Address correspondence and reprint requests to Björn H. Falkenburger or Jörg B. Schulz, Department for Neurodegeneration and Restorative Research, University of Göttingen, Waldweg 33, D-37073 Göttingen, Germany. E-mail: email@example.com or firstname.lastname@example.org
Aggregates of α-synuclein are the pathological hallmark of sporadic Parkinson’s disease (PD), and mutations in the α-synuclein gene underlie familial forms of the disease. To characterize the formation of α-synuclein aggregates in living cells, we developed a new strategy to visualize α-synuclein by fluorescence microscopy: α-synuclein was tagged with a six amino acid PDZ binding motif and co-expressed with the corresponding PDZ domain fused to enhanced green fluorescent protein (EGFP). In contrast to the traditional approach of α-synuclein-EGFP fusion proteins, this technique provided several-fold higher sensitivity; this allowed us to compare α-synuclein variants and perform time-lapse imaging. A C-terminally truncated α-synuclein variant showed the highest prevalence of aggregates and toxicity, consistent with stabilization of the α-synuclein monomer by its C-terminus. Time-lapse imaging illustrated how cells form and accumulate aggregates of α-synuclein. A substantial number of cells also reduced their aggregate load, primarily through formation of an aggresome, which could itself be cleared from the cell. The molecular chaperone Hsp70 not only prevented the formation of aggregates, but also increased their reduction and clearance, underlining the therapeutic potential of similar strategies. In contrast to earlier assumptions build-up, reduction and clearance of α-synuclein aggregation thus appear a highly dynamic process.
In a cell-free system, α-synuclein is natively unfolded; the monomer is stabilized by tertiary interactions, which are released by the disease-related mutations A30P and A53T (Bertoncini et al. 2005a). Interestingly, the disease-related mutations increase oligomerization, but not primarily the formation of larger fibrils as compared to WT (Conway et al. 2000). Whereas it has been investigated in some detail in cell-free systems, the process of α-synuclein aggregation has not yet been characterized in living cells. One reason has been the lack of methods to follow aggregation in real time.
Time-lapse imaging has been a valuable tool to investigate fundamental cellular processes and the formation of aggregates in a cellular model of Huntington disease (Arrasate et al. 2004). We therefore developed a new strategy to fluorescently label α-synuclein in living cells, making use of the strong and specific interaction of a PDZ binding motif with the corresponding PDZ domain. This interaction has been used to purify proteins in vitro (Reina et al. 2002), but so far not for visualization of proteins in vivo. The aim of the study was to characterize the aggregation of α-synuclein in living cells. Since a mouse model of Huntington’s disease has provided evidence that cells can also remove aggregates (Yamamoto et al. 2000), we were also interested in the potential clearance of α-synuclein aggregates, particularly since A53T but not WT α-synuclein has been shown to inhibit protein degradation pathways, such as the proteasome (Stefanis et al. 2001; Jiang et al. 2006) or chaperone-mediated autophagy (Cuervo et al. 2004).
Next to WT and the two disease-related mutations A30P and A53T, we also used a C-terminally deleted variant, which had shown increased aggregation in vitro (Bertoncini et al. 2005b). Furthermore, we wanted to characterize the in vivo effects of the molecular chaperone Hsp70, which has been shown to reduce the formation of α-synuclein fibrils observed in vitro (Dedmon et al. 2005) and the prevalence of aggregates in vivo (McLean et al. 2002; Klucken et al. 2004).
For the fusion protein of PDZ domain and EGFP (PDZ-EGFP), the PDZ1 domain of S-SCAM (accession NP_446073, a gift from the lab of Nils Brose, MPI-em, Göttingen) was cloned into pEGFP-N1, resulting in PDZ-EGFP (Fw: 5′-ATTAGATCTATGGGAACATTCCTCAGCACCAC-3′; Rev: 5′-ATTACCGGTTCCTCCGGATATGTGCCAT CTAGTTGG-3′). The corresponding PDZ binding motif (amio acids HSTTRV, from neuroligin-1, accession NP_446320) was added to the C-termini of α-synuclein variants by PCR (Fw: 5′-CTCAAGCTTATGGATGTATTCATG-3′; Rev: 5′-ATGGATATCTCACACTCTGGTGGTGCTGTGGGCTTCAGGTTCG-3′; ΔCRev: 5′-GTTGATATCTCACACTCTGGTGGTGCTGTGTGGGGCTCCTTCTTC-3′) and PDZ-EGFP was subcloned including its CMV promoter into the pcDNA3.1 (Invitrogen, Carlsbad, CA, USA) vectors containing the tagged α-synuclein variants. For expression of the α-synuclein-EGFP fusion proteins (WT-EGFP and ΔC-EGFP), α-synuclein variants were cloned into pEGFP-N1 (Invitrogen) using PCR (Primers: Fw: 5′-CTCAAGCTTATGGATGTATTCATG-3′; WTRev: 5′-CTTGATATCTCAGGCTTCAGGTTCGTAGTCTTGATAC-3′; ΔCRev: 5′-ACTGATATCTCATGGGGCTCCTTCTTCGGCAAG-3′). To express HA-tagged Hsp70, the coding sequence of human Hsp70 (accession NM-005345) was amplified and inserted into the vector pRRLsin.PPTs.hCMV (provided by L. Naldini, Institute for Cancer Research, Torino, Italy) with In-FusionTM PCR Cloning Kit (Clontech, Mountain View, CA, USA) according to the manufacturer’s instructions. Primers were Fw: (containing HA-tag) 5′-AGGGGGATCCACCGGTATGGCTTACCCATACGATGTTCCAGATTACGCTATGGCCAAAGCCGCGGCGATCG-3′; Rev: 5′-CCGCTTTACTTGTACACTAATCTACCTCCTCAATGGTGGGG-3′.
HEK293 cells were cultured in Dulbecco’s modified Eagle’s medium (PAN-Biotech, Aidenbach, Germany) with 10% fetal calf serum and 1% penicillin–streptomycin. Cells were transiently transfected using Metafectene (Biontex Laboratories, Martinsried, Germany), following the manufacturer instructions. For imaging, cells were grown on poly-l-lysine (Sigma, Munich, Germany) coated glass coverslips and used 24 h after transfection.
For immuno-cytochemistry, cells were washed once with phosphate-buffered saline (PBS), fixed for 15 min in 4% paraformaldehyde and 5% sucrose, washed twice in PBS and submerged for 60 min in blocking buffer, consisting of PBS, 0.1% TritonX (Sigma) and 5% goat serum (PAA, Cölbe, Germany). Monoclonal antibodies against α-synuclein (Abcam ab6176–100, 1 : 200), vimentin, γ-tubulin (both Sigma, 1 : 100), HA-tagged Hsp70 (Covance HA.11, 1 : 100) and the polyclonal OC antiserum (8 μg/μL) were incubated over night at 4°C in 10% blocking buffer. After three washing steps with blocking buffer, Alexa-555 conjugated secondary antibody (Invitrogen, 1 : 1000) was incubated at 20°C for 1.5 h. For staining of chromatin and F-actin, cells where submerged in PBS with 0.5 μg/mL of Hoechst 33258 (Invitrogen) and 1 : 500 of Alexa-568 conjugated phalloidin (Invitrogen) for 15 min at 20°C and washed three times in PBS. Coverslips were mounted on glass slides using mounting medium consisting of 24% w/v Glycerol, 0.1 M Tris-base pH 8.5, 9.6% w/v Mowiol 4.88 (Calbiochem, Darmstadt, Germany) and 2.5% w/v of DABCO (Sigma). To test membrane integrity, cells were washed in 37° Dulbecco’s modified Eagle’s medium with 20 mM HEPES and 1% penicillin–streptomycin, incubated (15 min, 37°C, 5% CO2) with 2 μg/mL propidium iodide (Sigma) or 1 : 500 CellTrace calcein-red-orange-AM (Invitrogen) and imaged in the same medium.
Imaging at 24 h was performed at 20°C using an inverted fluorescence microscope (DMI6000B, Leica Microsystems, Bensheim, Germany) with a 63x dry objective (HCX PL FLUOTAR, N.A. 0.7) and a Leica FX350 Camera. Confocal images were acquired using a Leica DMIRE2 microscope equipped with a TCS SP2 AOBS system (63x oil immersion objective) Time-lapse imaging was performed with a DMI6000B microscope (N PLAN 20x dry objective, N.A. 0.4) equipped with an incubator (37°C and 5% CO2) by acquiring images at defined positions every 15 min. Images were converted to Quicktime movies using ImageJ 1.38 (Sörenson compression, four frames per second).
For the quantification of EGFP distribution patterns at 24 h, in each experiment, three coverslips were stained for each construct, and the EGFP distribution of 100–300 cells per coverslip classified manually as ‘homogeneous,’‘aggresome only,’‘several aggregates’ or ‘preapoptotic.’ The numbers depicted in Figs 1, 2, 6 and 7 represent a summary of n = 3 to 11 independent experiments. Each experiment represents one datapoint. The number of independent experiments for each figure is depicted in the figure legend. The total number of cells evaluated for each genotype was WT: 1612 cells, A30P: 1922 cells, A53T 1344 cells, ΔC: 1841 cells, ctrl (PDZ-EGFP alone) 1137 cells, ΔC and Hsp70: 1037 cells, ctrl and Hsp70: 659 cells. For the quantification of time-lapse imaging 50–60 positions per genotype and experiment (4–5 independent experiments per genotype) were selected manually and transitions between EGFP distribution patterns rated. One cell could make several transitions in sequence (a → b → c → d). Transitions (a → b) were expressed either relative to the total number of cells followed in the experiment or relative to all transitions observed that departed from the same appearance, for example a → b/(a → a + a → b + a → c + a → d + a → e). Bars depicted in the graphs represent percentages of all EGFP positive cells as mean ± standard error of the mean. Comparisons were performed by one-way anova using GraphPad Prism 4.00 (GraphPad Software, San Diego, CA, USA). p-Values were derived from Bonferroni post hoc tests: n.s., not significant; *, p < 0.05; **, p < 0.01; ***, p < 0.001.
Visualization of α-synuclein aggregates in living cells
EGFP fusion proteins were used to visualize α-synuclein aggregates in living cells. Unfortunately, the number of cells with aggregates was very low when using a direct fusion protein of α-synuclein and EGFP (Fig. 1a top and Fig. 1b–e white bars), as described previously (McLean et al. 2001). We hypothesized that fusion of the relatively large EGFP (27 kDA) to the C-terminus of the relatively small α-synuclein (15 kDa) sterically hinders aggregation and toxicity and therefore tested the use of a PDZ binding motif to fluorescently label α-synuclein. PDZ domains can be found in many different proteins. Each PDZ domain specifically binds to a PDZ binding domain. Only the six amino acids of the PDZ binding motif from neuroligin-1 (Meyer et al. 2004) were added to the C-terminus of α-synuclein variants and cloned into a vector that also expressed a fusion protein of the corresponding PDZ1 domain from S-SCAM and PDZ-EGFP (Fig. 1a bottom).
Upon expression of tagged α-synuclein with PDZ-EGFP in HEK293 cells we observed four patterns of EGFP distribution: most cells showed a homogeneous distribution of EGFP throughout the cytosol and the nucleus (Fig. 1b). In some cells, we observed a single, prominent, usually perinuclear aggregate, surrounded by a basket of vimentin and gamma-tubulin filaments, which define it as an aggresome (Fig. 1c, g and h). A third category of cells showed more than one aggregate, most of which were small and scattered throughout the cytosol (Fig. 1d). In some of these cells, one aggregate clearly resembled an aggresome, as in Fig. 1d. In most cases, however, it was difficult to tell whether or not an aggresome was among the aggregates; therefore, all cells with more than one aggregate were included in the same category. Aggregates seen by PDZ-EGFP co-localized with aggregates observed with staining for α-synuclein in confocal images (Fig. 1i), indicating that they indeed represent aggregates of α-synuclein. A fourth category of rounded cells showed large amorphous aggregates and had lost stress fibers and focal adhesions (Fig. 1e). Nonetheless, these cells were impermeable to propidium iodide (not shown) and retained the viability marker calcein red-orange (Fig. 1j), indicating preserved membrane integrity. Time-lapse imaging confirmed that this appearance preceded the formation of apoptotic bodies (Fig. 1f). We therefore termed this category of cells preapoptotic.
The same categories were observed with the direct fusion protein of α-synuclein and EGFP, but using the PDZ-tag resulted in a sixfold higher prevalence of cells with aggregates (Fig. 1d). The percentage of cells with an aggresome was increased threefold (Fig. 1c); the percentage of apoptotic cells was increased tenfold (Fig. 1e). Some cleavage of EGFP was observed by western blot both for α-synuclein-EGFP as reported previously (McLean et al. 2001) and for PDZ-EGFP (not shown).
Comparison of α-synuclein variants at 24 h
We then used the PDZ-based labeling strategy to investigate the differences in aggregate formation and toxicity between four α-synuclein variants: WT, A30P, A53T, and a C-terminal deleted form consisting of amino acids 1-108 (ΔC). Expression of PDZ-EGFP alone served as negative control. Equal expression levels of the four α-synuclein variants were verified by quantitative PCR in three independent experiments (Supplementary material Fig. S1). Expression of tagged WT-α-synuclein with PDZ-EGFP resulted in significantly less homogeneous cells than PDZ-EGFP alone (Fig. 2a). Even less homogeneous cells were observed with the A30P, A53T, and ΔC mutations, in this rank order. There was no significant difference in the percentage of cells with a single aggresome (Fig. 2b), but the percentage of cells with many aggregates was significantly increased for A30P, A53T and ΔCα-synuclein as compared to PDZ-EGFP alone (Fig. 2c, p < 0.01, p < 0.001, p < 0.001, comparison not depicted for clarity). The percentage of preapoptotic cells was increased by co-expression of all α-synuclein variants, in the same rank order as the prevalence of aggregates, namely WT<A30P<A53T<ΔC (Fig. 2d, p < 0.01, p < 0.001, p < 0.001, p < 0.001). When compared to WT-α-synuclein A53T and ΔC (but not A30P) showed significant increases in the prevalence of aggregates and preapoptotic cells (Fig. 2c and d). C-terminal deletion of α-synuclein thus increased the prevalence of aggregates by 92% and increased the prevalence of pre-apoptotic cells by 102%, consistent with the in vitro results.
In order to gain some structural information about the aggregates visualized by the PDZ-EGFP method, we used an immune serum raised against a homogenous population of Aβ42 fibrils (‘OC’), which recognizes fibrils and fibrillary oligomeres but not random coil monomer or prefibrillar oligomers (Kayed et al. 2007). Interestingly, this antiserum recognized aggregates and aggresomes formed upon expression of ΔC but not WT α-synuclein (Fig. 3). This finding suggests that aggregates of WT-α-synuclein may be amorphous whereas aggregates of ΔC are fibrillar.
We then performed time-lapse imaging between 24 and 48 h after transfection to characterize the process of aggregate formation in more detail. This allowed us to follow for the first time in living cells the formation of protein aggregates by α-synuclein. Examples are depicted in Fig. 4 and as supplementary online videos. Cells were rated as either homogeneous, with a single aggresome, with small peripheral aggregates, with many or large aggregates (usually including an aggresome) or undergoing apoptosis. Figure 5 summarizes the results of counting transitions between these five categories. With WT-α-synuclein 50% of the cells formed visible aggregates during the time of imaging, which corresponds to 85% of cells that started homogeneous. Some cells first formed an aggresome, towards which then small peripheral aggregates were drawn, leading to a growing aggresome with peripheral aggregates (Fig. 4a, Video Clip S2). Other cells first formed small or reticulate aggregates in the periphery, which then grew into an amorphous aggregate (Fig. 4b). Both patterns of aggregation were equally frequent (Fig. 5a and f) and represent extremes of a continuum rather than strictly separated entities. However, progression of aggregate load was more likely for cells that started with small peripheral aggregates (92%) than for cells that started with a single aggresome (66%, compare Fig. 5c and h). Around 15% of all cells underwent apoptotic cell death during the observation period, evident by formation of apoptotic bodies (Fig. 1f and Video Clip S1). Most of the apoptotic cells were cells with many aggregates (81% of apoptotic cells), but homogeneous cells, cells with an aggresome or cells with a small number of peripheral aggregates also underwent apoptosis (6.5%, 10%, and 2.5% of apoptotic cells respectively).
Importantly, we not only observed formation and progression of aggregates, but also clearance of aggregates, either by a reduction of aggregate load to a single aggresome (Fig. 4c, Video Clip S3) or by an entire clearance of visible aggregates from the cytosol (Fig. 4d, Video Clip S4). As many as 24% of cells with many and large aggregates reduced their aggregate load to a single aggresome and 17% of cells with a single aggresome became homogeneous. Interestingly, clearance of the aggresome was more likely than clearance of small peripheral aggregates (Fig. 5b and g, respectively).
To compare aggregate formation of α-synuclein variants, we also performed time-lapse imaging of A53T and ΔCα-synuclein using the PDZ-based labeling strategy. There was a non-significant trend of reduced clearance of aggregates with WT>A53T>ΔC (Fig. 5g), but all other transitions were equally frequent between the α-synuclein variants. This suggests that the processes of aggregate formation, clearance and toxicity are not fundamentally different between α-synuclein variants. In contrast, the relative frequencies of aggregate formation, progression and clearance were markedly different upon co-expression of the molecular chaperone Hsp70.
Effect of Hsp70
To characterize the effects of Hsp70 on aggregation and toxicity of α-synuclein, we chose ΔC as the variant with most abundant aggregate formation and highest toxicity. Hsp70 was co-expressed with either PDZ-EGFP alone or with the plasmid expressing ΔC-α-synuclein and PDZ-EGFP (Fig. 6). EGFP distribution patterns were counted 24 h after transfection. Co-expression of Hsp70 with ΔC-α-synuclein and PDZ-EGFP led to a 58% reduction in the prevalence of aggregates (Fig. 6c) and to an even more pronounced 74% reduction in the prevalence of pre-apoptotic cells (Fig. 6d). In fact, Hsp70 reduced aggregation and toxicity of ΔC-α-synuclein virtually to the baseline levels observed with PDZ-EGFP alone. This effect on aggregation and toxicity was specific to aggregates induced by α-synuclein, since aggregation and toxicity observed with expression of PDZ-EGFP alone was unaltered by Hsp70. The prevalence of aggresomes was increased with Hsp70 (Fig. 6b). This increase was observed both with PDZ-EGFP alone and with ΔC-α-synuclein and PDZ-EGFP, indicating that induction of aggresomes by Hsp70 was independent of α-synuclein.
We then characterized the effects of Hsp70 on aggregate formation in more detail using time-lapse imaging (Fig. 5). Unexpectedly, Hsp70 did not increase the formation of an aggresome in homogeneous cells (Fig. 5a). Instead, it decreased the formation of small peripheral aggregates in homogeneous cells (Fig. 5f) and the likelihood of aggregate progression both from a single aggresome and from small peripheral aggregates to a state of many or large aggregates (Fig. 5c and h, respectively). Moreover, Hsp70 increased the likelihood of cells with many or large aggregates to reduce their aggregate load towards a single aggresome (Fig. 5d) and increased the clearance of both the single aggresome and small peripheral aggregates from the cytosol (Fig. 5b and g, respectively). The increased prevalence of aggresomes observed at the single timepoint (Fig. 6b) thus does not result from an increased formation of aggresomes in homogeneous cells but from increased aggregates reduction to a single aggresome. The likelihood of cells with large and many aggregates to undergo apoptosis was not reduced by Hsp70, indicating that Hsp70 does not inhibit the toxic effects of large aggregates directly. The decreased toxicity observed with Hsp70 is therefore likely to result mainly from the decreased prevalence of cells with many or large aggregates.
Inhibition of protein degradation
Finally, we wanted to gain insight into the processes that mediate formation and degradation of protein aggregates. We therefore pharmacologically inhibited (macro-)autophagy using 3-methyladenine (3MA), the proteasome and calpain using N-acetyl-leucyl-leucyl-norleucinal (ALLN) and lysosomal degradation using NH4Cl. We first evaluated the concentration-dependent toxicity of 3MA, ALLN, and NH4Cl in untreated HEK293 cells, and subsequently only used concentrations that were below the toxicity threshold. Interestingly, treatment with ALLN and NH4Cl showed different effects on aggresome formation and aggregation in cells with over-expression of EGFP versus WT-α-synuclein or ΔC-α-synuclein (Fig. 7): ALLN increased the prevalence of aggregates upon expression of WT-α-synuclein or ΔC-α-synuclein but not EGFP (Fig. 7g and k versus c). This increase was associated with an increase in the number of pre-apoptotic cells (Fig. 7 d, h and l). NH4Cl also increased aggregates and pre-apoptotic cells but to a much smaller degree. Moreover, this increase was not specific to α-synuclein expression because it also occurred with EGFP only. 3MA did not increase the prevalence of aggregates or pre-apoptotic cells. We therefore conclude that the proteasome is most important for the removal of aggregates upon expression of α-synuclein variants in our system.
We have developed a new strategy to fluorescently label α-synuclein and used it to characterize aggregate formation in living cells. Time-lapse imaging demonstrated that the formation of aggregates is not slow or irreversible, but rather a highly dynamic process with simultaneous generation, accumulation, reduction, and clearance of aggregates.
Removal of α-synuclein aggregates has not been observed in living cells before. The predominant path of aggregate reduction was through the aggresome. It was the ‘center of gravity’ to which all other aggregates were drawn (Video Clip S2), often resulting in the clearance of all other visible aggregates from the cytosol (Fig. 4d). Moreover, the aggresome itself was more likely to be cleared from the cytosol than other aggregates (Fig. 5b and g). Aggresomes are pericentriolar structures in which misfolded proteins are sequestered by an active, microtubule-dependent process (Johnston et al. 1998). Transport of GFP-chimera into the aggresome has been studied by time-lapse imaging before (Garcia-Mata et al. 1999). Consistent with an important role of aggresomes in the clearance of aggregates, aggresomes were observed in 60% of healthy cells but in only 10% of dying cells following over-expression of α-synuclein and synphilin-1 in 293T cells (Tanaka et al. 2004). Since Lewy bodies, the pathological hallmark of PD, share many properties with aggresomes, they most likely also represent a protective response of the cell and not a cause of cell death (Lee and Lee 2002; McNaught et al. 2002). Consequently, cognitive impairment in patients with Lewy body dementia did not correlate with the abundance of Lewy bodies, but with the extent of peripheral, neuritic aggregates of α-synuclein in the cortex (Kramer and Schulz-Schaeffer 2007).
The PDZ-based visualization was also used to compare aggregation and toxicity between α-synuclein variants. Consistent with previous results, the aggregation in vivo of WT-α-synuclein was relatively small (McLean et al. 2001; Jiang et al. 2006). The prevalence of aggregates and pre-apoptotic cells was higher for A53T and ΔC compared to WT at 24 h after transfection. Increased aggregation (Hasegawa et al. 2004) and toxicity (Stefanis et al. 2001) of A53T-α-synuclein has been described previously. The increased prevalence of aggregates observed with C-terminally deleted α-synuclein is consistent with the NMR-based finding that the C-terminus shields the aggregation-prone NAC region from oligomerization and aggregation (Bertoncini et al. 2005b), and of increased aggregation following other modifications of the C-terminus (McLean et al. 2001; Smith et al. 2005). The increased toxicity observed with ΔC is relevant since C-terminally deleted forms of α-synuclein have been observed in the brains of mice expressing full-length α-synuclein and in the brains of patients with synucleinopathies (Baba et al. 1998; Tofaris et al. 2003; Li et al. 2005; Liu et al. 2005). Interestingly, aggregates of ΔC but not WT α-synuclein were stained by a fibril-specific antiserum (Kayed et al. 2007), suggesting that a conformational difference may underlie the described functional differences.
Time-lapse imaging revealed no qualitative differences between α-synuclein variants and no significant quantitative differences in progression, reduction, clearance or toxicity of aggregates. The differences observed in Fig. 2 thus most likely result from a higher rate of aggregate formation with A53T and ΔC as compared to WT α-synuclein. This difference was not reflected in our time-lapse imaging results, but the estimate of aggregate formation from homogeneous cells was comparably inaccurate with this method. Homogeneous cells were often barely visible in the videos because the exposure time had to be kept constant for all images and was adjusted to follow aggregates that usually were bright fluorescent. As another possible explanation for the increased prevalence of aggregates with A53T and ΔCα-synuclein, we observed a non-significant trend for reduced clearance of aggregates. This finding would be in line with previous reports of impaired degradation pathways by A53T but not WT-α-synuclein (Stefanis et al. 2001; Cuervo et al. 2004; Jiang et al. 2006).
Even though the mean differences between α-synuclein variants were small, there was substantial heterogeneity between individual cells of the same experimental group. Our time-lapse imaging of individual cells clearly indicates that cells, which have already built up an aggresome and/or peripheral aggregates, have the endogenous capacity to clear them (Figs. 4 and 5). Blocking different protein degradation pathways (Fig. 7) suggests that the proteasome is most important for removal of α-synuclein aggregates. Interestingly, inhibiting the proteasome with expression of ΔC-α-synuclein increased the prevalence of cells with many aggregates but decreased the number of cells with an aggresome only (Fig. 7j and k, respectively). This could be explained by an impaired capacity to reduce its aggregate load (from a cell with many aggregates towards a cell with an aggresome only). The proteasome thus appear involved in this important reparation mechanism.
The ratio of aggregate progression, reduction and clearance in time-lapse imaging was significantly altered by addition of the molecular chaperone Hsp70 (Fig. 5). Consistently, the prevalence of aggregates and toxicity were also reduced in most previous studies (Auluck et al. 2002; McLean et al. 2002; Klucken et al. 2004). However, these previous studies could not differentiate between the prevention of aggregate production and a potential removal of aggregates. This difference is clinically relevant, because PD patients have already accumulated a significant aggregate load and loss of neurons at the time of diagnosis, which limits the therapeutic potential of purely preventive strategies. Our time-lapse experiments demonstrated for the first time that Hsp70 increased the clearance of protein aggregates, which occurred primarily through the aggresome. This underlines the therapeutic potential of Hsp70 or small molecules with similar effects. Hsp70 also increased the removal of small aggregates and aggresomes from the cytosol. Finally, Hsp70 decreased the formation of small peripheral aggregates in homogeneous cells and reduced the progression from single aggresomes to more and larger aggregates (Fig. 5d and f), possibly explained by the inhibition of aggregate formation observed in vitro (Dedmon et al. 2005). Consistent with heat-shock-induced aggregate removal through the aggresome, protection of dopaminergic neurons from lactacystin-induced cell death by heat-shock was inhibited by the microtubule depolymerizing agent nocodazole (Ahn and Jeon 2006). Expression of tagged ΔC-α-synuclein thus constitutes a tool to screen for small molecules that support aggresome formation and aggregate clearance and which therefore might be beneficial in PD patients.
The PDZ-based labeling strategy was the prerequisite to compare the prevalence of aggregates in living cells and to perform time-lapse imaging. We confirmed that aggregates observed with PDZ-EGFP can be stained for α-synuclein and compared PDZ-based labeling to direct α-synuclein-EGFP fusion proteins. Aggregates were sixfold more common with PDZ-based labeling than with direct fusion to EGFP (Fig. 1) and significant differences between α-synuclein were only observed with the PDZ-based strategy (Fig. 2) but not with direct fusion to EGFP (not shown). This indicates that the PDZ-based labeling is better suited to compare aggregation of α-synuclein variants than direct fusion to EGFP. We attribute this difference to sterical inhibition of biological effects by relative large EGFP. Only six amino acids were added for the PDZ-based strategy. Another recent approach to fluorescently label α-synuclein was to add the 14 amino acids of a tetracystein motif to the C-terminus of α-synuclein (Roberti et al. 2007). The tetracystein motif binds arsenic fluorophores, which are smaller than EGFP but cannot be encoded genetically; they have to be added exogenously to the cells of interest and cause substantial toxicity on their own. PDZ-based labeling thus appears better suited for future applications in intact animals or organotypic cultures.
Even though PDZ-based labeling thus appears to unmask aggregation and toxicity of α-synuclein as compared to the EGFP fusion protein, it also induces significant aggregation and toxicity on its own. This raises the question to what extent the findings refer to intrinsic properties of α-synuclein. However, adding an additional challenge is commonly used to study aggregation and the effects of α-synuclein in cells. These include iron (Hasegawa et al. 2004; Roberti et al. 2007), synphillin-1 (McLean et al. 2001; Shin et al. 2005; Smith et al. 2005), proteasome inhibition, MPP+ (McLean et al. 2001) (Lee et al. 2001), or H2O2 (Jiang et al. 2006). The additional challenge substitutes for the stresses acting upon α-synuclein during a lifetime and may even be necessary to resolve differences between α-synuclein variants in model systems, by lifting them from a floor effect. Moreover, we observed some qualitative differences between aggregates formed in the presence of α-synuclein as compared to controls: Only aggregates induced by α-synuclein were reduced by co-expression of Hsp70 (Fig. 6) and only aggregates induced by α-synuclein were increased by proteasome inhibition (Fig. 7).
As to the correlation of aggregation and toxicity, we observed cells with few or no aggregates that underwent apoptosis and cells with large amorphous aggregates that remained viable over the entire 24 h of imaging. Aggregates thus represent neither a necessary nor a sufficient condition for cell death. The prevalence of aggregates and toxicity was nonetheless highly correlated for the α-synuclein variants investigated. But of course, this correlation could mean either that aggregates are the cause of cell death or that an injured cell is less able to clear aggregates. Eventually, α-synuclein variants that either form aggregates but are not toxic or the inverse would be needed to further clarify the relationship between aggregates and cell death.
Using a new way to visualize α-synuclein in living cells, we have described formation and clearance of α-synuclein aggregates by time-lapse imaging. Clearance of aggregates occurred primarily through the aggresome. Our new labeling strategy offers an experimental approach to address individual steps of this process in greater detail. The molecular chaperone Hsp70 not only decreased aggregate formation but also increased removal of aggregates through the aggresome. This capacity of cells to remove small and even larger aggregates offers the therapeutic perspective to not only prevent but also reverse disease progression in PD and other disorders associated with protein aggregation.
Funded by Deutsche Forschungsgemeinschaft through the DFG-Research Center for Molecular Physiology of the Brain (CMPB) and by the Volkswagen-Stiftung. We thank J. Sebastian Richter and Chor Hoon Poh for their contribution to the project and Dr Charles G. Glabe for providing the fibril-specific antibody.