In the present chapter, we discuss the key findings on αsyn (α-synuclein) oligomers from a biophysical point of view. Current structural methods cannot provide a high-resolution structure of αsyn oligomers due to their size, heterogeneity and tendency to aggregate. However, a low-resolution structure of a stable αsyn oligomer population is emerging based on compelling data from different research groups. αsyn oligomers are normally observed during the formation of amyloid fibrils and we discuss how they are connected to this process. Another important topic is the interaction of αsyn oligomers and membranes, and we will discuss the evidence which suggests that this interaction might be essential in the pathogenesis of Parkinson's disease and other neurodegenerative disorders. Finally, we present a remarkable example of how small molecules are able to stabilize non-amyloid oligomers and how this might be a potential strategy to inhibit the inherent toxicity of αsyn oligomers. A major challenge is to link the very complex oligomerization pathways seen in clever experiments in vitro with what actually happens in the cell. With the tremendous developments in optical microscopy in mind, we believe that it will be possible to make this link very soon.
Introduction: the role of α-synuclein in Parkinson's disease
The interest in protein misfolding has increased spectacularly after it was established that the conversion of a range of soluble proteins into insoluble amyloid fibrils is linked to several diseases (see Chapter 1 by Louise Serpell). This link is direct in systemic amyloidosis, where excessive accumulation of amyloid fibrils in joints or organs leads to pathology. However, in neurodegenerative disorders such as PD (Parkinson's disease), the link is more indirect. In PD, amyloid fibrils consisting of the protein αsyn (α-synuclein) form intracellular deposits called LBs (Lewy bodies) and Lewy neurities, which appear to accompany the loss of dopaminergic neurons mainly in the part of the brain called the substantia nigra. Analogous observations have been made for Alzheimer's disease. Such observations initially led to the hypothesis that amyloid fibrils were responsible for the pathogenesis of PD and other neurodegenerative disorders. However, new observations have made a compelling case that pre-fibrillar oligomers, typically formed in the early stages of the fibril formation process, are the actual pathogenic culprits. This shift from amyloid fibrils to amyloid oligomers was inspired by the emergence of a similar oligomer hypothesis in Alzheimer's disease, where Aβ (amyloid β-peptide) was shown to form a range of oligomeric structures with cytotoxic properties . Subsequently, Lansbury, Fink and co-workers demonstrated that αsyn oligomers, and not mature amyloid fibrils, have membrane-permeabilizing properties [2−5]. Their toxicity has been demonstrated in vivo , and elevated levels of αsyn oligomers have been found in cerebrospinal fluid from PD patients  and in post-mortem brain extracts from patients with LB dementia. Consequently, αsyn oligomers are now one of the primary targets in PD drug development. αsyn oligomers are included in a clinical trial as potential biomarkers for the neurodegenerative disorder multiple system atrophy, known as a Parkinson-plus syndrome (ClinicalTrials.gov, NCT01485549), whereas Aβ oligomers have been targeted by the compound scyllo-inositol (ClinicalTrials.gov, NCT00934050).
αsyn consists of 140 residues and is intrinsically disordered, i.e. it lacks persistent secondary and tertiary structure . However, upon interaction with negatively charged membranes, residues 1–100 undergo a coil-helix conformational change . Under these conditions αsyn can co-exist both as a single long helical chain (extended conformation) and as two helices in a hair-pin arrangement (horse shoe) . This ability of αsyn to fold into membranes is believed to be essential for its physiological role. The basic N-terminal part (residues 1–60) initiates the interaction with membranes and leads to the subsequent folding of the NAC (non-amyloid β-peptide component) region (residues 61–95). The NAC region is hydrophobic and is known to constitute the core of amyloid fibrils . The C-terminus is highly acidic and unstructured. Apart from possible electrostatic interactions between the YEMPS region (residues 125–129) and other regions of the polypeptide, the C-terminus is disordered in both the free monomer, the membrane-bound monomer, the oligomer and the fibril structure . It has been suggested recently that αsyn could form a tetramer in vivo , but this remains a minority position, contradicted by a vast literature on the properties of natively unfolded αsyn as well as direct rebuttals which show αsyn to naturally form a disordered monomer .
Lashuel et al.  used EM (electron microscopy) to show that both αsyn and Aβ oligomers form annular ring-shaped oligomers (Figure 1). Combined with the observation that the oligomers were able to permeabilize synthetic membranes , this suggested that pore formation in cell membranes led to oligomer toxicity. This pioneering finding has later been supported for amyloid oligomers formed by other peptides and proteins (ABri, ADan, serum amyloid A and amylin) which are ring formed and permeabilize membranes . Note that sample preparation in EM and AFM (atomic force microscopy) involves drying which might induce structural artefacts. As discussed in the next section, solution studies suggest that the αsyn oligomer is ellipsoidal rather than ring shaped.
Amyloid oligomers of two disease-promoting mutations of αsyn (A30P and A53T), as well as the Arctic mutant of Aβ
A low-resolution structure of αsyn oligomers: compact core with diffuse shell
In this section we discuss the structure of a type of αsyn oligomer which has been studied thoroughly by the Subramaniam group [11,14,15], and which is highly similar (if not identical) to the oligomers that are studied in our research group [16–19]. The Subramaniam group form stable αsyn oligomers by incubating αsyn at a concentration of 1 mM for 18 h at room temperature under vigorous shaking, followed by a 2 h incubation at 37°C without shaking. It is an open question to define the biologically most relevant conditions to prepare oligomers. Most researchers agree to use PBS buffer pH 7.4, but other conditions can vary. Oligomers form spontaneously when monomeric αsyn is dissolved at high concentrations and left on ice , but higher yields are obtained by shaking the sample at the more physiological temperature of 37°C. Although the biological relevance of shaking can be questioned, it certainly leads to more reproducible aggregation, particularly in combination with the use of glass beads [21,22]. Subsequently, oligomers are purified by SEC (size-exclusion chromatography), similar to other established protocols [4,16]. Subramaniam and co-workers elegantly used a combination of sub-stoichiometric labelling and single-molecule photobleaching to count the number of monomers per oligomer, arriving at a number of ∼31 and at the same time showing the oligomer population to be monodisperse (i.e. only one type of oligomer) . By labelling individual positions with the fluorescent residue tryptophan, they were able to conclude that the N-terminus and NAC region are part of the oligomer core, whereas the C-terminus remains disordered .
In 2010, we used SAXS (small-angle X-ray scattering) to resolve the structure of αsyn oligomers that accumulate during fibril formation . We could resolve monomers, dimers, oligomers and fibrils in the fibril formation process and, by ab initio modelling, we determined the shape of the oligomers to be an ellipsoid with dimensions given in Figure 2 (left-hand panel). More recently, we have gone a step further. We purified the oligomers from samples with fibril formation, by incubating 840 μM αsyn for 5 h at 37°C with vigorous shaking and subsequent purification with SEC, similar to the approach of other groups [4,14,16]. This reasonably pure oligomer solution (<10% monomers) has provided a more detailed structural model. The oligomers consist of a rigid core with the same dimensions as our first SAXS model, but this core is covered by a 5 nm thick outer layer consisting of disordered polypeptides (Figure 2, right-hand panel) . The SAXS analysis together with a complementary SEC-MALLS (multi-angle laser light scattering) analysis estimate that the average oligomer is built up of ∼29 monomers , in strong agreement with the Subramaniam group . All of these size-estimation methods are independent of the shape and conformation of the molecules, giving more reliable data on the oligomer structure than SEC, DLS (dynamic light scattering), EM and AFM which either report on hydrodynamic radius (SEC and DLS) or are sensitive to drying artefacts (EM and AFM). Our revised structure of αsyn oligomers is in good agreement with the proposed micellar structure by the Subramaniam group where the less charged and more hydrophobic N-terminal and NAC regions forms the compact core, whereas the highly charged C-terminal forms a brush-like shell . This compact organization might explain the monodispersity and the remarkable stability of the αsyn oligomers.
Ellipsoid structure of the αsyn oligomer
The oligomer core is most likely to be organized in β-sheets [5,24]. Fourier transform infrared spectroscopy suggests that both Aβ and αsyn oligomers contain β-sheet structure. Although mature amyloids of Aβ and αsyn fibrils consist of parallel β-strands, oligomers appear to contain anti-parallel β-strands . This difference could arise in two ways. Either oligomers and fibrils belong to different aggregation pathways, or oligomers have to undergo structural rearrangements before they can become incorporated into fibril structures.
Note that there is not just one oligomer of αsyn. Different oligomers have been reported, containing different types of secondary structure, ranging from mainly α-helical to disordered, and of varying sizes, e.g. coexisting oligomers of ten and 15 monomers  (see below). Oligomers may also be induced by metal ions, lipids, alcohols and small molecules [26,27]. A tremendously important challenge is to determine which oligomer structures are relevant in vivo, and how they can be purified and stabilized for thorough analysis.
What is the role of αsyn oligomers in the process of fibril formation?
The amyloid fibril structure is a thermodynamically favourable state which has been suggested to be generic for all proteins. However, the mechanism of amyloid formation is not generic, but varies between proteins and is also highly dependent on solution conditions. A key question in this regard is whether oligomers, which are often observed during fibril formation, are compulsory precursors (on-pathway intermediates) for fibrils or rather dead-end species that are not incorporated into the fibrils. There is no simple answer to this; on- and off-pathway oligomers may co-exist and conflicting observations may also reflect different assembly processes under different conditions. However, it is generally believed that monomers are the elongating species in αsyn fibril formation, i.e. fibrils grow by addition of monomers to fibril ends (Figure 3). On-pathway oligomers are formed at an early stage. They cover the whole range of species between monomers and fibrils and are also commonly known as fibril nuclei, pre-fibrillar oligomers and protofibrils. Off-pathway oligomers can also be observed under fibril forming conditions. These belong to a separate aggregation pathway and would have to dissociate into monomers, or possibly smaller oligomers, to be able to enter fibril formation.
Schematic representation of the role of on- and off-pathway oligomers in the process of fibril formation
On-pathway αsyn oligomers
Cremades et al.  have combined single-molecule techniques with kinetic analysis to monitor the development of different αsyn oligomers in the initial phases of the fibrillation process. Before making the oligomers, αsyn monomers were labelled with two different fluorophors. This made it possible to identify oligomers in solution as follows: when the oligomers were prepared from these labelled monomers and their fluorescence was measured under very dilute conditions (where one detects only one or a few molecules at a time), oligomers gave rise to co-incident bursts of light from both fluorophors (since they contained both types of labelled monomers), whereas monomers only gave bursts from one fluorophor at a time. The size of the burst provided an estimate of the oligomer size. In addition, the authors were also able to make a rough size classification of oligomers based on the level of FRET (Förster resonance energy transfer) between the two fluorophors in the oligomer. The higher the FRET values, the closer the fluorophors are to each other, although analysis is complicated by the fact that there are several fluorophors in each oligomer. FRET values were shown to follow a Gaussian distribution of values which was consistent with four different oligomer distributions, denoted as Asmall (2–5-mers), Amed and Bmed (both 5–15-mers, with the B oligomers being slightly larger) and Blarge (∼15–150-mers). A and B refer to mid- and high-FRET values respectively. The difference in FRET values of A and B oligomers suggest that they are of different structure. As shown in Figure 4, A oligomers are just as sensitive to protease degradation as the monomer, indicating a highly flexible structure; B oligomers were much more protease resistant, suggesting compact β-sheet structure (consistent with higher FRET values), whereas mature fibrils were the most resistant. Asmall and Amed formed at similar rates and with hardly any lag time, whereas Bmed and Blarge showed a longer lag time. Thus B oligomers are likely to be formed from the A oligomers; A oligomers accumulate because they are formed more quickly from monomers (by nucleation) than they decay to B . In this model, αsyn follows NCC (nucleated conformational conversion) where non-amyloid oligomers are readily formed by incorporation of monomers into the growing oligomer and accumulate until they undergo an internal structural rearrangement (‘nucleated conversion’ without the participation of monomers) to amyloid oligomers competent of being elongated into mature fibrils. Both types of oligomer can subsequently grow by incorporating more αsyn monomers. This NCC model has also been proposed as a nucleation mechanism for prions  and Aβ .
Size classification of oligomers using FRET values
The experiments by Cremades et al.  provide the most detailed picture of the interconversion of different forms of αsyn oligomers under carefully controlled in vitro conditions. The obvious question is how this process occurs in the cell. Most likely, this question will be answered when we can apply fluorescence microscopy at single-molecule level in the cell using labelled αsyn, either prepared directly in the cell or introduced by e.g. electroporation .
Off-pathway αsyn oligomers
We have found that oligomers depicted in Figure 2 (right-hand panel) actually inhibit fibril formation and are unable to elongate into mature amyloid fibrils . Thus although these oligomers accumulate during the fibril formation process, they appear to be unproductive in this process. These oligomers can aggregate further into non-fibrillar aggregates suggesting that they belong to an aggregation process distinct from fibril formation. However, it cannot be ruled out that these oligomers might also be connected to fibril formation by processes similar to NCC. Off-pathway oligomers have also been observed when stabilized by e.g. small molecules; an example is given in the final section.
The variety of oligomers is the natural consequence of a process which has not undergone biological evolution to optimize formation of the end product . One current challenge is to understand to what extent off-pathway oligomers under different conditions might be able to undergo structural rearrangement and proceed in the process of fibril formation.
Oligomer–membrane interactions: the cause of toxicity?
Intriguingly, some conformational antibodies can recognize a range of different amyloid oligomers without recognizing the monomer or mature fibril form. This suggests that these oligomers have a common structure  which may be the basis for their cytotoxicity . Several oligomers have exposed hydrophobic regions and bind the hydrophobic probe ANS (8-anilinonaphthalene-1-sulfonic acid) . A common structural motif exposing ‘sticky’ surfaces may promote oligomer–membrane interactions and perturb the membrane [3,13,15,34,35]. It is possible that oligomer structural flexibility, leading to hydrophobic exposure and structural rearrangements following membrane interaction, is essential for toxicity.
The ability of αsyn oligomers to permeabilize membranes has been investigated intensely. An example of the permeabilization of synthetic membranes, composed of the anionic POPG lipid, by αsyn oligomers is shown in Figure 5. The degree of permeabilization is monitored by the fluorophore/quencher pair HPTS/DPX and the kinetics reveals that complete permeabilization (efflux/influx) is obtained within 5 s. Staining of the membrane shows that the form and size of the vesicles remains intact upon oligomer binding and permeabilization. The Subramaniam group has used single-tryptophan mutants to demonstrate that the N-terminus is involved in oligomer–membrane interactions, as is also the case for the monomer . We have confirmed recently how the N-terminus is essential for oligomer–membrane interaction and membrane permeabilization . The αsyn oligomer selectively binds to anionic lipids and preferentially to liquid-disordered-phase regions of the membranes  where the lipid bilayer is loosely packed and the hydrophobic membrane interior is more accessible . Thus interaction of αsyn oligomers with membranes seems to be governed by electrostatic interactions of the N-terminus with the membrane, combined with hydrophobic interactions of accessible hydrophobic patches in the αsyn oligomer structure with the membrane interior.
Confocal microscopy image of POPG giant unilamellar vesicles
Small molecules as potential drugs?
It is a tremendous challenge to develop drugs for neurodegenerative disorders due to the complex BBB (blood–brain barrier). The difficulty in delivering macromolecules such as antibodies and RNA aptamers has put focus on identifying small molecules which are able to cross the BBB and interact specifically with αsyn. Numerous small molecules with mono-, di- and tri-hydroxyphenyl groups inhibit protein aggregation while promoting oligomer formation. Examples include the neurotransmitter dopamine , the flavonoid baicalein obtained from herbal medicine  and the catechin EGCG (epigallocatechin gallate) from green tea . The effect of EGCG on αsyn oligomerization and fibril formation is a remarkable example of how small molecules can redirect aggregation pathways.
EGCG binds preferentially to disordered polypeptides, but also to structured regions at high concentrations. Wanker and co-workers have proposed a model where EGCG binds the αsyn monomer and redirects the protein from the process of fibril formation into sequential aggregation, leading to stable non-toxic oligomers  (Figure 6). These off-pathway oligomers are amorphous and show no β-sheet content. Moreover, EGCG is able to reorganize mature amyloid fibrils of αsyn and Aβ into amorphous and non-toxic oligomers .
Schematic representation of the fibril formation pathway and the EGCG-directed aggregation pathway
The effect of EGCG on the aggregation of αsyn and Aβ is only one of many possible strategies to prevent the formation of toxic amyloid oligomers. Another strategy is to use small molecules to stabilize the protein in the monomeric form; at the other extreme, pro-aggregators can shift the equilibrium towards the amyloid fibrils, which are believed to be less toxic .
To clarify the role of αsyn oligomers in the amyloid fibril formation process and their potential role in the pathogenesis of PD will be important for the understanding and treatment of not only PD, but also other related neurodegenerative disorders such as Alzheimer's disease and Huntington's disease. The major hurdle here is to define the cytotoxic species. Current research points towards oligomers as the prime suspects. Therefore it is crucial to establish which oligomers are relevant in the amyloid process in vivo, and whether it is a defined oligomer or rather a whole spectrum of different pre-fibrillar oligomers that are cytotoxic. Until then, it will be difficult to rationally design drug discovery programmes towards oligomers. For now, an alternative strategy might be the development of small molecules or nanoparticles that are specific towards αsyn, and which stabilize the monomer form in such a way that amyloid formation is completely inhibited without compromising the protein's underlying biological function. This, on the other hand, will not be trivial to accomplish.
αsyn oligomers are likely to be the cytotoxic species in the pathogenesis of PD.
A low-resolution structure is emerging for stable αsyn oligomers.
αsyn oligomers are observed as both on- and off-pathway in the fibril formation process.
Oligomers interact strongly with membranes and this is possibly the cause of neuronal damage in PD and other neurodegenerative disorders.
Small molecules, such as EGCG, redirect the aggregation pathway of αsyn.
We apologize to the many authors whose work we could not include because of limitations in the number of references. We thank Jørn Døvling Kaspersen and Jan Skov Pedersen for very fruitful collaborations on SAXS analysis of the αsyn oligomers. We are supported by the Michael J. Fox Foundation and the Danish Research Foundation (inSPIN).
1Present address: Department of Protein Structure and Biophysics, Novo Nordisk A/S, 2760 Måløv, Denmark.