclei. On the contrary, our data support the view that a generalized mitochondrial dysfunction within the brain results in a hypokinetic syndrome with widespread degeneration in basal ganglia and brain stem nuclei.


This study was supported by the Institut National de la Santé et de la Recherche Médicale (INSERM), Paris, and the National Parkinson's Disease Foundation, Miami. GH was funded by the Deutsche Forschungsgemeinschaft (DFG) and the Kompetenznetz Parkinson Syndrome funded by the German Bundesministerium für Bildung und Forschung (BMBF).