Liquid-Liquid Phase Separation of Tau Driven by Hydrophobic Interaction Facilitates Fibrillization of Tau

Abstract

Amyloid aggregation of tau protein is implicated in neurodegenerative diseases, yet its facilitating factors are poorly understood. Recently, tau has been shown to undergo liquid liquid phase separation (LLPS) both in vivo and in vitro. LLPS was shown to facilitate tau amyloid aggregation in certain cases, while being independent of aggregation in other cases. It is therefore important to understand the differentiating properties that resolve this apparent conflict. We report on a model system of hydrophobically driven LLPS induced by high salt concentration (LLPS-HS), and compare it to electrostatically driven LLPS represented by tau-RNA/heparin complex coacervation (LLPS-ED). We show that LLPS-HS promotes tau protein dehydration, undergoes maturation and directly leads to canonical tau fibrils, while LLPS-ED is reversible, remains hydrated and does not promote amyloid aggregation. We show that the nature of the interaction driving tau condensation is a differentiating factor between aggregation-prone and aggregation-independent LLPS.

Keywords: amyloid aggregation; electron paramagnetic resonance; hydrophobic interaction; liquid–liquid phase separation; tau.

Figure 1. Tau undergoes LLPS at high salt concentration.

A. Charge and Hopp-Woods hydrophobicity plot of 2N4R. Data points show the average values of consecutive 25 amino acids. Reference values of Glycine are shown as dashed lines. tau187 fragment (residues 255–441), four repeat domains (R1–R4, 244–368) and two native cysteines (C291, C322) are shown. B. Microscope images of tau187 and 2N4R at varying [NaCl]. 44 μM tau187 and 20 μM 2N4R were chosen to have same mass concentration. Scale bar length is 25 μm. C. Droplet merging of tau LLPS-HS. 450 μM tau187 with 2.2 M NaCl was used. Scale bar length is 50 μm. D. Effects of 4 wt% 1,6-hexanediol on different types of tau LLPS. Tau-RNA LLPS (+RNA) was prepared with 20 μM 2N4R and 40 μg/mL polyU RNA; Tau low salt simple coacervation (low salt) was prepared with 20 μM 2N4R at 5 mM NaCl; Tau high salt simple coacervation (high salt) was prepared with 100 μM tau187 with 4.75 M NaCl. Images were taken 10 minutes after sample preparation. Scale bar length is 25 μm.

Figure 2. FLIM fits of tau-RNA LLPS-complex coacervation (complex), and tau LLPS-high salt (high salt).

Fluorescence Lifetime Imaging Measurements were taken at 4 and 48 hours at the same resolution, scale bar = 25 μm. A-D, Fluorescent microscope images of tau-RNA droplets (25 μM 2N4R, 50 μg/mL RNA, 100 mM NaCl) and tau high salt droplets (25 μM 2N4R, 10% BODIPY-labeled, 4.75 M NaCl) before (4 hour) and after incubation (48 hour). Samples were prepared and incubated in 20 mM HEPES pH 7.0 at room temperature. Individual pixels were fitted with a 2-component exponential decay using the FLIMfit software tool developed at Imperial College London (Warren SC et al, PLOS ONE 2013), and the data were visualized using the higher lifetime component. The colors in the fluorescent images and corresponding histograms represent a pseudo color heat map that ranges from 3.0 to 4.8 ns, with blue representing low lifetimes and red representing high lifetimes. E-F. Histograms of fitted results in A-D. The histograms are normalized so that area under the curve is unity. The color under each histogram corresponds to the pseudo color of the heat map at the histograms max value. Solid line = 4 hours, Dashed line = 48 hours.

Figure 3.. Correlation of LLPS-high salt and amyloid aggregation.

A. Representative microscope images at t = 0 and ThT fluorescence at room temperature overnight of tau LLPS-high salt at various [NaCl]. 44 μM tau187 was used. B. Representative TEM image of tau LLPS-high salt sample after overnight incubation, compared with tau-heparin fibrils. C. Correlation of initial turbidity, measured by averaged OD₅₀₀ during t = 0 ~ 30 min, and final ThT fluorescence at t = 16 h of tau LLPS-high salt vs LLPS-complex coacervation (CC). 40 𝜇𝑀 ± 10% of tau187 was used with varying [NaCl]. For tau LLPS-high salt, [NaCl] varies from 0 ~ 4.75 M. For tau-heparin CC, [NaCl] varies from 0.020 ~ 0.080 M. For tau-RNA CC, [NaCl] varies from 0 ~ 0.150 M. Solids line and shadow show linear regression and its standard error, respectively. Samples come from 5 different batches of proteins. Each batch of protein contributes to 1~3 independent measurements. Each measurement is represented as one dot in the plot.

Figure 4. LLPS high salt undergoes dehydration.

A. Schematic diagram showing sensitive hydration shell in Electron Spin Echo Envelope Modulation (ESEEM) and Overhauser Nuclear Dynamic Polarization (ODNP). B. Representative 3 pulse-ESEEM of tau187 at site 313 at solution (before) and upon addition of 3.75 M NaCl (after). 22.5% Ficoll was used as glassing reagent. C. Representative ODNP cross-relaxivity parameter 𝑘_𝜎 of tau187 at various sites at solution (before) and upon addition of 3.75 M NaCl (after). 250 μM tau was used. For comparison, 𝑘_𝜎 of tau187 at site 322 upon tau-RNA complex coacervation was shown (322CC). 80% reduction of 𝑘_𝜎 has been reported at site 313, 322 and 404 for tau-heparin fibrils (Pavlova, et al 2016), and is shown as dashed line.

Figure 5.. Site specific dynamics of tau187 upon LLPS-high salt.

A. Representative X-band cwEPR lineshape of tau187 at site 322 at varying [NaCl]. Black lines show the experiment data and red lines show the fit. Spectra are shifted to avoid overlap. B. Rotational correlation time, 𝜏_𝑅, of tau187 at site 303, 313, 322 and 404 at varying [NaCl]. Data at low [NaCl] were fit with 1 component with the y-axis showing the 𝜏_𝑅. Data at high [NaCl] were fitted with 2 components with the area of the disk proportional to the percentage of each component.

Figure 6. Effects of P301L mutation on LLPS and amyloid aggregation of tau at high salt concentration.

A. Representative microscope images of tau187 wildtype (WT) and P301L mutants at varying [NaCl]. B. Turbidity, measured by OD₅₀₀, of corresponding samples in A. Data points come from independent measurements from 5 different batches of proteins. p-values were calculated from unpaired two sample t-test. C. ThT fluorescence of samples in A after overnight incubation at room temperature. Ribbons in C show standard deviation of readings from 5 different batches of proteins. D. Correlation of initial turbidity, measured by averaged OD₅₀₀ during t = 0 ~ 30 min, and final ThT fluorescence at t = 16 h of tau LLPS-high salt at varying [NaCl] using tau187 WT and P301L mutants. Solids line and shadow show linear regression and its standard error, respectively. Black points and line are duplicated from (Figure 3C). E. Half time, defined as the time when ThT fluorescence reach half maximum of WT and P301L tau187 LLPS-high salt at varying [NaCl]. In both D and E, 40 𝜇𝑀 ± 10% of tau187 was used. Samples come from 5 different batches of proteins. Each batch of protein contributes to 1~3 independent measurements. Each measurement is represented as one dot in the plot.