Fracture and Growth Are Competing Forces Determining the Fate of Conformers in Tau Fibril Populations

Abstract

Tau fibrils are pathological aggregates that can transfer between neurons and then recruit soluble Tau monomers by template-assisted conversion. The propagation of different fibril polymorphs is thought to be a contributing factor to phenotypic diversity in Alzheimer disease and other Tauopathies. We found that a homogeneous population of Tau fibrils composed of the truncated version K18 (residues 244-372) gradually converted to a new set of fibril conformers when subjected to multiple cycles of seeding and growth. Using double electron-electron resonance (DEER) spectroscopy, we observed that the distances between spin labels at positions 311 and 328 in the fibril core progressively decreased. The findings were corroborated by changes in turbidity, morphology, and protease sensitivity. Fibrils that were initially formed under stirring conditions exhibited an increased fragility compared with fibrils formed quiescently after multiple cycles of seeding. The quiescently formed fibrils were marked by accelerated growth. The difference in fragility and growth between the different conformers explains how the change in incubation condition could lead to the amplification of a minor subpopulation of fibrils. Under quiescent conditions where fibril breakage is minimal, faster growing fibrils have a selective advantage. The findings are of general importance as they suggest that changes in selective pressures during fibril propagation in the human brain could result in the emergence of new fibril conformers with varied clinicopathological consequences.

Keywords: Alzheimer disease; Tau protein (Tau); aggregation; amyloid; electron paramagnetic resonance (EPR); fibril; prion.

FIGURE 1.

Structural analysis of Tau fibrils by CW EPR. A, schematic of multicycle seeding reaction. K18 monomers are mixed with a 2-fold molar excess of heparin and allowed to form fibrils for 3 days while stirring. After sonication, the fibril seeds are mixed with fresh K18 monomers to produce a new generation of fibrils referred to as cycle 1. The procedure is repeated up to 15 times. In these templated reactions, fibrils are grown quiescently. At distinct steps, seeds are removed and used to form an independent set of fibrils for structural analysis. These fibrils have 2% spin-labeled K18 311/328 incorporated into their structure and are analyzed by EPR. The dilution is necessary to avoid spin-spin interactions along the fibril axis. B, CW EPR spectra of K18 fibrils from cycles 1, 5, 10, and 15 collected at a scan width of 150 G. The spectra are the first derivatives of the corresponding absorption spectra. The separation of outer peaks (67 G) indicates that the spin labels are immobilized in the fibril core.

FIGURE 2.

Structural evolution of K18 fibrils monitored by DEER and EM. DEER data were collected for Tau fibrils following different cycles of seeding. Data are analyzed by Tikhonov regularization. Background subtracted dipolar evolution curves in the time domain (A) and frequency domain (B) shown as black traces with best fits in red. C, distance distributions. The red arrow highlights changes of the fibril conformer at 4.8 nm. D, EM analysis of K18 fibrils. Cycle 1 fibrils have a distinct striated ribbon appearance with varying numbers of parallel filaments (diameter per filament = 7–8 nm), highlighted in the inset. Fibrils at later cycles are dominated by a typical twisted appearance (diameter ≈ 14 nm, helix periodicity = 90–180 nm). Scale bars, 100 nm. Panels from left to right represent data for fibrils from cycle 1, cycle 5, cycle 10, and cycle 15, respectively.

FIGURE 3.

Conformational stability of K19 fibrils revealed by DEER and EM. A, distance distributions for K19 fibrils collected after cycle 1 (black trace), cycle 5 (red trace), and cycle 10 (blue trace). In parallel, fibrils were stained with uranyl acetate and analyzed by transmission electron microscopy. Electron micrographs of fibrils from cycle 1 (B), cycle 5 (C), and cycle 10 (D). Scale bars, 100 nm. All fibrils exhibit the same ribbon-like morphology with different degrees of lateral associations. Individual filaments within these assemblies have a diameter of 7–8 nm.

FIGURE 4.

Structural analysis of K18 and K19 fibrils by turbidity and proteinase K digestion. The turbidity of K18 fibrils (A) and K19 fibrils (B) from different cycles was measured at 340 nm. All values represent average ± STD (n = 3 experiments). The decreased turbidity of K18 fibrils in later cycles agrees with changes in the fibril population. No such changes are observed for K19 fibrils. K18 fibrils (C) and K19 fibrils (D) from cycle 1 and cycle 10 (25 μm, monomer equivalents) were proteolyzed for 1 h at 22 °C with equal amounts of proteinase K (PK). The samples were analyzed by SDS-PAGE and Coomassie Blue staining. M, molecular weight marker. K18 fibrils from cycle 10 were more sensitive to degradation than fibrils from cycle 1. No differences in protease sensitivity were observed for K19 fibrils.

FIGURE 5.

K18 fibril fragilities determined using EM. A, cycle 1 and cycle 10 fibrils are analyzed by negative stain EM before (upper panels) and after sonication (lower panels). The fibrils are fractured under mild conditions in a bath sonicator. Scale bars, 200 nm. When subjected to identical stress, cycle 1 fibrils (diameter per filament = 7–8 nm) are distinctly shorter than cycle 10 fibrils (diameter ≈ 14 nm). B, length distributions for sonicated fibrils from cycle 1 (left panel) and cycle 10 (right panel). The average fibril lengths are 126 and 228 nm, respectively. AVG, average. The results indicate that cycle 1 fibrils are more fragile than cycle 10 fibrils.

FIGURE 6.

Analysis of Tau fibril growth rates. Negative stain EM images of cycle 1 (A) and cycle 10 (B) fibrils subjected to harsh sonication conditions with the sonicator tip immersed in the fibril solution. Scale bars, 200 nm. Length distributions of cycle 1 (C) and cycle 10 (D) fibrils. The lengths of 526 fibrils from cycle 1 and 493 fibrils from cycle 10 were measured using Image-J. The average lengths of cycle 1 and cycle 10 fibrils are 64 nm and 60 nm, respectively. AVG, average. E, 2% seeds (monomer equivalents) were mixed with 25 μm K18 monomers and 50 μm heparin. Fibril growth of cycle 1 fibrils (blue trace) and cycle 10 fibrils (red trace) was monitored by thioflavin T fluorescence. All values represent average ± S.E. (n = 6 experiments). The data indicate that cycle 10 fibrils elongate faster than cycle 1 fibrils.

FIGURE 7.

Examination of Tau seeding barrier for distinct conformations. 5% seeds (monomer equivalents) from cycle 1 and cycle 10 fibrils were mixed with 25 μm K18 or K19 monomers and 50 μm heparin. Fibrils were allowed to grow for 24 h at 37 °C. Equivalent amounts of pellets (P) and supernatants (S) were analyzed by SDS-PAGE and Coomassie Blue staining. Whereas K18 monomers grow onto K18 seeds (A), K19 monomers do not (B). The arrow in B refers to K18 protein bands that originate from the seeds. The data highlight that both populations of K18 fibril conformers are unable to recruit K19 monomers.

FIGURE 8.

Conformational selection based on fracture and growth. In a heterogeneous mixture of Tau fibril conformers, depicted as stacks of β-stranded segments (green arrows) in U- and L-shape conformations, efficient breakage (intense sonication) results in seeds with identical extensions along the fibril axis. Under these conditions the number of fibril ends that can recruit soluble monomers is the same for the different conformers giving the faster growing seed (U-shape) a selective advantage. Over repetitive cycles of fracture and growth this conformer will become the dominant species (upper reaction path). Inefficient breakage selectively fractures the more fragile conformer (L-shape). If repeated over multiple cycles (continuous stirring) this conformer will become the dominant species (lower reaction path). Note that the depicted conformers are only models. The real structures of Tau fibrils will be more complex and include additional β-sheets. The number of cycles required for evolving a dominant species will depend on the structural composition of the original ensemble and the specific parameters of fracture and growth. In cases where the fragilities and growth rates of different conformers are similar the populations may have multiple dominant fibril species. It is conceivable that within neuronal cells, molecular chaperones or other machineries could facilitate the breakage of fibrils and influence growth.