Structure/function of α‐synuclein in health and disease: rational development of animal models for Parkinson's and related diseases

α-Synuclein (αSYN) was originally identified in the electric organ of the electric eel Torpedo californica ( Maroteaux et al. 1988 ). It is now well documented that αSYN and its close homolog βSYN are abundantly expressed in presynaptic terminals of vertebrates. Experimental down-regulation of αSYN in knock-out mice and in primary neurones suggests that αSYN regulates synaptic neurotransmission, but the molecular mechanism is still poorly understood. A breakthrough was achieved when two different mutations in the αSYN gene were identified in autosomal-dominant hereditary Parkinson's disease (PD) patients. Soon thereafter, αSYN was found to be the major protein component in the hallmark lesions of PD [Lewy bodies (LBs) and Lewy neurites]. Lewy pathology is also characteristic for dementia with LBs (DLB), the LB variant of Alzheimer's disease, and neurodegeneration with brain iron accumulation type 1 (formerly known as Hallervorden–Spatz disease). Moreover, αSYN fibrils are deposited in (oligodendro)glial cytoplasmic inclusions (GCIs) of multiple system atrophy (MSA) patients. α-Synuclein specifically has an intrinsic propensity to form fibrils in vitro. PD-associated mutations and oxidative stress enhance αSYN aggregation. During fibril formation, the naturally unfolded αSYN undergoes conformational changes, transiently existing as a partially folded intermediate and protofibril. It is a matter of intense research to establish whether the folding intermediates and/or the LB fibrils exert neurotoxicity in human patients. To better understand the pathophysiology of αSYN, a number of transgenic animal models have been developed. Expression of human αSYN in Drosophila melanogaster caused age-dependent, dopamine-selective neurodegeneration and the formation of LB-like deposits. A variety of promoters have been used to express transgenic αSYN in mice. Aberrant accumulation of αSYN in cell bodies and swollen neurites were common observations. Some models displayed age-dependent enhancement of phenotype, such as formation of ubiquitin-positive inclusions, loss of dopaminergic synapses and locomotor deficits. Of great relevance are recent reports that amyloid seeded αSYN aggregation, whereas the non-amyloidogenic βSYN inhibited αSYN fibril formation both in vitro and in vivo. Still, additional genetic and/or environmental risk factors need to be considered in order to generate a complete mouse model of PD. The synucleins comprise a family of 15–20-kDa proteins including αSYN, βSYN and γSYN (also termed persyn) (Kahle et al. 2000a). The primary sequences span three regions (Fig. 1a). The amino-terminal half is characterized by five or six imperfectly conserved repeats (EKTKEGV, single-letter code). This region acquires α-helical secondary structure upon lipid binding (Eliezer et al. 2001), whereas the less conserved carboxy-terminus, in which acidic amino acid side chains are clustered, is highly flexible (Eliezer et al. 2001). A stretch of amino acids critically involved in αSYN aggregation (Giasson et al. 2001; Kahle et al. 2001) resides in the central, hydrophobic portion (called non-amyloid component [NAC]). (a) Schematic representation of synucleins. The imperfect KTKEGV repeats are numbered, the 6th repeat missing in βSYN is stippled. (b and c) Kyte–Doolittle hydropathy plots of αSYN (b) and βSYN (c). The NAC domain amino acid sequences are aligned in the middle, and are indicated by black boxes under each hydropathy plot. The natural deletion in βSYN is stippled. (dand e) Cortical LBs in the temporal cortex (arrows) in a DLB case showing strong αSYN immunoreactivity (d), but no labeling for βSYN (e). The absence of βSYN immunoreactivity in an LB (arrow) displacing a neuronal nucleus (arrowhead) is featured in the insert. The amino-terminal repeats of αSYN were predicted to form an α-helical structure similar to the lipid-binding domains of class A2 apolipoproteins (Davidson et al. 1998). Indeed, αSYN tightly binds to synthetic phospholipid vesicles, and preferentially binds to those with acidic head groups (Davidson et al. 1998; Jo et al. 2000). Mutational analysis of lipid-binding determinants (Perrin et al. 2000) revealed that the lipid-binding domain was broadly distributed over the amino-terminal half of αSYN. Amphipathic class A2 helix formation was crucial because replacement of the six amino-terminal threonines (predicted to lie along the hydrophobic face of the helix) with charged amino acids disrupted binding of mutant αSYN with vesicles. Recombinant full-length, but not amino-terminally truncated αSYN interacted with a crude membrane fraction from rat brain (Jensen et al. 1998). Both PD-associated mutations (A53T, Polymeropoulos et al. 1997; A30P, Krüger et al. 1998) reside in the lipid-binding amino-terminus of αSYN (Fig. 1a). The A30P mutation abolishes a helical moment in the amino-terminus (Bussell and Eliezer 2001), and was reported to disrupt αSYN binding to rat brain membranes (Jensen et al. 1998). The effect on binding to synthetic membranes was much less marked (Perrin et al. 2000; Narayanan and Scarlata 2001). In transfected cortical neurone cultures, PD mutations did not affect coupling of αSYN to lipophilic dyes (McLean et al. 2000). When transgenic mouse brain expressing human mutant [A30P]αSYN was subjected to sucrose gradient floatation experiments (Jensen et al. 1998), the mutant protein was found in the soluble as well as in the floating fractions, as was the endogenous αSYN (Kahle et al. 2000b). Thus, it is unlikely that αSYN mutations cause PD as a result of a loss of membrane binding. Subcellular fractionation of rat, mouse and human brain revealed that αSYN and βSYN were enriched in the synaptosomal fractions, but were dissociated from the synaptic vesicles (SVs) upon further purification (Maroteaux and Scheller 1991; Shibayama-Imazu et al. 1993; George et al. 1995; Kahle et al. 2000b). Thus, the interactionof synucleins with SVs appears to be reversible. Phosphorylation of synucleins (Nakajo et al. 1993; Okochi et al. 2000; Proninet al. 2000) might modulate their association with SVs, as is the case for synapsin (Huttner et al. 1983). However, αSYN binds to synthetic phospholipid membranes regardless of the phosphorylation state of αSYN (Narayanan and Scarlata 2001). This finding does not rule out thepossibility that there is a phosphorylation-dependent interaction with putative synuclein binding protein(s) on the surface of SVs. Our attempts to identify such synuclein binding protein(s) on rat brain SVs using chemical crosslinkers were hampered by the formation of oligomeric synuclein ladders (Okamoto, Kahle and Haass, unpublished results). In cultured rat hippocampal neurones, αSYN associated with synaptic vesicles (Withers et al. 1997; Murphy et al. 2000). Synucleins showed a delayed expression and translocation from the soma upon differentiation in culture (Withers et al. 1997; Murphy et al. 2000). In mature synapses, down-regulation of αSYN with antisense oligonucleotides decreased the number of SVs that were not directly docked to the presynaptic membrane (Murphy et al. 2000). Although no anatomical malformation of synapses was found in αSYN knock-out mice, synaptic function was abnormal in these animals (Abeliovich et al. 2000). When recording dopamine release from striatal slice cultures in response to paired electrical pulses, the normal depression of the second pulse was significantly decreased in the αSYN–/– mice (Abeliovich et al. 2000). Consequently, dopamine release into the synaptic cleft was abnormally high. Over time simple diffusion may drain striatal dopamine, which was indeed slightly but significantly reduced in the αSYN–/– mice (Abeliovich et al. 2000). αSYN was reported to bind to the presynaptic dopamine transporter, causing translocation of the dopamine transporter to the cell surface where it accelerated dopamine uptake (Lee et al. 2001a). Conversely, knocking out αSYN may cause reduced mobilization of the presynaptic dopamine transporter. This would decrease thebinding sites for amphetamine, possibly accounting for the attenuation of amphetamine-induced locomotion seen in αSYN–/– mice (Abeliovich et al. 2000). Moreover, inefficient synaptic re-uptake of dopamine may result from disrupted αSYN recruitment of dopamine transporters. Reduced clearance of dopamine from the synaptic cleft might lead to an overshoot of dopamine in the paired pulse depression model of Abeliovich et al. 2000). Thus, synucleins are not essential for synapse formation, but rather play a role in the maintenance of synaptic function. The molecular mechanisms remain to be further elucidated. It will be interesting to investigate the influence of synucleins on other neurotransmitters. Finally, the extensive colocalization of αSYN and βSYN (Kahle et al. 2000b) raises the question if and how these two closely related synucleins act in synergy, and/or if there are distinct pools of αSYN and βSYN SVs with specific functions. Soon after the genetic link to PD was discovered, αSYN was identified as the major component of LBs and Lewy neurites, the diagnostic brain lesions in PD patients (Spillantini et al. 1997). Moreover, αSYN antibodies stained the Lewy pathology of patients with DLB (Fig. 1d), LB variant of Alzheimer's disease and neurodegeneration with brain iron accumulation type-1 (Spillantini et al. 1997; Wakabayashi et al. 1997; Arawaka et al. 1998; Baba et al. 1998; Irizarry et al. 1998; Takeda et al. 1998; Wakabayashi et al. 1999). In the peripheral (sympathetic) nervous system, αSYN-positive Lewy pathology has been described for patients with pure autonomic failure (Arai et al. 2000; Kaufmann et al. 2001). αSYN-positive LBs were also detected in familial Alzheimer's disease and in patients with Down syndrome (Lippa et al. 1998; Lippa et al. 1999). Although the NAC fragment of αSYN was isolated from senile plaques (Uéda et al. 1993), it has been questioned if αSYN is an integral component of amyloid deposits (Bayer et al. 1999; Culvenor et al. 1999). Rather, dystrophic neurites within and around neuritic plaques accumulate αSYN (Wirths et al. 2000), a feature reproduced in a mouse model (Tg2576) of amyloidosis (Yang et al. 2000). Thus, in addition to the condensation of αSYN into somal LBs, αSYN forms neuritic pathology. Lewy neurites are specific lesions composed of αSYN fibrils, whereas in dystrophic neurites and neuroaxonal spheroids, αSYN accumulates along with other axonally transported proteins. Finally, αSYN was identified as the building block of the GCIs that characterize MSA (Gai et al. 1998; Tu et al. 1998; Wakabayashi et al. 1998). It is not known how the neuronal, presynaptic protein αSYN appears in the oligodendrocytes of MSA patients. Some speculate that αSYN leaks out of damaged axons and passes through the surrounding myelin, ultimately aggregating into GCIs. Alternatively, oligodendrocytes in the brain regions affected by MSA (cerebellum and striatum) may aberrantly express αSYN, which over time aggregates into GCIs. Post-translational events appear to stimulate α-synucleinopathy. Advanced glycation endproducts were proposed to enhance αSYN crosslinking within LBs (Münch et al. 2000). A well-studied risk factor for PD, namely oxidative stress, was implicated in Lewy pathology by the recent development of specific antibodies against oxidized αSYN. Using these reagents, Giasson et al. could demonstrate oxidative modification of αSYN in LBs, Lewy neurites and GCIs (Giasson et al. 2000). Oxidative conditions enhanced the aggregation of αSYN in vitro (see below). Very recently, Iwatsubo and colleagues found αSYN in LBs to be hyperphosphorylated at S129 (Fujiwara et al. 2001), the major phosphoacceptor site of αSYN (Okochi et al. 2000; Pronin et al. 2000). It will be interesting to study the physiological (regulatory?) role of αSYN phosphorylation, and the pathological events leading to its hyperphosphorylation in αSYN-containing inclusions. The fact that αSYN inclusions are mostly ubiquitinated implies a connection between αSYN metabolism and the ubiquitin/proteasome degradation system (Hershko and Ciechanover 1998). The most straightforward model is that defective proteasomal degradation raises the intracellular αSYN concentration above a threshold concentration necessary for fibril formation (see below). Although some evidence supports this model, it appears too simplistic at present. αSYN is a stable protein (Bennett et al. 1999; Okochi et al. 2000; Tofaris et al. 2001), and findings of elevated steady-state levels of αSYN in the presence of proteasome inhibitors were inconclusive (Bennett et al. 1999; Ancolio et al. 2000; Rideout et al. 2001; Tofaris et al. 2001). Moreover, even though αSYN and ubiquitin are in very close apposition inside LBs (Sharma et al. 2001), covalent binding of ubiquitin to αSYN was never demonstrated. In fact, because αSYN is a naturally unfolded protein, it may not depend on ubiquitination for proteasomal targeting. The naturally unfolded αSYN may enter the proteasome in a non-ubiquitinated state via initial binding to Tat binding protein 1 (Ghee et al. 2000), a subunit of the 700-kDa proteasome activator complex. The core 20S proteasome lacks the 700-kDa proteasome activator complex, but it nevertheless efficiently and specifically degraded recombinant αSYN (Tofaris et al. 2001). In PD patients, fibrillar αSYN might no longer be degraded by the proteasome, but rather inhibit the proteasome analogous to aggregates of huntingtin fragments and the cystic fibrosis protein (Bence et al. 2001). Reduced proteasomal activity was detected in post-mortem brains of PD patients (McNaught and Jenner 2001). Thus, a vicious cycle of impaired clearance of αSYN and additional ubiquitinated LB components may culminate in Lewy pathology and neurone death. The fact that αSYN fibrils form the hallmark lesions of PD and related disorders warrants detailed structural analysis of the conversion of native αSYN to amyloid fibrils in vitro. As discussed above, the amino-terminus of αSYN has a slight α-helical moment, and the helical structure is stabilized in the presence of phospholipid vesicles. This might represent the physiologically relevant conformation. In solution, purified αSYN is naturally unfolded (Weinreb et al. 1996) and because it has no stable secondary structure, αSYN does not denature even upon boiling, which accounts for its heat stability (Weinreb et al. 1996; Kim et al. 2000). However, conditions of elevated temperature or low pH induce transient conformational changes that have been interpreted as folding intermediates of the aggregation pathway (Uversky et al. 2001a). As αSYN aggregation proceeds, mature fibrils with the typical amyloid β-sheet structure become predominant (Conway et al. 2000a). Thus, three stable conformational states exist for αSYN: lipid-bound helix, unfolded in solution and amyloid fibrils. It is of prime importance to understand the interconversion of these conformational states in terms of kinetics, folding intermediates and pathological risk factors. of purified recombinant αSYN above a concentration to the formation of amyloid fibrils et al. 1998; Giasson et al. 1999; et al. 1999; et al. et al. 2000). have the as those isolated from brains of patients from PD, DLB and MSA (Spillantini et al. In βSYN is not in LBs (Fig. and is of aggregation in vitro. and experiments to the of the for aggregation of αSYN (Giasson et al. 2001; Kahle et al. 2001). The to amino acids of αSYN in the NAC region is in Kyte–Doolittle plots of αSYN and βSYN that these form a in the NAC domain of αSYN and in this region the of mutant αSYN proteins to in vitro (Giasson et al. 2001; Kahle et al. 2001). temperature accelerated αSYN fibril formation et al. 1998; et al. 2001a). these of hydrophobic may of the partially folded intermediate structure, ultimately fibril formation (Uversky et al. 2001a). relevant risk factors for PD, namely oxidative conditions and as well as have been found to αSYN et al. et al. 2000; et al. also a partially folded structure on αSYN (Uversky et al. mutations enhanced αSYN (Conway et al. 1998; et al. but not via of the partially folded intermediate et al. 2001). and colleagues have the pathological of PD mutant αSYN to enhanced formation of αSYN that be isolated by and appear as in the (Conway et al. 2000b). The of αSYN might the between aggregation and the of dopaminergic neurones in the that accounts for PD including dopamine form covalent with αSYN (Conway et al. 2001). Although this modification inhibited the formation of mature αSYN may be because of formation of (Conway et al. 2001). αSYN but the mature fibrils, were reported to vesicles, a mechanism of neurotoxicity et al. 2001). of vesicles a cycle may that ultimately the dopaminergic Consequently, the formation of LBs may be a of the affected neurone to αSYN When the LB the of a LB the nucleus of a neurone in LBs may block proteins and inhibit proteasomal The may in neurones, for in the cortex where the number of mature LBs with a of namely dementia et al. 2000; et al. 2000). Thus, the of αSYN and neurodegeneration to be further elucidated. The animal models for PD on the neurones using and and Although these models were to study dopaminergic cell relevant for PD, Lewy pathology was not αSYN in mice and et al. 2000; et al. 2000). somal accumulation of αSYN was in both models et al. 2000; et al. 2000), but did not the formation of LB-like inclusions. recently, the was in a rat model that displayed both dopaminergic neurone loss as well as cytoplasmic αSYN inclusions et al. 2000). αSYN fibril formation in vitro (Uversky et al. and in cell culture (Lee et al. by et al. 2000; et al. is to cause oxidative stress that αSYN aggregation et al. et al. 2000). on the genetic and link of αSYN to PD, a number of transgenic animal models have been Drosophila LB-like pathology and dopaminergic cell which caused an age-dependent locomotor in the and 2000). This model an interesting for the of risk for PD. To generate the promoters from and protein were used to express human and PD mutant αSYN the brains of transgenic mice. A common finding was the specific accumulation of transgenic human αSYN in cell bodies and pathological neuritic (Kahle et al. et al. 2000; et al. 2000; Kahle et al. 2001). experiments with and mouse brains revealed that a portion of the transgenic αSYN was (Kahle et al. 2001). is a feature diagnostic for human LB et al. 1998; Culvenor et al. 1999; et al. 2000; et al. 2000; Kahle et al. 2001). In the non-amyloidogenic βSYN was soluble in these (Kahle et al. 2001). In a highly expressing mouse inclusions of αSYN were found in the of some neurones, but also in for LBs, namely at the and inside the nucleus et al. 2000). ubiquitination was in and mice et al. 2000; et al. 2000). αSYN mice also locomotor et al. 2000; et al. 2000; Giasson et al. et al. et al. in model with some of dopaminergic et al. 2000). In the loss of locomotor in the mice was caused by of et al. 2000). This indicated a effect of the most in this expression of transgenic human αSYN in the dopaminergic neurones in the was achieved with the et al. 2001; et al. 2001). In these mice, the as in the were The somal accumulation of αSYN in dopaminergic neurones transgenic mice was of mice et al. 2000). However, transgenic αSYN mice did not to in terms of number of neurones and striatal dopamine levels et al. 2001). Thus, even expression of αSYN does not lead to Lewy pathology in the model, with the absence of LBs in human patients from caused by et al. 1999). The to in an does not with the initial in transgenic mice. that and neurotoxicity of αSYN. It is that failure of this mechanism to Lewy pathology and Consequently, risk factors exist that induce αSYN aggregation and cell in LB disease patients. It is of that Lewy pathology is associated with pathology. A of plaques and αSYN aggregation was by and mice with mice that expressed amyloid protein two different Alzheimer's disease mutations et al. 2001). The mice αSYN inclusions mice and the inclusions fibrillar as to the described for the mice. The amino amyloid αSYN aggregation in vitro et al. 2001), but it is not if and how this that resides within vesicles et al. and in plaques aggregation of the αSYN. The fact that βSYN lacks a aggregation in the NAC and the extensive colocalization of αSYN and βSYN the that βSYN inhibited αSYN aggregation. Indeed, βSYN inhibited αSYN aggregation in vitro et al. 2001). When human αSYN and βSYN were in transgenic mouse the number of αSYN inclusions was reduced with mice et al. 2001). Thus, the of αSYN and βSYN may be a the of αSYN to In DLB patients, αSYN expression with decreased βSYN expression was which may the αSYN aggregation et al. 2001). Consequently, βSYN and might become for Parkinson's disease used to be a for a disease when the PD mutation was to the αSYN This to the of αSYN as the major fibrillar protein in LBs and Lewy neurites of PD and related MSA could be to the of which are to as of the aggregation of synucleins the into the and structural of pathological fibril of this may lead to the of will be in the animal models of α-synucleinopathy. remain to be elucidated. is the of αSYN in which for the of neurones αSYN stability and is the between fibril formation and cell cortical LB formation, and is the molecular of sites of αSYN aggregation within these brain is there of the neuronal protein but not is the physiological role of The research which are to the to other caused by of unfolded such as the 2001).