Special Reviews

Molecular Pathways of Neurodegeneration in Parkinson's Disease

See allHide authors and affiliations

Science  31 Oct 2003:
Vol. 302, Issue 5646, pp. 819-822
DOI: 10.1126/science.1087753


Parkinson's disease (PD) is a complex disorder with many different causes, yet they may intersect in common pathways, raising the possibility that neuroprotective agents may have broad applicability in the treatment of PD. Current evidence suggests that mitochondrial complex I inhibition may be the central cause of sporadic PD and that derangements in complex I cause α-synuclein aggregation, which contributes to the demise of dopamine neurons. Accumulation and aggregation of α-synuclein may further contribute to the death of dopamine neurons through impairments in protein handling and detoxification. Dysfunction of parkin (a ubiquitin E3 ligase) and DJ-1 could contribute to these deficits. Strategies aimed at restoring complex I activity, reducing oxidative stress and α-synuclein aggregation, and enhancing protein degradation may hold particular promise as powerful neuroprotective agents in the treatment of PD.

PD is the most common neurodegenerative movement disorder. It affects more than 0.1% of the population older than 40 years of age (1). Clinically, most patients present with a motoric disorder and suffer from slowness of movement, rest tremor, rigidity, and disturbances in balance. A number of patients also suffer from anxiety, depression, autonomic disturbances, and dementia. Although there are effective symptomatic therapies, there are no proven neuroprotective or neurorestorative therapies (2).

Loss of dopamine (DA) neurons in the substantia nigra pars compacta (SNC) leads to the major clinical symptoms of PD, but there is widespread neuropathology and the SNC only becomes involved toward the middle stages of the disease (3). Lewy bodies (LBs) and dystrophic neurites (Lewy neurites) are a pathologic hallmark of PD and classically are round eosinophilic inclusions composed of a halo of radiating fibrils and a less defined core (4). LBs are thought to be a pathognomonic feature of PD, but recent studies suggest that some forms of PD do not have LBs (5). Ultrastructurally, LBs are composed of 10- to 14-nm amyloid-like fibrils (4) and α-synuclein, which can polymerize into ∼10-nm fibrils in vitro and is the primary structural component of the LB (6).

The etiology of PD is still not fully understood, but genetic analyses, epidemiologic studies, neuropathologic investigations, and new experimental models of PD are providing important new insights into the pathogenesis of PD (1, 5, 79). At least 10 distinct loci are responsible for rare Mendelian forms of PD (Table 1) (5). Genes linked to PD have opened up new and exciting areas of research, brought spectacular advances, and created a renaissance in PD research. Despite the genetic advances, PD is primarily a sporadic disorder with no known cause (1). As such, this review focuses on the more common sporadic form of PD and highlights discoveries in the genetic causes of PD that have led to advances in our understanding of the pathogenesis of sporadic PD.

Table 1.

Loci and genes linked to familial PD or implicated as genetic causes for PD. NA, not assigned.

GeneMode of inheritanceLocusChromosomal location
α-synuclein Autosomal dominant PARK1 (View inline) 4q21-q23
Parkin Autosomal recessive PARK2 (View inline) 6q25.2-27
Unknown Autosomal dominant PARK3 (View inline) 2p13
α-synuclein Autosomal dominant PARK4 (View inline, View inline) 4q
UchL1 Autosomal dominant PARK5 (View inline) 4p14
Unknown Autosomal recessive PARK6 (View inline) 1p35-p36
DJ-1 Autosomal recessive PARK7 (View inline) 1p36
Unknown Autosomal dominant PARK8 (View inline) 12p11q13.1
Unknown Autosomal recessive PARK9 (View inline) 1p36
Unknown Late-onset susceptibility gene PARK10 (View inline) 1p32
NR4A2 Susceptibility gene NA (View inline) 2q22-23
Synphilin-1 Susceptibility gene NA (View inline) 5q23.1-23.3
Tau Susceptibility gene NA (View inline) 17q21

Deficits in Mitochondrial Complex I

Ultimately, the advances in the treatment of PD that will have the greatest impact will require understanding of the pathogenesis of sporadic PD. Studies in human postmortem material indicate that reactive oxygen species are important in the pathogenesis of sporadic PD (10, 11). There are also consistent findings of decrements in mitochondrial complex I (complex I) (1214). Impaired complex I activity leads to free radical stress and makes neurons vulnerable to glutamate excitotoxicity (12). Cytoplasmic hybrid cell lines (cybrids) from patients with sporadic PD indicate that deficits in complex I can be stably transmitted (15). Maternal descendants in families with maternal patterns of inheritance of PD have lower complex I activity, increased reactive oxygen species production, and increased radical scavenging enzyme activities, consistent with the notion that defects in complex I play a role in PD (16). Furthermore, a single-nucleotide polymorphism leading to a nonconservative amino acid change from threonine to alanine within the NADH (reduced form of nicotinamide adenine dinucleotide) dehydrogenase 3 of complex I leads to a significantly reduced risk of developing PD in Caucasians; this finding provides genetic evidence that alterations in complex I activity play a role in the pathogenesis of sporadic PD (17).

Several epidemiologic studies suggest that pesticides and other environmental toxins that inhibit complex I are involved in the pathogenesis of PD (12). 1-Methyl-4-phenyl-1,2,3,6-tetrahydropyridine (MPTP) inhibits complex I and replicates most features of sporadic PD (18). MPTP selectivity for DA neurons is due to the active metabolite, 1-methyl-4-phenyl pyridium (MPP+), which is then concentrated in DA neurons via the DA transporter, where it kills DA neurons through inhibition of complex I (18). In aged nonhuman primates, MPTP leads to intracellular proteinaceous inclusions that are filamentous (19) and stain for α-synuclein (20). The common herbicide 1,1′-dimethyl-4,4′-5 bipyridinium (paraquat) coupled with the administration of the fungicide manganese ethylenepistithiocarbamate (maneb) leads to selective degeneration of DA neurons (21). Paraquat is a complex I inhibitor with structural similarity to MPP+, and it leads to up-regulation and aggregation of α-synuclein (22). Rotenone, another complex I inhibitor that is a widely used insecticide and fish poison, produces Parkinsonism in rats when administered intravenously or subcutaneously. In contrast to MPTP and paraquat, rotenone is not concentrated in DA neurons, yet it induces selective DA cell death, which suggests that DA neurons are uniquely sensitive to complex I impairments (23). Furthermore, the slow and chronic nature of rotenone toxicity leads to LB-like intraneuronal filamentous protein inclusions containing α-synuclein and ubiquitin that are remarkably similar to authentic LBs (2325).

The fact that three complex I inhibitors cause dopaminergic cell death and induce the formation of LB-like filamentous inclusions containing α-synuclein as well as the systemic defect in complex I (either through genetic or acquired alterations in PD) suggests that impairments in complex I may be central to the pathogenesis of DA neuronal demise in sporadic PD. Most, if not all, known neuronal cell death pathways play a role in complex I–mediated cell death, including excitotoxicity, reactive oxygen species–induced injury, caspase-dependent and caspase-independent apoptosis, necrosis, and inflammation-induced injury (2, 18, 26, 27).


A role for α-synuclein in PD was fueled by the discovery of missense mutations in the α-synuclein gene as a cause of autosomal dominantly inherited PD (28). An Ala53 → Thr (A53T) mutation resulting from a G to A transition at position 209 was identified in a large Italian-American Greek kindred (28), and an Ala30 → Pro (A30P) mutation resulting from a G to C transition at position 88 was identified in a small German kindred (29). A genomic triplication of wild-type α-synuclein leading to overexpression of wild-type α-synuclein is the disease-causing mutation in familial PD in the Iowan kindred and the PARK4 locus (30). Genetic variability in the α-synuclein gene is also a risk factor for the development of PD (31). Even though mutations in α-synuclein are a very rare cause of hereditary PD, its apparent role as the major structural feature of the LB has placed α-synuclein at center stage in the pathophysiology of PD (6).

α-Synuclein is an intrinsically unfolded or natively unfolded protein that assembles into LB-like filaments, whereas its related homologs β- and γ-synuclein do not (32). Residues 71 to 81 of α-synuclein are essential for filament assembly (33). α-Synuclein appears to be associated with membrane compartments in cultured cells and brain tissue through interactions with acidic head groups of phospholipids (34). Membrane-bound α-synuclein may play an important role in fibril formation (34, 35).

A number of cellular and transgenic models of α-synuclein–induced neurodegeneration have been reported (36). None of the mammalian transgenic models fully recapitulate PD, but they have been useful for studying synucleinopathy-induced neurodegeneration. Abnormal accumulation of detergent-insoluble α-synuclein (37) and abnormal proteolytic processing of α-synuclein appear to be associated with neurodegeneration (38). The A53T α-synuclein mutant causes significantly greater in vivo toxicity as compared with other α-synuclein variants (38). α-Synuclein induces fibrillization of the microtubule-associated protein, tau, and coincubation of tau and α-synuclein synergistically promotes fibrillization of both proteins (39). Amyloid-like α-synuclein and tau filamentous inclusions co-occur in human neurodegenerative disease and in α-synuclein transgenic mice, which suggests that the interaction between α-synuclein and tau promotes their fibrillization and drives the formation of pathologic inclusions in human neurodegenerative diseases (39). Interestingly, polymorphisms in the tau gene are significantly associated with an increased risk of developing PD (40).

In most transgenic models, the development of fibrillar α-synuclein–containing inclusions is associated with neurodegeneration. However, in one line of transgenic mice, motoric impairment and loss of dopaminergic terminals are observed in the presence of nonfibrillar α-synuclein inclusions (41), raising the possibility that the α-synuclein protofibril, an intermediate in the fibrillization process, may be pathogenic (42). Support for the role of protofibrils in PD is provided by the observation that crossing α-synuclein transgenic mice that exhibit nonfibrillar α-synuclein inclusions with β-synuclein transgenic mice results in amelioration of inclusions, dopaminergic terminal loss, and behavioral abnormalities (43). Interestingly, protofibril formation is promoted by catecholamines, particularly DA (44). The failure of transgenic mice that overexpress the protofibrillogenic A30P α-synuclein to exhibit neurodegeneration provides strong in vivo evidence that protofibrillar α-synuclein is not the primary toxic moiety of α-synuclein (38). Indeed, only when the A30P mutant forms inclusions and fibrils do transgenic flies and mice exhibit neurodegeneration (4547). One finding consistent with the notion that the fibrillar form of α-synuclein is more toxic is that β-amyloid (Aβ) promotes α-synuclein fibrillar inclusions in bigenic transgenic mice overexpressing mutant human Aβ precursor protein and α-synuclein, thus creating a more severe phenotype (48).

Fibrillization and aggregation of α-synuclein may play a central role in neuronal dysfunction and death of neurons in PD. Familial-associated mutants of α-synuclein have a greater propensity to aggregate than does wild-type α-synuclein, which accounts for the development of PD in families harboring these mutations (6, 32, 42, 48). Simple overexpression of a fibrillogenic protein in transgenic animals probably accounts for disease in these animal models, as genomic triplication of wild-type α-synuclein in humans leads to familial PD (30).

How does wild-type α-synuclein aggregate in sporadic PD? Derangements in complex I clearly lead to aggregation and accumulation of α-synuclein (20, 2225), and other forms of oxidative stress promote α-synuclein aggregation (49). Oxidative damage appears to play a role in the aggregation of α-synuclein in sporadic PD, as there is selective α-synuclein nitration in synuclein lesions in PD and related disorders (50). The defect in complex I is probably upstream from derangements in α-synuclein in sporadic PD, as cybrid cell lines expressing mitochondrial DNA from persons with the A53T α-synuclein mutation do not manifest complex I deficiency, in contrast to maternally inherited or sporadic PD (51). α-Synuclein may play a critical role in the pathogenesis of complex I–induced DA neuron injury, as mice lacking the gene for α-synuclein are resistant to the toxic effects of the complex I inhibitor MPTP (52). These observations suggest that inhibition of complex I creates an environment of oxidative stress that ultimately leads to aggregation of α-synuclein and the subsequent death of DA neurons (Fig. 1).

Fig. 1.

Complex I deficiency may be central to sporadic PD. Dysfunction of complex I leads to increased oxidative stress, free radical formation, and reduction in adenosine triphosphate (ATP) formation. Decrements in ATP lead to membrane depolarization and contribute to excitotoxic neuronal injury and further free radical–mediated injury involving nitric oxide (NO) and peroxynitrite (ONOO-) and a feedforward cycle of increasing oxidative stress and injury. Slow and chronic complex I deficiency leads to accumulation and aggregation of α-synuclein, which leads to dysfunction of the proteasome and contributes to cell death. Familial-associated mutations in α-synuclein bypass complex I deficiency, but they promote α-synuclein accumulation and aggregation. Parkin appears to be a multipurpose neuroprotective agent that may allow the cell to more readily handle proteasomal impairment. Loss of parkin function may decrease the cells' ability to deal with proteasomal dysfunction. DJ-1 may function as a chaperone, and its absence may also decrease the cells' ability to deal with proteasomal dysfunction.

Impairment in the Ubiquitin-Proteasomal System

In addition to selective decrements in complex I and oxidative stress in sporadic PD, impairment in the ubiquitin-proteasomal system (UPS) and proteolytic stress may also contribute to SNC dopaminergic pathology (53, 54). Structural and functional deficits in the 26/20S proteasome (53, 54) and accumulation and aggregation of potentially cytotoxic proteins, such as α-synuclein, in DA neurons in patients with sporadic PD are consistent with this notion. Rescue of the motoric and pathologic features in transgenic flies expressing normal and mutant α-synuclein by transgenic or pharmacologically induced overexpression of chaperones provides further support for a role of protein mishandling in PD (55, 56). Chaperones may play a role in sporadic PD, as LBs in human postmortem tissue immunostain for chaperones (55). Mutations in DJ-1 cause autosomal recessive PD, and recent structural studies indicate that DJ-1 may have similarities to the bacterial HSP31 homologs, suggesting that it may function to alleviate protein misfolding by interacting with early-unfolding intermediates (57).

The linkage of two genes—those encoding parkin (58) and ubiquitin C-terminal hydrolase-L1 (UCH-L1) (59)—within the UPS in hereditary PD also supports the notion that derangements in the UPS may contribute to the demise of DA neurons. Mutations in parkin were first linked to autosomal recessive–juvenile Parkinsonism (ARJP) (58). Mutations in parkin are a major cause of autosomal recessive PD and are considered to be one of the major causes of familial PD (42). The parkin protein has a ubiquitin-like domain at its N terminus, and it has a pair of RING finger motifs surrounding a RING finger domain at its C terminus (60). Like many other RING finger–containing proteins, parkin appears to function as an E3 ubiquitin protein ligase (61, 62). E3 ubiquitin protein ligases confer substrate specificity in the ubiquitination process and work in concert with the ubiquitin-activating E1 and ubiquitin-conjugating E2 enzymes (63). Polyubiquitinated proteins are recognized by the 26S proteasome subunit and are targeted for degradation (63). Most familial-associated mutations in parkin are defective in E3 ubiquitin protein ligase activity, which suggests that disruption of the E3 ubiquitin protein ligase activity of parkin is the cause of autosomal recessive PD. Because the loss of parkin's E3 ligase activity may cause ARJP, great importance has been placed on the identification of protein substrates of parkin (61, 62). Dysfunction of the proteasomal processing of one or more of these proteins may lead to dopaminergic cell loss (64). Drosophila lacking parkin exhibit reduced life-span, locomotor defects, and male sterility (65). The locomotor defects derive from apoptotic cell death of muscle subsets, whereas the male sterile phenotype derives from a spermatid individualization defect at a late stage of spermatogenesis. The earliest manifestation of muscle degeneration and a prominent characteristic of the individualized spermatids in parkin mutants is mitochondrial pathology; thus, mitochondrial impairment may trigger the selective loss of DA neurons observed in ARJP (65).

Most patients with parkin mutations lack LB pathology (60). Thus, it has been suggested that the function of parkin may promote the formation of LB inclusions (53, 60, 64, 66). Alternatively, parkinmediated neurodegeneration may proceed through mechanisms distinct from those that occur in PD with LBs (64). Native α-synuclein, the major species in brain, does not appear to be a parkin substrate, as parkin fails to interact with and ubiquitinate native α-synuclein (66). Instead, parkin interacts with the α-synuclein interacting protein synphilin-1, and through this interaction it may promote the formation of LBs (66). Alternatively, parkin may interact with a minor O-linked glycosylated form of α-synuclein (67). Interestingly, DA neuronal death that results from ectopic expression of human α-synuclein is mitigated by coexpression of human parkin (68). Furthermore, flies coexpressing parkin and α-synuclein exhibit a sharp reduction (relative to flies expressing α-synuclein alone) in the abundance of α-synuclein–positive grain-like structures and ubiquitin-positive LB-like neurites (69). Thus, parkin may act specifically on aberrant α-synuclein deposits (64). Because parkin does not directly interact with native α-synuclein, alterations in its structure— either posttranslational modifications (such as O-glycosylation and phosphorylation) or conformational changes (protofibrils/fibrils or oligomers)—may be the signal that promotes an interaction between α-synuclein and parkin. However, parkin's protection against toxicity associated with mutant α-synuclein is not due to a direct parkin–α-synuclein interaction, but instead parkin rescues the toxic effects of mutant α-synuclein through rescuing impaired proteasome function (68). It has been suggested that parkin is a multipurpose neuroprotectant (64).

DA neurons seem to be particularly vulnerable to proteasome inhibition (68, 70). Overexpression of α-synuclein inhibits the proteasome and sensitizes cells to toxicity induced by proteasomal inhibitors (71, 72). Aggregated α-synuclein binds directly to the proteasome and inhibits ubiquitin-dependent proteasomal function (73). A model for the pathogenesis of sporadic PD emerges from these findings. If complex I inhibition is central to PD pathogenesis, it would set in motion a process leading to α-synuclein aggregation, which would bind and inhibit the proteasome. Inhibition of the proteasome would lead to failure to clear proteins targeted for degradation by the UPS, ultimately resulting in the demise of DA neurons (Fig. 1). Parkin would participate in the detoxification of proteins that accumulate by rescuing proteasomal function (64), and DJ-1 might participate in the detoxification of proteins through its putative chaperone function (57).

References and Notes

View Abstract

Navigate This Article