Familial and idiopathic Parkinson's disease (PD) is associated with the abnormal neuronal accumulation of α-synuclein (aS) leading to β-sheet-rich aggregates called Lewy Bodies (LBs). Moreover, single point mutation in aS gene and gene multiplication lead to autosomal dominant forms of PD. A connection between PD and the 14-3-3 chaperone-like proteins was recently proposed, based on the fact that some of the 14-3-3 isoforms can interact with genetic PD-associated proteins such as parkin, LRRK2 and aS and were found as components of LBs in human PD. In particular, a direct interaction between 14-3-3η and aS was reported when probed by co-immunoprecipitation from cell models, from parkinsonian brains and by surface plasmon resonance in vitro. However, the mechanisms through which 14-3-3η and aS interact in PD brains remain unclear. Herein, we show that while 14-3-3η is unable to bind monomeric aS, it interacts with aS oligomers which occur during the early stages of aS aggregation. This interaction diverts the aggregation process even when 14-3-3η is present in sub-stoichiometric amounts relative to aS. When aS level is overwhelmingly higher than that of 14-3-3η, the fibrillation process becomes a sequestration mechanism for 14-3-3η, undermining all processes governed by this protein. Using a panel of complementary techniques, we single out the stage of aggregation at which the aS/14-3-3η interaction occurs, characterize the products of the resulting processes, and show how the processes elucidated in vitro are relevant in cell models. Our findings constitute a first step in elucidating the molecular mechanism of aS/14-3-3η interaction and in understanding the critical aggregation step at which 14-3-3η has the potential to rescue aS-induced cellular toxicity.
Parkinson's disease (PD) is the second most common neurodegenerative disorder after Alzheimer's disease affecting ∼1–2% of the population over 65 years (1). Characteristics of this disease are the progressive death of the dopaminergic neurons in the substantia nigra pars compacta and the presence of protein inclusions, termed Lewy bodies (LBs) in surviving neurons (2). Although PD etiology is mainly idiopathic, ∼10% of PD cases are caused by mutations in single genes. Among the several genes associated with PD, mutations in the SNCA gene, which encodes for the protein α-synuclein (aS), cause autosomal dominant and late onset forms of PD. Specifically, aS single point mutations (A30P, E46K, H50Q, G51D and A53T) and SNCA gene duplication or triplication are all associated with familial and aggressive forms of PD. Moreover, polymorphisms around the SNCA locus lead to an increased risk of developing sporadic PD (3–6).
Gene mutations are not the only link between aS and PD; this protein is also the major component of LBs, in which it is mainly found in a β-sheet-rich fibrillar form (7). In its monomeric form, aS is a highly soluble intrinsically disordered protein (IDP, 8–11) predominantly localized in the presynaptic terminals of the central nervous system. aS can associate with membranes, acquiring an α-helical conformation upon interaction (12). Recent studies (13,14) proposed a tetrameric alpha-helical aS native structure under physiological conditions in the solution, which is still object of debate (15). While a complete description of aS physiological function is still not available, there is mounting evidence for aS playing a role in synaptic vesicles recycling and docking (16), neurotransmitter release (17) and SNARE-complex assembly (18).
The pathway of aS fibril formation has been characterized in vitro and it is similar to that of other known amyloid-forming proteins. In vitro aS aggregation proceeds through a nucleation-dependent mechanism, in which soluble prefibrillar oligomers are formed during an initial ‘lag’ phase (19,20). This phase is followed by the rapid formation of insoluble mature fibrils, which resemble those found in LBs (20,21). While the structure of aS amyloid fibrils has been extensively studied in vitro with several biochemical and biophysical techniques (8,22), aS oligomers, which were proposed to be the most toxic among aS misfolded and aggregated species (23), are poorly characterized, because of their transient nature and structural heterogeneity (24,25).
The study of aS aggregation in cell models is an arduous task, mainly because aS overexpression alone does not lead to the formation of easily detectable aggregates in cultured cells. Luk and collaborators recently developed a PD cellular model for aS aggregation (26), based on the introduction of aS fibrillar fragments (termed seeds) obtained by sonication of aS mature fibrils. When introduced in the cytoplasm of HEK293T cells overexpressing aS, seeds were competent to trigger the aggregation process (26).
Starting from the observation that aS oligomerization and aggregation are related to aS-induced toxicity in PD, several molecules were studied to verify their possible interference with the aS aggregation process. Specifically, molecular chaperones are now raising interest as modulators of the aggregation mechanisms that lead to neurodegeneration. Molecular chaperones have been proved to assist protein folding and to modulate protein-degradation pathways, both mechanisms associated with the pathogenesis of PD (27). Moreover, several studies showed that some molecular chaperones, i.e. heat-shock protein 70, β-crystallin and DJ-1, could specifically interfere with the aS aggregation process in vitro and in PD cellular models (28–33).
In this context, the possibility that the 14-3-3 family of chaperone-like proteins could come into play was explored. 14-3-3s are a family of highly conserved acidic proteins expressed in the cytoplasm of all eukaryotic cells and represent ∼1% of total soluble brain proteins. The human genome contains seven distinct 14-3-3 genes termed β, γ, ε, η, σ, τ and ζ, encoding for proteins which commonly form hetero- or homodimers (34). They are able to associate with several binding partners (35) and are involved in the regulation of a wide range of cellular processes, including signaling, cell cycle control, apoptosis, exocytosis, cytoskeletal rearrangements, transcription and enzyme activity (34).
The basis of molecular recognition by 14-3-3 proteins is a phosphorylated motif on their binding partners that defines this family of proteins as phosphoserine/threonine binding molecules (36). However, interactions between 14-3-3 proteins and the C-terminus of non-phosphorylated target proteins were also reported (37). The physiological significance of the different isoforms in the 14-3-3 family is still object of debate: it is not yet known whether they have distinct and specialized functions, or whether they are under control of temporal and tissue-specific regulation only (38). As molecular chaperones, 14-3-3 proteins have been suggested to play a role in PD. Isoforms γ, ε and η were suggested to be localized at the synaptic junction and bound to the synaptic membrane (35), as aS does. More importantly, some 14-3-3 isoforms have been associated with the PD-related proteins parkin, LRRK2 and aS (39,40).
The interrelation between 14-3-3 proteins and aS has been analyzed in several studies: the two proteins share sequence similarity (38); some 14-3-3 isoforms were found in LBs in human PD (34,41,42) and co-localized with aS in a A53T transgenic mouse model for PD (43). Co-immunoprecipitation of aS and 14-3-3 (β and ε isoforms) from rat brain homogenate showed that the two proteins can be associated in both cytosolic and membrane fractions (38). Soluble aS and 14-3-3 complexes (∼54–83 kDa) were also found in human primary neurons cell culture after aS overexpression (44). aS and the isoform η were co-immunoprecipitated from the substantia nigra of PD patients (40). Interestingly, the latter study reported that aS is able to sequester 14-3-3η from the parkin-14-3-3η complex, interfering with the regulation of parkin by the chaperone. Moreover, Yacoubian et al. (45) observed differential neuroprotective effects of the various isoforms of 14-3-3 in PD models. The only 14-3-3 isoform found associated with aS in parkinsonian brains is 14-3-3η. However, no evidence of a stable and strong interaction between 14-3-3η and aS has been reported in vitro (40), suggesting that particular modification(s) or condition(s) are needed for the interaction between the two proteins.
Herein, we provide a comprehensive study of the effects exerted by 14-3-3η on the aS amyloidogenesis process both in vitro and in cell models. We first show how the kinetics and the final products of the aS in vitro aggregation process are heavily influenced by the presence of 14-3-3η. We present evidence that, while 14-3-3η is unable to bind monomeric aS, it strongly interacts with the oligomeric aS aggregates which occur during the early aggregation stages of aS amyloidogenesis, drastically diverting the aggregation process even when present in sub-stoichiometric amounts relative to aS. However, when the level of aS is overwhelmingly higher than 14-3-3η, the fibrillation process becomes a sequestration mechanism for 14-3-3η. We finally show how the above in vitro processes are also relevant in cell model systems, investigating 14-3-3η overexpression rescue toxicity effects in aS aggregation cell models in relation to the stage of fibril formation.
Sub-stoichiometric amounts of 14-3-3η influence aS aggregation kinetics
The interrelation between 14-3-3 proteins and aS is strongly suggested by several reports in the literature. To explore the functional consequences of this interrelation we first assessed whether dimeric 14-3-3η is able to influence aS aggregation, using a fluorescence method described by Luk et al. (46). Briefly, both aS and 14-3-3η were covalently functionalized with an Oregon Green (OG) probe to produce their fluorescent forms aS* and 14-3-3η*. Progression of the in vitro amyloidogenesis was followed as variation of the fluorescence polarization (FP) value, which correlates with the dimension acquired by the aggregates being formed. The addition of a small fraction of either aS* or 14-3-3η* to their non-fluorescent counterparts did not affect their aggregation properties (46).
This method was first used to monitor the unperturbed aggregation behavior of aS in the absence of 14-3-3η, and of 14-3-3η in the absence of aS (Fig. 1). As expected, the time dependence of the FP signal measured for the aggregation experiments of the aS*/aS samples shows a characteristic sigmoidal increase (46). The latter occurs after an induction (‘lag’) phase of highly variable duration (Fig. 1, full squares). The same procedure was then applied to 14-3-3η-homodimer (14-3-3η) solutions containing 14-3-3η* homodimers, resulting in flat FP-versus-time traces lacking any detectable feature (Fig. 1, empty squares). Two types of FP-monitored aggregation experiments of [aS:14-3-3η homodimer] mixtures at a molar ratio of [4 : 1] were performed, only differing in the type of FP probe (either aS* or 14-3-3η*). Regardless of which probe was added to the aggregation mixture, the resulting FP traces showed no significant variation from the starting value (Fig. 1, empty circles for aS* and full diamonds for 14-3-3η*) suggesting that 14-3-3η inhibits aS aggregation.
The aggregation assays described above were repeated employing Thioflavin T (ThT) fluorescence monitoring methods, yielding compatible results. Samples only containing aS showed the rapid formation of ThT-positive aggregates after a lag phase of highly variable duration, whereas the presence of 14-3-3η resulted in a negligible increase in ThT fluorescence (Supplementary Material, Fig. S1).
14-3-3η Redirects the aS amyloidogenic pathway toward non-fibrillar end products
To assess the morphology of the end products of aS aggregation experiments with 14-3-3, aliquots were taken from the aggregation mixtures mentioned above at the end of each respective time courses and examined via TEM. As expected (8), the end products of aS aggregations in the absence of 14-3-3η were mature amyloid fibrils having lengths from approximately a few tens of nanometers to several microns (Fig. 2A). Conversely, the aggregation mixtures containing both aS and 14-3-3η in a 4 : 1 molar ratio were shorter and apparently thinner (Fig. 2B) sub-fibrillar aggregates (hence termed ‘Non-Fibrillar Products’, NFPs).
End products from the aggregation mixtures were also independently examined via AFM microscopy. The resulting images were then analyzed via a semi-automated procedure yielding a statistically robust morphometric characterization of the aggregates. AFM morphometry of the mature amyloid fibrils produced by aS in the absence of 14-3-3η yielded values which are in good agreement with existing literature (47,48). The average apparent diameter of mature aS fibrils deposited on mica and measured in air was found to be 7.3 ± 0.9 nm (Fig. 3A). An undulation was often detectable in the height profiles measured along the individual fibrillar axes. The profiles in which the periodicity was measurable were pooled and fitted with a sigmoidal function with a period of 118 nm. All the observed mature aS fibrils were largely straight, with very low local curvature values (Supplementary Material, Fig. S2).
When subjected to the same AFM-based morphometry measurements, the NFPs produced by aggregation mixtures containing aS and 14-3-3η homodimers at a stoichiometric ratio of 4 : 1 gave significantly different values. The apparent diameter of NFPs was measured to be 1.9 ± 0.6 nm (Fig. 3G), and they exhibited comparatively high local curvatures with respect to typical aS mature fibrils (Supplementary Material, Fig. S2). It is worthwhile mentioning that local curvature distributions obtained from AFM and TEM images of both mature aS fibrils and NFPs were essentially identical. When further increasing the amount of 14-3-3η in the aggregation assays, reaching aS:14-3-3η stoichiometric ratios of 1 : 1 or 1 : 4, the obtained NFPs show identical diameters compared with the NFPs obtained at a stoichiometric ratio of 4 : 1, suggesting that a saturation regime has been reached (Supplementary Material, Fig. S3).
To evaluate the effect of 14-3-3η on the aS saturation concentration during the aggregation reaction, we roughly evaluated proteins concentration at the beginning and at the end of the aggregation assay by loading the supernatant (obtained by a 2 h 20 000g centrifugation) of the reaction onto a sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS–PAGE). The resulting data, reported in Supplementary Material, Figure S4, show that the amount of soluble aS is lower at the end of the reaction suggesting that 14-3-3η is able to stabilize aS oligomeric species and co-aggregate with them.
14-3-3η is incorporated into the non-fibrillar aS products
The localization of 14-3-3η in the products of the 4 : 1 aS:14-3-3η homodimers aggregation was visualized via immuno-electron microscopy (immuno-TEM). Briefly, aliquots from the aggregation mixtures were deposited on EM grids and incubated first with a rabbit polyclonal primary antibody against 14-3-3η, then with a gold nanoparticle (NP)-coupled secondary antibody against rabbit IgG. In the resulting TEM images, the immunolabeled gold NPs are almost exclusively localized on NFPs (Fig. 4). The sequence homology between aS and 14-3-3 called for a control experiment performed by applying primary antibody against 14-3-3η to mature aS fibrils which showed almost no cross reactivity as witnessed by the absence of gold NPs in the recorded EM fields (Supplementary Material, Fig. S5B).
To further establish whether 14-3-3η was included within the NFPs, we analyzed the fluorescence emission properties of OG-labeled aS* and 14-3-3η* in the NFPs. Two independent 4 : 1 aS:14-3-3η aggregation experiments similar to those mentioned above were performed, one of which contained aS* as a fluorescent reporter, while the other contained 14-3-3η*. As expected, both aggregations produced exclusively NFPs end products, as verified by AFM imaging. The fluorescence emission spectra of aS* and 14-3-3η* were completely superimposable for the NFPs, and different from those of free monomeric aS* and free dimeric 14-3-3η* (Supplementary Material, Fig. S6A). It is also worth mentioning that the fluorescence emission spectra of aS* included in mature aS fibrils obtained in the absence of 14-3-3η were different from the spectra obtained from aS*-containing NFPs (Supplementary Material, Fig. S6B).
The secondary structure content of NFPs was also analyzed via circular dichroism (CD) spectroscopy. The CD spectra of Mixtures containing NFPs clearly showed a α-helical signal. Aggregated aS molecules are expected to mostly contribute to the beta component of the CD spectra as shown in the Supplementary Material, Figure S7A and previously reported for each of its described aggregation products (21). In the case of the NFPs, the residual α-helical signal is likely to be due to unreacted 14-3-3η, which maintains its native folding (Supplementary Material, Fig. S7B) and conceals the contribution of the NFPs aggregation products (Supplementary Material, Fig. S7C).
14-3-3η Does not interact with monomeric or fibrillar aS
The above results prompted the question of whether monomeric aS could interact with 14-3-3η as the first step in the processes that lead to the formation of NFPs. The ability of dimeric 14-3-3η to form stable complexes with monomeric aS was first investigated via native polyacrylamide gel electrophoresis (PAGE). Mixtures containing monomeric aS and dimeric 14-3-3η in different aS:14-3-3η molar ratios (ranging from 4 : 1 to 1 : 2) were incubated and analyzed via native PAGE. While bands of the individual proteins were clearly observable in the gels, no additional bands from putative strong complexes were detected (Supplementary Material, Fig. S6).
Nuclear magnetic resonance (NMR) spectroscopy was then employed to explore the possibility of a weak binding between monomeric aS and dimeric 14-3-3η that could go undetected in the biochemical assay described above. Specifically, we investigated the putative effect of 14-3-3η interaction on the [1H-15N] heteronuclear single quantum coherence (HSQC) spectrum of 15N-labeled aS. The superposition of the [1H-15N] HSQC spectra of 15N-labeled monomeric aS, obtained in the absence (green) and in the presence (red) of unlabeled 14-3-3η at a 1 : 3 aS: 14-3-3η molar ratio (Supplementary Material, Fig. S9) and the analysis of weighted chemical shift changes as a function of residue number showed that there was no significant perturbation of the aS backbone 1H and 15N chemical shifts in the presence of 14-3-3η. Therefore, NMR measurements offer no evidence of a direct weak interaction between monomeric aS and the 14-3-3η homodimer.
To evaluate a possible direct interaction between the 14-3-3η homodimer and the fibrillar aS, preformed mature aS fibrils were incubated with a 4 : 1 excess of 14-3-3η (calculated on the starting monomeric aS concentration) for 48 h at 37°C and examined via AFM imaging. As shown in Supplementary Material, Figure S10, this procedure did not cause either fibril fragmentation nor curvature changes.
Minimal effective aS:14-3-3η molar ratio required to generate NFPs
In order to assess the stage at which the aS amyloidogenic reactions performed in the absence or in the presence of the 14-3-3η homodimer diverge, several aliquots from aggregation mixtures containing aS only or aS with the addition of 14-3-3η at different molar ratios were taken at regular intervals and analyzed via AFM. The starting mixtures, containing only monomeric aS (or aS plus 14-3-3η) were virtually indistinguishable by AFM imaging. The first detectable oligomeric aggregates emerged in both cases after 6–12 h and appeared in AFM images as small spherical objects having diameters of a few nanometers. In our hands, both (i) the time elapsed from the start of the aggregation reaction to the appearance of the first detectable aggregates and (ii) their apparent diameters estimated via AFM imaging proved to have very broad, non-mutually correlating distributions, either in the presence or in the absence of 14-3-3η. Conversely, both types of mixtures produced, usually within 48 h, the first stable aggregates (either short mature aS fibrils or NFPs). Both were characterized by sharp distributions of morphometric properties similar to those mentioned in the previous section. In all recorded AFM and TEM images, no evidence for the formation of NFPs by aS in the absence of 14-3-3η was ever found.
To evaluate the minimal effective aS:14-3-3η molar ratio required to steer the amyloidogenic reaction pathway toward the production of NFPs, the same procedure outlined in the previous paragraph was repeated on reaction mixtures containing decreasing amounts of 14-3-3η relative to aS (aS:14-3-3η dimer molar ratios of 4 : 1, 7 : 1, 12 : 1, 20 : 1, 24 : 1 and 30:1). In all cases, the distribution of the diameters of the first observable aggregates appearing at ∼48 h of incubation (Fig. 3B–G) contains one or both of the two peaks corresponding to the diameter of mature aS fibrils (∼7.5 nm) and of NFPs (∼2.0 nm). When more than one 14-3-3η dimer for every 20 aS monomers was present, NFPs were the predominant products of aggregation (Fig. 3D–G). At lower relative concentrations of 14-3-3η, the aggregation products contained an increasing proportion of mature aS fibrillar fragments (Fig. 3B and C), while for molar ratios smaller than 30 : 1 no contribution from the NFPs peak was detectable and the resulting diameter distributions were indistinguishable from those of mature aS fibrils.
14-3-3η interferes with aS ‘seeded’ aggregation
To evaluate whether the aS/14-3-3η interaction is of relevance also at later stages of aggregation, we analyzed the well-documented process of fibril elongation occurring via the accretion of monomeric or oligomeric aS on preformed fibrillar ‘seeds’ (20) both in the absence and in the presence of dimeric 14-3-3η. To this end, 4 : 1 aS:14-3-3η aggregation mixtures were triggered with two types of seeds obtained by subjecting mature aS fibrils to different sonication protocols: (i) protofilamentous/protofibrillar fragments (hence ‘Type 1 seeds’) and (ii) mature fibrillar fragments (‘Type 2 seeds’). Briefly, Type 1 seeds are obtained by more vigorous sonication with respect to Type 2 seeds, and contain smaller aggregates. Most importantly, whereas Type 1 seeds mainly contain sub-fibrillar aggregates such as short protofibrils and protofilaments, Type 2 seeds are fragments of mature aS fibrils, as verified via TEM and AFM imaging (Supplementary Material, Fig. S11A and C).
Seeded aggregation experiments were first performed in the presence of either Type 1 or Type 2 seeds but in the absence of 14-3-3η. In these experiments, the amount of added seeds was 20% of the total aS concentration at the start of the aggregation. After 48 h of incubation, aliquots were collected from aggregation mixtures and examined via TEM and AFM, revealing that long mature aS fibrils were present in both cases (Supplementary Material, Fig. S11B), and in much larger amounts than the control unseeded aS aggregation experiments. The same seeded aggregation experiments were then repeated in the presence of 14-3-3η with a starting aS : dimeric 14-3-3η relative concentration of 4 : 1 (calculated on the total aS present in the mixture, i.e. aS monomer plus seeds). Aliquots taken at 48 h of aggregation showed that Type 1 seeds seemed to exert no effect on aS fibril formation when compared with unseeded aggregation results (Fig. 6A, left panel), being that only short mature fibrils were observed, while Type 2 seeds had promoted the formation of large amounts of long mature aS fibrils in spite of the presence of 14-3-3η (Fig. 5B, left panel).
The localization of 14-3-3η in the products of seeded aggregation after 48 h of incubation was then examined via the same immuno-TEM assay previously performed on NFPs. While no specific placement of gold NPs on the mature fibrils induced by Type 1 seeds was evident (Fig. 5A, right panel), virtually all of the gold NPs observed in samples containing Type 2 seeds were in close contact with long mature fibrils (Fig. 5B, right panel). To rule out the possibility that the observed specific placement of NPs was due to non-specific interaction between aS mature fibrils and the secondary antibody, the immuno-TEM staining was performed on fibrils of sole aS obtained using type 2 seeds, with no anti-14-3-3η primary antibody, revealing the total absence of specific fibril recognition by immunolabeled NPs (Supplementary Material, Fig. S5A). Moreover, aS mature fibrils obtained by unseeded aggregation proved to be completely devoid of NPs when subjected to the full immuno-TEM protocol (Supplementary Material, Fig. S5B).
The influence of 14-3-3 proteins on aS aggregation is isoform-dependent
To investigate the isoform specificity of the interaction between aS and 14-3-3 proteins, all the six remaining 14-3-3 isoforms (β, γ, ε, σ, τ and ζ) were cloned and expressed. All individual isoforms of 14-3-3 were then added to aS aggregation mixtures at an aS:14-3-3 molar ratio of 4 : 1, as previously described for the η isoform. After 48 h of incubation, the respective aggregation products were analyzed via AFM morphometry, revealing a complex isoform-specific behavior (Supplementary Material, Fig. S12). Aggregations performed in the presence of isoforms ε, σ, and ζ led to the formation of long, mature fibrils characterized by morphological parameters (i.e. average diameter of ∼7 nm and a ∼100 nm periodicity) compatible with those of canonical aS fibrils obtained in their absence. More specifically, the apparent diameters of fibrils induced by the ε, σ and ζ isoforms were measured to be 7.6 ± 0.7 nm (ε), 7.8 ± 2.6 nm (σ) and 7.5 ± 0.9 nm (ζ), while those obtained in the absence of 14-3-3 were 7.3 ± 2.1 nm. Isoforms η and τ led to the formation of NFPs with undistinguishable diameter distributions (averaging 2.5 ± 0.6 nm for both isoforms) and high local curvatures. Finally, aggregates formed in the presence of isoform γ appeared as long fibrils, with a diameter different from that of canonical aS fibrils (4.7 ± 0.8 nm).
Effects of 14-3-3η on aS toxicity in cells
To evaluate whether the effects of 14-3-3η on aS aggregation observed in vitro can be relevant in the cell, aS C-terminally fused with enhanced green fluorescent protein (aS-EGFP) was expressed in HEK293T cells. As aS overexpression by itself does not lead to protein massive aggregation (26), we triggered aggregation by addition of Type 1 seeds to cells overexpressing aS-EGFP, as previously described (26,49,50). As shown in Figure 6A and B, addition of aS seeds induced substantial aS aggregation in the cytoplasm and these aggregates appeared as amyloid fibrils by TEM (Fig. 6E). Moreover, aS aggregates in part co-localize with 14-3-3η co-expressed with aS-EGFP (Fig. 6C and D).
We next evaluated whether aS overexpression in the presence or absence of Type 1 seeds is toxic to cells and whether 14-3-3η modifies the cellular toxicity. We counted aS-EGFP positive cells as readout of cell viability at 12, 24 and 36 h. As shown in Figure 6F, overexpression of aS-EGFP led to increased toxicity with respect to control EGFP after 24 and 36 h (**P < 0.01 at 24 h and P < 0.001 at 36 h by one-way analysis of variance (ANOVA) test). Importantly, overexpression of 14-3-3η significantly attenuated aS-mediated cell toxicity (**P < 0.01 at 24 h and at 36 h by one-way ANOVA test), possibly by reducing the amount of aS oligomers/early-stage intermediates, which cannot be appreciated by confocal imaging.
We next investigated the effect of Type 1 seeds addition to the cell medium of cells overexpressing aS-EGFP in the presence or absence of 14-3-3η. We found that cells overexpressing aS-EGFP displayed significant decreased viability when exposed to Type 1 seeds compared with untreated cells (***P < 0.001 at 24 h and at 36 h by one-way ANOVA test compared with aS-EGFP overexpressing cells with no seeds), indicating that these amyloid aggregates are cytotoxic. Interestingly, overexpression of 14-3-3η did not increase the number of aS-EGFP positive cells at 24 and 36 h when treated with seeds (Fig. 6F), likely indicating that 14-3-3η is unable to exert a protective effect on this late-stage type of fibrils.
14-3-3 Proteins influence in vitro aS aggregation
14-3-3 chaperone-like proteins were identified in several independent studies as a recurrent component of the LBs (41,42,51,52) or as direct interactors of aS in parkinsonian brains (40,43). Since other chaperones were shown to interfere with the aS aggregation process (28–33), among the 14-3-3 isoforms we chose to focus on 14-3-3η and to verify if it is able to affect the aS aggregation process. In fact, this isoform seemed to have the most promising characteristics: it was found associated with aS in parkinsonian brain (40) and is up-regulated in A53T mice models (53), suggesting a degree of specificity of its involvement when compared with other 14-3-3 isoforms.
FP kinetics results suggested that 14-3-3η inhibits fibril formation and/or leads to the formation of smaller objects compared with mature fibrils. The analysis of the aggregation process through ThT fluorescence emission over time suggested that the chaperone reduces the β-sheet content in the aggregation products with respect to aS canonical fibrils. The observed differences in the kinetic properties have also a parallel in the diversity of aggregation products of aS in the presence of 14-3-3η. AFM analysis of the aS aggregation products revealed that 14-3-3η redirects the aS amyloidogenic pathway toward NFPs. The morphology of the NFPs is characterized by diameter and curvature distributions that are correlated with the aS:14-3-3η stoichiometric ratio in the aggregation assay.
Nevertheless, we analyzed the effects of all the other isoforms on aS aggregation by AFM and found that some other isoforms beside 14-3-3η can have an influence on aS aggregation and that different isoforms have different effects. Specifically, the presence of τ isoforms in aS aggregation assays led to the formation of characteristic NFPs, while the γ isoform led to the formation of protofibrillar structures whose average diameter was ∼5 nm (54). All the remaining isoforms apparently did not affect aS fibril formation in vitro. This heterogeneous behavior suggests a degree of specificity that has to reside in the molecular details of the 14-3-3 isoforms.
Identification of the aggregation stage at which aS and 14-3-3η interact
The following objective was to define at which step of the aS aggregation process the interaction between aS and the 14-3-3η occurs. NMR results did not support the occurrence of binding between monomeric aS and 14-3-3η in solution in vitro. Sato et al. (40) observed an aS/14-3-3η interaction through co-immunoprecipitation of the proteins from parkinsonian brains; moreover, the binding was measured in vitro by means of SPR. However, in both experiments aS oligomeric species could have been present (as toxic species in parkinsonian brains and as contaminant in monomeric aS solution) (8) and may rationalize the positive results obtained by those authors for the interaction between aS and 14-3-3η (40).
Our AFM measurements showed that the aS : 14-3-3η interaction does not occur at the mature fibril stage, and that 14-3-3η retains its ability to influence the course of the aggregation up to an aS : 14-3-3η molar ratio of 30 : 1. Beyond this threshold, the chaperone is no longer able to modify aS aggregation products in a way detectable by AFM. Overall, these results suggest that the interaction is more likely to occur at an intermediate stage of the aggregation process, during which early aS oligomeric species are formed. We thus propose that 14-3-3η homodimers during aS aggregation are able to stabilize and co-aggregate with a class of early aS oligomeric aggregates.
The mean size reported in the literature for aS oligomers varies as a consequence of their heterogeneity, but also in relation to the different techniques used to measure it, ranging from 10 to 16 molecules when estimated by small angle X-ray scattering data modeling (55), to oligomers comprising 20–24 aS molecules as estimated by EM and scanning transmission EM, to more than 42 aS molecules for protofibrils when evaluated by gel filtration (24). More recently, detailed studies performed using single-molecule fluorescence techniques led to the determination of the dimensions of the small (<10 aS monomers) oligomers formed at the beginning of the aggregation process, that over time convert to a wider ensemble of oligomeric species constituted by 2 to ∼100 aS molecules (56). Another group showed by single-molecule photobleaching that the aS oligomer they obtained in vitro were a well-defined species constituted by 31 aS monomers (57).
The size of the oligomers competent to co-aggregate with 14-3-3η could not be inferred from our experimental data, and would probably be more appropriately described as a range of sizes.
Characterization of the mode of interaction between 14-3-3η and aS oligomers
While our TEM-immunogold data show that NFPs include 14-3-3η in addition to aS, the structural features that govern this inclusion are yet to be defined. For the chaperone Hsp70 (58), a hydrophobic interaction has been invoked to explain the mode of binding with its molecular partners. This type of interaction might also occur in the case presented here, since the hydrophobicity of the aS NAC region represents a driving feature in its fibrillation (59). The signal arising from the helical component of 14-3-3η in CD experiments is likely to be due to unreacted 14-3-3η, which overwhelms the contribution of the NFPs aggregation products. An attempt to determine if the quaternary structure of 14-3-3η is conserved in NFPs was made by evaluating the presence of dimers by TEM-immunogold, but it led to inconclusive results. As mentioned above, although not described in its structural details, the interactions that hold together aS and 14-3-3η in the NFPs are isoform specific. Moreover, it is worth noting that in our in vitro reactions aS is not phosphorylated, hence this modification cannot be a discriminating factor in the interaction even though it characterizes most of the binding partners of 14-3-3 proteins.
The role of 14-3-3η in the late steps of the aS aggregation
To study the effect of 14-3-3η on the latest stages of the aS aggregation process, we performed ‘seeded’ aggregation assays. In those experiments, the presence of aS fibril seeds rapidly promotes the formation of mature fibrils. We produced and characterized two types of seeds: protofibrillar (Type 1) and fibrillar (Type 2) seeds. 14-3-3η was found to be able to interfere only with the aggregation promoted by Type-1 seeds. Conversely, Type 2 seeds promoted aS aggregation even in the presence of 14-3-3η, leading to the formation of the canonical aS fibrils. These fibrils were shown to incorporate 14-3-3η molecules, as shown via immuno-TEM. These results, together with the critical aS:14-3-3η stoichiometric ratio threshold beyond which 14-3-3η ceases to have an effect on aggregation as measured via AFM morphometry, suggest that when the amount of aS is too high, or when the aggregation process has entered the mature fibril elongation step, the chaperone is not able anymore to influence aS aggregation. This point is of particular interest if considered together with the critical aS : 14-3-3η stoichiometric ratio, beyond which 14-3-3η does not have an effect on aggregation anymore. In fact, duplication and triplication of the SNCA gene have as sole effect an increase of the amount of aS, possibly to a level that might overwhelm the chaperone capacity to redirect fibrillation. Interestingly, stoichiometric or excess amounts of dimeric 14-3-3η yield the same NFPs as in the aS:14-3-3η stoichiometric ratio 4 : 1, suggesting the reaching of a saturation regime.
14-3-3η Behavior in aS aggregation cell models
As previously shown (26), simple aS overexpression in HEK293T cells does not lead to the formation of aS oligomers detectable with conventional microscopy techniques in the HEK293T cells, but under comparable conditions aS oligomeric aggregates were observed using other advanced fluorescence microscopy methods (60–62). Similarly, we could not directly observe the formation of NFPs in the HEK293T cells overexpressing both aS and 14-3-3η by confocal microscopy. However, 14-3-3η overexpression rescues aS induced toxicity in these aS-overexpressing HEK293 cells, suggesting that NFPs that may form in cells are either less toxic compared with aS oligomers or more susceptible to cellular clearance machinery.
Conversely, we observed no rescue due to 14-3-3η overexpression in cells in which aS aggregation was triggered by seeds, which leads to the formation of large intracellular inclusions. This result suggests that the chaperone cannot affect fibril formation when the momentum of the aggregation process is too large.
As it can be derived from the in vitro experiments, 14-3-3η molecules can be sequestered in aS mature fibrils when the aggregation process is potentiated by the presence of seeds. In such a situation, the chaperone molecules are avalanched by aS fibrillation and the sequestered molecules are not competent to reroute the amyloidogenic process. However, we were not able to fully prove that 14-3-3η overexpression in HEK293T cells is effective in slowing down aS inclusions formation.
Being that 14-3-3η has not been found in LBs (52), it seems reasonable that this isoform is recruited in the fibrils in a more specific way, compared with the other isoforms trapped in the LBs. Therefore, when the fibrils become part of the LBs in PD brains, 14-3-3η is no more detectable by immunohistochemical methods, while the other isoforms, i.e. ε, γ, ζ and τ, could be simply engulfed in the LBs but not within the fibrils. Interestingly, it has been recently proposed that one of the toxic mechanisms ascribable to pathogenic protein aggregation (in the reported case a chimerical fibril forming paradigm) is the sequestration of cytoplasmic proteins with essential cellular functions (63). Considering the data presented on 14-3-3η sequestration by aS aggregates, it seems plausible that this mechanism occurs also when aS aggregation is triggered by seeds in HEK293T cells. The mild up-regulation of the 14-3-3η gene observed in a A53T transgenic mice (53) suggests that neurons may react to the sequestration of the chaperone through 14-3-3η overexpression.
In this framework, the role of aS seeds seems to be particularly important because the presence or the acquisition from the extracellular media of a few seeds may be responsible for a faster sequestration of 14-3-3η in the aggregates, further speeding the aggregation and the progression of the disease. The spreading of aS aggregation to healthy neurons as reported in PD cellular and animal models (64,65) relies on the direct or indirect transfer of seeds from unhealthy neurons to healthy ones.
In summary, we propose that 14-3-3η interacts with aS with different modes. Under non-pathological physiological conditions, 14-3-3η is able to bind to aS oligomers and re-route the aggregation process toward the formation of NFPs. When aS aggregation enters into its last stage of mature fibril elongation (due to an abnormally high aS concentration or to the presence of seeds), 14-3-3η is sequestered into the fibrillar aS aggregates and is no longer able to counteract the process. The sequestration of 14-3-3η is probably irreversible, since the added free chaperone is not able to disassemble aS fibrils. This epilog for 14-3-3η thus ultimately leads to enhanced aS aggregation, toxicity and neurodegenerative potential.
The interplay between 14-3-3η and aS seems to be governed by the relative amount of the two proteins, by the aS aggregation state and by the stage or speed of the aggregation process. This observation is particularly important for the evaluation of the effects exerted in a cellular model by 14-3-3s in the early stages of protein aggregation and neurodegeneration, in which the chaperones may still be effective. However, the disease progression may not be hindered by 14-3-3s, when the amount of aS is overwhelming the chaperone and/or when the aggregation process progressed too far. In this situation, chaperone proteins contrasting protein aggregation might be trapped into amyloid inclusions and consequently deprive the cell of all their other physiological functions.
Some of the mechanisms we observed (e.g. the alternative aggregation pathway and rescue of aS-induced toxicity caused by 14-3-3η, as well as its sequestration into the mature fibrillar aS aggregates occurring in specific conditions) might also be relevant for the interaction of 14-3-3s with other amyloidogenic proteins (i.e. tau and huntingtin (66,67)). Given that the interactions between several chaperones and their binding partners are governed by similar hydrophobicity-driven mechanisms, it seems reasonable to think that some of them might rescue the toxicity of pathogenic protein aggregation following routes similar to those employed by 14-3-3η.
MATERIALS AND METHODS
Expression and purification of aS
Wild-type (WT) human aS and the C141 mutant suitable for fluorophore labeling were cloned in pET-28a plasmid (Novagen). The mutated form was obtained by mutagenic PCR, introducing a triplet codifying for cysteine at the end of the sequence (C141). Both proteins were expressed in Escherichia coli BL21(DE3) strain and recovered from the periplasm. Briefly, bacteria were grown to an OD600nm of 0.3–0.4 and induced with 0.1 mm isopropyl β-d-1-thiogalactopyranoside (IPTG). After 5 h, cells were collected by centrifugation and recombinant proteins recovered from the periplasm by osmotic shock as previously described (68). Subsequently, the periplasmic homogenate was boiled for 15 min and the soluble aS-containing fraction was subjected to a two-step (35 and 55%) ammonium sulfate precipitation. The pellet was then resuspended, extensively dialyzed against 20 mm Tris–HCl, pH 8.0, loaded into a 6 ml Resource Q column (Amersham Biosciences) and eluted with a 0–500 mm gradient of NaCl. Proteins were then dialyzed against water, lyophilized and stored at −20°C.
Expression and purification of 14-3-3 proteins
Human 14-3-3η was cloned in pQE50 plasmid (Qiagen) in fusion with a C-terminal His-tag. All other six isoforms were subcloned in pET-28a plasmid (Novagen) in fusion with an N-terminal His-tag starting from a pEBG series of plasmids that is a kind gift of Dario Alessi (MRC, Dundee).
All 14-3-3 proteins were subsequently expressed in E. coli in BL21(DE3) strain. Bacteria were grown at 37°C to an OD600nm of 0.7, then induced with 0.5 mm IPTG and grown overnight. Cells were harvested by centrifugation and the pellet resuspended in phosphate buffer. Phenylmethylsulfonyl fluoride 100 mm and a cocktail of protease inhibitors were added to the cells 1:100 (v/v) and cells were subjected to one French Press cycles (Constant Systems Ltd). After that, the cell homogenate was centrifuged and the supernatant loaded onto a Ni2+ affinity column and eluted with a 20–300 mm linear gradient of imidazole in 20 min. The elution peak was dialyzed against phosphate-buffered saline (PBS), then incubated overnight at 22°C with thrombin (according to manufacturer's instructions, Amersham Biosciences) and loaded again onto a Ni2+ affinity column to separate the protein of interest from the cleaved His-tag. The flow-through was collected, concentrated and loaded into a Superdex 200 column (GE Healthcare Life Sciences). The eluted protein was stored at 4°C with 3 mm dithiothreitol (DTT) and 0.02% NaN3 as preservative or frozen in liquid nitrogen and stored at −80°C for long-term storage. Protein purity, integrity and dimerization were checked after purification and/or storage, by SDS–PAGE and by size exclusion chromatography.
15N-labeled aS was expressed in E. coli BL21 (DE3) cells in M9 minimal medium with 15NH4Cl as the sole nitrogen source. Unlabeled 14-3-3η was produced as previously described. NMR experiments were performed in PBS buffer (pH 7.4), with 10% D2O (v/v), 0.02% NaN3 (w/v) and 5 mm DTT. [1H–15N]-HSQC spectra were recorded at 283 K on a 600 MHz Bruker spectrometer equipped with a triple resonance probe. NMR data were processed and analyzed using NMRPipe (69) and Sparky (T. D. Goddard and D. G. Kneller, University of California, San Francisco) software packages.
Fluorescent labeling of aS and 14-3-3η
The aforementioned aS-C141 mutant was used to obtain fluorescent-labeled aS. Labeling was performed adding a 5-fold molar excess of tris(2-carboxyethyl)phosphine to aS-C141 dissolved in 20 mm Tris buffer (pH 7.0). After 30 min of incubation, Oregon Green 488 maleimide (Molecular Probes, Invitrogen) was added to the protein in a 5 : 1 excess stoichiometric ratio, and the reaction was left at 45°C for 4 h. The conjugated aS (aS*) was separated from the unreacted protein, fluorophore, and reducing agent by reverse phase chromatography (C4 column, Phenomenex). 14-3-3η was mixed with a 20-fold molar excess of Oregon Green 488 isothiocyanate F2FITC (Molecular Probe, Invitrogen) in PBS, 20 mm sodium bicarbonate, pH 9.0. The fluorescent-labeled 14-3-3*-containing solution was incubated for 5 h at 25°C and then dialyzed overnight against PBS, 5 mm DTT to eliminate the excess fluorophore. The absorbance of aS* and 14-3-3η* was measured at 496 nm with a UV-visible spectrophotometer (Agilent 8453), while their concentration was calculated considering the molar extinction coefficient of the fluorophore (εOG = 81 000 M−1 cm−1).
Protein aggregation assays
Prior to aggregation, monomeric aS solutions were ultra-filtered with a 100 kDa cut-off Vivaspin (Sartorius) filter to remove residual unwanted oligomeric aggregates. All aggregation experiments were carried out at 37°C in PBS supplemented with 0.05% (w/v) sodium azide and 5 mm DTT (when not stated differently), providing a constant agitation at 1000 rpm. The sample volume was 200 μl and the reaction vessel was either a 96-well polycarbonate plate (fluorescence-based methods) or PCR vials (all other in vitro experiments). Aggregation experiments monitored via fluorescence techniques were performed at a 70 μm starting concentration of monomeric aS, while all other experiments were performed at 20 μm aS. In some of the experiments 14-3-3η was added to the aggregation mixture to afford specific aS:14-3-3 stoichiometric ratios (please refer to the Results section). When required, fluorescent-tagged aS* or 14-3-3η* (see above) were added, diluted at molar ratios of 1 : 100 (aS*:aS) or 1 : 25 (14-3-3η* : 14-3-3η). In seeded aggregation experiments, preformed fibrillar or protofibrillar seeds (see below) constituted 20% of the total starting aS concentration (e.g. 16 μm monomeric aS + 4 μm seeds for experiments at 20 μm starting aS).
Sonication of aS mature fibrils and seed classification
Fragments of aS fibrils were obtained via sonication of mature aS fibril samples. Two different sonication protocols reliably transformed the mature fibrils into two types of seeds, as verified via AFM morphometry performed on the sonication products (see Fig. S11C). The first protocol involved sonication of the aS fibrils (in PBS, refrigerated by immersion in an ice bath) with a tip sonicator (Sonics Vibracell VCX-750) equipped with a 3 mm tapered tip operated at 20% (150 W) power output. Total sonication time was 120 s, obtained with 24 successive 5 s sonication pulses alternated with 5 s pauses. This procedure caused a thorough fragmentation of fibrils into small protofibrillar and protofilamentous aggregates (Type 1 seeds, see Supplementary Material, Fig. S11A, left panel). As determined via AFM imaging, the vast majority of Type 1 seeds have an apparent diameter of <5 nm, which is significantly smaller than the diameter of mature aS fibrils. The second protocol differed from the first only in the type of tip sonicator employed (Labsonic U equipped with a 4 mm tip) and sonication power output (50 W), and afforded mature fibrillar fragments (Type 2 seeds, see Supplementary Material, Fig. S11A, right panel). Type 2 seeds present an apparent diameter distribution peaked at 7.8 nm, in good accord with the diameter of pre-sonication aS mature fibrils measured under the same conditions (∼7.5 nm).
The amyloid fragments thus obtained were directly used as nucleation seeds in aggregation assays performed both in vitro and in cells. Seed concentration was estimated as an equivalent monomer concentration by difference between the initial monomer concentration and the final concentration of the supernatant, after pelleting fibrils.
Fluorescence polarization and emission spectroscopies
Aggregation assays performed on 96-well plates as described above were monitored with FP measurements at 535 nm, conducted at intervals of 6–10 h via a plate reader (DTX 880 Multimode Detector, Beckham Coulter). The excitation wavelength was 490 nm. OG fluorescence spectra in the different aggregation samples were collected with a PTI QuantaMaster C60/2000 spectrofluorimeter (Photon Technology International, Inc.).
Thioflavin T (ThT) binding assay
ThT binding assays were performed adapting a protocol as described elsewhere (70) using a microfiltered (cutoff 0.22 μm) 20 μm ThT solution in PBS (pH 7.4). Constant volume aliquots of 5 μl of protein samples were taken at regular intervals during aggregation assays performed in a 96-well plate to reproduce the aggregation conditions of FP kinetics. Aliquots were diluted into the ThT-containing buffer (final volume 100 μl). Fluorescence emission measurements were conducted on a plate reader (Viktor, Perkin Elmer), at 25°C using an excitation wavelength of 450 nm and recording the ThT fluorescence emission at ∼480 nm to quantify the amount of β-sheets in the proteinaceous material.
All CD experiments were carried out at room temperature on a JASCO J-715 spectropolarimeter equipped with HELLMA quartz cells with Suprasil® windows and an optical path length of 0.1 cm. CD spectra were acquired and processed using the manufacturer software. All spectra were recorded in the wavelength range of 200–250 nm, using a bandwidth of 2 nm and a time constant of 2 s at a scan speed of 50 nm/min. The signal : noise ratio was improved by accumulating four scans. Spectra were acquired on aggregated samples diluted 10 times with ultrapure water and then concentrated with a Vivaspin500 5000 MWCO (Sartorius Stedim Biotech) to the original volume in order to reduce salt concentration. Since part of the examined samples was a mixture of two proteins, spectra were not normalized to protein concentration.
TEM and immuno-TEM
TEM samples were prepared by adsorbing a 15 μl aliquot taken from the relevant aS aggregation batch onto a carbon-coated copper grid, and applying a 0.05% uranyl acetate solution for negative staining. For immuno-TEM, grids were first incubated with a rabbit polyclonal primary antibody against 14-3-3η (ABCAM) previously diluted 1 : 200 in PBS, then washed in PBS, and finally incubated with a gold-coupled secondary antibody against rabbit IgG (Sigma Anti-Rabbit IgG – Gold antibody, diluted 1:30 in PBS). After a second washing step in PBS, the grids were negatively stained as described above. TEM pictures were taken on a Tecnai G2 12 Twin instrument (FEI Company, Hillsboro, OR, USA). For electron microscopy on cell samples, HEK293T cells were transfected with aS and treated with seeds, cultured for 4 days, fixed overnight in 0.1 M sodium cacodylate buffer at pH 7.4 containing 2.5% glutaraldehyde and then processed and embedded in LR White resin (Polysciences, Warrington, PA, USA). Ultrathin sections were stained with uranyl acetate for imaging.
Atomic force microscopy imaging was performed either (i) in tapping mode with Ultrasharp NSC15/AlBS silicon probes having a nominal resonant frequency of 325 kHz (Mikromasch, Tallin, Estonia) on a NanoScope IIIa system or (ii) in ‘PeakForce tapping’ mode with Scanasyst-Air probes (Bruker, Mannheim, Germany) on a Nanoscope V system. Both Nanoscope controllers were equipped with a Multimode head and a type-E piezoelectric scanner (Bruker, Mannheim, Germany). Ten microliters of sample were deposited on freshly cleaved mica (RubyRed Mica Sheets, Electron Microscopy Sciences, Fort Washington, USA) and left to adsorb for 5 min at room temperature (∼20°C). The mica surface was then rinsed with ∼500 μl of MilliQ H2O (Millipore Simplicity) at the same temperature and dried with dry nitrogen. In most experiments, the sample was diluted ∼10 times with PBS then equilibrated at RT for 10 min prior to deposition in an attempt to minimize overlap of individual aS aggregates on the surface. Multiple images from successive depositions were then pooled and digitalized to obtain statistically significant morphological measurements as outlined below.
Aggregate recognition and morphometry in AFM and TEM images
Fibril-containing AFM images have been analyzed via a semi-automated procedure in order to identify and characterize fibrillar objects. The images were first processed through a pipeline of ImageJ (71) plugins: each image was first de-speckled to remove ‘salt-and-pepper’ noise, then its background was subtracted via a rolling-ball algorithm using a radius of 60 nm (72). Gaussian-blur was then subtracted with a standard deviation of 60 nm and a mask weight of 0.7, after which the image was skeletonized (73). Candidate fibrillar objects were identified by searching the image skeleton for short unbranched segments which were then grown in both directions following their skeletons until their termini were reached. Fibrillar objects in contact with the edges of the image or other overlapping objects were discarded. Each remaining object was then interpolated through a 2D cubic B-spline, from which its contour length and curvature profile were readily obtained. Height profiles were obtained by bilinear interpolation of the heights from the de-speckled and background-subtracted image.
Fibril-containing TEM images were analyzed through a semi-automated procedure consisting of several steps. First, the images were convolved with a Gaussian kernel of standard deviation 2.5. The normalized image gradient was then computed, after which a user-interactive segmentation procedure based on the curvature-aware livewire algorithm (74,75) was applied to identify the skeleton of the fibril objects. Fibril curvature profiles were then extracted from the fibril skeletons using the same procedure as for the AFM images.
Cell cultures and transient transfections
aS tagged with EGFP (aS-EGFP) was obtained subcloning aS cDNA in pEGFP-N1 vector (Clonetech). aS cDNA was amplified by PCR with specific primers that introduced KpnI and BamHI recognition sites at the 5′ and 3′ of the sequence, respectively. The pEBG14-3-3η plasmid was a kind gift of Dario Alessi (MRC, Dundee). HEK293T cells were cultured at 37°C in 5% CO2 in DMEM (GIBCO, Invitrogen) supplemented with 10% FBS, 50 U/ml penicillin and 50 μg/ml streptomycin. The pEGFP-aS and pEBG14-3-3η plasmids (or the correspondent empty vectors) were used for HEK293T transient transfections, performed in optimum medium using polyethylenimine as transfection reagent, when cells were at a density of ∼5 × 105 cells/cm2. To assure that almost all the cells overexpressing aS-EGFP were also overexpressing 14-3-3η in coexpression, the plasmidic DNA ratio was always maintained 2 : 3. When needed, 1 μm aS seeds were added to the medium during transfection. After 12 h, transfected cells were washed in PBS and the medium was changed. aS and 14-3-3η overexpression and transfection efficiency were evaluated, respectively, by western blot (WB) analysis, after cell lysis and protein separation on SDS–PAGE, and by immunocytochemistry, using a mouse monoclonal anti-aS antibody (Cell Signaling) and a rabbit polyclonal anti-14-3-3 proteins antibody (Santa Cruz Biotechnology, Inc.).
Live imaging and immunocytochemistry
For live imaging cells were observed with an inverted widefield microscope (Leica DMI 4000B) every 12 h after transfection for three times. To assure the significance of our results, we acquired at least five fields per culture and the experiments were repeated three times independently. At the end of the time course, cells were fixed with methanol at −20°C, permeabilized with PBS 0.1% Triton and blocked with PBS with 5% FBS. Blocked cells were then stained with a rabbit polyclonal primary antibody against 14-3-3η proteins (Santa Cruz Biotechnology, Inc.) diluted 1 : 100 in PBS. The counterstaining was done with Alexa Fluor 568 goat anti-rabbit IgG (Invitrogen). Cells were also stained with Hoechst 33242 (Invitrogen) to visualize the nuclei. Higher-resolution fluorescence microscopy images were acquired with a confocal microscopy (Nikon Eclipse E6000).
The statistical analysis for immuno-TEM images, i.e. frequency counts and fits, was performed using OriginPro8 (OriginLab). The statistical analysis on cell viability assay was performed using GraphPad Prism. A one-way ANOVA with Tukey post hoc test was performed and P-value considered statistically significant was <0.05, represented on the histograms with a single ‘*’. When the P-value was <0.01, the graphical representation was ‘**’, while ‘***’ were drawn for P < 0.001.
B.S. acknowledges the Human Frontier Science Program (ref RGP0010/2011). E.G, L.B. Michael J. Fox Foundation.
The authors thank F. Caicci and F. Boldrin of the Electron Microscopy Facility, Department of Biology, University of Padova, for the assistance with the electron microscopy experiments. B.S. acknowledges the Human Frontier Science Program (ref RGP0010/2011).
Conflict of Interest statement. None declared.