Secondary nucleating sequences affect kinetics and thermodynamics of tau aggregation

Abstract

Tau protein was scanned for highly amyloidogenic sequences in amphiphilic motifs (X)(n)Z, Z(X)(n)Z (n ≥ 2), or (XZ)(n) (n ≥ 2), where X is a hydrophobic residue and Z is a charged or polar residue. N-Acetyl peptides homologous to these sequences were used to study aggregation. Transmission electron microscopy (TEM) showed seven peptides, in addition to well-known primary nucleating sequences Ac(275)VQIINK (AcPHF6*) and Ac(306)VQIVYK (AcPHF6), formed fibers, tubes, ribbons, or rolled sheets. Of the peptides shown by TEM to form amyloid, Ac(10)VME, AcPHF6*, Ac(375)KLTFR, and Ac(393)VYK were found to enhance the fraction of β-structure of AcPHF6 formed at equilibrium, and Ac(375)KLTFR was found to inhibit AcPHF6 and AcPHF6* aggregation kinetics in a dose-dependent manner, consistent with its participation in a hybrid steric zipper model. Single site mutants were generated which transformed predicted amyloidogenic sequences in tau into non-amyloidogenic ones. A M11K mutant had fewer filaments and showed a decrease in aggregation kinetics and an increased lag time compared to wild-type tau, while a F378K mutant showed significantly more filaments. Our results infer that sequences throughout tau, in addition to PHF6 and PHF6*, can seed amyloid formation or affect aggregation kinetics or thermodynamics.

FIG 1

(A) Schematic of full-length tau showing the 20 sequences which fit the motifs (X)_nZ or Z(X)_nZ (n≥2) or (XZ)_n (n≥2), where X is a hydrophobic residue and Z is a charged or polar amino acid. The model assumes that these sequences have a high propensity to act as nucleating sequences in amyloid formation. (B) Amyloidogenic propensities for sequences calculated by taking the sum of the absolute values of constituent amino acid propensities (40).

FIG 2

TEMs of high propensity peptide sequences in MOPS-Cl-HMWH buffer. Scale bar represents 100 nm.

FIG 3

(A) Far-UV CD spectra of 100 μM N-acetylated peptide amides in water or MOP-Cl-HMWH buffer (pH 7.2). (B) CD of AcPHF6, Ac¹⁰VME (MOPS-Cl-HMWH), the calculated mean residue ellipticity of a 1:1 stoichiometric mixture, and the observed ellipticity of the mixture. (C) as in (B) for AcPHF6 and AcPHF6*. (D) Difference between calculated and observed ellipticity for the peptides. A positive interaction is one in which the observed CD of a 1:1 mixture is more negative above (higher wavelength) the isodichroic point and more positive below this point. In contrast, a negative interaction infers that a smaller fraction of the sample exists in an aggregated state when the two peptides are mixed.

FIG 4

Aggregation kinetics of peptides and peptide mixtures in MOPS-Cl-HMWH followed by fluorescence of ThS at 490 nm (440 nm excitation). AcPHF6* was 100 μM. (A) AcPHF6* (filled circles), AcPHF6*:Ac³⁷⁵KLTFR at 2:1 (filled squares), 1:1 (filled triangles) and 1:2 stioichiometry (crosses), in MOPS-Cl, pH 7.2. Solid lines represent best fits to the data, assuming a single Gompetz growth curve model for pure AcPHF6* and AcPHF6*:Ac³⁷⁵KLTFR mixtures at less than 1:1 stoichiometry. The AcPHF6*:Ac³⁷⁵KLTFR mixture at 1:2 stoichiometry was modeled using a sum of Gompetz growth curves (Eq. [2]). (B) as in (A) with AcPHF6* and AcPHF6*:Ac³⁴³KLDFK mixtures in phosphate-Cl buffer at pH 3.0. (C) Aggregation kinetics of AcPHF6 (filled circles), AcPHF6* (filled triangles), and a 1:1 stoichiometric mixture of AcPHF6:AcPHF6*. Solid lines represent best fits to a single Gompetz growth curve, in the case of pure peptides, or a sum of two Gompetz growth curves, in the case of the mixture. The dashed line represents a sum of fitted curves for AcPHF6 and AcPHF6*, assuming no interaction between the peptides.

FIG 5

Plots of parameters obtained from fitting kinetic data for AcPHF6* and AcPHF6*:Ac³⁷⁵KLTFR and AcPHF6*:Ac³⁴³KLDFK mixtures. (A) Log(k_app) for first aggregation event of AcPHF6* and AcPHF6*:Ac³⁷⁵KLTFR mixtures at pH 7.2 (filled diamonds) and for AcPHF6* and AcPHF6*:Ac³⁴³KLDFK mixtures (exempting AcPHF6*:Ac³⁴³KLDFK at 1:2 stoichiometry) at pH 3.0 (open diamonds). (B) Lag time for first aggregation event of AcPHF6* and AcPHF6*:Ac³⁷⁵KLTFR mixtures. Data could be fit with an r² > 0.97.

FIG 6

2N4R lysine mutants show variations from WT tau. Mutant forms (M11K, F378K, and Y394K (diamond, triangle and square respectively)) and WT (circle) 2N4R tau were assayed by LLS over the course of 240 minutes (first 100 minutes shown) (A). Data was fit to the Gompertz growth curve (M11K, F378K, Y394K and WT (solid, long/short dash, short dash and long dashed lines respectively)). Representative TEM images of WT (B), M11K (C), F378K (D), and Y394K (E) were collected at 3600x magnification. Images were quantitated using Image Pro Plus for filament number (F), length (G), and mass (H). Data in F-H are averages of 5 fields ± SEM (p ≤ 0.0001 for bars marked with asterisks). Scale bar represents 1 μm.

FIG 7

Hypothetical model for AcPHF6 and Ac³⁷⁵KLTFR in hybrid steric zipper. At 1:1 stoichiometry antiparallel layers of parallel in-register chains of AcPHF6 and Ac³⁷⁵KLTFR continue to form amyloid, albeit at a reduced rate of aggregation.