14-3-3τ as a Modulator of Early α-Synuclein Multimerization and Amyloid Formation

The aggregation of α-synuclein (αS) plays a key role in Parkinson’s disease (PD) etiology. While the onset of PD is age-related, the cellular quality control system appears to regulate αS aggregation throughout most human life. Intriguingly, the protein 14-3-3τ has been demonstrated to delay αS aggregation and the onset of PD in various models. However, the molecular mechanisms behind this delay remain elusive. Our study confirms the delay in αS aggregation by 14-3-3τ, unveiling a concentration-dependent relation. Utilizing microscale thermophoresis (MST) and single-molecule burst analysis, we quantified the early αS multimers and concluded that these multimers exhibit properties that classify them as nanoscale condensates that form in a cooperative process, preceding the critical nucleus for fibril formation. Significantly, the αS multimer formation mechanism changes dramatically in the presence of scaffold protein 14-3-3τ. Our data modeling suggests that 14-3-3τ modulates the multimerization process, leading to the creation of mixed multimers or co-condensates, comprising both αS and 14-3-3τ. These mixed multimers form in a noncooperative process. They are smaller, more numerous, and distinctively not on the pathway to amyloid formation. Importantly, 14-3-3τ thus acts in the very early stage of αS multimerization, ensuring that αS does not aggregate but remains soluble and functional. This offers long-sought novel entries for the pharmacological modulation of PD.


ThT aggregation assay
Figure S1.αS aggregation curves in absence (grey) and presence (blue) of 14-3-3τ.The thioflavin-T intensity is plotted relative to the standard deviation of the background signal (σstd) as function of time.The threshold to determine the lag time is given at σstd = 20, plotted in green.Each condition represents the time average of a triplicate and the error bars represent their standard deviation.A. The black curve corresponds to the aggregation of 100 μM αS.The blue curves correspond to the aggregation of 100 μM αS in presence of 5 μM (light blue), 10 μM, 15 μM, 20 μM and 25 μM (increasingly darker blue) 14-3-3τ.The presence of 14-3-3τ delays the aggregation lag time, with higher concentrations of 14-3-3τ resulting in a stronger delay.B. The αS aggregation curves corresponding to the aggregation of 100 μM (black), 95 μM, 90 μM, 85 μM, 80 μM and 75 μM (light grey) total αS.The aggregation lag time does not change significantly between αS concentrations of 75 μM and 100 μM.

Thermal shift assay
Figure S2.Change in melting temperature (ΔTm) determined in a thermal shift assay of 14-3-3τ in presence of the known 14-3-3 binding partner Estrogen receptor alpha (ERα) or full-length αSyn-WT, relative to 14-3-3τ alone (53 °C).We observe an increased melting temperature (Tm) in presence of ERα, indicating interaction (positive control), while we observe no change in Tm in presence of αSyn-WT, indicating no interaction.Experiments are performed in duplicate three times.ΔTm is calculated per experiment.

Single molecule burst analysis
Figure S3. A. Normalized cumulative histograms of the passage times at low total αS concentration (grey) and high total αS concentration in absence of 14-3-3τ (blue), or B. presence of 10 μM 14-3-3τ (green).Error bars represent the standard deviation from three different measurements.

Figure S4.
Average passage times at low total αS concentration (grey) and high total αS concentration (blue), corresponding to the data in Figure S3A.We find 639 ± 9 μs and 681 ± 6 μs for low and high total αS concentration, respectively.The increase in 〈  〉 shows an increased population of slower diffusing, and hence larger species at high αS concentration.Error bars represent the standard deviation from three different measurements.

Robustness of MST experiments and model fits.
To study the robustness of the results derived from MST experiments on the multimerization of αS, we prepared and measured three independent samples of 50 nM αS-AF488 in presence of increasing amounts of αS-WT, performed at 37 °C.The MST results were fitted to the self-assembly model and parameters derived from the fits are presented in Figure S5.Additionally, we studied the influence of each measured individual data point in a single measurement series on the fitted MST multimerization data.We therefore performed 5 repeats of the MST measurements on one sample containing 50 nM αS-AF488 and increasing amounts of αS-WT at 23 °C.We fitted the resulting average MST multimerization curve (Figure S6A) and the individual MST multimerization curves (Figure S6B-F) to the self-assembly model.The shaded areas, representing the 0.5 th and 3 rd percentile confidence intervals of the fits, are similarly shaped independent of the exact values obtained for σ and   .
Next we shuffled the data sets and created 500 additional MST multimerization curves from the 5 repeats by randomly selecting data points for each αS-WT concentration.These data sets were again fitted to the self-assembly model.Indeed, the resulting values of σ and   follow the confidence interval areas (Figure S6) as expected.
Together, the data presented in Figure S5 and Figure S6 evidences the robustness of the experiments and data analysis.Note that the shape of the confidence intervals may represent the characteristics of the multimerization curve even more accurately than the exact fitted values.A change in the shape of the confidence intervals, like we observe in presence of increasing amount of 14-3-3τ (Figure 5A), therefore also shows the different multimerization characteristics.
Finally, we find that the standard deviation of the measured MST response at given αS-WT following from the 5 repeats is smaller than the standard deviation determined based on the intra-variability for an individual measurement (see Methods section).Hence, the intra-variability serves as a fair, fast and easy approximation of the actual variance of the measurement, or may even overestimate it.

Fluorescence correlation spectroscopy (FCS) experiments on αS30
To obtain a rough indication for the aggregation number of the αS multimers formed in the SM and MST experiments, we determined the diffusion passage time of αS30, a higher order αS species of known aggregation number.We labelled αS30 with AF488 as described for the labelling of αS30-488/568 (see Materials and Methods), but with approximately 1 αS-AF488 per 3 αS30.FCS measurements were performed on the same setup as the single molecule burst detection experiments (see Materials and Methods).Fluorescence fluctuation traces were measured for 180 s.We calculated the fluorescence autocorrelation curve for lag times between 0.03 ms and 1000 ms.We fitted this curve with a pure diffusion model and determine a diffusion time of 900 μs for αS30.This analysis was performed with the SymphoTime64 software (PicoQuant).

Negative control effect on αS multimerization
We used PEG20k as a negative control.PEG20k does not interact with αS and should therefore not affect the multimerization at the micromolar concentrations typically used for interaction studies.

Figure S9
. Normalized fluorescence emission of αS30-488/568 in absence (grey) and presence of 2 μM 14-3-3τ (black).The sample was excited at 460 nm and the emission is normalized to the maximum fluorescence intensity peak at 517 nm.Addition of 14-3-3τ results in a higher FRET signal, indicating compaction of αS30 upon binding with 14-3-3τ.

Figure
Figure S6. A. The self-assembly model fit result of 50 nM αS-AF488 + (3 nM -110 μM) αS-WT performed at 23 °C.The dark and light blue shaded areas indicate the 0.5 th and 3 rd percentile confidence intervals, respectively.The fit is performed on the average MST response of a 5 times repeated measurement at 60% IR laser power.The white dots represent fit results performed on 500 MST multimerization curves with data randomly selected from the 5 repeats.Brighter dots indicate multiple occurrences.B-F.Results of fits to the 5 individual repeats.The black dashed line indicates the minimum and maximum critical concentration at given σ based on the 3 rd percentile confidence intervals of B-F.

Figure S7 .
Figure S7.FCS autocorrelation curve for αS30.The normalized G(τ) is plotted as function of the correlation lag time (τ).The curve is normalized to G(τ = 0.03 ms).The curve was fitted to a single species diffusion model with a diffusion time of 900 μs (red).

Figure S8 .
Figure S8.Value of σ for αS multimerization in absence (grey) and presence (blue) of either 14-3-3τ or PEG20k at comparable concentrations.The dashed line indicates the transition between cooperative and non-cooperative multimerization.Presence of 14-3-3τ results in non-cooperative multimerization of αS.A comparable concentration of PEG20k does not change αS multimerization.The error bars indicate the 0.5 th confidence interval for the fitted σ.