Filament assembly conditions

Cryo-EM statistics

Cryo-EM analysis of PAD12 tau filaments

A Schematic of PAD12 tau. The amino-terminal inserts N1 (residues 44–73), and N2 (74–102) are shown in grey, the proline-rich region (151-243) is in light grey, the microtubule-binding repeats are in purple (R1, 244-274), blue (R2, 275-305), green (R3, 306-336) and yellow (R4, 337-368) and the carboxy-terminal domain (369-441) is shown in orange. The twelve phosphomimetic mutations of PAD12 are indicated with vertical lines. B Cryo-EM micrograph of PHFs composed of a 1:1 mixture of 0N3R and 0N4R PAD12 tau. Arrows show PHFs (blue) and single protofilaments with the Alzheimer tau fold (light blue) C Cross-sections of cryo-EM reconstructions perpendicular to the helical axis, with a thickness of approximately 4.7 Å for assembly reactions with 0N3R PAD12 tau, 0N4R PAD12 tau and a 1:1 mixture of 0N3R and 0N4R PAD12 tau. Inserts show pie charts with the particle distribution per filament types (PHFs in blue; single protofilaments with the Alzheimer fold in light blue; single protofilaments with the CTE fold in yellow; discarded solved particles (coloured) and discarded filaments in grey). D Cryo-EM density map of the 1:1 0N3R:0N4R PAD12 tau mixture in transparent grey, superimposed with its refined atomic model. E Main chain trace of the atomic model shown in D (blue), overlaid with the model of PHFs from AD brain (PDB-ID: 6hre) (grey).

Purification of 0N3R and 0N4R PAD12 tau

A Schematic of protein purification and assembly procedure. B Coomassie-stained SDS-PAGE (4-20%) after cation exchange chromatography of 0N3R PAD12 tau. B SDS-PAGE after size exclusion chromatography (SEC) of 0N3R PAD12 tau. C SEC profile of 0N3R PAD12 tau. D-F As A-C, but for 0N4R PAD12 tau.

Fourier shell correlation curves

A Fourier shell correlation curves between two independently refined half-maps (black); between the atomic model fitted in half-map 1 and half-map1 (blue); between that the atomic model fitted in half-map 1 and half-map 2 (yellow) and between the atomic model fitted in the sum of the two half-maps and the sum of the two half-maps (red), for the PAD12 PHF structure assembled from a 1:1 mixture of PAD12 0N3R and 0N4R tau (left) and from PAD12 0N3R tau, seeded with filaments that were extracted from the brain of an individual with AD (middle) and 297-441 Δ392-395 (right).

Cryo-EM analysis of tau297-391 and PAD12 tau PHFs

A Cross-sections of cryo-EM reconstructions perpendicular to the helical axis, with an approximate thickness of 4.7 Å and pie charts showing the distribution of filament types (blue: PHFs; light blue: single protofilaments with the Alzheimer fold (not shown); for freeze-thawed filaments (flash frozen in liquid nitrogen, thawed at RT), for filaments assembled in an Eppendorf tube, for filaments assembled from PAD12 tau0-391, for filaments assembled from PAD12 tau151-391 and for filaments assembled from 0N3R PAD12 with glutamate phosphomimetics instead of aspartate. B-E Cryo-EM overviews of PHFs composed of residues tau297-391 (B), a 1:1 mixture 0N3R and 0N4R PAD12 tau (C), tau 0-391 (D) and 151-391 PAD12 3R tau (E). From left to right: grid square overview, foilhole overview, and acquisition image. Arrows indicate clumped aggregates of filaments (pink) and ice particles (blue).

Quantification of PAD12 tau filament types with time

Hierarchical classification of filament segments (left) based on their assigned 2D class averages (vertical axis) and corresponding filament IDs (horizontal axis). The pie charts on the top right show the relative amounts of different filament types. Pink represents PHFs and ochre shows singlets. 2D class averages on the bottom right show examples of the boxed-out filament segments from the dendrogram, with PHFs in pink boxes and singlets in ochre boxes. 2D class averages in grey boxes are examples that were not included in the quantification of filament polymorphs. A After 4 days of shaking. B After one week of shaking. C After one week of shaking followed by two weeks of quiescent incubation.

reproducibility of PHF formation

The percentage of PHFs formed in seven replicates of the assembly reaction with 0N3R PAD12 tau, with one week of shaking, followed by two weeks of quiescent incubation. Five replicates yielded 100% PHFs; one outlier contained 75% PHFs and 25% singlets. The mean percentage of PHFs is 95.3%, with a standard deviation of 9.4%.

EM analysis of 0N3R PAD12+4, PAD12-4 and PAD12+/−4 tau constructs

A Schematic of tau sequence as in Figure 1A, with the four extra mutations of PAD12+4 in blue and the four mutations that were removed from PAD12-4 in red. B Negative stain EM of filaments formed with PAD12+4 (left), PAD12-4 (middle) and PAD12+/−4 (right) 0N3R tau. C Cryo-EM reference-free 2D classes of filaments assembled from 0N3R PAD12-4. D Reference-free 2D classes of filaments assembled from 0N3R PAD12-4.

Labelling of pre-assembled PAD12 tau filaments

A Cartoon showing that filaments can be labelled via NHS-ester chemistry, which specifically targets primary amines in lysine residues. B Immuno-EM showing biotinylated tau is labelled with streptavidin-coated 10 nm gold particles. C Cryo-EM reconstructions of PAD12 3R tau (top) labelled with different fluorophores. The circular inset shows a cross-section of the corresponding cryo-EM reconstruction, and the pie charts shows the distributions of the different filament types (PHFs in solid colours; singlets in lighter colours). The coloured pellets (bottom) indicate successful labelling of the filaments.

In vitro seeded assembly with PAD12 tau

A Schematic of experimental approach for the seeded assembly of PAD12 0N3R. Small amounts of brain material are used to seed the assembly of PAD12 0N3R (round 1). The filaments formed from round one are used as seeds for a second round. B ThT fluorescence profile of the AD-seeded (purple), second-round seeding (pink) and non-seeded control (yellow) N=9. The circles are individual measurements (normalised for each reaction). C Cross-sections of cryo-EM reconstructions perpendicular to the helical axis, with a thickness of approximately 4.7 Å and pie charts showing the distribution of filament types for the first (left) and second (right) round of seeding (pink/purple PHFs, yellow single protofilaments with the Alzheimer fold (not shown); grey discarded filaments). D Cryo-EM density map of AD-seeded 0N3R PAD12 tau (transparent grey) with the superimposed fitted atomic model. E Main chain trace of in vitro seeded PHF (purple) overlaid with AD PHF (grey; PDB-ID: 6hre).

Cellular seeding with recombinant tau filaments

A Box plot showing the number of detected seeds, which were normalised to the number of cells and compared to mock-treated control cells (n ≥ 10,000 cells/condition analysed). Graph represents mean values; error bars represent standard deviation. B Images from control (-seed) and cells seeded with 0.25 µg of assembled tau297-391 PHFs, PAD12 0N3R PHFs and PAD12 0N3R:0N4R PHFs. Fixed cells were stained against HA for labelling over-expressed tau297-391 (green) and Hoechst (blue) for labelling of the nucleus. Scale bar, 50 µm. C Images from control cells without the addition of seeds (-seed) and cells seeded with 0.25 µg of PAD12 0N3R PHFs that were pre-labelled with DyLight-488. Fixed cells were stained against HA for labelling over-expressed tau (red) and Hoechst (blue) for labelling nuclei. Scale bar, 50 µm.

Cellular seeding with AD brain-derived tau and with PAD12 0N3R tau filaments labelled with DyLight-488

A, B Box plots showing the number of detected seeds which were normalised to the number of cells, and compared to mock-treated control cells (n ≥ 10,000 cells/condition analysed) for cells seeded with AD-brain-derived seeds (A) and PAD12 0N3R DyLight-488-labelled filaments (B). Graphs represent mean values; error bars represent standard deviations. C Images from control cells without the addition of seeds (-seed) and cells seeded with insoluble tau from 200 µg of sAD brain tissue or 0.25 µg PAD12 0N3R tau filaments that were pre-labelled with DyLight-488. Fixed cells were stained against HA for labelling overexpressed tau (green) and Hoechst dye (blue) for labelling nuclei. Scale bar, 50 µm.

NMR and cryo-EM of C-terminal tau297-441, tau297-441 PAD12 and tau297-441 Δ392-395

A Peak height differences for selected residues in HSQC spectra as shown by normalised peak intensity. Values for tau297-391 are shown in gold, tau297-441 in black, PAD12 tau297-441 in lilac and tau297-441 Δ392-395 in blue. The grey dashed line box indicates the FIA region (residues 301-316). Lilac dashed lines are PAD12 mutations and blue dashed lines are the residues deleted in the Δ392-395 tau construct. Full residue information is shown in Figure 6 – supplementary figure 5C B The cryo-EM structure of tau297-441 Δ392-395. Cryo-EM density map in transparent grey with the superimposed fitted atomic model in blue. All filaments were of the same type.

NMR spectroscopy of wild-type and PAD12 tau151-391

A Chemical shift perturbation (CSP) map of the peak location differences between wild-type and PAD12 tau151-391, with mutated residues shown in lilac. B Secondary chemical shift analysis of the backbone Cα and Cβ resonances. Residues that have a preference for helical torsion angles have positive values, and those with an extended backbone preference have negative values. Secondary chemical shifts for tau297- 391 are shown in gold, wild-type tau151-391 in black and PAD12 tau151-391 in lilac. C Heteronuclear NOE (hetNOE) values for wild-type and PAD12 tau151-391 are shown in black and lilac, respectively. HetNOE values are sensitive to backbone motion on the picosecond timescale. Residues within a permanent secondary structure element will have hetNOE values of ∼0.8; lower or negative values indicate increased backbone flexibility. D Normalised peak intensity differences in HSQC spectra are shown for values of tau151-391 in black and PAD12 tau151-391 in lilac. The dashed box highlights residues that form the FIA and dashed lilac lines indicate mutated residues. Residues that we were unable to assign in the tau151- 391 or the PAD12 tau151-391 constructs are highlighted with a purple or black asterisk, respectively.

HSQC peak assignment of wild-type and PAD12 tau151-391

Assigned 700 MHz 15N–1H heteronuclear single quantum coherence (HSQC) spectrum of left, wild-type tau151-391 (black), and right PAD12 tau151-391 (lilac).

NMR spectroscopy of wild-type and PAD12 tau297-441

A Secondary shift analysis of the backbone Cα and Cβ chemical shifts. Residues that have a preference for helical torsion angles have positive values, and those with an extended backbone preference have negative values. Secondary chemical shifts for tau297-391 are shown in gold, tau297-441 in black and PAD12 tau297-441 in lilac. B Chemical shift perturbation (CSP) map of the peak location differences between wild-type and PAD12 tau297-441. C Heteronuclear NOE (hetNOE) values for wild-type and PAD12 tau297-441 are shown in black and lilac, respectively. HetNOE values are sensitive to backbone motion on the picosecond timescale. Residues within a permanent secondary structure element will have hetNOE values of ∼0.8; lower or negative values indicate increased backbone flexibility. D Normalised peak intensity differences in HSQC spectra are shown for values of tau297-391 in gold, tau297-441 in black and PAD12 tau297-441 in lilac. The dashed box highlights residues that form the ordered core of the FIA and dashed lilac lines indicate mutated residues.

HSQC peak assignment of wild-type and PAD12 tau297-441

Assigned 700 MHz 15N–1H heteronuclear single quantum coherence (HSQC) spectrum of, from left to right, tau297-441 (black), overlaid and PAD12 tau297-441 (lilac). Residues from the PHF6 region (306-311, VQIVYK) and the C-terminal IVYK motif (392-395) are indicated in the central overlayed panel.

NMR spectroscopy of tau297-441Δ392-395

A Secondary chemical shift analysis of the backbone Cα and Cβ resonances. Residues that have a preference for helical torsion angles have positive values, and those with an extended backbone preference have negative values. Secondary chemical shifts for wild-type tau297-441 in black, PAD12 tau297-441 in lilac and tau297-441Δ392-395 in blue. B Heteronuclear NOE (hetNOE) values for wild-type 297-441, PAD12 tau297-441 and tau297-441Δ392-395 are shown in black, lilac and blue, respectively. HetNOE values are sensitive to motion on the picosecond timescale. Residues within a permanent secondary structure element will have hetNOE values of ∼0.8; lower or negative values indicate increased backbone flexibility. C Normalised peak intensity differences in HSQC spectra are shown for values of tau297-391 in gold, tau297-441 in black, PAD12 tau297-441 in lilac and tau297-441Δ392-395 in blue. The dashed box highlights residues that form the FIA, the dashed lilac lines indicate mutated residues, and the dashed blue lines indicated the truncated residues.

HSQC peak assignment of tau297-441Δ392-395

Assigned 700 MHz 15N–1H heteronuclear single quantum coherence (HSQC) spectrum of tau297-441Δ392-395 (blue), overlaid with tau297-441 (black). Residues 392-395 of tau297-441 are indicated in red.

Hairpin model of phosphorylated tau filament assembly

In unmodified tau monomer, residues 392IVYK395 (orange) in the carboxy-terminal domain interact with residues 306VQIVYK311 (purple) in the core-forming region of tau. A similar interaction between the amino-terminal domain of tau (green) may also exist, but the residues involved in this interaction remain unknown. Upon phosphorylation (red) of residues in the amino-terminal and the carboxy-terminal domains of tau, these interactions are disrupted, which then leads to filament nucleation through inter-molecular interactions of residues 306VQIVYK311 in the FIA.