Abstract
The assembly of tau into amyloid filaments is associated with more than twenty neurodegenerative diseases, collectively termed tauopathies. Cryo-EM structures of brain-derived tau filaments revealed that specific structures define different diseases, triggering a quest for the development of experimental model systems that replicate the structures of disease. Here, we describe twelve phosphomimetic, serine/threonine to aspartate, mutations in tau, which we termed PAD12, that collectively induce the in vitro assembly of full-length tau into filaments with the same structure as that of paired helical filaments extracted from the brains of individuals with Alzheimer’s disease. Solution-state nuclear magnetic resonance spectroscopy suggests that phosphomimetic mutations in the carboxy-terminal domain of tau may facilitate filament formation by disrupting an intramolecular interaction between two IVYK motifs. PAD12 tau can be used for both nucleation-dependent and multiple rounds of seeded assembly in vitro, as well as for the seeding of tau biosensor cells. PAD12 tau can be assembled into paired helical filaments under various shaking conditions, with the resulting filaments being stable for extended periods of time. They can be labelled with fluorophores and biotin. Tau filaments extracted from the brains of individuals with Alzheimer’s disease brains have been known to be made of hyperphosphorylated and abnormally phosphorylated full-length tau, but it was not known if the presence of this post-translational modification is more than a mere correlation. Our findings suggest that hyperphosphorylation of tau may be sufficient for the formation of the Alzheimer tau fold. PAD12 tau will be a useful tool for the study of molecular mechanisms of neurodegeneration.
Introduction
The assembly of tau into amyloid filaments characterises a group of neurodegenerative diseases that are called tauopathies. Alzheimer’s disease (AD) is the most common tauopathy. Whereas in most tauopathies tau is the only protein that assembles into filaments, in AD intracellular assemblies of tau co-exist with extracellular plaques of filamentous amyloid-β.
Six tau isoforms, with lengths ranging from 352 to 441 amino acids, are expressed in the adult human brain by alternative mRNA splicing from a single gene (MAPT). The tau sequence (Figure 1A) comprises an amino-terminal projection domain (residues 1-150), a proline-rich region (residues 151-243), four microtubule-binding repeats (R1-R4; residues 244-368) and a carboxy-terminal domain (residues 369-441). Tau isoforms vary by the incorporation of zero, one, or two inserts of 29 amino acids at the amino-terminal domain (0N, 1N or 2N isoforms) and the presence or absence of the second microtubule-binding repeat, resulting in three-or four-repeat (3R or 4R) tau isoforms (Goedert et al., 1989). A mixture of all six tau isoforms is present in AD and chronic traumatic encephalopathy (CTE), whilst only 3R tau assembles in Pick’s disease, and only 4R tau assembles in progressive supranuclear palsy (PSP), corticobasal degeneration (CBD), globular glial tauopathy (GGT) and argyrophilic grain disease (AGD).
Electron cryo-microscopy (cryo-EM) revealed that specific folds of assembled tau define different tauopathies (Shi et al., 2021). Residues from the microtubule-binding repeats plus approximately 10 residues from the carboxy-terminal domain are ordered in tau filaments from the different diseases. The remaining residues are disordered, forming a fuzzy coat that surrounds the ordered core (Wischik et al., 1988). The tau folds of AD and CTE comprise residues from R3 and R4 and the carboxy-terminal domain. The Alzheimer tau fold is also found in posterior cortical atrophy, Down’s syndrome (Fernandez et al., 2024; Ghosh et al., 2024), some prion protein amyloidoses (Hallinan et al., 2021), familial British dementia, familial Danish dementia, primary age-related tauopathy and cases of frontotemporal dementia caused by missense mutations V337M and R406W in MAPT (Qi et al., 2024b). The CTE tau fold is also found in subacute sclerosing panencephalitis, the amyotrophic lateral sclerosis/Parkinsonism-dementia complex of Guam and the Kii peninsula (Qi et al., 2023), as well as in vacuolar tauopathy (Qi et al., 2024a). The Pick fold comprises in addition residues 254-274 from R1, whereas folds of the 4R-tauopathies also comprise all residues (275-305) from R2 (Scheres et al., 2023). Intriguingly, with the exception of a familial tauopathy caused by a P301T mutation in tau (Schweighauser et al., 2024), only a single tau fold has been observed for each disease, whereas multiple individuals with a given disease always had the same filaments. The consistent structure of tau filaments across individuals with the same tauopathy supports the prion hypothesis (Prusiner, 1998), in which tau filaments propagate their structure through templated misfolding.
The specific tau folds that form and propagate in the human brain may be influenced by the cellular and biochemical environments where they assemble. Model systems used to study tauopathies should therefore replicate the same tau folds and biochemical properties observed in diseased human brains. However, this has proven challenging. For instance, when recombinant wild-type (WT) full-length tau is assembled in vitro, it requires the addition of negatively charged molecules (Goedert et al., 1996; Kampers et al., 1996; Pérez et al., 1996; Wilson and Binder, 1997), but the structures of tau filaments assembled in the presence of heparin or RNA are unlike those extracted from human brains (Abskharon et al., 2022; Lövestam et al., 2022; Zhang et al., 2019). Similarly, attempts to reproduce disease-specific tau structures by over-expressing full-length human tau in SH-SY5Y cells, combined with seeding using human brain-derived tau filaments, have not yet succeeded in faithfully mimicking the structures of AD or CBD (Tarutani et al., 2023). Additionally, mice over-expressing 4R tau with the P301S mutation under the control of different promoters form tau filaments with different structures, suggesting that different promoters may cause filament assembly in distinct cell types, affecting which tau fold forms (Schweighauser et al., 2023; Zhao et al., 2024). None of the tau filament structures from mouse models were identical to those observed in human brains.
We previously reported that, under shaking conditions, the truncated tau construct comprising residues 297-391 (tau297-391) (Al-Hilaly et al., 2020; Novak et al., 1993) assembles into paired helical filaments (PHFs), which are the main filament type in AD (Lövestam et al., 2022). The addition of sodium chloride to the assembly reaction of tau297-391 led to the formation of CTE filaments. Subsequent time-resolved cryo-EM studies revealed that formation of the disease-specific structures occurs through many intermediate amyloid filaments, with a common first intermediate amyloid (FIA) observed in the assembly of AD and CTE filaments (Lövestam et al., 2024). But unlike filaments isolated from AD brains which contain full-length tau (Lee et al., 1991), filaments of tau297-391 lack the fuzzy coat that is typical of tau filaments inside brain cells.
PHF-tau from the brains of individuals with AD is hyperphosphorylated and abnormally phosphorylated (Grundke-Iqbal et al., 1986; Iqbal et al., 2016); this precedes filament assembly to which it has been linked. Hyperphosphorylation of tau may also accelerate aggregation indirectly by detaching tau from microtubules. Mass spectrometry and epitope mapping of antibodies that were raised against PHFs from AD brains (Greenberg et al., 1992; Mercken et al., 1992) led to the identification of multiple sites, all in the fuzzy coat of tau, that are hyperphosphorylated in PHFs from AD brains (Lee et al., 2001). Monoclonal antibody AT270 recognises phosphorylation of threonine 181 (T181); AT8 recognises phosphorylation of serine 202 (S202) and threonine 205 (T205), AT100 recognises phosphorylation of T212, S214 and T217, AT180 recognises phosphorylation of T231 and S235, and PHF-1 recognises phosphorylation of S396 and S404. A comprehensive mass spectrometry map of post-translational modifications confirmed that these residues are hyperphosphorylated in assembled tau from AD brains (Wesseling et al., 2020).
Based on these observations, we previously designed four phosphomimetic mutations in the carboxy-terminal domain of tau (S396D, S400D, T403D and S404D) that allowed the assembly of recombinant tau297-441 into filaments with the Alzheimer fold, albeit only with a single protofilament (Lövestam et al., 2022). A similar construct was also shown to be suitable for the production of large amounts of isotope-labelled protein for NMR, but again it yielded filaments with the Alzheimer protofilament fold, but with protofilament packings distinct from those in PHFs (Duan et al., 2024b). Additionally, a tau construct with phosphomimetic mutations in the proline-rich region (S202E, T205E, S208E), as well in the carboxy-terminal region (S396E, S400E, T403E, S404E) was found to assemble, but the filaments formed were again distinct from those found in AD (Mammeri et al., 2024).
We hypothesised that the introduction of additional phosphomimetic mutations in the proline-rich region of tau might facilitate the assembly of recombinant full-length tau. Based on the epitopes of antibodies and the mass spectrometry data described above, we mutated eight residues within the proline-rich region of tau (T181, S202, T205, T212, S214, T217, T231 and S235) to aspartate. Combined with the four previously described mutations in the carboxy-terminal domain of tau, this resulted in twelve phosphomimetic mutations in full-length tau, nine of which were in serine/threonine-proline motifs (Figure 1A). We show that these mutations, which we refer to as twelve phosphomimetics of AD, or PAD12, allow both nucleation-dependent and seeded assembly of full-length recombinant tau into PHFs, thereby opening new avenues for studying the role of tau’s fuzzy coat in the molecular mechanisms of disease. These findings indicate that the hyperphosphorylation of tau at specific sites may be sufficient for the formation of the Alzheimer tau fold from full-length protein.
Results
Assembly of PAD12 full-length tau into PHFs
0N3R and 0N4R versions of PAD12 tau were expressed in E. coli and purified, followed by in vitro assembly under shaking conditions (Figure 1A; Figure 1 – figure supplement 1; Methods). Assembly was carried out with 500 rpm orbital shaking (2 min on; 1 min off) for 7 days at a protein concentration of 100 µM in a buffer containing 100 mM potassium phosphate at pH 7.2, 400 mM potassium citrate and 4 mM tris(2-carboxyethyl) phosphine (TCEP). We used cryo-EM to determine the structures of the resulting filaments (Figure 1B-D; Figure 1 – figure supplement 2; Table 1). 0N3R-PAD12 tau primarily formed paired helical filaments (PHFs) (66%), with a minor population (14%) of single protofilaments with the Alzheimer fold. The remainder of the filaments, including false positives from automated filament picking (Lövestam and Scheres, 2022), were discarded during image processing. 0N4R-PAD12 tau assembled predominantly into filaments consisting of a single protofilament with the CTE fold. It is possible that small amounts of sodium chloride in the protein sample led to the formation of protofilaments with the CTE fold. The observation that part of the second microtubule-binding repeat is ordered, forming cross-beta packing against the third repeat, may be related to this construct forming filaments with only a single protofilament.
Because tau filaments from AD brains contain equimolar amounts of 3R and 4R tau (Goedert et al., 1992), we also assembled a 1:1 mixture of 0N3R-PAD12 and 0N4R-PAD12 tau under the same conditions. Again, we observed predominantly PHFs (59%), with a minority of singlets with the Alzheimer tau fold (11%). The PHFs were identical to those found in AD brains, with a backbone root mean square deviation (r.m.s.d.) of 1.4 Å (Figure 1E).
PHFs of PAD12 tau are not sticky and form under various shaking conditions
While several groups have replicated our finding that tau297-391 assembles into PHFs, filament formation has required further optimisation of the assembly conditions and the formation of other filament types has been reported (Duan et al., 2024b; Glynn et al., 2024). We previously showed that multiple filament types exist in the assembly reactions that are on-pathway to form PHFs (Lövestam et al., 2024). Another possible reason for the difficulties in reproducibility may be that the assembly of tau297-391 is sensitive to shaking conditions, with different structures forming when shaking is performed at 200 or at 700 rpm (Lövestam et al., 2022). The physical forces inside the assembly reaction vessel due to shaking are likely more difficult to control than the biochemical components of the reaction mixtures. For example, different shaking machines, or differently shaped vessels with different volumes, may all lead to different forces. The observation that the assembly of 100 μl of 0N3R-PAD12 tau inside an Eppendorf LoBind microcentrifuge tube, with orbital shaking at 500 rpm in an Eppendorf ThermoMixer C, led to the formation of PHFs with the Alzheimer fold (Figure 1 -figure supplement 3A) suggests that the assembly of full-length PAD12 tau is less dependent on the physics of shaking than the assembly of tau297-391.
Another disadvantage of tau297-391 PHFs is that they tend to clump together, especially after prolonged storage or freeze-thawing. Clumping of filaments complicates cryo-EM structure determination and may interfere with subsequent experiments, such as binding studies of small-molecule compounds or seeding experiments in cell culture and in animals. In contrast to tau297-391, almost all micrographs of full-length PAD12 tau showed individually dispersed filaments, even after assembled filaments were kept at 4°C for two months, or flash-frozen at -196°C and then thawed at RT (Figure 1 - figure supplement 3B,C). The observation that tau 0-391 and 151-391 3R PAD12 tau constructs also did not clump together (Figure 1 – figure supplement 3D,E) suggests that the absence of an amino-terminal fuzzy coat contributes to the stickiness of tau297-391 filaments.
Different phosphorylation patterns lead to different filaments
To test whether the formation of AD PHFs depends on the specific PAD12 mutations, we tested another three 0N3R tau constructs with different mutations. In the first construct (PAD12-4), we removed four of the eight phosphomimetic mutations in the proline-rich region, leaving only T181D, S202D, T217D, and S235D. In the second construct (PAD12+4), we added four phosphomimetic mutations that also had high levels of phosphorylation by mass spectrometry of AD tau (Wesseling et al., 2020) (T153D, S237D, S262D and T263D). In the third construct (PAD12+/-4), we removed the same four mutations as in the first construct and added the same four new mutations as in the second construct, resulting in the same net charge as in the original PAD12 construct. Assembly of the 0N3R versions of these constructs, under the same orbital shaking conditions as before, led to the formation of filaments in all three cases. However, the filaments formed with the PAD12+4, PAD12-4 or PAD12+/-4 constructs had morphologies that were distinct from those of PHFs; they did not twist and we were thus not able to determine their structures (Figure 2).
PAD12 tau filaments can be labelled
Next, we explored the ability to label filaments of PAD12 tau using NHS-ester chemistry, which specifically targets primary amines in lysine residues. Fluorescently labelled filaments may be useful, for example, to follow seeding reactions in cells by optical microscopy. We labelled pre-assembled 0N3R-PAD12 tau filaments with DyLight-488. We observed a fluorescent yellow-green pellet and fluorescent bands on SDS-PAGE Figure 3A-B). Cryo-EM of the resulting filaments confirmed that the PHF structure remained intact (Figure 3C).
We also explored whether filaments of PAD12 tau can be biotinylated using the same NHS-ester chemistry. Biotinylation is used for protein selection and for studying protein-protein interactions by proximity labelling. Immuno-gold EM with 10 nm gold-conjugated streptavidin confirmed that pre-assembled 0N3R-PAD12 tau filaments can be extensively biotinylated (Figure 3D).
In vitro seeded assembly with PAD12 tau
Next, we investigated whether AD brain-derived tau filaments could seed the assembly of PAD12 tau, and if this seeding could be sustained over more than one round. We introduced sonicated AD brain-derived tau filaments to a solution of 40 mM (∼1.5 mg/mL) 0N3R-PAD12 tau and performed in vitro assembly under orbital shaking at 500 rpm (2 min on; 1 min off) for 48 h in a buffer containing 20 mM HEPES at pH 7.3, 4 mM potassium phosphate at pH 7.2, 300 mM sodium citrate and 4 mM TCEP. Under the same conditions, but without the addition of AD seeds, we observed no filament formation within 48 hrs. Although we only used extracts from 0.8 mg of AD brain tissue for reactions that contained 60 mg of recombinant (unlabelled) PAD12 tau, we observed no lag phase in Thioflavin T (ThT) fluorescence upon addition of the seeds. Instead, ThT fluorescence increased linearly directly after the seeds were added and fluorescence plateaued after 48 h (Figure 4A). Cryo-EM analysis confirmed that the seeded tau filaments were predominantly (73%) PHFs (with a backbone r.m.s.d. of 1.3 Å with PDB-ID 6hre), along with a minority of single protofilaments with the Alzheimer fold (8.5%) (Figure 4B-D).
It was recently reported that in vitro assembly of wildtype 0N3R tau can be seeded with large amounts of AD brain-derived filaments (Duan et al., 2024a). The seeded filaments were shown to be PHFs, but they were not capable of seeding the assembly of wildtype 0N3R tau in a second round of seeded assembly. To test whether this is also the case for PAD12 tau, we used filaments from the first round at an estimated ratio of 1:2,500 to recombinant (unlabelled) PAD12 tau for a second round of seeded assembly. We again observed a linear increase in ThT fluorescence directly after addition of the seeds, and cryo-EM analysis of the seeded aggregates still showed PHFs (Figure 4A,B), indicating that PAD12 tau can be used for multiple rounds of seeded aggregation.
Seeding of tau reporter cells with PAD12 PHFs
We then used a biosensor cell line that overexpresses hemagglutinin (HA-)tagged tau297-391 in HEK293T cells (see Methods) and compared its seeding response to increasing amounts of PHFs made with tau297-391, PAD12 0N3R or PAD12 0N3R:0N4R (Figure 5A,B). These cells also yielded robust seeding, in a concentration-dependent manner, when exposed to seeds extracted from AD brains (Figure 5 – Figure supplement 1A). Comparing the recombinant PHFs of tau297-391 and PAD12 0N3R or PAD12 0N3R:0N4R tau, we observed efficient seeding at low concentrations (∼5 ng) of PAD12 PHFs, whereas tau297-391 did not lead to seeding at the same concentrations. At higher concentrations, seeding with tau297-391 PHFs was less efficient and showed greater variability among replicates than seeding with PAD12 PHFs. Using appropriate excitation and emission settings to visualise seeds of PAD12 0N3R filaments that were labelled with DyLight-488 inside the biosensor cells, we observed colocalization of the original seeds with the HA-297-391 puncta (Figure 5C; Figure 5 - Figure supplement 1B,C).
Phosphomimetic mutations in the carboxy-terminal domain affect the FIA region
We used solution-state nuclear magnetic resonance (NMR) to explore the effects of PAD12 mutations on the tau monomer. Because of size limitations imposed by this technique, we used two truncated tau constructs. The first construct comprised residues 151-391 of 4R tau, with or without the eight phosphomimetic mutations of PAD12 in the proline-rich region. The second construct comprised residues 297-441 of 4R tau, with or without the four phosphomimetic mutations of PAD12 in the carboxy-terminal domain.
The effects of the phosphomimetic mutations in the proline-rich region of the tau151-391 construct were subtle. Chemical shift perturbations (CSPs) between wildtype and PAD12 tau151-391 were largest around residues 198-222, which harboured mutations at S202, T205, S212, S214 and T217, and around residues 237–244, which were adjacent to the mutations at T231 and S235 (Figure 6 - figure supplement 1A; Figure 6 – figure supplement 2). Analysis of the backbone Cα and Cβ chemical shifts (which are sensitive to back bone torsion angles) revealed little difference in secondary structure propensity in the 297-391 region between wildtype tau151-391, PAD12 tau151-391 and tau297-391 (Figure 6 - figure supplement 1B), or in the proline-rich region of wildtype and PAD12 tau151-391 (although its 26 prolines may prevent any potential alterations to secondary structure). Heteronuclear NOE (HetNOE) analysis (Figure 6 - figure supplement 1C), which reports on picosecond time scale backbone mobility, revealed increased rigidity of PAD12 tau151-391 compared to wildtype tau151-391, predominately around the mutation sites. Line-shape analysis, based on peak intensities (Figure 6 - figure supplement 1D), showed some differences in the proline-rich region, where relative peak attenuation indicated greater conformational sampling for wildtype tau151-391 than for PAD12 tau151-391. However, the subtle differences between these two constructs do not suggest an obvious reason for the increased propensity of the PAD12 tau151-391 to form filaments.
By contrast, comparison of wildtype and PAD12 tau297-441 revealed interesting differences. Secondary chemical shift analyses (Figure 6A; Figure 6 – figure supplement 3) suggested that residues 393–404, which harbour the four mutation sites, had an increased propensity to adopt extended conformations in the PAD12 construct, possibly because of electrostatic repulsion between the negatively charged aspartates. Residues 405–407 exhibited a strong preference for helical backbone torsion angles, indicating the presence of a turn-like structure, in the PAD12 construct, but not in wildtype tau297-441. The effect of the phosphomimetic mutations on these residues was also echoed in the CSPs (Figure 6B). HetNOE analysis suggests that the PAD12 construct is more rigid on the picosecond timescale than wildtype tau297-441 between residues 305-311 and 398–412 (Figure 6C). Finally, line shape analysis revealed that residues 297-319 of wildtype tau297-441 displayed reduced peak intensities compared to tau297–391 and PAD12 tau297–441 (Figure 6D). Residues 392-404 of wildtype tau297-441 also showed similar attenuated peak intensities compared to PAD12 tau297-441.
Residues 302-316 form the ordered core of the FIA in the assembly of tau297-391 into PHFs or CTE filaments. We previously hypothesised that partial rigidity of these residues in tau297-391 monomers may reduce the entropic cost of nucleating new filaments (Lövestam et al., 2024). The observation that, by hetNOE analysis, the same region appears more rigid in PAD12 tau297-441 than in wildtype tau297-441 may contribute to the ability of the former to assemble into filaments. In addition, it could be that the reduced peak intensities for residues 297-319 and 392-404 stem from conformational exchange broadening as a result of a transient intramolecular interaction between their respective IVYK motifs in repeat 3 (residues 308-311) and in the carboxy-terminal region (residues 392-395). Such an interaction, which may resemble that between the two protofilaments of the FIA, could inhibit filament formation in the absence of phosphomimetic mutations. A similar interaction, which was also impeded by phosphomimetic mutations in the carboxy-terminal domain (S396E and S404E), has been hypothesised to occur between the microtubule-binding repeat region of tau and its carboxy-terminal domain based on FRET measurements (Jeganathan et al., 2008). Much earlier, phosphorylation of tau had already been observed to lead to a pronounced change in electrophoretic mobility, which was suggested to reflect a conformational change (Lindwall and Cole, 1984).
Discussion
Tau filaments that are extracted from the brains of individuals with AD using sarkosyl contain full-length, hyperphosphorylated tau of all six isoforms (Goedert et al., 1992). By contrast, the assembly of full-length recombinant tau into filaments requires the presence of negatively charged cofactors. We previously showed that heparin-induced in vitro assembly of full-length recombinant tau yields structures that are different from those of filaments extracted from AD brains (Zhang et al., 2019), but that truncated tau297-391 can be assembled into PHFs in the absence of cofactor (Lövestam et al., 2022). We also reported that the introduction of four phosphomimetic mutations in the carboxy-terminal domain of tau297-441 yields individual protofilaments with the Alzheimer fold. In this paper, we show that an additional eight phosphomimetic mutations in the proline-region rich region are sufficient for the assembly of recombinant full-length tau into PHFs. These findings show for the first time the nucleation-dependent assembly of full-length recombinant tau into PHFs.
The proline-rich region of unmodified tau is positively charged. Our findings with the PAD12+4, PAD12-4 and PAD12+/-4 constructs show that the assembly of full-length tau into PHFs is not merely dependent on the introduction of negative charges in this region: which residues are mutated matters too. This is in agreement with studies by others using pseudo-phosphorylation of recombinant tau and phosphorylation of recombinant tau by some protein kinases (Haase et al., 2004; Lee et al., 2001). We cannot exclude the possibility that each construct requires its own optimisation of assembly conditions and that constructs with modifications other than those of PAD12 could be made to form PHFs too. Previous studies have reported that phosphorylation or phosphomimetic mutations of different subsets of the PAD12 mutations could lead to the assembly of tau in vitro and in cells (Alonso et al., 2010; Despres et al., 2017). However, for none of these studies are cryo-EM structures available to show that the resulting filaments were PHFs. The definition of a minimal set of phosphomimetic mutations that is necessary for the spontaneous assembly of full-length tau into PHFs would require the systematic removal of PAD12 mutations, each with their own optimisation of assembly conditions and cryo-EM structure determination to verify that the correct structures are formed. While such work may yield further insights in the future, we feel that the assembly of PAD12 full-length tau into PHFs warrants dissemination before such explorations are performed.
Like tau filaments in disease, and unlike filaments of tau297-391, PAD12 tau filaments have a fuzzy coat. The role of tau’s fuzzy coat in the molecular mechanisms of disease remains poorly understood. For example, AD-seeded PAD12 tau filaments can be used as seeds in a second round of in vitro seeded assembly, whereas AD-seeded unmodified tau filaments were reported to be seeding-incompetent (Duan et al., 2024a), even though both tau constructs formed PHFs in the first round of seeding. It is also possible that the fuzzy coat affects the interactions between tau filaments and cellular components, such as extracellular receptors for the uptake of tau filaments (De La-Rocque et al., 2021), or macromolecular complexes that degrade them (Koopman et al., 2022). Moreover, the fuzzy coat may interfere with the binding of candidate ligands for positron emission tomography or biologics designed to target the ordered cores of tau filaments in disease. Further research to explore the roles of the fuzzy coat and its post-translational modifications is required, and the ability to make PHFs with a fuzzy coat from recombinant PAD12 tau will facilitate this.
Hyperphosphorylation of tau disrupts its ability to interact with microtubules and has been implicated in filament assembly (Iqbal et al., 2016). Our NMR results suggest that phosphorylation of the fuzzy coat, in particular in the carboxy-terminal domain, has a direct effect on the conformation of the residues of tau that become ordered in the FIA (Lövestam et al., 2024). Phosphorylation of the fuzzy coat may facilitate spontaneous filament assembly by reducing the entropic cost of nucleating new filaments. In addition, the negative charges of hyperphosphorylated tau in filaments may actively recruit soluble tau monomers with less phosphorylation through electrostatic attraction. This could explain why the assembly of wildtype full-length tau with large amounts of AD brain-derived seeds worked for a single round of seeding, with the resulting filaments being unable to seed the assembly of wildtype full-length tau in a second round (Duan et al., 2024a). It is striking that phosphorylation patterns of tau from AD brains exclude residues from the ordered core of the PHF. In agreement, no cryo-EM density for phosphorylation groups is visible in the ordered cores of any of the reported tau filaments from postmortem brains (Scheres et al., 2023). Because the ordered cores of PHFs comprise the third and fourth microtubule-binding domains of tau, it is possible that these residues are protected from phosphorylation, while tau is bound to microtubules. In fact, tau molecules with phosphorylation of the residues in the ordered core of PHFs may not be able to form filaments, due to the size of the phosphate groups and the repulsive forces between their charge. PAD12 tau may thus represent a mimic of the partially phosphorylated tau that disassembles from microtubules, after residues in the amino-and carboxy-terminal domains get phosphorylated and residues in its microtubule-binding domains remain unmodified.
It could be that different post-translational modification patterns are important for the assembly of tau into protofilament folds that are specific for the other tauopathies. To explore this, similar approaches as described here for AD could be applied to the other diseases, provided proteomics data on tau filaments from these diseases becomes available. In the meantime, the present findings suggest that hyperphosphorylation of tau is sufficient for the formation of the Alzheimer fold. It remains to be seen if other post-translational modifications of tau can also give rise to PHFs. The ability to form PHFs from recombinant PAD12 tau will enable further research into the molecular mechanisms of tau aggregation and its role in AD. The observations that PAD12 tau filaments can be labelled, and that individual filaments are stable in solution for weeks, will facilitate their use in the seeding of tau in cells and in animals, as well as for high-throughput binding assays with potential therapeutic or diagnostic compounds, and their structure determination. Understanding how and why tau adopts specific folds in the different tauopathies, and whether or how these folds affect the different diseases, may provide new avenues for therapeutic development.
Materials and Methods
Protein purification
Tau constructs in pRK172 containing 0N4R or 0N3R cDNA tau were made using in vivo assembly (García-Nafría et al., 2016) and transformed into BL21-CodonPlus (DE3)-RIPL competent cells (Agilent) for expression. Transformed cells from one plate were resuspended in 2xTY with 5 mM magnesium chloride and 100 mg/L ampicillin; inoculated in 4 L of 2xTY with 5 mM magnesium chloride and 100 mg/L of ampicillin; grown to an optical density of 0.8 at 600 nm; and induced with 0.6 mM isopropyl β-D-1-thiogalactopyranoside (IPTG) for 3 h at 37 °C. For 15N and 13C-labeled tau, bacteria were grown in isotope-enriched M9 minimal medium containing 1 g/L of [15N]ammonium chloride and 2 g/L of [13C]glucose (Sigma), supplemented with 1.7 g/L yeast nitrogen base (Sigma). Expression was induced with 0.8 mM IPTG at 18 °C overnight. Cells were harvested by centrifugation for 35 min at 4 °C 4400 g. Pelleted cells were flash frozen in liquid nitrogen and snapped into a beaker with a stirring bar. Per 6 L culture, 50-60 ml of Buffer A (50 mM MES at pH 5.5-6.5, with 50 mM sodium chloride, 10 mM ethylenediaminetetraacetic acid (EDTA), 5 mM magnesium chloride, 10 mM dithiothreitol (DTT), 0.03 mM chymostatin, 0.1 mM phenylmethylsulphonyl fluoride (PMSF), supplemented with cOmplete EDTA-free Protease Inhibitor Cocktail (Roche), 40 µg/mL DNAse I (Sigma) and 10 µg/mL bovine pancreas RNAse (Sigma) were added. The pH of the buffer was adjusted to be one unit lower than the isoelectric point (pI) of each tau construct used. Cells were lysed by sonication (3s on; 6s off at 45 % amplitude for 5 min in a Sonics VCX-750 Vibra Cell Ultra Sonic Processor). Lysed cells were further supplemented with 40 µg/ml DNAse (Sigma) and 10 µg/ml RNAse (Sigma) and left to stir for 10 min at room temperature. Cell lysates were centrifuged at 20,000 g for 35 min at 4°C. The supernatant was loaded onto a HiTrap CaptoS column (GE Healthcare) and eluted with a 1 M sodium chloride gradient (Buffer A + 1 M sodium chloride). Fractions were analysed by SDS-PAGE and protein-containing fractions were precipitated using 0.38 g/mL of ammonium sulphate at 4°C for 45 min. Precipitated protein was pelleted at 20,000 g for 30 min at 4°C, and resuspended in 10 mM potassium phosphate buffer at pH 7.2, with 10 mM DTT, and loaded onto a Superdex 75 pg 16/600 size exclusion column (GE Healthcare) using a flow rate of 1 mL/min. Protein-containing fractions were pooled and concentrated using a vivaspin 3 kDa concentrator (Merck) until protein concentrations reached 0.2-1 mM.
In vitro assembly
Protein samples were thawed on ice and filtered (Costar® Spin-X® Centrifuge Tube Filters, 0.22 µm). The membranes of the filters were washed with 20 µl water before applying the sample. Protein sample concentrations after filtering were measured using a NanoDrop™ spectrophotometer. Assembly reactions were prepared in Eppendorf® Protein LoBind tubes. All buffers were filtered. Reactions were prepared by subsequently mixing water, buffering agent (from a 1 M HEPES stock at pH 7.28, or a 1 M phosphate buffer stock at pH 7.2), TCEP (from a 100 mM stock), salt (from a 5 M sodium chloride, 1 M potassium chloride, 1 M potassium citrate, 1 M sodium citrate, 1 M sodium malate) or ATP (from a 750 mM stock), protein, and lastly ThT (from a 150 µM stock). Wells in a 384-well plate were flushed with 100 µL water prior to setting up the reaction which were prepared in batch. Aliquots of 30-40 µL were dispensed in each well, making sure no bubbles were present. Each reaction had an empty well next to it to prevent cross-contamination by evaporation. Shaking conditions were 2 min on, 1 minoff, at 500 rpm orbital shaking at 37°C, reading ThT fluorescence every 10 min.
Seeded assembly in vitro
Frozen frontal cortex from individuals with AD, the same case as used in (Fitzpatrick et al., 2017), were thawed at room temperature, and 100 mg tissue was homogenised in 10 vol extraction buffer (20 mM Tris, pH 7.4, 5 mM EGTA, 5 mM EDTA, 800 mM sodium chloride, 10 % sucrose and 1 % sarkosyl) and incubated at 37 °C for 30 min with shaking. The samples were centrifuged at 20,000 g for 15 min at room temperature, followed by ultracentrifugation of the supernatants at 150,000 g for 30 min. The pellets were resuspended in 100 µl of extraction buffer and incubated at 37 °C for 3 h with shaking at 500 rpm (Eppendorf ThermoMixer C). The samples were then centrifuged at 20,000 g for 20 min, and the supernatants ultracentrifuged at 150,000 g for 30 min at room temperature. Pellets were resuspended in 50 µl of 20 mM Tris at pH 7.4, with 100 mM sodium chloride and stored at 4 °C. One µl was diluted in 99 µl of buffer 20 mM HEPES pH 7.28, and sonicated using a UP200St with VialTweeter operating at 200 W. This served as a x100 stock solution of seeds (at ∼2 µg of original brain tissue per µl of stock solution). A total of 0.4 µl of the AD stock was used per seeding reaction in a 40 µl reaction, corresponding to a total amount of 0.02 µg of brain tissue per µl of the seeded assembly reactions. Assembly reactions were prepared as described above, with the addition of seeds preceding that of ThT.
Lentivirus-mediated generation of tau biosensor cells
HEK293T cells were purchased from ATCC (CRL-3216) and were maintained in High Glucose GlutaMAX Pyruvate Dulbecco’s modified Eagle medium (DMEM) (Thermo Fisher Scientific, #31966047) supplemented with 10% foetal bovine serum (FBS) (Thermo Fisher Scientific, # 10270106), 100 U/mL penicillin, 100 µg/mL streptomycin and grown at 37 °C in 95% O2 / 5% CO2. HEK293T cells stably expressing tau297-391 with an N-terminal human influenza hemagglutinin (HA-) tag were generated as described (Elegheert et al., 2018). Briefly, the construct was amplified using the following pair of primers 4R-HA_I297_EcoRI_fwd: CTGACTGACTGAGAATTCgccaccATGTACCCATACGATGTTCCAGATTACGCTATCAAACACGTCCCGGGAGGC and 4R-E391_XhoI_rev: CTGACTGACTGActcgagTctactaCTCCGCCCCGTGGTCTGTC. The resulting PCR products were cloned into the lentiviral vector pHR_SFFV (Addgene, #79121) using the restriction enzymes EcoRI and XhoI (New England Biolabs). Lentiviral particles were produced using the packaging plasmid psPAX2 (Addgene, #12260) and the VSV-G envelope expressing plasmid pMD2.G (Addgene, #12259). All three plasmids were co-transfected in a 1:1:1 ratio into empty HEK293T cells using Lipofectamine 3000 (Thermo Fisher Scientific, # L3000015) according to the manufacturer’s instructions. After 3 days, cell culture supernatants containing lentiviral particles were collected and passed through a 0.45 µm filter. The clarified supernatants were diluted by addition of 25% vol fresh medium and used to transduce HEK293T cells with Polybrene (Merck, #TR-1003-G) at a final concentration of 10 µg/mL for 72 h.
Cell seeding assays
Before use, recombinant filaments were sonicated in a water-bath (QSonica) for 15 s at 50% amplitude; brain-derived seeds were sonicated during the sarkosyl-extraction procedure. Brain-derived seeds were prepared from 1-2 g of frontal cortex, which was homogenised in 15 vol of extraction buffer. The samples were then centrifuged at 10,000 g for 10 min at 4 °C and the supernatants passed through a 70 µm cell strainer. The clarified homogenates were centrifuged at 150,000 g for 1 h at 4 °C, and the pellets resuspended in 700 µl extraction buffer per g tissue, followed by sonication with a Microson XI-2000 Ultrasonic Cell Disruptor (Misonix) for 20 s and centrifugation at 10,000g for 10 min. The supernatants were diluted threefold in 50 mM Tris-HCl, pH 7.4, containing 150 mM NaCl, 10% sucrose and 0.2% sarkosyl, followed by centrifugation at 150,000g for 1 h. The final pellets were resuspended in 50 µl/g tissue of 20 mM Tris-HCl, pH 7.4, 100 mM NaCl.
Cell seeding experiments were performed as described, with minor modifications (McEwan et al., 2017). Approximately 15,000 cells were plated in black 96-well plates that were pre-coated with poly-D-lysine (Merck, #A-003-E, final coating concentration of 50 µg/ml) and left to adhere overnight. The next day, cells were rinsed with PBS and added to 100 µl Opti-MEM medium (Thermo Fisher Scientific, #31985062) containing the indicated amounts of recombinant or brain-derived assemblies in complex with 1 µl of lipofectamine 2000 (Thermo Fisher Scientific, # 11668019). Cells were incubated at 37 °C for 1 h and the lipofectamine-mediated delivery of the assemblies were stopped by the addition of 100 µl DMEM containing 10% FBS. Two days after the addition of seeds, the cells were fixed with cold methanol for 3 min at room temperature and incubated overnight at 4 °C with an anti-HA antibody (diluted 1:2000) (Biolegend,). Cells were then rinsed three times with PBS and incubated for 1 h at room temperature with an Alexa647-conjugated goat anti-mouse antibody (diluted 1:1000) (Thermo Fisher Scientific). After rinsing three times with PBS, cell nuclei were stained with 1 µg/ml Hoechst dye (Thermo Fisher Scientific, #H3570) for 10 min and images were acquired at 405 and 647 nm on a Ti2-E High Content Microscope (Nikon) using the 10x objective. For the experiments with the Dy-Light488-labelled assemblies, images were acquired at 488 nm. Nine fields per well were read in a horizontal serpentine acquisition mode with a 10x objective and the downstream analysis was performed using the Fiji software (Schindelin et al., 2012). For nuclear counting, the images acquired at 405 nm were locally subtracted for background using the Rolling ball algorithm, and cells were segmented based on nuclear staining using the Median filter and Find Maxima tools, with the option of “Segmented Particles above lower threshold” activated. Seeded aggregates were detected in the 488 nm images and quantified using the ComDet plugin (Katrukha, 2020) in Fiji. Positive puncta were determined by an approximate particle size of 6 pixels, while the intensity threshold was variable between experiments and was based on detecting the minimum number of aggregates in the unseeded condition. Finally, the relative levels of seeding were calculated as the number of aggregates in each field was normalised to the corresponding number of cells and was then compared to the untreated control.
Labelling of pre-assembled filaments
Filaments were labelled as described in the provider protocol description. A 50 µg aliquot of DyLight™ 488 NHS Ester (Thermo ScientificTM) was dissolved in 20 µl HEPES buffer (pH 7.28), followed by the addition of 50 µL of filament solution. The reactions were incubated for 1 h at room temperature in the dark. For biotinylation, 1 mg of EZ-Link™ Sulfo-NHS-Biotin (Thermo ScientificTM) was dissolved in 300 uL dimethylsulfoxide (DMSO). Then, 0.5 µL of the biotin solution was added to 20 µl of filament solution, maintaining a 5:1 (biotin:tau) molar ratio. The reaction was incubated for 30 min at room temperature.
Subsequently, both reactions were ultracentrifuged at 100,000 g for 20 min at room temperature. Pellets were resuspended in 20 mM HEPES buffer (pH 7.28) with 100 mM potassium citrate, using 50 µl and 20 µl for the DyLight-labelled filaments and the biotinylated filaments, respectively. To confirm successful labelling, 3.5 µg of DyLight labelled filaments were analysed via SDS-PAGE (4-20%), followed by exposure at a wavelength of 488 nm. Biotinylation was validated by immuno-EM.
Negative stain EM and Immuno-EM
For negative stain EM, samples were diluted 10-fold (to ∼4 µM) in 20 mM HEPES at pH 7.3, and applied to glow-discharged carbon grids for 1 min, blotted, washed with 4 µl water, blotted and stained with 2% uranyl acetate for 2 min, blotted and imaged.
For immuno-EM, after applying filaments to the grid for 2 min, they were blotted and blocked using 0.5% (fish gelatin) in PBS for 5 min, washed with 200 µl of water, and incubated with 50 µl streptavidin-coated with 10 nm gold nanobeads (Sigma) in a ratio of 1:20 in blocking buffer, incubated for 30 min, further washed with 200 µl of water, blotted and stained with 2% uranyl acetate for 2 min, before being blotted and imaged by transmission electron microscopy at room temperature.
Cryo-EM data acquisition
Protein sample aliquots of 3 µL were applied to glow-discharged holey carbon grids (Quantifoil Au R1.2/R1.3 300 mesh), blotted with filter paper, and plunge frozen into liquid ethane using an FEI vitrobot Mark IV (100 % humidity, 4 °C).
The cryo-EM images in Figure 1 – figure supplement 3 were recorded on TFS Glacios using a Falcon 3 direct electron detector. All other cryo-EM images were recorded on Krios G1, Krios G2 and Krios G4 (Thermo Fisher Scientific) electron microscopes. Images on the Krios G1 microscope were recorded using a Gatan K3 and a Gatan energy filter with a slit width of 20 eV. Images on the Krios G2 were recorded on a Falcon-4i Camera (Thermo Fisher Scientific). Images on the Krios G4 were recorded on a Falcon-4i camera and a Selectris X (Thermo Fisher Scientific) energy filter with a slit width of 10 eV. Images were recorded at a dose of 30-40 electrons/Å2, using EPU software (Thermo Fisher Scientific). Images from the Gatan K3 were saved in tiff. Images on the Falcon 4i camera were saved as EER movies and converted to tiff format using an EER-grouping of 34 or 40 frames, to give a dose fractionation of approximately one electron/Å2.
Cryo-EM image processing
Raw micrograph movies were gain-corrected, aligned and dose-weighted using RELION’s motion correction (Zivanov et al., 2019). Contrast transfer function (CTF) parameters were estimated using CTFFIND-4.1 (Rohou and Grigorieff, 2015). Helical reconstruction was performed in RELION-5.0. Filaments were picked manually or automatically using a modified version of Topaz (Lövestam and Scheres, 2022). Picked particles were extracted in box sizes of 1024, 768 or 512 pixels and down-scaled to 256, 128 or 64 pixels for initial 2D classifications. We performed reference-free 2D classification for 35 iterations with 150-200 classes, ignoring the CTF until the first peak, to assess overall filament quality, presence of different polymorphs and crossover distances. Polymorphs were separated by the hierarchical clustering approach (Lövestam et al., 2024). Additional 2D classifications were run for each identified cluster, iterating the procedure until homogenous populations of 2D classes were generated. Initial 3D references were generated by 2D class averages using relion_helix_inimodel2d (Scheres, 2020). Selected particles were re-extracted in boxes of 384 pixels for initial 3D refinement. Subsequently, 3D classifications and 3D auto-refinements were used to optimise helical parameters and improve the reconstructions, applying symmetry where necessary. Bayesian polishing and CTF refinements were used to increase resolutions (Zivanov et al., 2020, 2019). Final maps were sharpened using standard post-processing procedures in RELION, and reported resolutions were estimated using a threshold of 0.143 in the Fourier shell correlation (FSC) between two independently refined half maps (Chen et al., 2013; Scheres and Chen, 2012). Further details of cryo-EM structure determination are given in Table 1.
Nuclear magnetic resonance
Unless otherwise stated, Tau construct data sets were collected at 278K using a Bruker Avance II+ spectrometer operating at a 700 MHz proton frequency and fitted with a 5mm TCI triple resonance cryoprobe. All samples were prepared in 50mM potassium phosphate (pH 7.4) with 150mM NaCl, 10 mM DTT, and 5% D2O as a lock solvent.
Backbone NH, N, Cα, Cβ and C’ resonances of both tau151-391 and tau297-441 with and without phosphomimetic mutations were assigned using isotopically enriched (15N/13C) 300 mM samples. Standard 3D datasets were acquired as pairs to provide own and preceding carbon connectivities, using between 18-39% non-uniform sampling (NUS) to aid faster data acquisition. Both the HNCO & HN(CA)CO, HNCA & HN(CO)CA and CBCA(CO)NH & HNCACB complimentary 3D datasets were collected with 2048, 80 and 128 complex points in the 1H, 15N and 13C dimensions, respectively. Additional 15N connectivities were established with an (H)N(COCA)NNH experiment with 2048, 80 and 128 complex points in the 1H, direct and indirect 15N dimensions, respectively. C’-detect experiments c_hcacon_ia3d and c_hcanco_ia3d (Bruker) were also collected to aid backbone assignment with 1024, 64 and 128 points collected in the direct 13C, 15N and indirect 13C dimensions respectively. These experiments benefitted, in the case of the 151-391 PTM samples, from the increased sensitivity of a 700 MHz TXO, X-detect optimised cryoprobe.
All raw NMR data were processed using Topspin versions 3.2 or 4 (Bruker) or, if required, NMRPipe (Delaglio et al., 1995) with compressed sensing for reconstruction of NUS data (Kazimierczuk and Orekhov, 2011) and were analysed using NMRFAM-Sparky or POKY (Lee et al., 2021) and MARS (Jung and Zweckstetter, 2004).
Secondary structure preferences of individual tau residues were derived from a secondary chemical shift analysis. Random coil Cα and Cβ chemical shift values for the tau constructs primary sequence were calculated according to (Kjaergaard et al., 2011; Kjaergaard and Poulsen, 2011; Schwarzinger et al., 2001), with appropriate corrections for the same experimental conditions (pH and temperature). Subtraction of these values from the experimentally derived values yielded ΔCα and ΔCβ. When ΔCβ is subtracted from ΔCα, a negative value indicates that the residue resides in an extended backbone conformation, whereas positive values suggest a helical preference.
The change in relative peak position as a result of the phosphomimetic mutations (Chemical Shift Perturbations: CSP) was determined using the following equation:
To monitor any relative changes in peak intensities between tau constructs, peak heights extracted from processed HSQC spectra were normalised to the carboxy-terminal residue of each construct and all other relative peak intensities were adjusted accordingly. The picosecond timescale backbone dynamic properties of each tau construct were probed with interleaved 2D HetNOE experiments (Bruker) with a recovery delay of 5 s.
Data availability
Cryo-EM maps and atomic models for PHFs formed with a mixture of 0N3R:0N4R PAD12 tau and for PHFs formed with 0N3R PAD12 tau and seeded with filaments extracted from the brain of an individual with AD have been deposited in EMDB and the PDB (see Table 1).
Acknowledgements
We thank Jake Grimmett, Toby Darling and Ivan Clayson for help with high-performance computing; David Li and Max Wilkinson for helpful discussions; and Tony Crowther for critical reading of the manuscript. This work was supported by the facilities for Biophysics, Electron Microscopy, NMR and Scientific Computing of the Medical Research Council (MRC) Laboratory of Molecular Biology. This work was supported by the MRC, as part of United Kingdom Research and Innovation (UKRI) [MC_U105184291 to M.G. and MC_UP_A025-1013 to S.H.W.S.].
Additional information
Author contributions
S.L. performed biochemistry, filament assembly and cryo-EM data acquisition; T.K. performed cellular seeding assays; S.L. and S.H.W.S. analysed cryo-EM data; J.L.W. and S.M.V.F. performed NMR data acquisition and analysis; M.G. and S.H.W.S. supervised the project. All authors contributed to the writing of the manuscript.
Additional Information
For the purpose of open access, the MRC Laboratory of Molecular Biology has applied a CC BY public copyright licence to any Author Accepted Manuscript version arising.
References
- Cryo-EM structure of RNA-induced tau fibrils reveals a small C-terminal core that may nucleate fibril formationProc Natl Acad Sci U S A 119https://doi.org/10.1073/pnas.2119952119
- Tau (297-391) forms filaments that structurally mimic the core of paired helical filaments in Alzheimer’s disease brainFEBS Lett 594:944–950https://doi.org/10.1002/1873-3468.13675
- Phosphorylation of tau at Thr212, Thr231, and Ser262 combined causes neurodegenerationJ Biol Chem 285:30851–30860https://doi.org/10.1074/jbc.M110.110957
- High-resolution noise substitution to measure overfitting and validate resolution in 3D structure determination by single particle electron cryomicroscopyUltramicroscopy 135:24–35https://doi.org/10.1016/j.ultramic.2013.06.004
- Knockin’ on heaven’s door: Molecular mechanisms of neuronal tau uptakeJ Neurochem 156:563–588https://doi.org/10.1111/jnc.15144
- NMRPipe: a multidimensional spectral processing system based on UNIX pipesJ Biomol NMR 6:277–293https://doi.org/10.1007/BF00197809
- Identification of the Tau phosphorylation pattern that drives its aggregationProceedings of the National Academy of Sciences 114:9080–9085https://doi.org/10.1073/pnas.1708448114
- Alzheimer’s Disease Seeded Tau Forms Paired Helical Filaments Yet Lacks Seeding PotentialJournal of Biological Chemistry 300https://doi.org/10.1016/j.jbc.2024.107730
- Milligram-scale assembly and NMR fingerprint of tau fibrils adopting the Alzheimer’s disease foldJournal of Biological Chemistry 300https://doi.org/10.1016/j.jbc.2024.107326
- Lentiviral transduction of mammalian cells for fast, scalable and high-level production of soluble and membrane proteinsNat Protoc 13:2991–3017https://doi.org/10.1038/s41596-018-0075-9
- Cryo-EM structures of amyloid-β and tau filaments in Down syndromeNat Struct Mol Biol 31:903–909https://doi.org/10.1038/s41594-024-01252-3
- Cryo-EM structures of tau filaments from Alzheimer’s diseaseNature 547:185–190https://doi.org/10.1038/nature23002
- IVA cloning: A single-tube universal cloning system exploiting bacterial In Vivo AssemblySci Rep 6https://doi.org/10.1038/srep27459
- Cryo-EM structures reveal tau filaments from Down syndrome adopt Alzheimer’s disease foldActa Neuropathol Commun 12https://doi.org/10.1186/s40478-024-01806-y
- Reconstitution of the Alzheimer’s Disease Tau Core Structure from Recombinant Tau297-391 Yields Variable Quaternary Structures as Seen by Negative Stain and Cryo-EMBiochemistry 63:194–201https://doi.org/10.1021/acs.biochem.3c00425
- Assembly of microtubule-associated protein tau into Alzheimer-like filaments induced by sulphated glycosaminoglycansNature 383:550–553https://doi.org/10.1038/383550a0
- Tau proteins of Alzheimer paired helical filaments: abnormal phosphorylation of all six brain isoformsNeuron 8:159–168
- Multiple isoforms of human microtubule-associated protein tau: sequences and localization in neurofibrillary tangles of Alzheimer’s diseaseNeuron 3:519–526
- Hydrofluoric acid-treated tau PHF proteins display the same biochemical properties as normal tauThe Journal of biological chemistry 267
- Abnormal phosphorylation of the microtubule-associated protein tau (tau) in Alzheimer cytoskeletal pathologyProc Natl Acad Sci U S A 83:4913–4917https://doi.org/10.1073/pnas.83.13.4913
- Pseudophosphorylation of tau protein alters its ability for self-aggregationJ Neurochem 88:1509–1520https://doi.org/10.1046/j.1471-4159.2003.02287.x
- Structure of Tau filaments in Prion protein amyloidosesActa Neuropathol 142:227–241https://doi.org/10.1007/s00401-021-02336-w
- Tau and neurodegenerative disease: the story so farNat Rev Neurol 12:15–27https://doi.org/10.1038/nrneurol.2015.225
- Proline-directed Pseudo-phosphorylation at AT8 and PHF1 Epitopes Induces a Compaction of the Paperclip Folding of Tau and Generates a Pathological (MC-1) Conformation*Journal of Biological Chemistry 283:32066–32076https://doi.org/10.1074/jbc.M805300200
- Mars -- robust automatic backbone assignment of proteinsJ Biomol NMR 30:11–23https://doi.org/10.1023/B:JNMR.0000042954.99056.ad
- RNA stimulates aggregation of microtubule-associated protein tau into Alzheimer-like paired helical filamentsFEBS Lett 399:344–349https://doi.org/10.1016/s0014-5793(96)01386-5
- ekatrukha/ComDet: ComDet 0.5.3Zenodo https://doi.org/10.5281/zenodo.4281064
- Accelerated NMR spectroscopy by using compressed sensingAngew Chem Int Ed Engl 50:5556–5559https://doi.org/10.1002/anie.201100370
- Random coil chemical shift for intrinsically disordered proteins: effects of temperature and pHJ Biomol NMR 49:139–149https://doi.org/10.1007/s10858-011-9472-x
- Sequence correction of random coil chemical shifts: correlation between neighbor correction factors and changes in the Ramachandran distributionJ Biomol NMR 50:157–165https://doi.org/10.1007/s10858-011-9508-2
- How do protein aggregates escape quality control in neurodegeneration?Trends Neurosci 45:257–271https://doi.org/10.1016/j.tins.2022.01.006
- A68: a major subunit of paired helical filaments and derivatized forms of normal TauScience 251:675–678https://doi.org/10.1126/science.1899488
- Neurodegenerative tauopathiesAnnu. Rev. Neurosci 24:1121–1159https://doi.org/10.1146/annurev.neuro.24.1.1121
- POKY: a software suite for multidimensional NMR and 3D structure calculation of biomoleculesBioinformatics 37:3041–3042https://doi.org/10.1093/bioinformatics/btab180
- Phosphorylation affects the ability of tau protein to promote microtubule assemblyJ Biol Chem 259:5301–5305
- Assembly of recombinant tau into filaments identical to those of Alzheimer’s disease and chronic traumatic encephalopathyElife 11https://doi.org/10.7554/eLife.76494
- Disease-specific tau filaments assemble via polymorphic intermediatesNature 625:119–125https://doi.org/10.1038/s41586-023-06788-w
- High-throughput cryo-EM structure determination of amyloidsFaraday Discuss 240:243–260https://doi.org/10.1039/d2fd00034b
- Structures of AT8 and PHF1 phosphomimetic tau: Insights into the posttranslational modification code of tau aggregationProc Natl Acad Sci U S A 121https://doi.org/10.1073/pnas.2316175121
- Cytosolic Fc receptor TRIM21 inhibits seeded tau aggregationProc. Natl. Acad. Sci. U.S.A 114:574–579https://doi.org/10.1073/pnas.1607215114
- Monoclonal antibodies with selective specificity for Alzheimer Tau are directed against phosphatase-sensitive epitopesActa Neuropathol 84:265–272https://doi.org/10.1007/BF00227819
- Molecular characterization of the minimal protease resistant tau unit of the Alzheimer’s disease paired helical filamentEMBO J 12:365–370https://doi.org/10.1002/j.1460-2075.1993.tb05665.x
- Polymerization of tau into filaments in the presence of heparin: the minimal sequence required for tau-tau interactionJ. Neurochem 67:1183–1190https://doi.org/10.1046/j.1471-4159.1996.67031183.x
- PrionsProc Natl Acad Sci U S A 95:13363–13383https://doi.org/10.1073/pnas.95.23.13363
- Tau filaments with the chronic traumatic encephalopathy fold in a case of vacuolar tauopathy with VCP mutation D395GActa Neuropathol 147https://doi.org/10.1007/s00401-024-02741-x
- Tau filaments with the Alzheimer fold in cases with MAPT mutations V337M and R406WbioRxiv https://doi.org/10.1101/2024.04.29.591661
- Tau filaments from amyotrophic lateral sclerosis/parkinsonism-dementia complex adopt the CTE foldProc Natl Acad Sci U S A 120https://doi.org/10.1073/pnas.2306767120
- CTFFIND4: Fast and accurate defocus estimation from electron micrographsJ. Struct. Biol 192:216–221https://doi.org/10.1016/j.jsb.2015.08.008
- Amyloid structure determination in RELION-3.1Acta Crystallogr D Struct Biol 76:94–101https://doi.org/10.1107/S2059798319016577
- Prevention of overfitting in cryo-EM structure determinationNat. Methods 9:853–854https://doi.org/10.1038/nmeth.2115
- Molecular pathology of neurodegenerative diseases by cryo-EM of amyloidsNature 621:701–710https://doi.org/10.1038/s41586-023-06437-2
- Fiji: an open-source platform for biological-image analysisNat Methods 9:676–682https://doi.org/10.1038/nmeth.2019
- Sequence-dependent correction of random coil NMR chemical shiftsJ Am Chem Soc 123:2970–2978https://doi.org/10.1021/ja003760i
- Cryo-EM structures of tau filaments from the brains of mice transgenic for human mutant P301S TauActa Neuropathologica Communications 11https://doi.org/10.1186/s40478-023-01658-y
- Novel tau filament folds in individuals with MAPT mutations P301L and P301TbioRxiv https://doi.org/10.1101/2024.08.15.608062
- Structure-based classification of tauopathiesNature 598:359–363https://doi.org/10.1038/s41586-021-03911-7
- Cryo-EM structures of tau filaments from SH-SY5Y cells seeded with brain extracts from cases of Alzheimer’s disease and corticobasal degenerationFEBS Open Bio 13:1394–1404https://doi.org/10.1002/2211-5463.13657
- Tau PTM Profiles Identify Patient Heterogeneity and Stages of Alzheimer’s DiseaseCell 183:1699–1713https://doi.org/10.1016/j.cell.2020.10.029
- Free fatty acids stimulate the polymerization of tau and amyloid beta peptides. In vitro evidence for a common effector of pathogenesis in Alzheimer’s diseaseAm J Pathol 150:2181–2195
- Structural characterization of the core of the paired helical filament of Alzheimer diseaseProc. Natl. Acad. Sci. U.S.A 85:4884–4888
- Heparin-induced tau filaments are polymorphic and differ from those in Alzheimer’s and Pick’s diseasesElife 8https://doi.org/10.7554/eLife.43584
- Cryo-EM structures reveal variant Tau amyloid fibrils between the rTg4510 mouse model and sporadic human tauopathiesCell Discov 10https://doi.org/10.1038/s41421-023-00637-w
- Estimation of high-order aberrations and anisotropic magnification from cryo-EM data sets in RELION-3.1IUCrJ 7:253–267https://doi.org/10.1107/S2052252520000081
- A Bayesian approach to beam-induced motion correction in cryo-EM single-particle analysisIUCrJ 6:5–17https://doi.org/10.1107/S205225251801463X
Article and author information
Author information
Version history
- Preprint posted:
- Sent for peer review:
- Reviewed Preprint version 1:
Copyright
© 2025, Lövestam et al.
This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.
Metrics
- view
- 1
- downloads
- 0
- citations
- 0
Views, downloads and citations are aggregated across all versions of this paper published by eLife.