Introduction
Parkinson's disease is a common age-associated degenerative disorder characterized by rigidity, difficulty in initiating movements and a resting tremor1, 2. At autopsy, the brain is characterized by loss of basal ganglia neurons, predominantly dopaminergic neurons of the substantia nigra. The underlying cause of this degeneration is unknown, so current treatments for the disease are based on ameliorating the motor symptoms, either by replacement of the deficient neurotransmitter (for example, by L-DOPA therapy) or, less commonly, by direct surgical perturbation of the affected neuronal circuitry. A possible clue to the aetiology of Parkinson's is the presence in some of the nigral neurons that remain at autopsy of fibrillar cytoplasmic inclusions known as Lewy bodies3. In contrast, Lewy bodies in cortical neurons characterize diffuse Lewy body disease (DLBD), a relatively common dementia often mistaken for Alzheimer's disease4. Whether Lewy bodies are a cause or a consequence of neuronal death in Parkinson's disease and DLBD, or are an unrelated epiphenomenon, has yet to be determined. This question is reminiscent of the unresolved role of fibrillar deposits (amyloid plaques and nuclear inclusions, respectively) in the aetiology of Alzheimer's and Huntington's diseases5, 6.
A clue from familial Parkinson's
The genetic contribution to idiopathic late-onset Parkinson's disease
is difficult to determine because of the existence of a significant presymptomatic
phase that has been revealed by positron emission tomographic (PET) imaging;
neuropathological examination shows that loss of around 60% of the
nigral dopaminergic neurons can apparently be tolerated. It is, however, clear
from studies of identical twins that cases in which one twin is diagnosed
before the age of 50 have a significant genetic component7.
The first Parkinson's disease gene to be identified was linked to a
rare, autosomal-dominant, early-onset form of the disease (FPD; other genes
have been reported subsequently, see Box 1)8, 9.
This gene encodes the protein 
-synuclein, which is widely expressed
in the brain. Subsequent to reports of the genetic linkage, immunohistochemical10, 11 and biochemical12 studies of Lewy bodies from
idiopathic cases showed that wild-type -synuclein is a major component
of the insoluble fibrillar portion. The fact that the disease-associated gene
encodes the fibrillar protein is again reminiscent of other neurodegenerative
diseases, including Alzheimer's and Huntington's (
Box 2), and suggests that a gain of toxic function may be linked to -synuclein
fibrillization.
An 
-synuclein knockout mouse
-Synuclein is expressed throughout the brain at high levels13, but there is little information on its normal function. Unlike
most proteins of 140 amino acids, neither wild–type -synuclein
nor the two disease–linked mutants fold into structured, globular forms
in vitro; the protein is intrinstically disordered or 'natively
unfolded'14, 15. This unusual behaviour raises the question
of how an unstructured cytoplasmic protein evades rapid degradation. It may
be that -synuclein exists in vivo as a complex with another
protein or proteins; one possible candidate has been identified11. -synuclein
is a presynaptic protein that appears to be associated with vesicular structures16, 17 and has been linked to learning, development and plasticity18. Thus, despite apparently convergent evidence that points to a
toxic gain of -synuclein function being responsible for Parkinson's
disease, the possibility that disease results, in full or in part, from the
loss of functional -synuclein must be investigated. Knockout
mice, in which the -synuclein gene has been removed, represent
one loss–of–function model. These mice appear to develop normally
and exhibit slightly altered stimulus–dependent dopamine release, suggesting
that -synuclein is a negative regulator of dopamine release19. The brains of the knockout mice, like those of Parkinson's
patients, are characterized by reduced levels of striatal dopamine, but, unlike
Parkinson's brains, they do not contain Lewy bodies, nor are they characterized
by neuronal or synaptic loss (Table 1). Although it
is possible that these differences may be related to the slow progression
of the disease relative to the lifespan of the mouse, it is more likely that
some or all of the features of Parkinson's disease are a consequence
of -synuclein–dependent toxicity, casting further suspicion on
the process of fibrillization.
![Table 1 - Comparison of human Parkinson|[rsquo]|s disease with|[ndash]|synuclein
transgenic and knockout animals.](/common/images/table_thumb.gif)
-synuclein-expressing transgenics
The gain of toxic function scenario inspired two recent animal models of
Parkinson's disease – transgenic mice20 and
Drosophila21 (see Table 1). These
animal models recreate certain features of the disease by expression of human
wild–type -synuclein (in the case of the flies, both mutant proteins
have similar effects). In both cases, cytoplasmic inclusions that resemble
Lewy bodies, to different degrees, were detected. Both reports note a rough
correlation between the transgene expression level and inclusion formation;
in the mouse, the correlation is with the number of inclusions, in the flies,
with the rate at which fibrillar inclusions appear. In the mouse model, the -synuclein
inclusions have no detectable fibrillar substructure, unlike those in the
'symptomatic' fly (see below) and those in post–mortem Parkinson's
and DLBD brains. Thus the mouse inclusions, as well as diffuse -synuclein
deposits detected in 'presymptomatic' fly brains, may represent
a prefibrillar stage. Importantly, both the mice and the flies have age-dependent
pathological and behavioural dopaminergic abnormalities; synaptic loss in
the mice and cell loss in the flies – and motor deficiencies in both
organisms (Table 1). The onset of a motor phenotype
was in both cases correlated to inclusion formation; in the case of the mice,
the expression level of wild-type -synuclein correlated with the frequency
of inclusions and with the extent of the phenotype. Similarly, in the case
of the flies, the appearance of inclusions parallelled the onset of motor
abnormalities, consistent with -synuclein fibrillization being critical
for disease.
-synuclein oligomerization
The transformation of -synuclein from its normal soluble state to
the disease-associated fibrillar state involves coupled changes in its conformation
(increased 
-sheet content) and its quaternary structure (oligomerization
and subsequent fibrillization). There is no evidence for a stable, structured -synuclein
monomer; that is, the -sheet conformation is not long-lived in the absence
of oligomerization (this also seems to be the case for many other fibrillogenic
proteins)22. The FPD-linked mutations do not significantly alter
the conformation of -synuclein, suggesting that they may instead influence
the intermolecular interactions that drive formation of the ordered fibrillar
form15. Our own studies of in vitro fibril formation
by -synuclein, and also by the A protein of Alzheimer's
disease, demonstrate that it does not follow a simple one-step transition,
but rather a complex process that involves one or more discrete intermediates,
termed protofibrils (Fig. 1)15, 23, 24, 25, 26.
The possible existence of these prefibrillar species in vivo offers
a simple explanation for the fact that fibrillar -synuclein is not
detected in 'symptomatic' transgenic mice20; that
is, a protofibril, rather than the fibril itself, may be pathogenic. This
scenario (Fig. 2) is also consistent with the in
vitro behaviour of the A30P variant of -synuclein, which is linked
to early–onset Parkinson's disease9. A30P accumulates
in an oligomeric form, as it forms fibrillar species more slowly than does
wild-type -synuclein27. We have characterized several -synuclein
oligomers that are structurally related to the fibril in that they, like fibrils
but unlike the natively unfolded monomer, contain –sheet structure
(J-C. Rochet and P.T.L., unpublished observations). These oligomers are much
smaller than fibrils and appear early in the fibrillization process (Fig. 1)26, 27.
Figure 1: Images obtained by atomic force microscopy (AFM) of several discrete 
-synuclein
oligomers; placed in a pathway that is consistent with the time–course
of -synuclein fibrillization and parallels the pathway of A
fibrillization,
which has been extensively studied.
![Figure 1 : Images obtained by atomic force microscopy (AFM) of several discrete |[alpha]|-synuclein
oligomers; placed in a pathway that is consistent with the time|[ndash]|course
of |[alpha]|-synuclein fibrillization and parallels the pathway of A|[bgr]| fibrillization,
which has been extensively studied.](/ncb/journal/v2/n7/thumbs/ncb0700_E115_f1.jpg)
The first step involves the conversion of natively unfolded monomeric -synuclein
to an apparently spherical (height is 3–5 nm) oligomeric form that is
rich in –sheet structure (by circular dichroism, J-C. Rochet and
P.L., unpublished). This is the step that seems to be sensitive to both mutations
linked to Parkinson's disease27. It is not difficult to
imagine how age-associated or toxin-induced2 deficits in energy
metabolism could lead to increased concentrations of cytoplasmic synuclein
(the proteasome is ATP-dependent) and/or the inability to prevent oligomerization
(many chaperones are ATP-dependent). Subsequent steps seem to involve a series
of sphere–sphere annealing steps, leading to chain-like assemblies (height
is 3–5 nm) that can either fuse head-to-tail to produce an annular
protofibril (height is 3–5 nm), or can anneal side-to-side to produce,
by a cooperative mechanism, large fibrils (height. is around 8 nm and length
is much greater than chain protofibrils). The blue scale bar = 250 nm.
Figure 2: A scenario, based on the actual species shown in Fig. 1, that offers one explanation for the imperfect correlation between fibrils/Lewy bodies and Parkinson's disease.
![Figure 2 : A scenario, based on the actual species shown in Fig.
1, that offers one explanation for the imperfect correlation between
fibrils/Lewy bodies and Parkinson|[rsquo]|s disease.](/ncb/journal/v2/n7/thumbs/ncb0700_E115_f2.gif)
This scenario represents one extreme version of a group of related possible
mechanisms, in that it holds that the fibrils are inert. In fact, a low level
of toxicity (per -synuclein molecule) may be associated with the fibrillar
state. Two critical steps that could be targeted by small drug-like molecules
are highlighted in red and green. Inhibition of the initial formation of -sheet
containing spherical protofibril (red step) might be expected to decrease
the number of Lewy bodies and delay onset of Parkinson's-like symptoms.
In contrast, inhibition of the protofibril-to-fibril transition (green step)
might be expected to decrease the number of Lewy bodies while accelerating
onset of disease symptoms, by causing accumulation of a toxic species. Finally,
the effect of inhibition of protofibril circularization (blue step) may help
to determine whether the annular protofibril is the toxic species or a harmless
off-pathway reservoir for toxic protofibrillar forms.
The proposal that a prefibrillar intermediate or an alternative assembly
state of -synuclein is toxic also explains two observations made on
pathological examination of human brains. The first is that the substantia
nigra dopaminergic neurons containing Lewy bodies appear to be 'healthier,'
by morphological and biochemical criteria, than neighbouring neurons28, 29; the second is that it is not uncommon to observe 'incidental'
Lewy bodies at autopsy of aged individuals who had no symptoms of Parkinson's
or other neurodegenerative diseases30. Thus, fibrillar inclusions
may sequester toxic species and/or divert -synuclein from toxic assembly
pathways. Careful studies of pathogenesis in animal models of related neurodegenerative
diseases also support the notion that a nonfibrillar species, the formation
of which is linked to fibrillization, is pathogenic. First, two transgenic
mouse models of Alzheimer's disease exhibit Alzeimer's-like abnormalities
before fibrillar plaques can be detected31, 32. Second, at the
time of birth, a transgenic mouse model of Creutzfeldt-Jakob disease contains
abnormal forms of the prion protein (PrP), albeit not the form correlated
with full-blown disease (PrPSc)33. As the animals
age, neurologic abnormalities can be detected and, subsequent to that time,
PrPSc appears in the brains of these animals. Third, several
cellular and animal models of Huntington's disease and other triplet-expansion
diseases have demonstrated that experimental perturbations that reduce the
number of nuclear inclusions that are detectable by light microscopy increase
the severity of disease-related abnormalities (see Box 2).
Unfortunately, direct detection of the prefibrillar intermediates that may
be responsible for these effects has proved extremely difficult, because the
A- and -synuclein-derived oligomeric species formed and characterized
in vitro (Fig. 1) are transient, too small to be
reliably detected in tissue by light and electron microscopy, and relatively
unstable to dilution and other conditions used to extract abnormal protein
from tissue.
The missing link
The mechanism by which abnormal -synuclein oligomers may cause dysfunction
and death of dopaminergic neurons is the missing link in the model of the
Parkinson's disease pathogenic cascade depicted in Fig.
2 (ref. 2). It is important to remember
that the toxic pathway was not necessarily optimized by evolution, but may
be a survivor of a selective process that would disfavour its interference
with reproduction5. Thus, the cell-type selectivity of Parkinson's
disease degeneration, a hallmark of the disease, may merely be a consequence
of the fact that dopamine is, itself, neurotoxic, so that any disruption of
its packaging and secretion could result in cell death. Alternatively, it
may be that the dopaminergic cytoplasm presents the optimal environment for -synuclein
oligomerization or that the target of the toxic species is most highly expressed
in these neurons. Structured -synuclein oligomers could exhibit potent
interactions with certain targets that present repeating interacting motifs,
such as membrane surfaces, proteoglycans and other oligomeric proteins. These
binding partners will be difficult to identify, because the popular methods
for identifying protein-protein interactions in pathways (for example, yeast
two-hybrid assay) are designed to discover monomer-monomer complexes. Although
it is not necessary to understand the mechanism of oligomer toxicity in order
to devise effective therapies for Parkinson's disease, as once the toxic
species has been identified, oligomerization itself can be targeted, elucidation
of this pathway will provide new downstream targets for therapeutic intervention.
Aetiology of transgenic animals
The animal models of Parkinson's disease allow the timing of -synuclein
oligomerization and fibrillization to be compared to disease-associated phenomena
and testable aetiological models of disease to be constructed. The promise
of these models is that they can be used to distinguish cause and effect experimentally,
by observing the consequences of various perturbations on the progression
of the disease and its associated phenomena. If it is possible to 'uncouple'
a particular phenomenon (for example, -synuclein fibrillization) from
disease progression, then it is unequivocally an epiphenomenon. The repeated
failure to do so suggests, but does not prove, a direct involvement of that
phenomenon in the pathway.
Two classes of perturbations can be envisioned. The first are endogenous
perturbations from altered expression of existing or mutant genes, and the
second are exogenous perturbations from drug-like compounds. Using classical
Drosophila genetic methods, mutants can be selected on the basis of their
ability to suppress or enhance disease in the fly34. Suppressor
genes are candidate protective factors, and the products of these genes are
potential protein therapeutics. Alternatively, using classical screening methods
of medicinal chemistry, drug-like molecules can be selected on the basis of
their ability to modify a particular disease-associated phenomenon in vitro
. For example, molecules could be selected that inhibit -synuclein
fibrillization, while allowing oligomerization (Fig. 2).
According to the speculative model presented above, these molecules should
accelerate disease onset and progression by causing accumulation of the toxic
oligomer. Molecules of this class would, however, inhibit appearance of fibrillar
Lewy-body-like inclusions. A similar effect may account for the effects of
certain proteins that inhibit the appearance of visible inclusions on experimental
triplet repeat disease (see Box 2).
Therapeutics and diagnostics
The rapid progression from identification of the first Parkinson's
disease gene in 1997 to animal models in early 2000 holds great promise that,
in the next few years, the pathogenesis of the disease might be elucidated
to the extent that target proteins and processes can be identified and new
therapeutic approaches developed. The successful implementation of any approach
based on the earliest pathogenic events, such as -synuclein oligomerization,
may, however, depend on the detection of very early preclinical disease. The
possibility of identifying Parkinson's disease presymptomatically is
likely to result from rapid advances in two distinct areas of research. First,
genetic screening may allow the identification, and eventually the detection,
of susceptibility factors for the disease, and thus allow the identification
of individuals who are at increased risk of contracting it. Second, imaging
methods such as PET that are already able to detect presymptomatic dopaminergic
deficits will become more practical and less expensive, allowing secondary
screens of the at-risk populations. Although economic considerations may prevent
widespread screening for presymptomatic Parkinson's disease, these advances
are likely to be critical for the development of novel therapeutics.
