Synergistic influence of phosphorylation and metal ions on tau oligomer formation and coaggregation with α-synuclein at the single molecule level

Mass spectroscopy showed that recombinant protein tau is of high purity and does not contain significant amounts of cleavage products (Figure 1A). Notably, fluorescence labeling of tau with Alexa dyes did not alter the protein’s electrophoretic mobility, though mass spectroscopy showed an increase in tau’s molecular weight after labeling (data not shown). We confirmed in vitro phosphorylation of human recombinant tau by demonstrating a typical band shift of the protein in western blot (see Figure 1B) [17]. We further evaluated whether the SIFT method can be employed to detect phosphorylated protein tau by labeling with Alexa-488 tagged antibodies AT-8 and T46. While T46 detects both phosphorylated and unphosphorylated tau, the AT-8 antibody requires the protein to be phosphorylated at specific sites (Goedert et al, 1995). The Aβ specific antibody 6E10 was used as a negative control. SIFT analysis shows that both T46⁴⁸⁸ and AT-8⁴⁸⁸, but not 6E10⁴⁸⁸ bind to phosphorylated tau⁶⁴⁷ (pTau, Figure 1C), while only T46⁴⁸⁸ binds to mock phosphorylated tau⁶⁴⁷ (mTau, also see materials and methods section).

Our previous studies revealed that the organic solvent DMSO and the metal ion aluminium (Al³⁺) induce the formation of distinct tau oligomer species [49]. In order to evaluate the influence of protein phosphorylation on the effects of these aggregation inducers, we employed GSK-3β to create phosphorylated tau (pTau) and compared its aggregation behavior to mock phosphorylated tau (mTau). Our findings show that oligomerization of both pTau and mTau can be induced by 1% DMSO, with mTau showing a higher rate of aggregation than pTau (p < 0.001, Figure 2A). Conversely, in presence of 10 μM Al³⁺, pTau oligomerization exceeds mTau (p < 0.05), with the overall aggregation level being distinctly higher compared to DMSO for both pTau and mTau (also see Additional files 1 and 2).

2D histograms of detected aggregates show the formation of distinct oligomer species in presence of DMSO and Al³⁺ (Figure 2B). In presence of DMSO, tau protein shows only moderate aggregation, while Al³⁺ enhances the formation of larger oligomers with higher fluorescence intensity.

Fluorescence intensity distribution analysis (FIDA) of tau oligomers induced by DMSO yielded larger aggregates for mTau (mean 76 molecules) than for pTau (mean 26 molecules). In presence of Al³⁺, large tau oligomers comprising on average 240 (pTau) and 220 (mTau) molecules occurred within 15 minutes, while the overall aggregation level was higher for pTau (Figure 2A, for aggregation dynamics also see Additional file 3). The combination of DMSO and Al³⁺ increased aggregation levels compared to Al³⁺ alone predominantly for mTau (also see Additional files 1 and 2).

Furthermore, we examined the SDS stability of oligomers generated by the different enhancers (Figure 2C, D). While DMSO induced aggregates dissolve upon addition of the ionic detergent SDS in a final concentration of 0.2%, more than 50% of the Al³⁺ induced oligomer signal remains stable 100 min after addition of SDS.

In conclusion, Al³⁺ promotes rapid formation of detergent resistant oligomers, preferably of phosphorylated protein tau. Conversely, DMSO induced aggregation is diminished upon tau phosphorylation.

Since the role of tau and α-syn coaggregation is increasingly recognized, we investigated the effect of tau⁴⁸⁸ phosphorylation on its coaggregation with α-syn⁶⁴⁷. As previous work has shown a major influence of Fe³⁺ on α-syn oligomerization [36‐38], we also used Fe³⁺ at a final concentration of 10 μM in our assay.

In these experiments, 1% DMSO promoted the coaggregation of tau and α-syn in a similar manner for pTau and mTau (Figure 3A). On average, coaggregates were composed of 11 pTau + 8 α-syn monomers or 15 mTau + 12 α-syn monomers, respectively, as determined by FIDA analysis. Fe³⁺ also had a promotional influence on coaggregation, which was higher for pTau (p < 0.01, on average 53 pTau + 10 α-syn and 39 mTau + 3 α-syn monomers per oligomer). Upon combining DMSO and Fe³⁺, an additional effect exceeding the influence of one inducer alone can be seen (also see Additional files 4 and 5), while no significant difference in aggregation levels of pTau and mTau was detectable. Similar to our tau oligomerization experiments, Al³⁺ was the strongest inducer of tau and α-syn coaggregation, with pTau showing higher coaggregation activity than mTau (p < 0.001, Figure 3A; on average 131 pTau + 115 α-syn and 137 mTau + 80 α-syn monomers per oligomer). This difference was still detectable when combining Al³⁺ with DMSO, though to a lesser extent (p < 0.05). Mixed oligomers comprising both tau and α-syn appear scattered around the bisectrix of these histograms, while homogeneous aggregates are located close to the axes, as illustrated in Figure 3D.

Concerning oligomer stability, 0.2% SDS reduced the number of coaggregates formed in presence of every tested agent (Figure 3B, C). Again, coaggregates formed in presence of metal ions proved to be more resistant to SDS than those induced by DMSO, though to an overall lesser extent than the homogeneous tau aggregates. Oligomer sizes before and after addition of SDS as determined by FIDA analysis are provided in Additional file 6. The presence of Al³⁺-induced, SDS stable mixed oligomers was further validated applying gel filtration and fluorescence spectroscopy. In these experiments, Al³⁺ induced oligomers were separated from monomers by gel filtration. SDS stable mixed tau and α-syn oligomers eluted prior to monomers, as demonstrated by cross-correlation and SIFT-2D analysis (see Additional file 7).

Thus, our data show increased tau and α-syn coaggregation after tau phosphorylation in presence of trivalent metal ions, and that the heterogenic oligomers generated by Fe³⁺ and Al³⁺ are more stable to SDS treatment than those formed by DMSO.

Giasson et al. showed that tau and α-syn induce each other’s filament formation, and that some of the fibrils formed in presence of both proteins comprise tau and α-syn segments [41]. Such findings suggest that these segments are either assembled by end to end annealing of short filament fragments, or that such fragments act as seeds for both proteins. To evaluate the effect of tau phosphorylation on the interaction of tau oligomers with monomeric α-syn, tau⁴⁸⁸ was first incubated in presence of 1% DMSO, 10 μM Fe³⁺, 10 μM Al³⁺, and combinations of these agents for 90 minutes to generate pre-formed tau oligomers. α-Syn⁶⁴⁷ monomers were then added to the assay and measurements were conducted (Figure 4A).

Contrary to the experiments with tau monomers, DMSO had a stronger effect on the coaggregation of pre-formed mTau oligomers with monomeric α-syn compared to pTau oligomers (p < 0.05, Figure 4A). Under these conditions, coaggregates contained on average 20 pTau + 2 α-syn monomers or 27 mTau + 5 α-syn monomers, respectively.

pTau oligomer coaggregation with monomeric α-syn exceeded mTau seven- to ninefold in presence of Fe³⁺ (p < 0.01; on average 197 pTau + 14 α-syn or 175 mTau + 3 α-syn monomers per oligomer) and Al³⁺ (p < 0.001; on average 151 pTau + 73 α-syn or 154 mTau + 35 α-syn monomers per oligomer). This difference was markedly reduced upon combining Fe³⁺ and 1% DMSO, which again yielded a synergistic effect compared to the two inducers alone (also see Additional files 8 and 9). The combination of Al³⁺ and 1% DMSO also reduced the difference between pTau and mTau to a two-fold excess of pTau coaggregation (p < 0.01).

SDS (0.2%) reduced the number of oligomers formed in presence of every tested inducer (Figure 4B). Especially aggregates induced by metal ions alone showed enhanced resistance to SDS treatment compared to those formed in presence of DMSO (Figure 4C).

It has been demonstrated that in vitro phosphorylation of human protein tau by GSK-3β can be verified using SDS-PAGE by demonstrating a complete band shift of phosphorylated tau compared to the unphosphorylated protein [17]. We verified efficient tau phosphorylation indicated by a complete band shift compared to the unphosporylated protein (Figure 1B). In addition, we investigated whether fluorescently labeled antibodies can be employed in the SIFT method to verify tau phosphorylation at nanomolar protein concentrations, allowing for possible diagnostic implementations of this method in the future. Our data corroborate that tau phosphorylation was successful and that antibodies can be employed in the SIFT method to identify proteins and posttranslational modifications such as phosphorylation at nanomolar protein concentrations (Figure 1C). Such an assay may provide a valuable tool for future diagnostic applications e.g. the detection of novel specific CSF biomarkers of neurodegeneration.

In vitro studies have yielded conflicting results regarding the influence of tau phosphorylation on its assembly to filaments. To date, both inhibitory and promotional effects of GSK-3β mediated phosphorylation on tau filament assembly were found in absence or presence of polyanionic aggregation inducers [17‐20, 50, 51]. Notably, for Al³⁺ induced tau aggregation, phosphorylation was consistently found to be enhancing or, in some experiments, even a prerequisite [24‐27].

However, the methods applied in these studies, such as electron microscopy, thioflavine fluorescence, laser light scattering, SDS-PAGE and western blot, are of limited use in directly examining single molecule interactions at the oligomer level. Furthermore, these techniques often require high concentrations of both protein and aggregation inducer that might not depict physiological conditions, especially given the fact that tau protein readily polymerizes at high concentrations without any inducer [52, 53]. In addition, the majority of studies investigating brain Al³⁺ concentrations did not detect concentrations exceeding 10 μg/g brain mass (dry weight) in AD brains [54], which approximately corresponds to 60 μmol/kg (wet weight), while many studies examining the influence of Al³⁺ on tau aggregation in vitro applied Al³⁺ concentrations in the milimolar range [25, 27, 55].

In this study, we applied confocal single molecule fluorescent techniques to investigate the influence of protein phosphorylation by GSK-3β on tau oligomer formation at the single molecule level at nanomolar concentrations. Exposure to 10 μM Al³⁺ induced the rapid formation of large tau oligomers, while DMSO at a final concentration of 1% led to the formation of smaller oligomers. The large Al³⁺ triggered oligomers contained an average of 220 to 240 molecules as shown by FIDA analysis, and were resistant to treatment with 0.2% SDS, which in contrast readily dissolved the smaller oligomers formed in presence of DMSO. The oligomer sizes observed here were comparable with data presented earlier [49]. Interestingly, phosphorylation by GSK-3β yielded an increase in Al³⁺ induced tau oligomer formation, while oligomer size was comparable for pTau and mTau.

Thus, our data identify tau phosphorylation and physiological concentrations of the metal ion Al³⁺ as synergistic inducers of the formation of SDS resistant tau oligomers even at nanomolar protein concentrations. These findings substantiate the hypothesis that Al³⁺ may play a role in the formation of neurotoxic oligomers even at early stages of neurodegeneration.

Co-deposits of tau and α-syn have been found in several neurodegenerative diseases, including AD, PSP, CBD and DLB [39, 40]. Since interactions of different proteins are difficult to monitor in vitro, only few studies investigating the interaction of tau and α-syn have been published so far. It was demonstrated that in mixed micromolar solutions of α-syn and tau, both proteins’ minimal concentrations for filament assembly are decreased [41, 56]. Moreover, Giasson et al. detected filaments that comprised both tau and α-syn, proving that the two proteins directly interact in mixed solutions [41]. A more recent study showed that α-syn fibrils can be taken up by tau overexpressing cells and induce the formation of phosphorylated triton-insoluble tau oligomers [57]. In vivo studies further demonstrated that tau deposits can be found in mice overexpressing pathologic human α-syn mutations [58].

In this study, we demonstrate that coaggregation of tau and α-syn takes place even at nanomolar protein concentrations, is strongly induced by trivalent metal ions, and differentially modulated by tau’s phosphorylation status. In contrast to the findings of Giasson et al., we did not detect tau and α-syn coaggregation in the absence of aggregation inducers. This apparent discrepancy is most likely explained by the differences in protein concentrations (up to 1000-fold) and the shorter observation time [41]. However, our data demonstrate that tau and α-syn coaggregation occurs even at nanomolar protein concentrations in presence of aggregation inducers. As in our tau aggregation assay, Al³⁺ had the most pronounced effect on coaggregation of α-syn with tau monomers and oligomers. Al³⁺ induces rapid coaggregation of tau and α-syn (see Additional file 3). Fe³⁺ and DMSO also induce coaggregation of the two proteins, though to an overall lesser extent. In our experiments, tau phosphorylation by GSK-3β strongly enhanced the formation of mixed oligomers induced by Al³⁺ or Fe³⁺. Moreover, the mixed oligomers resulting from metal ion induced coaggregation proved to be more resistant to SDS treatment than those formed in presence of DMSO.

We further provide data from FIDA analysis and gel filtration experiments demonstrating the presence and size of mixed tau and a-syn oligomers (see Additional files 6 and 7). Earlier studies applying atomic force micoscropy and SIFT established FIDA-based quantification of oligomer sizes as a reliable method [38].

These results extend the current model of tau and α-syn interaction at early stages of neurodegeneration. As the cytoplasmic concentration of unbound tau and α-syn is low in pre-clinical neurodegeneration, the presence of pro-aggregatory factors may be crucial in the generation of early mixed oligomers. We identify trivalent metal ions and tau phosphorylation by GSK-3β as potential inducers of such oligomer formation even at nanomolar protein concentrations. A time-dependent increase of oligomer concentrations may then induce self-propagating and cross-seeding mechanisms as demonstrated by Giasson and Waxman [41, 57].

Proteins were labeled with fluorescent dyes Alexa-488-O-succinimidylester and Alexa-647-O-succinimidylester, respectively, as described previously [48]. Tau isoform 5 was incubated with a four-fold molar excess of Alexa-488 or Alexa-647 in 100 mM NaHCO₃ for 24 h at room temperature. For α-syn, a two-fold molar excess of dye was used. Antibodies T46 (Invitrogen, Carlsbad, CA), AT-8 (Pierce Endogen, Rockford, IL) and 6E10 (Acris Antibodies, Hiddenhausen, Germany) first underwent buffer exchange to PBS pH 7.0. Since antibody concentrations after buffer exchange were unknown, concentrations of Alexa-488-O-succinimidylester were determined empirically. Unbound dye was separated using a PD10 desalting column. Fluorescently labeled proteins are further referred to as tau⁴⁸⁸, tau⁶⁴⁷, and α-syn⁶⁴⁷.

Tau phosphorylation was conducted as described previously [17]. Fluorescently labeled protein tau was incubated with Glycogen Synthase Kinase 3β (Sigma, Saint Louis, MO) in a ratio of 0,018 U GSK-3 β / pmol tau in buffer containing 33 mM tris pH 7.5, 40 mM hepes pH 7.64, 100 mM NaCl, 5 mM EGTA, 3 mM MgCl₂, 2 mM ATP, 180 mM sucrose, 0.13 mM PMSF, 0.67 mM benzamidine, 0.067% mercaptoethanol and 0.02% Brij-35. Samples were incubated for 20 h at 30 °C under constant shaking. As control, fluorescently labeled protein tau was incubated in buffer containing all components except for the enzyme, and is further referred to as mock phosphorylated tau (mTau). Protein phosphorylation was confirmed by sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) and western blot using antibody T46 as described previously [60]. Phosphorylated protein tau (pTau) featured a typical band shift compared to mTau, as described previously [17] (Figure 1B).

Stock solutions of tau⁴⁸⁸, tau⁶⁴⁷, and α-syn⁶⁴⁷ were prepared for a final concentration of 10 molecules per focal volume, equivalent to a concentration of 10 nM to 20 nM. For coaggregation experiments, a 1:1 ratio of tau⁴⁸⁸ and α-syn⁶⁴⁷ was used. Prior to each experiment, tau samples were centrifuged at 100.000 g for 30 min to remove pre-formed aggregates, and all protein stock solutions were then controlled for pre-existing aggregates by SIFT [49]. Only samples free of pre-formed aggregates were used. All experiments were conducted in 96 well plates with a cover slide bottom, in a total sample volume of 20 μl per well. Plates were covered with adhesive film to obviate evaporation. For aggregation and coaggregation experiments, protein stock solutions were added to wells containing 50 mM tris buffer pH 7.0, 10 μM AlCl₃, 10 μM FeCl₃ or 1% dimethyl sulfoxide (DMSO), and measurements were started immediately. For coaggregation of tau⁴⁸⁸ oligomers with α-syn⁶⁴⁷ monomers, samples of tau⁴⁸⁸ were prepared as described above and pre-incubated for 90 min. Subsequently, α-syn⁶⁴⁷ monomers were added and measurements were started immediately. To investigate oligomer stability, a 10-fold SDS stock solution for a final concentration of 0.2% was added to each well after each experiment, and measurements were repeated. Each well was measured four times, resulting in a total duration of each experiment of 60 to 120 minutes, depending on experimental layout.

FCS and SIFT were performed on an InsightReader (Evotec-Technologies, Hamburg, Germany) with dual color excitation at 488 and 633 nm as described previously [48, 49, 61]. Data was collected separately for each excitation wavelength by two single-photon detectors. Analysis was performed using FCSPP evaluation software version 2.0 (Evotec-Technologies), allowing auto-correlation, cross-correlation and fluorescence intensity distribution (FIDA) analysis, and SIFT-2D evaluation software (Evotec-Technologies). For SIFT-2D analysis, photons were added up in time intervals (bins) of 40 μs and illustrated in a 2D scatterplot. Scatterplots depicted in this paper contain all photons of all consecutive measurements of one sample. Photon-weighted SIFT aggregation and coaggregation analysis was performed as described previously [49]. As cross-correlation detects small oligomeric protein aggregates with higher sensitivity, only cross-correlation data is depicted for experiments on tau aggregation and coaggregation with α-syn (for SIFT data see Additional files 1, 4, and 7).

Mass spectroscopy was performed according to established protocols to verify the purity of tau protein stock solution [61].

Normal distribution of data was determined by Shapiro-Wilk test. If normal distribution was confirmed, a two-sided student’s t-test preceded by Levene’s test for equality of variance was performed. A paired student’s t-test was used for experiments on SDS-resistance of oligomers. If normal distribution was not confirmed, Mann–Whitney U-test was performed. Bonferroni-adjustment for multiple testing was done where appropriate. Data is demonstrated as average of all independent samples, and a paired student’s t-test was performed including all independent samples. Error bars in figures show the standard error of the mean.