2.1 Fluorescent dyes for amyloid fibrils

In this section, we will present an overview of the fluorescent dyes used to study the time dependence of amyloid fibril formation. The fluorescence emission of the dye Thioflavin T (ThT) increases upon binding to β-sheet structures. Therefore, measuring ThT fluorescence intensity variations as a function of time in an aggregation assay has allowed the monitoring of β-sheet formation and thereby the fraction of aS fibrils. ThT is typically excited at 450 nm, and the emission spectrum is recorded between 460 and 550 nm, with a maximum at 480 nm. The fluorescence intensities of ThT and Thioflavin S (ThS) are usually measured in vitro (LeVine, Reference Levine1993) and in cells or tissue (Luk et al. Reference Luk, Song, O'Brien, Stieber, Branch, Brunden, Trojanowski and Lee2009), respectively. However, ThS can also be used in vitro; there are two main differences between the two dyes, which are both derived from the same chemical precursor: ThT undergoes a change in its emission spectrum and is positively charged under binding conditions, whereas ThS has a negative charge under the same conditions and its emission spectrum only increases in intensity in the presence of β-sheets (LeVine, Reference Levine1993). ThT is preferred over ThS for in vitro assays because the shifted ThT emission peak in the presence of amyloid is more easily distinguished from the background fluorescence of free dye in solution. To our knowledge, it is still unclear why ThT is not used to stain cells or tissues, but it may be related to its reduced binding specificity in a complex environment.

The binding mechanism of ThT to amyloids is not fully understood; Krebs et al. (Reference Krebs, Bromley and Donald2005) proposed that the ThT molecules intercalate perpendicularly to the backbone of aS in a β-sheet conformation and therefore are parallel to the long axes of protofilaments, protofibrils or fibrils. However, even though the binding between aS fibrils and ThT shows high specificity, no binding constants or evidence for the mode of binding aS fibrils have been reported, although the binding constants were estimated using absorbance spectroscopy data of ThT and fibril solution for the amyloids formed by lysozyme (i.e., 7.5 × 10⁶ and 5.6 × 10⁴ M⁻¹) (Sulatskaya et al. Reference Sulatskaya, Kuznetsova and Turoverov2011). A crystallographic structure of ThT bound to β-2 microglobulin amyloid-like aggregates allowed the definition of the photophysical properties of ThT when bound to β-sheets (Wolfe et al. Reference Wolfe, Calabrese, Nath, Blaho, Miranker and Xiong2010). The authors proposed that the fluorophore's electronic ground state distribution is stabilized sterically and electrostatically by β-sheets, that emission due to non-radiative relaxation is minimized, and that consequentially the quantum yield increases. However, it should be mentioned that despite its use in several studies to evaluate the kinetic properties of aS aggregation, emission measurements of ThT fluorescence often show variable emission efficiency and low reproducibility. These difficulties can only in part be ascribed to the intrinsic variability in the kinetics of the aS aggregation process. Indeed, it has been shown that ThT micelles can be present in solution and enhance fluorescence intensity (Khurana et al. Reference Khurana, Coleman, Ionescu-Zanetti, Carter, Krishna, Grover, Roy and Singh2005), even in the absence of the β-sheet structure. It has also been proposed that the surfaces responsible for the interaction with ThT can be shielded upon conversion of protofilaments to protofibrils and fibrils, causing a reduced ThT emission intensity due to decreased binding (Krebs et al. Reference Krebs, Bromley and Donald2005). These findings suggest that this method may not provide a precise determination of the kinetic properties of amyloid fibril formation, not only in the case of aS, in which part of the variability is due to aS, but also for the other amyloidogenic proteins.

Furthermore, the presence of other exogenous compounds, such as molecular chaperones or molecules designed to interfere with aS aggregation, can induce a significant bias in ThT intensity measurements (Hudson et al. Reference Hudson, Ecroyd, Kee and Carver2009), reducing the reliability of this method for the evaluation of aggregation enhancers or inhibitors. Finally, ThT is likely unable to properly bind β-sheets smaller than a fibrillar structure, preventing the detection of aggregation intermediates, which are, according to the latest hypothesis, the most toxic species in vivo (Winner et al. Reference Winner, Jappelli, Maji, Desplats, Boyer, Aigner, Hetzer, Loher, Vilar, Campioni, Tzitzilonis, Soragni, Jessberger, Mira, Consiglio, Pham, Masliah, Gage and Riek2011) (see Section 5.3).

The second most common extrinsic dye used to study amyloid fibrils formation is Congo Red. This dye, when observed through a polarized filter, shows birefringent emission upon binding to amyloid fibrils and has been used in particular to identify amyloids in post-mortem brain tissues. Moreover, the absorbance spectrum of fibril-bound Congo Red shows a red-shift compared with the free form of the dye (Conway et al. Reference Conway, Harper and Lansbury2000). Potential caveats are that Congo Red has been shown to induce oligomerization of native proteins and to bind to partially folded conformations of several proteins, not exclusively to amyloid fibrils (Khurana et al. Reference Khurana, Uversky, Nielsen and Fink2001). For these reasons, Congo Red uses in the study of the kinetic properties of the aS aggregation process or in the detection of amyloid fibrils in vitro has to be approached with care. Up-to-date reviews are available for most of the suitable dyes used to detect amyloid fibrils, and these describe in depth the binding modes and the possible applications of these assays for studying protein aggregation (Groenning, Reference Groenning2010).

However, the potential pitfalls in the use of extrinsic dyes to study the subsequent steps in the aggregation process and the limited reproducibility of data concerning aS fibril formation (Giehm et al. Reference Giehm, Lorenzen and Otzen2011a) propelled the development of alternative methods to evaluate aS oligomerization and fibril formation.

2.2 Monitoring aggregation through the intrinsic properties of aS

2.3 Fluorophores covalently bound to aS monomers

To study the dynamics of aS aggregation by monitoring variations in fluorescent properties, the protein has to be modified. In these paragraphs, we will survey studies in which aS has been covalently bound to specific fluorophores. The binding is usually done through thiol groups by insertion of a cysteine in a specific position in the aS sequence, exploiting the fact that aS does not have any cysteines in its native sequence. In most cases, the cysteine is added at the C-term to limit the effects that both the amino acid and the fluorophore could have on aS aggregation properties. Modification in other regions, i.e., the NAC, could be more informative about aS rearrangements but would also more likely to be invasive because this region is the site of the primary interactions that drive aS fibrillization.

2.4 Novel approaches

The approaches mentioned above are well established and provided important information about the aS aggregation process. However, most of these studies manifest some limitations: ThT experiments have limited reproducibility, evaluation of aS fluorescence variations requires modifications of the aS sequence, and most of the studies, for example, the ThT, DLS and absorbance variation experiments, do not allow the characterization of the whole aS aggregation process.

Therefore, investigators embarked on the exploration of alternative techniques to achieve a more thorough characterization of the kinetic properties of aS fibrillization. The favoured approaches are those capable of tackling the early stages of the process, which are the less known stages, and of providing several types of information during the same measurement.

In 2008, a new fluorescence aggregation assay, which provides information on oligomeric species and on amyloid fibrils, was developed (Thirunavukkuarasu et al. Reference Thirunavukkuarasu, Jares-Erijman and Jovin2008). Pyrene-labelled A19C, A90C and A140C aS mutants were monitored in an aggregation assay, and several fluorescence parameters were measured. This assay provides a wide range of information, including the detection of the early stages during aS aggregation process, the decrease of the monomer fraction and the increase of the molecular masses in solution can be easily detected. These parameters are also commonly measured using other probes, but a progressive decrease in the sample's polarity, an increase in the steady-state anisotropy and differences in intramolecular binding can be characterized at the same time, providing a wider set of information concerning the aS aggregation process.

More recently, Yushchenko et al. (Reference Yushchenko, Fauerbach, Thirunavukkuarasu, Jares-Erijman and Jovin2010) introduced a new method that is particularly sensitive to the early intermediates of the aggregation process. They measured excited-state intramolecular proton transfer (ESIPT) in the A140C aS mutant conjugated to a thiol-reactive (maleimide) probe (MFC), based on 2-(2-furyl)-3-hydroxychromone FC (aS-MFC). aS-MFC has two emission bands, the intensities of which were denoted T* and N*; the authors showed that the intensity ratio T*/N* increased upon aggregation and the T* band exhibited a red-shift, meaning that the probe was confined to a less protic and less polar environment. The aggregation reaction can be followed continuously from the early steps of the process using this technique. Moreover, ESIPT measurement of aS-MFC can provide a multiparameter set of data, which is more informative in comparison to other conventional assays.

Variations in the fluorescent properties of a dansyl probe were also used to study the aggregation of aS (Yap et al. Reference Yap, Pfefferkorn and Lee2011). The authors prepared four cysteine mutants (G7C, V26C, L100C and Y136C) on which to place the dansyl moiety in the aS sequence. The dansyl reagent is particularly suitable to study proteins in solution because it is sensitive to the local environment and can act as fluorescence energy transfer acceptor, i.e., it is able to provide a great deal of information by measuring the same fluorescence signal in a single experiment. 1.5% of dansyl-derivatized aS was used to characterize wt aS aggregation kinetic, by monitoring variations in dansyl emission intensity and light scattering. These data suggested that the first steps of oligomerization mainly involve the ends of the protein – the C- and N-term disordered regions.

A very reliable method to study aS aggregation kinetic was recently proposed by Cremades et al. (Cremades et al. Reference Cremades, Cohen, Deas, Abramov, Chen, Orte, Sandal, Clarke, Dunne, Aprile, Bertoncini, Wood, Knowles, Dobson and Klenerman2012). They used single-molecule two-colour coincidence detection (TCCD) and Forster resonance energy transfer (FRET) experiments to estimate the change in oligomer fraction during the aS aggregation process. aS was conjugated to two different fluorophores (Alexa Fluor 488 and 633) and a time series TCCD analysis was performed on different aliquots taken from the aggregation mix. The presence of oligomers was detected in correspondence of fluorescence emission bursts for both colours. Characterization of different oligomeric species belonging to different aggregation pathways was performed over time, leading to the identification of four lag phases and oligomerization rates for these species, before conversion into fibrillar species. This method, first demonstrated in 2006 by the same group (Orte et al. Reference Orte, Birkett, Clarke, Devlin, Dobson and Klenerman2008), is sensitive and able to detect oligomeric aggregation intermediates during the aggregation process, although it requires an experimental apparatus specifically designed for the collection of this type of data.

It was recently shown that aS oligomers and aggregates develop an intrinsic fluorescence emission that can be used in a quantitative assay to monitor amyloid fibril growth (Pinotsi et al. Reference Pinotsi, Buell, Dobson, Kaminski Schierle and Kaminski2013). The major advantage of this method is that it is label-free: monitoring the variations in the intrinsic emission allows the study of the kinetics of aggregation without the introduction of extrinsic dyes that can interfere with the process. The authors of the paper report that the effects of lacmoid on aS fibrillization were properly revealed by measuring the intrinsic fluorescence emission, whereas the measurement of ThT fluorescence led to the incorrect conclusion that lacmoid can hinder the aggregation process because lacmoid interacts with ThT and alters its fluorescence properties. Interestingly, this method could be extended to the characterization of other amyloidogenic proteins.

These different and novel methods could seem less approachable compared with the traditional ones, but they provide more refined information about the process studied. However, these methods do suffer from the drawback that few of the required experimental setups exist, hindering robust replication by multiple-independent laboratories.

2.5 Aggregation inhibitors and enhancers

Early on, aS fibrillization has been suggested as a possible drug target to slow or stop neurodegeneration, and it has been shown to be strongly affected by various factors and molecules that can either enhance or inhibit the process.

The fibril elongation rate was shown to be directly proportional to protein concentration (Wood et al. Reference Wood, Wypych, Steavenson, Louis, Citron and Biere1999), which may explain the pathogenicity associated with SNCA gene multiplication (Singleton et al. Reference Singleton, Farrer, Johnson, Singleton, Hague, Kachergus, Hulihan, Peuralinna, Dutra, Nussbaum, Lincoln, Crawley, Hanson, Maraganore, Adler, Cookson, Muenter, Baptista, Miller, Blancato, Hardy and Gwinn-Hardy2003) that leads to an increased dosage of aS.

The kinetic parameters of fibrillization for the pathological mutants were compared with those of the wt: A53T showed an increased rate of aggregation and a shorter lag phase, whereas A30P displayed a longer lag phase and a slower growth rate, leading to a delay in the midpoint of the fibrillization transition (Li et al. Reference Li, Uversky and Fink2001). In addition, the third mutant E46K, discovered after A30P and A53T, showed an increased fibrillization rate, similar to A53T (Fredenburg et al. Reference Fredenburg, Rospigliosi, Meray, Kessler, Lashuel, Eliezer and Lansbury2007), with a lag phase twice as long (∼8 h) compared with the wt one. Published data concerning the most recently discovered aS pathological mutants are not yet available, but the emerging picture implies that the aS mutants, more than the aggregation process per se, are the culprits in neurodegeneration.

Covalent chemical modifications of aS, such as nitration, oxidation (associated with an inflammatory response) and phosphorylation, lead to variations in both the rates and the final products of the aggregation process. Methionine oxidation reduces aS aggregation propensity in vitro: the presence of a 4-fold molar excess of oxidized protein (Uversky et al. Reference Uversky, Yamin, Souillac, Goers, Glaser and Fink2002) completely inhibits fibril formation. It has also been shown that the presence of a sub-stoichiometric concentration of nitrated aS in the aggregation assembly interferes with the fibrillization process (Yamin et al. Reference Yamin, Uversky and Fink2003). However, aS with tyrosine nitration has been detected in LB, suggesting that interfering with aggregation leads to the sequestration of the nitrated aS into the aggregates (Duda et al. Reference Duda, Giasson, Chen, Gur, Hurtig, Stern, Gollomp, Ischiropoulos, Lee and Trojanowski2000; Giasson et al. Reference Giasson, Duda, Murray, Chen, Souza, Hurtig, Ischiropoulos, Trojanowski and Lee2000).

In addition, the fraction of aS phosphorylated in position 87 (pSer87) and 129 (pSer129) is enhanced in LBs and in the brains of patients affected by synucleinopathies (Fujiwara et al. Reference Fujiwara, Hasegawa, Dohmae, Kawashima, Masliah, Goldberg, Shen, Takio and Iwatsubo2002; Paleologou et al. Reference Paleologou, Oueslati, Shakked, Rospigliosi, Kim, Lamberto, Fernandez, Schmid, Chegini, Gai, Chiappe, Moniatte, Schneider, Aebischer, Eliezer, Zweckstetter, Masliah and Lashuel2010). Of relevance to aggregation, it has been shown that pSer87, generated in vitro by casein kinase 1 (CK1) phosphorylation, inhibits aS fibrillization (Paleologou et al. Reference Paleologou, Oueslati, Shakked, Rospigliosi, Kim, Lamberto, Fernandez, Schmid, Chegini, Gai, Chiappe, Moniatte, Schneider, Aebischer, Eliezer, Zweckstetter, Masliah and Lashuel2010); similarly, Paleologou et al. demonstrated that CK1-mediated phosphorylation of aS hampers aS aggregation (Paleologou et al. Reference Paleologou, Schmid, Rospigliosi, Kim, Lamberto, Fredenburg, Lansbury, Fernandez, Eliezer, Zweckstetter and Lashuel2008), whreas Fujiwara et al. (Reference Fujiwara, Hasegawa, Dohmae, Kawashima, Masliah, Goldberg, Shen, Takio and Iwatsubo2002) reported that pSer129 aS, phosphorylated using casein kinase 2 (CK2), shows an increased fibrillization propensity. It is noteworthy that Ser129 is not in the core of aS fibrils and can also be phosphorylated after fibrillization in a physiological environment (Paleologou et al. Reference Paleologou, Schmid, Rospigliosi, Kim, Lamberto, Fredenburg, Lansbury, Fernandez, Eliezer, Zweckstetter and Lashuel2008). Interestingly, aS is phosphorylated at Ser129 not only by casein kinases but also by Polo-like kinase 1, 2 and 3, both in vitro and in vivo, with different phosphorylation efficiencies (Mbefo et al. Reference Mbefo, Paleologou, Boucharaba, Oueslati, Schell, Fournier, Olschewski, Yin, Zweckstetter, Masliah, Kahle, Hirling and Lashuel2010). However, the relationship between phosphorylation and aggregation is still an open question and needs clarification.

Divalent and trivalent metal ions (for example, Al³⁺, Cu²⁺, Fe³⁺ or Pb²⁺) have been shown to strongly accelerate the aS aggregation rate in vitro, although their relevance to the process in vivo remains to be proven (Bisaglia et al. Reference Bisaglia, Tessari, Mammi and Bubacco2009). Similarly, polyanions (Cohlberg et al. Reference Cohlberg, Li, Uversky and Fink2002; Liu et al. Reference Liu, Uversky, Munishkina, Fink, Halfter and Cole2005) or polycationic species (Goers et al. Reference Goers, Uversky and Fink2003) accelerate the aS aggregation rate in vitro. The effect of these compounds depends on their concentration and, in the case of polymers, also on their length, and suggests a critical role for electrostatic interactions in modifications of the aggregation rate. It has been proposed that the masking of negative charges at the C-term could lead to the collapse of aS to a more amyloidogenic conformation.

Other molecules, such as catechins or chaperones (Bieschke et al. Reference Bieschke, Russ, Friedrich, Ehrnhoefer, Wobst, Neugebauer and Wanker2010; Dedmon et al. Reference Dedmon, Christodoulou, Wilson and Dobson2005a), or properly designed peptides (El-Agnaf et al. Reference El-Agnaf, Paleologou, Greer, Abogrein, King, Salem, Fullwood, Benson, Hewitt, Ford, Martin, Harriott, Cookson and Allsop2004; Sievers et al. Reference Sievers, Karanicolas, Chang, Zhao, Jiang, Zirafi, Stevens, Munch, Baker and Eisenberg2011), can slow or inhibit aS aggregation by direct interaction with the monomer or with other aggregation intermediates.

2.6 Kinetic models

From the plethora of data available in the literature on the time dependence of aS (but also of other amyloid proteins) fibrillization in vitro, investigators attempted to develop kinetic models to describe and predict aS aggregation behaviour.

The proposed models are based on different types of assumptions and mathematical analysis (for an exhaustive review and the details of the mathematical treatments, see Morris et al. Reference Morris, Watzky and Finke2009), but none of them, to our knowledge, emerge as more significant than the others. This result is most likely due to variability in the data, which depend on the chosen experimental conditions, as well as to the complexity of the described process and the consequent complexity of the proposed models. Furthermore, the real test that could validate a kinetic model is its application to the data on aS aggregation in a cellular environment, a possibility that is hindered by the sparse experimental data on the process in live cells (see Section 6).

Each kinetic model provides a fitting of the data based on different assumptions about the mechanism that leads to oligomer and fibril formation in vitro, but the limiting step remains the intrinsically low reproducibility of aS aggregation kinetic data. In the sigmoid curve, the slope at t₅₀ and the length of the lag phase change significantly (lag phase length varies by two orders of magnitude) by varying the incubation conditions (Pronchik et al. Reference Pronchik, He, Giurleo and Talaga2010). In particular, agitation (Pronchik et al. Reference Pronchik, He, Giurleo and Talaga2010) and shear forces due to mechanical influences, such as centrifugation (Bhak et al. Reference Bhak, Lee, Hahn and Paik2009), have been identified as the most relevant parameters in the intrinsic variability that affect the aggregation process. Pronchik et al. performed aggregation experiments to determine a direct correlation between shaking rate and fibrillization rate. In the literature, three hypotheses have been proposed: (i) a faster fibrillization rate upon faster shaking is due to an increased rate of fibril fragmentation and the consequent generation of more active nucleation sites; (ii) vigorous mixing eliminates local concentration gradients and promotes fibril elongation, but is limited by the mass transfer rate; and (iii) in the presence of hydrophobic–hydrophilic interfaces, aS behaves as a surfactant, i.e., it concentrates, attempting to minimize the exposure of its hydrophobic residues to water solution by clustering, and acquires an aggregation-prone conformation. By aggregating aS under gentle agitation and mixing with beads at different concentrations and with different surface hydrophobicity, the authors observed that only the hydrophobic–hydrophilic interface hypothesis is consistent with their results: they showed how the concentration (the actual active surface area per aS molecule) and the type (polytetrafluorethylene, polymethylmethacrylate and glass) of interface directly correlate with aggregation rates. Based on their results, they propose that the irreproducibility often observed in aS kinetic data is due to the variability in heterogeneous air–water interfaces during agitation. Moreover, they suggested that aS partitioning at these interfaces causes an increase in local protein concentration and drives the equilibrium towards conformations that could more easily prompt amyloidogenesis.

Relatively high concentrations (more than 1 mg/ml ∼ 70 μM) and/or the presence of aS seeds can further enhance reproducibility (Giehm & Otzen, Reference Giehm and Otzen2010; Giehm et al. Reference Giehm, Lorenzen and Otzen2011a), but these parameters can act only from the second or third stage of the process (Fig. 2), when oligomers (or seeds) form and fibrils are growing in length. However, the main source of variability remains the first step of the aggregation when aS undergoes conformational rearrangements whose probability seems to be extremely low. For this reason, we suggest that the reproducibility issue cannot be completely overcome at this point, partially because of technical limits but also because of the intrinsic properties of aS.

Two recent papers explore the possibility of accelerating the aS fibrillization process by preparing chimaeric aS dimers (Pivato et al. Reference Pivato, De Franceschi, Tosatto, Frare, Kumar, Aioanei, Brucale, Tessari, Bisaglia, Samori, De Laureto and Bubacco2012) or by inducing dimerization in a mutated aS through an inducible dimerizing domain (Roostaee et al. Reference Roostaee, Beaudoin, Staskevicius and Roucou2013). In this second case, the induction of dimerization between the C-term of two aS molecules led to an increase in the aggregation rate compared with the wt aS.

In Pivato et al. the comparison between different aS dimers is reported: aS cys mutants were designed to form aS dimers bound through disulphide bonds between the two N- or C-termini (respectively, NN or CC dimers). Moreover, a tandem dimer, consisting of two aS molecules (NC dimer), and a double-core dimer, consisting of two NAC regions (DC dimer), were designed. Interestingly, CC and DC dimers showed faster aggregation kinetics when studied by ThT compared wyth the wt, whereas NN and NC dimers exhibited a negligible increase in ThT fluorescence intensity over time, indicating slower aggregation. These latest results suggest that it is not only the increase in the local concentration of aS molecules due to dimerization but also the chosen constraints in terms of the relative orientation of aS molecules in these dimers that strongly affect the kinetics of the aggregation process.

A final comment on the kinetics of aS aggregation is on the stirring issue: most of the literature regarding aS aggregation in vitro describes experiments performed while shaking. This consideration led to a question on the relevance of stirring for the aggregation process in cells and in PD models more generally. In our opinion, stirring should be seen as an efficient accelerator of the aggregation process and a price to pay to reduce the aggregation time frame to values experimentally approachable. It remains clear that from a physiological point of view, stirring has no relevance, but stirring in vitro allows experiments to fast forward through the years that aS seems to require to aggregate in parkinsonian brains.

3.1 Monomeric aS structural analysis

3.2 aS conformation and aggregation on membranes

Recently, a new aspect has been proposed to be relevant in the aggregation process: the accumulation of aS on membrane surfaces. When monomeric aS interacts with vesicles or membranes, the N-term of the protein acquires an α-helical conformation that can be a bent helix, an extended helix or an antiparallel helix-turn-helix structure, depending on the membrane properties. Conversely, the C-term, as amply demonstrated, remains unstructured (Borbat et al. Reference Borbat, Ramlall, Freed and Eliezer2006; Bussell et al. Reference Bussell, Ramlall and Eliezer2005; Davidson et al. Reference Davidson, Jonas, Clayton and George1998; Ferreon et al. Reference Ferreon, Gambin, Lemke and Deniz2009; Georgieva et al. Reference Georgieva, Ramlall, Borbat, Freed and Eliezer2010; Jao et al. Reference Jao, Hegde, Chen, Haworth and Langen2008; Middleton & Rhoades, Reference Middleton and Rhoades2010; Trexler & Rhoades, Reference Trexler and Rhoades2009; Ulmer et al. Reference Ulmer, Bax, Cole and Nussbaum2005). For amyloidogenic proteins, but not for aS, a direct conversion was observed between the α-helical conformation that most of the proteins acquire upon membranes binding (Butterfield & Lashuel, Reference Butterfield and Lashuel2010) and the β-sheet structures typical of amyloid oligomers and fibrils. Furthermore, an increase in fibril formation was observed when aS was aggregated in the presence of brain-derived membranes (Lee et al. Reference Lee, Choi and Lee2002) and anionic detergents or fatty acids micelles (Necula et al. Reference Necula, Chirita and Kuret2003), suggesting that membranes also play an important role in aS aggregation. Interestingly, it has been proposed that the binding of aS to the membrane increases the local concentration, favouring the triggering of the aggregation process (Auluck et al. Reference Auluck, Caraveo and Lindquist2010).

Reynolds et al.  (Reference Reynolds, Soragni, Rabe, Verdes, Liverani, Handschin, Riek and Seeger2011) showed by supercritical angle fluorescence microscopy (SAF) that aS in a sub-physiological concentration in solution tends to aggregate in the presence of negatively charged supported lipid bilayers. FRET images of donor-labelled (Dy647-N-hydroxysuccinimide) supported lipid bilayers and acceptor-labelled (Cy7-monofunctional NHS ester dye molecules) aS were performed by SAF microscopy, with the total protein–lipid ratio going from 1:21.6 (200 nM) to 1:1.44 (3 μM). Most of the protein remains in solution, and initially, the interaction between aS and the lipids led to the switch of the protein to the α-helical conformation previously mentioned. Later on, a structural rearrangement occurs, leading to the formation of aggregates on the lipid bilayer of a smaller part of the aS molecules, while the majority of the protein is still in solution. The aggregates were also observed by cryo scanning electron microscopy, showing a size varying from a few micrometres to tens of micrometres and displaying a morphology typical of amyloid fibrils. It seems that the ability of aS to fulfil its physiological function(s) at the membranes depends on the protein–lipid ratio and on the type of lipids. It could be that under pathological conditions, aS binding to the membrane is no longer protective and leads to aS aggregation on the membranes, which can damage membranes through different possible mechanisms. The aS aggregation described here is time and concentration dependent: at nM concentration, aS aggregates led to membrane damage, destabilization, thinning and, more unexpectedly, lipid extraction from the membrane, while at μM concentration, disruption of the bilayer occurred. A solid-state NMR (ss-NMR) study on aS aggregation in the presence or absence of anionic phospholipids showed that fibrils are formed in both cases, but when lipids are present, the fibril structure exhibits major structural perturbations at the N-term (Comellas et al. Reference Comellas, Lemkau, Zhou, George and Rienstra2012). Therefore, the authors proposed a model for aS aggregation in the two different cases, accounting for the revealed structural differences.

The increase in the local concentration of aS at the membranes can induce not only increased aggregation but can also trigger membrane permeabilization induced by monomeric aS that assembles into oligomers at the membranes upon the application of a voltage (Tosatto et al. Reference Tosatto, Andrighetti, Plotegher, Antonini, Tessari, Ricci, Bubacco and Dalla Serra2012). These oligomers are ordered and show a recursive structure similar to the pores formed by some bacterial toxins. The relationship between these objects and the previously mentioned aggregates formed at the membranes and observed by other groups remains unclear.

4.1 Physical methods for obtaining aS oligomers

Several physical methods have been applied to aS oligomer ‘production’. In a first case, lyophilization of recombinant proteins after purification and their subsequent resuspension led to the formation of stable aS oligomers that were analysed by different biochemical and biophysical methods (Apetri et al. Reference Apetri, Maiti, Zagorski, Carey and Anderson2006; Lashuel et al. Reference Lashuel, Petre, Wall, Simon, Nowak, Walz and Lansbury2002). Fragmentation of aS fibrils by supercooling was also used to obtain an enriched preparation of aS oligomers with electrophysiological activity on artificial membranes and whose structural features were extensively characterized in two papers (Kim et al. Reference Kim, Cho, Riedel, Fernandez and Zweckstetter2008, Reference Kim, Cho, Kumar, Maier, Siebenhaar, Becker, Fernandez, Lashuel, Benz, Lange and Zweckstetter2009). In particular, oligomer morphologies were characterized by AFM and TEM, whereas their secondary structure, a β-sheet non-fibrillar structure, was studied by solid-state NMR.

The sonication of recombinant aS fibrils results in the formation of so-called ‘seeds’. Although seeds are heterogeneous objects of variable size, they turned out to be an extremely useful tool in aggregation experiments in which the introduction of seeds as preformed nucleation centres into the aggregation mix led to more reproducible results (Giehm & Otzen, Reference Giehm and Otzen2010; Wood et al. Reference Wood, Wypych, Steavenson, Louis, Citron and Biere1999). Moreover, aS seeds have been used to trigger aS aggregation in the cellular compartments using cell lines or primary neuronal cultures (Luk et al. Reference Luk, Song, O'Brien, Stieber, Branch, Brunden, Trojanowski and Lee2009; Nonaka et al. Reference Nonaka, Watanabe, Iwatsubo and Hasegawa2010; Volpicelli-Daley et al. Reference Volpicelli-Daley, Luk, Patel, Tanik, Riddle, Stieber, Meaney, Trojanowski and Lee2011; Waxman & Giasson, Reference Waxman and Giasson2010), as will be discussed more in detail in Section 6.1.

4.2 Chemical oligomers

The presence of specific chemical compounds in the aggregation mix was shown to influence the aS aggregation process, leading to the formation of off-pathway oligomerization products through the formation of covalent bonds. Among these products, particularly interesting are the oligomers formed in the presence of docosahexaenoic acid (DHA) (De Franceschi et al. Reference De Franceschi, Frare, Pivato, Relini, Penco, Greggio, Bubacco, Fontana and De Laureto2011) or 4-hydroxy-2-nonenal (HNE) (Nasstrom et al. Reference Nasstrom, Fagerqvist, Barbu, Karlsson, Nikolajeff, Kasrayan, Ekberg, Lannfelt, Ingelsson and Bergstrom2011; Qin et al. Reference Qin, Hu, Han, Reaney, Di Monte and Fink2007b). DHA is a fatty acid present in high abundance in the brain regions where LBs are present, whereas HNE is a lipid peroxidation product that can modify proteins, most likely by forming adducts or aggregates in parkinsonian brains (Yoritaka et al. Reference Yoritaka, Hattori, Uchida, Tanaka, Stadtman and Mizuno1996). De Franceschi et al. obtained covalent oligomers after incubation of aS with DHA at the molar ratio of 1:50; these oligomers are characterized by annular and spherical morphologies, as detected by AFM and TEM. Moreover, FTIR and circular dichroism (CD) analysis showed a high content of α-helical and random structures but almost no β-sheet structure, suggesting that these oligomers are not on-pathway intermediate aggregates of mature amyloid fibrils. DHA can therefore modulate the properties of the covalent aS oligomers, which were shown to be toxic to SH-SY5Y neuroblastoma cells (De Franceschi et al. Reference De Franceschi, Frare, Pivato, Relini, Penco, Greggio, Bubacco, Fontana and De Laureto2011).

HNE–aS oligomers also showed an annular morphology, as detected by AFM (inner diameter 30–50 nm, outer diameter 80–100 nm), but also a curved protofibrillar structure 100–200 nm long and 2–4 nm high. However, in contrast to DHA–aS oligomers, their secondary structure is characterized by a shift in the CD spectra to wavelengths typical of a β-sheet conformation (Nasstrom et al. Reference Nasstrom, Fagerqvist, Barbu, Karlsson, Nikolajeff, Kasrayan, Ekberg, Lannfelt, Ingelsson and Bergstrom2011). HNE–aS oligomers are internalized by SH-SY5Y, which show a reduced viability induced by the oligomers’ toxicity.

Another physiologically relevant molecule responsible for the formation of covalent oligomers of aS is DA. DA is particularly important in PD pathogenesis because the prominent death of dopaminergic neurons in parkinsonian brains suggests that this neurotransmitter may play a key role in the toxic events that lead to cell death and neurodegeneration. Bisaglia et al. showed that incubation of aS and DA leads to the formation of aS oligomeric species that are stable when separated in a sodium dodecyl sulphate (SDS)–PAGE (Bisaglia et al. Reference Bisaglia, Mammi and Bubacco2007). The structure of aS/DA oligomers was also studied by groups (Cappai et al. Reference Cappai, Leck, Tew, Williamson, Smith, Galatis, Sharples, Curtain, Ali, Cherny, Culvenor, Bottomley, Masters, Barnham and Hill2005; Rekas et al. Reference Rekas, Knott, Sokolova, Barnham, Perez, Masters, Drew, Cappai, Curtain and Pham2010) using SAXS and CD; size-exclusion chromatography allowed separation of aS/DA dimers, trimers and higher molecular weight oligomers. Dimers and trimers showed an increased content of β-sheets and turns, while SAXS data permitted the elaboration of a low-resolution model for the aS/DA oligomers shape. It is worth noting that 3,4-dihydroxyphenylacetaldehyde, i.e., the monoamine oxidase metabolic product of DA, can form SDS insoluble oligomeric adducts with aS (Burke et al. Reference Burke, Kumar, Pandey, Panneton, Gan, Franko, O'Dell, Li, Pan, Chung and Galvin2008).

Interestingly, the presence of metal ions in the aggregation assay also induces the formation of SDS-resistant aS oligomers, whose main characteristic is again their annular or spherical structure (Kostka et al. Reference Kostka, Hogen, Danzer, Levin, Habeck, Wirth, Wagner, Glabe, Finger, Heinzelmann, Garidel, Duan, Ross, Kretzschmar and Giese2008; Lowe et al. Reference Lowe, Pountney, Jensen, Gai and Voelcker2004). The type of metal ion appears to affect the dimensions and shape of the aS oligomers formed: spherical particles (0.4–8 nm) are obtained from aS in the presence of copper(II), iron(III) and nickel(II); and larger spherical oligomers (5–8 nm) result from the aggregation of aS with magnesium(II), cadmium(II) and zinc(II). Moreover, calcium(II) and cadmium(II) lead to the formation of annular oligomers with diameters, respectively, of 70–90 and 22–30 nm (Lowe et al. Reference Lowe, Pountney, Jensen, Gai and Voelcker2004). To summarize, aS oligomeric species generated in vitro have to be taken cum grano salis. The numerous studies reported above suggest that the heterogeneity observed can hardly be ascribed only to the intrinsic heterogeneity of the aS aggregation products and is most likely strongly affected by the experimental conditions under which aggregation assays are performed. On this ground, it appears difficult to firmly ascribe a pathological function to these oligomers without robust proof of their presence and activity in PD cell or animal models or, even better, in parkinsonian brains.

The key future challenge for PD researchers is to conceive of an experimental method to observe these oligomers in action within the cell to determine which among the several aS oligomeric species are truly relevant for the pathology.

5.1 Fibril structure

5.1.2 NMR and EPR to probe aS fibril structure

Among the aS aggregation products, amyloid fibrils are the best characterized, being stable in solution and easy to isolate. They share several structural characteristics with amyloid fibrils formed by other aggregating proteins (Dobson, Reference Dobson1999) linked to neurodegenerative disorders, such as AD, Huntington's disease and amyotrophic lateral sclerosis (Ross & Poirier, Reference Ross and Poirier2004).

A first attempt to define the structure of fibrillar aS was accomplished by X-ray diffraction (Serpell et al. Reference Serpell, Berriman, Jakes, Goedert and Crowther2000). Serpell et al. measured X-ray diffraction patterns of dried fibrils obtained from wt aS, from a deletion mutant consisting of the first 120 residues, and from the pathological mutants A30P and A53T. All samples showed a ring shape at a spacing of ∼0.47 nm, and the wt also showed a weak diffraction at ∼1.0–1.1 nm; the X-ray data suggest a β-sheet structure for all these fibrils, with the β-strands perpendicular to the fibril axis.

More resolved structural information has been acquired by nuclear and electron magnetic resonance techniques, which were applied to the characterization of the aS fibril structure. The main goal was to define both the structures of the fibril core and of the N- and C-terms to ultimately define a model for aS amyloid fibrils.

The structural features of aS fibrils were first studied by site-specific spin labelling and by electronic paramagnetic resonance (EPR) (Der-Sarkissian et al. Reference Der-Sarkissian, Jao, Chen and Langen2003). This work was based on a nitroxide spin label, used as reporter molecule, whose structural environment and proximity to other reporter molecules can be deduced from the modulation of its EPR properties. EPR spectral analysis allowed the determination of loops, buried sites and secondary structure elements in amyloid fibrils. The data suggested that aS fibrils have a core region (including not only the NAC domain) organized in an ordered, in-register, parallel structure, consisting of 68 amino acids from residues 34 to 101. The N-term region showed a more heterogeneous and less ordered structure, while the C-term was described as completely unfolded.

In an ss-NMR study (Heise et al. Reference Heise, Hoyer, Becker, Andronesi, Riedel and Baldus2005), the authors used correlation experiments, reverse isotope labelling and novel strategies for the characterization of the aS fibril structure. The residue-specific resonance assignments of 48 amino acids in the aS core region (from residues 38 to 95) were obtained using two-dimensional (2D) and 3D correlation experiments. These data also allowed determination of the structure of the β-sheet core region, constituted in the proposed model by six β-strands. Within this core region, a turn and a loop were identified (at residues 66–68 and 83–86). Moreover, the authors found, in accordance with the previous EPR study (Der-Sarkissian et al. Reference Der-Sarkissian, Jao, Chen and Langen2003), that the N-term starting from residue 22 is quite rigid, while the C-term is mobile and has no secondary structure. Through the differences in ¹³C and ¹⁵N shifts found for the amino acids between E83 and I88 and for the side-chain atoms, Heise et al. identified two different types of fibrils: a relatively ordered form and a disordered one (Heise et al. Reference Heise, Hoyer, Becker, Andronesi, Riedel and Baldus2005). The parallel TEM imaging experiments performed suggested that the differences in the structure detected by ss-NMR correspond to at least two different fibril morphologies (see Section 4.2).

An EPR study was performed on fibrils obtained from 83 different point mutation spin-labelled derivatives to define more precisely the fibril boundaries of the core region and to determine its secondary structure elements (Chen et al. Reference Chen, Margittai, Chen and Langen2007). Consistent with previous work, Chen et al. found a core packed region that extended from amino acids 36 to 98 and, within this region, some mobile and flexible areas. Moreover, they confirmed the fibrils’ parallel arrangement, in which the β-strand planes are perpendicular to the fibril axis.

In 2008 Vilar et al. (Reference Vilar, Chou, Luhrs, Maji, Riek-Loher, Verel, Manning, Stahlberg and Riek2008) presented new NMR experiments on aS fibrils. Hydrogen–deuterium (H/D) exchange measurements suggested, as expected, that most of the residues from 30 to 110 were not exposed to the solvent. Moreover, H/D exchange showed biphasic behaviour consisting of two exchange rates, consistent with the presence of two distinct conformations in the sample, as previously reported (Heise et al. Reference Heise, Hoyer, Becker, Andronesi, Riedel and Baldus2005). Residues 1–27 and 104–140 appeared flexible and showed no secondary structure, as indicated by their fast exchange rates. ss-NMR experiments on ¹⁵N,¹³C-labelled aS fibrils allowed the assignment of 70% of the residues in the core region (35–95), but no heterogeneity was detectable with this technique in the samples. Based on ss-NMR and EPR experiments performed by other groups and on their own work, Vilar et al. proposed a new structural model, in which the core of the fibrils (from residues 30 to 110) is a β-sandwich consisting of five β-strands with different possible packings and solvent exposure, while the N- and C- terms are unstructured. This model also accounts for the data on the proteolysis resistance of the fibrils in solution (Pivato et al. Reference Pivato, De Franceschi, Tosatto, Frare, Kumar, Aioanei, Brucale, Tessari, Bisaglia, Samori, De Laureto and Bubacco2012; Qin et al. Reference Qin, Hu, Han, Hong and Fink2007a).

In a recent work, Comellas et al. by assigning via ss-NMR 106 residues, versus the 62 residues presented in a previous ss-NMR study (Heise et al. Reference Heise, Hoyer, Becker, Andronesi, Riedel and Baldus2005), demonstrated that the core of the fibril goes from residues 38–96 (Comellas et al. Reference Comellas, Lemkau, Nieuwkoop, Kloepper, Ladror, Ebisu, Woods, Lipton, George and Rienstra2011). The novelty of their model is due to the identification of a motif repeated twice in the fibrils’ secondary structure, consisting of a long β-strand and two short β-strands, from residues 38 to 66 and from 68 to 96. Moreover, the authors could identify structural features in both the N- and C- terms (Comellas et al. Reference Comellas, Lemkau, Nieuwkoop, Kloepper, Ladror, Ebisu, Woods, Lipton, George and Rienstra2011), in disagreement with the previous NMR and EPR works, in which these parts of the protein were described as completely unfolded. More recently, the same research group published a comparison between each pathological mutant and wt aS fibrils in terms of structural features, based on an ss-NMR study. The authors showed that, despite the slower aggregation rate of A30P mutants, the resultant amyloid fibrils have chemical shifts and secondary structure similar to the wt aS (Lemkau et al. Reference Lemkau, Comellas, Kloepper, Woods, George and Rienstra2012), whereas the structures of A53T and E46K fibrils show some small differences around the mutated region and significant variations compared with the wt aS (Lemkau et al. Reference Lemkau, Comellas, Lee, Rikardsen, Woods, George and Rienstra2013). Given these latest results, the mutants’ pathogenicity may be attributable not only to the properties of the aggregation intermediates but also to the characteristics of the final products of the aggregation process.

Over the past decade, structural data has provided a satisfactory description of the aS fibrillar core and allowed the elaboration of models that describes the aS monomer structure within the fibril (Fig. 3).

Fig. 3.

Schematic representation of the arrangement of an individual aS molecule within a fibril, following the results obtained by Comellas et al. by solid-state NMR. N-term and C-term do not present a define structure, but still present some structural features, in disagremeent with part of the previous works. Conversly, the central part of the protein is folded into six β-strands: solid-state NMR allowed also the identification a repeated motif, constituted by one long and two short β-strands (from amino acids 38 to 66 and from 68 to 96). Structural features (turns, loops and β-strands) and the associated number of residues are reported in the schema. Fibril axis is perpendicular to the plane that contains the folded protein (Comellas et al. Reference Comellas, Lemkau, Nieuwkoop, Kloepper, Ladror, Ebisu, Woods, Lipton, George and Rienstra2011).

The latest model clearly suggests that the pathological mutations E46K and A53T are structured, while the missense mutation A30P is located in an unstructured loop; moreover, the N- and C-terms are not completely unfolded, as hypothesized in the first studies, but present some minor structural elements that could be crucial, as we will see in the next section, for the formation of mature fibrils.

5.2 Fibril imaging

TEM and AFM are the main imaging techniques used to study aS fibrils and to determine their morphology, in particular, their mean diameters, mean lengths and periodicity. Optical imaging, even in its more recent developments, such as stimulated emission depletion microscopy (STED), still has not achieved a sufficient resolution compared with the actual aS fibril dimensions.

5.3 Fibril modelling

Taking into account most of the structural data on aS fibrils present in the literature, it has been possible to recapitulate the results in a structural schema representing the most probable folding of aS molecules within the fibrils (Fig. 3), which is in good agreement with the putative aS β-sheet propensity calculated from aS primary sequence (Fig. 2) (Suk et al. Reference Suk, Lokappa and Ulmer2010; Zibaee et al. Reference Zibaee, Jakes, Fraser, Serpell, Crowther and Goedert2007). Moreover, the assembly of aS molecules in fibrils was proposed to be a general hierarchical mechanism, from protofilaments to protofibrils to fibrils.

However, the recognition of polymorphisms in fibrils samples by NMR, EPR and microscopy techniques and the differences in the quantitative analysis on fibril diameters, even when performed with the very same technique (Sections 4.1 and 4.2), suggest that even small changes in the general aggregation conditions and in the sample preparation can lead to large variations in the results obtained.

In terms of differences between the fibrils obtained in vitro from wt aS and from the pathological mutants, it seems that there are some structural differences around the mutation regions, at least for A53T and E46K, while morphological or dimensional differences can be found for all of the mutants.

An interesting observation arose from an aggregation assay performed with chimaeric aS mutants, in which β-sheet-prone regions in the aS sequence were shuffled (Suk et al. Reference Suk, Lokappa and Ulmer2010). The authors showed that even after scrambling the aS sequence, some of the mutants, but not all, were still able to form fibrils. This result suggested that the formation of aS fibrils is associated with the possible spatial rearrangements of the β-sheets composing the structure. Moreover, the formation of the initial nucleation centre is affected by the presence of gaps between the different β-sheet regions.

Our opinion on this issue is that many of the structural and morphological differences among fibrils reported in the literature may not only depend on the different aggregation intermediates, including protofilaments, protofibrils and fibrils (see Section 3.2), present in the same sample but may also be, for the multiple amyloidogenic mechanisms of fibril formation, strongly affected by aggregating conditions that are still uncharacterized but that can result in different final aggregation products. Interestingly, all pathological mutations in the aS sequence, excluding A30P, are located in the core region of the fibril, suggesting that they can have some consequences not only in the assembly of aS fibrils but also in the formation of the oligomeric species.

If even small differences in the protofibrillar and fibrillar aggregation products could be relevant to their potential effects in vivo, a fine and detailed characterization of even the tiniest dissimilarities is pivotal for designing the most appropriate therapeutic strategy.

Moreover, an in-depth characterization of aS fibrils purified from LBs extracted from parkinsonian brains or from cellular models for aS fibril formation is still lacking, although TEM images of aS fibrils have been available since the discovery of aS presence in LBs (Spillantini et al. Reference Spillantini, Crowther, Jakes, Hasegawa and Goedert1998). These data will allow the exploitation and testing of the knowledge acquired through in vitro studies in physiologically relevant models.

6.1 Aggregation kinetics and aggregate properties in cells

Several studies employ immunocytochemistry and the conventional fluorescence microscopy technique to identify aS aggregation in cells. These assays require cells transiently transfected with aS or stable cell lines overexpressing aS. Moreover, aS aggregation in cells is triggered by exogenous agents, such as recombinant aS seeds or oxidative stress (Luk et al. Reference Luk, Song, O'Brien, Stieber, Branch, Brunden, Trojanowski and Lee2009; Nonaka et al. Reference Nonaka, Watanabe, Iwatsubo and Hasegawa2010; Roberti et al. Reference Roberti, Bertoncini, Klement, Jares-Erijman and Jovin2007; Waxman & Giasson, Reference Waxman and Giasson2010). A few days after treatment (from 2 to 7 depending on the experimental conditions), cells are fixed and immunostaining is performed to visualize aS with fluorescence microscopy.

6.2 aS aggregates toxicity

As gene dosage of SNCA is linked to pathology and aS is deposited in LBs, it seems reasonable that point mutations in SNCA act through a gain of function rather than through a loss of function.

Several different toxic mechanisms have been correlated to the aS aggregation process and to its oligomeric and fibrillar products. Given the heterogeneity of the aggregation products, it is reasonable to expect that cytotoxicity may occur through different mechanisms.

aS could be defined as a polymorphic protein, plastically changing its conformation, secondary structure and interactors (proteins or lipids). Given this premise, it is not surprising to find a large variety of noxious effects ascribed to aS in its different observed conformations.

Among the most studied toxicity mechanisms ascribed to aS is the ability of monomeric or oligomeric aS to interact in a disruptive way with lipid membranes (Reynolds et al. Reference Reynolds, Soragni, Rabe, Verdes, Liverani, Handschin, Riek and Seeger2011) or to form well-characterized pores similar to those observed in the presence of pore-forming toxins (Kim et al. Reference Kim, Cho, Kumar, Maier, Siebenhaar, Becker, Fernandez, Lashuel, Benz, Lange and Zweckstetter2009; Tosatto et al. Reference Tosatto, Andrighetti, Plotegher, Antonini, Tessari, Ricci, Bubacco and Dalla Serra2012; Zakharov et al. Reference Zakharov, Hulleman, Dutseva, Antonenko, Rochet and Cramer2007). To explore the pore-forming capacity of aS in its various forms in intact cells, aS was overexpressed in both SH-SY5Y and HEK293 cells, in which aS can aggregate. Specifically, aggregate-like aS species observed in SH-SY5Y by fluorescence microscopy were localized in part of the cell membrane, with a concomitant increase in conductance (Feng et al. Reference Feng, Federoff, Vicini and Maguire-Zeiss2010). Increased conductance was also independently found in HEK293T cells overexpressing aS (Tsigelny et al. Reference Tsigelny, Bar-On, Sharikov, Crews, Hashimoto, Miller, Keller, Platoshyn, Yuan and Masliah2007). Based on these convergent observations, and in agreement with in vitro studies (Kim et al. Reference Kim, Cho, Kumar, Maier, Siebenhaar, Becker, Fernandez, Lashuel, Benz, Lange and Zweckstetter2009; Schmidt et al. Reference Schmidt, Levin, Kamp, Kretzschmar, Giese and Botzel2012; van Rooijen et al. Reference Van Rooijen, Claessens and Subramaniam2009a), increased membrane conductance could be predicted to result in ion leakage with expected deleterious effects. Formation of membrane pores may support Na⁺ diffusion from the extracellular environment to the cytoplasm, driven by its favourable electrochemical potential, with consequent unwanted electrical responses in neuronal cells. Ca²⁺ may also diffuse into the cytoplasm from the extracellular space, causing cytotoxicity. Furthermore, under the assumption that aS oligomers induce an increase in plasma membrane conductance, it is reasonable to predict that they could also damage intracellular membranes, i.e., vesicles containing neurotransmitters (Mosharov et al. Reference Mosharov, Staal, Bove, Prou, Hananiya, Markov, Poulsen, Larsen, Moore, Troyer, Edwards, Przedborski and Sulzer2006) and mitochondrial membranes. The latter have been independently reported to be damaged by aS overexpression. Indeed, aS overexpression in HeLa cells induces mitochondrial fission and variation in mitochondria morphology (Nakamura et al. Reference Nakamura, Nemani, Azarbal, Skibinski, Levy, Egami, Munishkina, Zhang, Gardner, Wakabayashi, Sesaki, Cheng, Finkbeiner, Nussbaum, Masliah and Edwards2011). The authors could not verify that the mitochondrial-associated aS was oligomeric, but they were able to demonstrate aS oligomer interaction with mitochondrial-like membranes in vitro. More to the point, aS overexpression in PC12 cells (a pheocytochroma-derived dopaminergic cell line) leads to mitochondrial depolarization and subsequent mitochondrial-mediated cell death (Smith et al. Reference Smith, Jiang, Pei, Tanaka, Morita, Sawa, Dawson, Dawson and Ross2005; Tanaka et al. Reference Tanaka, Engelender, Igarashi, Rao, Wanner, Tanzi, Sawa, Dawson, Dawson and Ross2001). Although a few studies suggest the formation of oligomeric aS pores in vitro, their existence in vivo still awaits direct experimental evidence.

The presence of toxic aS forms affects the physiology not only of membranes but also of other cellular structures. In this respect, aS overexpression was shown to induce endoplasmatic reticulum (ER) stress in PC12 cells (Smith et al. Reference Smith, Jiang, Pei, Tanaka, Morita, Sawa, Dawson, Dawson and Ross2005) and Golgi fragmentation in COS-7 cells (Gosavi et al. Reference Gosavi, Lee, Lee, Patel and Lee2002). The observed ER stress induced by aS was further confirmed by the finding of aS oligomers within the ER in an A53T transgenic mouse model of PD (Colla et al. Reference Colla, Jensen, Pletnikova, Troncoso, Glabe and Lee2012).

Both the ubiquitin–proteasome system and the autophagic lysosomal pathway have been suggested as targets of aS aggregation products. These systems are involved in the clearance of monomeric, oligomeric or aggregated aS, among other proteins, and have been shown to be impaired in cell and animal models of aS aggregation, possibly because they are saturated by an excessive amount of misfolded protein. However, the literature regarding this field is wide and sometimes controversial (Cuervo et al. Reference Cuervo, Stefanis, Fredenburg, Lansbury and Sulzer2004; Ebrahimi-Fakhari et al. Reference Ebrahimi-Fakhari, Wahlster and Mclean2012; Mak et al. Reference Mak, Mccormack, Manning-Bog, Cuervo and Di Monte2010; Plotegher & Civiero, Reference Plotegher and Civiero2012).

A related aS toxicity mechanism that has been recently proposed is the sequestration of essential proteins during the aggregation process. As proof of concept, this sequestration mechanism was shown for amyloid-forming chimaeric proteins in a recent paper (Olzscha et al. Reference Olzscha, Schermann, Woerner, Pinkert, Hecht, Tartaglia, Vendruscolo, Hayer-Hartl, Hartl and Vabulas2011). An analogous process was shown to occur for aS in primary neurons treated with seeds, in which synaptic proteins are sequestered into the aggregates (Volpicelli-Daley et al. Reference Volpicelli-Daley, Luk, Patel, Tanik, Riddle, Stieber, Meaney, Trojanowski and Lee2011), and can be predicted for other chaperone proteins present in cell cytoplasm. Molecular crowding induced by aS aggregates and the ability of aS to co-aggregate with other proteins or chaperones (Dedmon et al. Reference Dedmon, Christodoulou, Wilson and Dobson2005a; Xu et al. Reference Xu, Kao, Lee, Song, Jin and Yankner2002) found in LBs (Shirakashi et al. Reference Shirakashi, Kawamoto, Tomimoto, Takahashi and Ihara2006; Xia et al. Reference Xia, Liao, Cheng, Duong, Gearing, Lah, Levey and Peng2008) further substantiate this hypothesis.

Finally, it has been shown that neuroinflammatory mechanisms (Alvarez-Erviti et al. Reference Alvarez-Erviti, Couch, Richardson, Cooper and Wood2011a; Maguire-Zeiss & Federoff, Reference Maguire-Zeiss and Federoff2010), oxidative stress and ROS production (Cremades et al. Reference Cremades, Cohen, Deas, Abramov, Chen, Orte, Sandal, Clarke, Dunne, Aprile, Bertoncini, Wood, Knowles, Dobson and Klenerman2012; Freeman et al. Reference Freeman, Cedillos, Choyke, Lukic, Mcguire, Marvin, Burrage, Sudholt, Rana, O'Connor, Wiethoff and Campbell2013) can be induced by the presence of aS oligomers and fibrils.

Interestingly, Codolo et al. showed that recombinant aS fibrils trigger inflammation in human monocytes (Codolo et al. Reference Codolo, Plotegher, Pozzobon, Brucale, Tessari, Bubacco and De Bernard2013), likely through a pathway that is general for needle-shaped objects, such as amyloid beta fibrils (Halle et al. Reference Halle, Hornung, Petzold, Stewart, Monks, Reinheckel, Fitzgerald, Latz, Moore and Golenbock2008) as well as asbestos and carbon nanotubes (Palomaki et al. Reference Palomaki, Valimaki, Sund, Vippola, Clausen, Jensen, Savolainen, Matikainen and Alenius2011). In addition, a specific form of aS oligomer directly released from cells was shown to activate microglia (Kim et al. Reference Kim, Ho, Suk, You, Michael, Kang, Joong Lee, Masliah, Hwang, Lee and Lee2013), while other aS aggregates obtained from recombinant aS did not result in an inflammatory response. Interestingly, both papers suggest that the activation of this process is mediated by an interaction with the membrane protein toll-like receptor 2.

Although it is widely accepted that aS aggregation is toxic to cells, albeit that the nature of these toxic species is still controversial, another possibility is that aS aggregation instead causes a decrease in the available functional aS and a consequent loss of function, which leads, in turn, to cell death. aS has in fact been suggested to play an important role at the presynapse through interaction with the SNARE complex during vesicle exocytosis and neurotransmitter release (Burre et al. Reference Burre, Sharma, Tsetsenis, Buchman, Etherton and Sudhof2010; Nemani et al. Reference Nemani, Lu, Berge, Nakamura, Onoa, Lee, Chaudhry, Nicoll and Edwards2010). Indeed, a recent report showed that aS oligomerization is responsible for the inhibition of SNARE-mediated vesicles docking (Choi et al. Reference Choi, Choi, Kim, Yang, Lai, Kweon, Lee and Shin2013).

Figure 4 summarizes the most important toxic mechanisms related to aS aggregation in cells that could be addressed by biophysical methods. For example, organelles’ integrity in the presence of aS aggregates could be studied by TEM, FRET or other microscopy techniques (Nakamura et al. Reference Nakamura, Nemani, Wallender, Kaehlcke, Ott and Edwards2008, Reference Nakamura, Nemani, Azarbal, Skibinski, Levy, Egami, Munishkina, Zhang, Gardner, Wakabayashi, Sesaki, Cheng, Finkbeiner, Nussbaum, Masliah and Edwards2011), or electrophysiological studies could monitor membrane permeabilization due to aS oligomers (Kim et al. Reference Kim, Cho, Kumar, Maier, Siebenhaar, Becker, Fernandez, Lashuel, Benz, Lange and Zweckstetter2009). These major advancements in biophysics could provide new tools for addressing these open issues.

Fig. 4.

Summary of the most recurrent toxic mechanisms proposed related to aS aggregation and transmission: membrane damages (Feng et al. Reference Feng, Federoff, Vicini and Maguire-Zeiss2010; Mosharov et al. Reference Mosharov, Staal, Bove, Prou, Hananiya, Markov, Poulsen, Larsen, Moore, Troyer, Edwards, Przedborski and Sulzer2006), mitochondrial morphology alterations, dysfunctions and damages (Devi et al. Reference Devi, Raghavendran, Prabhu, Avadhani and Anandatheerthavarada2008; Nakamura et al. Reference Nakamura, Nemani, Azarbal, Skibinski, Levy, Egami, Munishkina, Zhang, Gardner, Wakabayashi, Sesaki, Cheng, Finkbeiner, Nussbaum, Masliah and Edwards2011; Smith et al. Reference Smith, Jiang, Pei, Tanaka, Morita, Sawa, Dawson, Dawson and Ross2005; Tanaka et al. Reference Tanaka, Engelender, Igarashi, Rao, Wanner, Tanzi, Sawa, Dawson, Dawson and Ross2001), ER stress (Smith et al. Reference Smith, Jiang, Pei, Tanaka, Morita, Sawa, Dawson, Dawson and Ross2005), oxidative stress and ROS production (Cremades et al. Reference Cremades, Cohen, Deas, Abramov, Chen, Orte, Sandal, Clarke, Dunne, Aprile, Bertoncini, Wood, Knowles, Dobson and Klenerman2012; Freeman et al. Reference Freeman, Cedillos, Choyke, Lukic, Mcguire, Marvin, Burrage, Sudholt, Rana, O'Connor, Wiethoff and Campbell2013) and clearance mechanisms impairment (Cuervo et al. Reference Cuervo, Stefanis, Fredenburg, Lansbury and Sulzer2004; Ebrahimi-Fakhari et al. Reference Ebrahimi-Fakhari, Wahlster and Mclean2012; Mak et al. Reference Mak, Mccormack, Manning-Bog, Cuervo and Di Monte2010). Among the aS aggregates transmission mechanism exosome transmission (Alvarez-Erviti et al. Reference Alvarez-Erviti, Seow, Schapira, Gardiner, Sargent, Wood and Cooper2011b; Danzer et al. Reference Danzer, Kranich, Ruf, Cagsal-Getkin, Winslow, Zhu, Vanderburg and Mclean2012), axonal transport (Freundt et al. Reference Freundt, Maynard, Clancy, Roy, Bousset, Sourigues, Covert, Melki, Kirkegaard and Brahic2012) and nanotubes transport (Gousset et al. Reference Gousset, Schiff, Langevin, Marijanovic, Caputo, Browman, Chenouard, De Chaumont, Martino, Enninga, Olivo-Marin, Mannel and Zurzolo2009) are reported.

A relevant question that remains unanswered in the etiopathogenesis of PD is when aS oligomerization or aggregation is the cause of toxicity and when toxicity is, instead, the result of upstream pathological events that affect aS physiological functionality and induce aS aggregation.

6.3 Misfolded aS transmission in cell models

Braak staging for PD indicates that LBs first appear in the olfactory bulb and that only later in disease progression do they form in other brain regions (Braak et al. Reference Braak, Del Tredici, Rub, De Vos, Jansen Steur and Braak2003). This result suggests that the transmission of misfolded or aggregated aS from unhealthy to healthy neurons could be the culprit for the spread of the pathology and the progression of the disease. This hypothesis gained further support from clinical evidence (Kordower et al. Reference Kordower, Chu, Hauser, Freeman and Olanow2008; Li et al. Reference Li, Englund, Holton, Soulet, Hagell, Lees, Lashley, Quinn, Rehncrona, Bjorklund, Widner, Revesz, Lindvall and Brundin2008) showing the presence of LBs in naïve neurons grafted in parkinsonian brains 10 years after surgery.

The evaluation of this hypothesis requires many of the experimental approaches that fall in the realm of biophysics. Several cellular studies have now confirmed that oligomeric aS can transfer from a donor to a receiving cell (Desplats et al. Reference Desplats, Lee, Bae, Patrick, Rockenstein, Crews, Spencer, Masliah and Lee2009; Hansen et al. Reference Hansen, Angot, Bergstrom, Steiner, Pieri, Paul, Outeiro, Melki, Kallunki, Fog, Li and Brundin2011). Although some results suggest that the transfer may occur via endocytosis, they did not provide any direct evidence.

The transmission of fragments of recombinant aS fibrils was found to be responsible for enhanced aS aggregation. The fragments serve as ‘seeds’ and trigger aS aggregation in the recipient cells (Nonaka et al. Reference Nonaka, Watanabe, Iwatsubo and Hasegawa2010), as evaluated by TEM and fluorescence microscopy imaging.

Volpicelli-Daley et al. (Reference Volpicelli-Daley, Luk, Patel, Tanik, Riddle, Stieber, Meaney, Trojanowski and Lee2011) further showed that pre-formed recombinant aS fibrils were taken up by cultured primary mouse neurons. These fibrils affected neural excitability and connectivity, as reported by calcium imaging of the hippocampal neuronal networks. Moreover, aS inclusion formation caused both the sequestration of some synaptic proteins and neuronal cell death. Volpicelli-Daley et al. further showed intracellular propagation of fibrils from soma to neurites and vice versa using microfluidic devices as precast templates to guide the growth of neurites.

In vivo fibril propagation was recently documented by immunohistological experiments: transmission from neuron to neuron in mouse brains after fibril injection induced the spread of PD pathology to untreated brain regions. Interestingly, these mice showed some of the symptoms that can be ascribed to PD (Luk et al. Reference Luk, Kehm, Carroll, Zhang, O'BRIEN, Trojanowski and Lee2012).

Despite these recent studies, the precise, and most likely multiple, mechanisms that cause aS spreading from neuron to neuron or aS uptake by cells from the extracellular space are not clear. Several different hypotheses have been proposed: (i) direct cell-to-cell transfer due to the increased protein concentration in the extracellular space, possibly caused by cell death; (ii) exosomes or vesicles that protect the protein from the action of proteolytic enzymes and carry it to recipient cells; (iii) molecular chaperones that mediate aS transmission; (iv) misfolded aS that moves through the slow movement of axonal transport; and (v) membrane nanotubes that interconnect cells over long-distances serving as highways for aS aggregates (Steiner et al. Reference Steiner, Angot and Brundin2011).

Hints about the possible mechanism came from two recent papers. The first proposed that overexpressed aS can be transmitted from transfected SH-SY5Y cells to normal SH-SY5Y cells by exosomes (Alvarez-Erviti et al. Reference Alvarez-Erviti, Seow, Schapira, Gardiner, Sargent, Wood and Cooper2011b; Danzer et al. Reference Danzer, Kranich, Ruf, Cagsal-Getkin, Winslow, Zhu, Vanderburg and Mclean2012). The second showed that fibrillar aS can be transferred along axons via anterograde axonal transport, with a velocity that was inversely proportional to the aggregate size (Freundt et al. Reference Freundt, Maynard, Clancy, Roy, Bousset, Sourigues, Covert, Melki, Kirkegaard and Brahic2012).

Among the other mechanisms, tunnelling nanotubes, which allow the movement of organelles and vesicles (Rustom et al. Reference Rustom, Saffrich, Markovic, Walther and Gerdes2004), seem to be promising candidates, mostly because of their proven involvement in prion protein spreading. Indeed, it has been shown, by time-lapse confocal imaging of live neurons, that tunnelling nanotubes are responsible for the transfer of endogenous and exogenous prion proteins confined in lipid vesicles (Gousset et al. Reference Gousset, Schiff, Langevin, Marijanovic, Caputo, Browman, Chenouard, De Chaumont, Martino, Enninga, Olivo-Marin, Mannel and Zurzolo2009).

This emerging interest is related not only to the mechanisms that govern the transfer of aS but also to the definition of those aS species that are specifically transferred. The latter represent the true key to understanding the pathological process as those species are likely to carry the structural information required to spark aS aggregation in healthy neurons. These unanswered questions are perfect candidates for future biophysical investigations