. Cryo-EM structures of tau filaments from Alzheimer's disease. Nature. 2017 Jul 13;547(7662):185-190. Epub 2017 Jul 5 PubMed.


Please login to recommend the paper.


  1. It is hard to overstate how important this new paper from Goedert, Scheres, and colleagues is for the Alzheimer’s disease field. From a clinical perspective, these new structures will have significant implications for how compounds are designed to detect and eliminate AD-associated tau fibrils (Kolb and Andrés, 2017). From a mechanistic perspective, this paper represents a fountainhead for a structural understanding of how different tau-associated phenotypes (or tau prion strains) arise from the structural heterogeneity of tau amyloid fibrils (Sanders et al., 2016).

    Notably and surprisingly, the authors show that AD-associated tau fibrils, which include straight filaments (SFs) and paired helical filaments (PHFs), feature identical monomeric core segments and structures. In other words, differences in β-sheet alignment and amino acid identity in protease-resistant cores of the two types of filaments are not responsible for their morphological differences!

    So what accounts for the distinctive shapes of the two types of AD-associated tau fibrils commonly seen in electron micrographs? Rather than core structure, the authors convincingly demonstrate that differences between tau amyloid filaments can be attributed to packing between cores. In other words, small tweaks in how monomers within the fibril associate with one another has drastic effects on the fibril shape. This fits amazingly well with previous work (Boluda et al., 2015; Clavaguera et al., 2013; Kaufman et al., 2016; Sanders et al., 2014) on patient-derived tau strains, which demonstrated that most AD patients likely feature a single common strain (Sanders et al., 2014), driven by a homogenous underlying amyloid structure that propagates throughout the brain to drive pathogenesis.

    In other words, despite morphological differences between filaments—paired, helical, or straight—both can result from a single structure, which is defined by the identity of the core itself.

    I am quite excited to see future cryo-EM-derived structures of tau fibrils obtained from postmortem tissues from patients with distinct tauopathies. Based on previous work from the Goedert, Lee/Trojanowski, and Diamond labs, it seems highly likely that distinct structures (or prion strains) will be associated with distinct diseases. Moreover, I expect to see that patients with identical syndromes will feature common structures. A structural understanding of strains associated with various diseases will provide fundamental insights into what accounts for phenotypic differences between distinct tauopathies.

    With that in mind, and with my mind elsewhere in the luxurious pastures of phase separation, I’d like to pose a few questions that I look forward to the field answering:

    1. Do all AD patients feature a common tau fibril structure? It will be critical to analyze fibrils obtained from numerous patients at various stages of disease to conclusively demonstrate this. Moreover, it will be important to show that fibrils isolated from distinct brain regions of a single patient’s brain feature similar structures, thus confirming that these stably propagate between neurons and along networks.
    2. What is responsible for rapidly progressive AD? Do these individuals feature unique tau structures/strains? If so, what accounts for this? It is quite possible that fragility of individual fibrils could drive progression rate, similarly to what has been observed for classic prion diseases (Legname et al., 2006) as well as beneficial yeast prions (Tanaka et al., 2006). 
    3. Will we be able to conclusively demonstrate that unique tauopathies are associated with disease-specific structures? Once obtained, can we create models that recapitulate the selective neuronal vulnerability and morphologically distinct inclusion types associated with various diseases? Recent work (Boluda et al., 2015; Clavaguera et al., 2013; Kaufman et al., 2016) from the previously mentioned labs represents an essential first step. However, homogenous purification and definition of specific structures are of critical importance!
    4. What is the link between Aβ and this very specific tau structure/strain? This has been a nagging question ever since the field identified tau and Aβ in AD brains. Yet, in my opinion, little progress has been made toward solving this vexing question. It is fascinating how the vast majority of patients with tau pathology are of the Alzheimer’s type. Solving why the other tau structures/strains are not more common is a problem of fundamental importance.

    Finally, I’d like to state that I greatly appreciate the authors’ acknowledgement that this work was only made possible by decades of support from their institution and collaborators. Solving big problems requires tremendous time, effort, resources, and perseverance. This work is evidence that such approaches to science pay enormous dividends. 


    . Differential induction and spread of tau pathology in young PS19 tau transgenic mice following intracerebral injections of pathological tau from Alzheimer's disease or corticobasal degeneration brains. Acta Neuropathol. 2015 Feb;129(2):221-37. Epub 2014 Dec 24 PubMed.

    . Brain homogenates from human tauopathies induce tau inclusions in mouse brain. Proc Natl Acad Sci U S A. 2013 Jun 4;110(23):9535-40. PubMed.

    . Tau Prion Strains Dictate Patterns of Cell Pathology, Progression Rate, and Regional Vulnerability In Vivo. Neuron. 2016 Nov 23;92(4):796-812. Epub 2016 Oct 27 PubMed.

    . Tau Positron Emission Tomography Imaging. Cold Spring Harb Perspect Biol. 2017 May 1;9(5) PubMed.

    . Continuum of prion protein structures enciphers a multitude of prion isolate-specified phenotypes. Proc Natl Acad Sci U S A. 2006 Dec 12;103(50):19105-10. PubMed.

    . Distinct tau prion strains propagate in cells and mice and define different tauopathies. Neuron. 2014 Jun 18;82(6):1271-88. Epub 2014 May 22 PubMed.

    . Prions and Protein Assemblies that Convey Biological Information in Health and Disease. Neuron. 2016 Feb 3;89(3):433-48. PubMed.

    . The physical basis of how prion conformations determine strain phenotypes. Nature. 2006 Aug 3;442(7102):585-9. Epub 2006 Jun 28 PubMed.

    View all comments by David Sanders
  2. This paper by Fitzpatrick et al. is a seminal accomplishment in two fields—structural biology and neurodegeneration research, both achieved at the MRC LMB in Cambridge, U.K.

    For structural biology, it is an example of the recent revolution in high-resolution cryo-electron microscopy and image reconstruction. It started out ~1975 when Nigel Unwin and Richard Henderson showed that unstained cryo-preserved membrane proteins retained their native structure at atomic resolution, provided that they were imaged at low electron doses to minimize radiation damage (Unwin and Henderson, 1975). The structures could be solved by computer-based averaging over many molecules and thus revealed structures from “invisible images.” Many years of development in electron microscopy, computer and detector technology, and procedures of image processing have now opened the possibility of solving the structures of a wide range of biomolecules at high resolution, rivalling those obtained by X-ray crystallography (Scheres, 2012). 

    In neurodegeneration research, one of the key questions was the origin and structure of the “paired helical filaments” (Kidd, 1963) that appear in the brains of Alzheimer‘s patients. The major component turned out to be the microtubule-associated protein tau (Grundke-Iqbal et al., 1986). This protein had been discovered by Marc Kirschner and colleagues (Weingarten et al., 1975; Lee et al., 1988), who noted two key properties, stabilization of axonal microtubules and an unusually hydrophilic composition, making it the prototype of a natively unfolded and highly soluble protein. It is therefore still a mystery why this protein  aggregates into semi-ordered filaments, yet the distribution of these filaments in the brain serves as a reliable marker of the progression of AD (the “Braak stages” proposed by Heiko and Eva Braak, 1991). In several important papers around 1988 onwards, the group at LMB Cambridge (Aaron Klug, Claude Wischik, Michel Goedert, Tony Crowther, and colleagues) described key structural and biochemical properties of tau and tau filaments, including the six isoforms of human tau, the core of PHFs and their “fuzzy coat,” and low-resolution reconstructions from electron micrographs of negatively stained tau fibers from Alzheimer brain (PHF and straight filaments). Three decades later, the new publication confirms the basic conclusion and extends them to near-atomic resolution of the core of tau fibers.

    Our lab has been active in tau research since ~1988, much of it on structural, biochemical, and cellular aspects of tau. The new structure of Fitzpatrick et al. now offers a chance to revisit earlier views, as discussed below:

    Core of PHFs and amyloidogenic motifs: The tau sequence contains two hexapeptide motifs with high β-propensity, 306-VQIVYK-311 at the beginning of R3 (termed PHF6) and 275-VQIINK-280 (=PHF*) at the beginning of R2 (von Bergen et al., 2000). Their extended conformation is already visible by NMR in soluble tau (Mukrasch et al., 2009), and the corresponding peptides can be crystallized to reveal the amyloid-like packing into cross-β-sheets (Sawaya et al., 2007). Both motifs strongly promote aggregation of tau, yet the new structure reveals only the motif VQIVYK in R3. Thus it appears that there is another β-structured part in the filament comprising R2 which may be too heterogeneous to show up in the map. Note that in the filaments from this patient, R2 is present in only 50 percent of the tau molecules because of alternative splicing of exon 10 (encoding R2). However, even filaments consisting entirely of 4-repeat tau show this heterogeneity, in contrast to R3, as judged by site-directed EPR (Margittai and Langen, 2006). The missing β-structured part of R2 might become visible in future studies of tau filaments from 4R-tauopathies. A surprising feature is the pronounced appearance of repeats R4 and R5. R4 also harbors an amyloigenic hexapeptide motif, 337-VEVKSE-342 (=PHF6**), with an extended conformation already noticeable in soluble tau. The sequence of R5,also termed R' (Gustke et al., 1994) shows a lower sequence homology than the other repeats, but its initial hexapeptide KKIET also shows up as β-structure in the map. Frequently used recombinant constructs comprising the repeat domains such as K19 (=R1+R3+R4) or K18 (=R1+R2+R3+R4) end with the motif 369-KKIE-372 of repeat R5, which may be one reason why they can be assembled into filaments resembling PHFs.

    Tau mutations: Most of the mutations known to cause tauopathies in humans occur around the repeat domain, in particular in or near the hexapeptide motifs. One likely mode of action is to enhance the tendency for tau aggregation. Examples are N279K, ΔK280 (in R2), P301L, P301S (just upstream of R3), V337M (in R4), K369I (in R5). Alternatively, these mutations could affect the binding of tau to microtubules or other tau effectors, since the hexapeptide motifs are also involved in those interactions (Kadavath et al., 2015). 

    Protofilaments: The term "paired helical filament" was coined by Kidd  to describe the appearance of stained filaments from AD brain tissue. However, a fraction of filaments did not have the twisted appearance and were therefore called "straight filaments" (Crowther, 1991). Since the packing of tau molecules was not known there has been a debate in the literature on whether the straight filaments (SF) consisted of one or two protofibrils (Pollanen et al., 1997; Wegmann et al., 2010), especially since it was not possible to observe protofilaments fraying apart (unlike, say, the case of microtubules, where fraying protofilaments are readily visible). The new structure puts the issue to rest by clearly revealing two protofilaments, consistent with Crowther's 1991 image reconstruction.

    Mass distribution: Because tau is highly soluble it does not self-assemble readily, except in special conditions (Wille et al., 1992) or with cofactors distinct from the cellular milieu (Goedert et al., 1996). The fibers assembled in vitro show some heterogeneity, which caused a debate whether they had the same packing of molecules as PHFs from AD brain tissue. We therefore determined the molecular mass per length of AD-PHFs and filaments reassembled in vitro from recombinant tau with different domain compositions (von Bergen et al., 2006). This showed that all filaments contained ~four molecules per nm length, independent of the size of the tau subunits. This agrees well with the structure of Fitzpatrick et al., which contains two protofilaments, with 0.47nm spacing between molecules in each.

    Oxidation of tau: Tau contains either one cysteine (C322 in 3-repeat tau) or two (C291+C322 in 4-repeat tau). Cells have a reducing environment, but aging cells are compromised such that oxidation can occur. When studying the effects of oxidation we noticed that 4-repeat tau assembles poorly, whereas 3-repeat tau assembles much more readily than controls where the cysteines had been substituted by Ala. The explanation was that oxidized 4-repeat tau assumes a folded conformation via an intrachain disulfide crosslink between C291 and C322 which is not well integrated into a fiber, whereas oxidized 3-repeat tau can form interchain disulfide bridges to generate dimers and then polymers (Schweers et al., 1995). This interpretation is consistent with the new structure: Although it shows only C322 (because R2 is not resolved), its position precludes an interaction with C291 in the assembled filament. By contrast, C322 is able to make a disulfide bond with another molecule in the axial direction, thus promoting the templated assembly of the filament. The existence of such a disulfide bond has been verified by solid state NMR (Daebel et al., 2012). By implication, the structure also illustrates how blocking of SH goups (e.g., by the aggregation inhibitor methylene blue) can antagonize the formation of PHFs (Akoury et al., 2013). 

    Phosphorylation of tau: In neurons, aggregation of tau is usually accompanied by phosphorylation, but does not depend on it. Tau contains multiple phosphorylation sites, most of them outside the repeat domain. The predominant sites within the repeats are the serines in the KXGS motifs (X=I or C) in R1-R4 (phosphorylation mainly of S262, S356 in the KIGS motifs, whereas S293, S324 in the KCGS motifs are minor sites). They can be targeted by kinases MARK, SADK, PKA, and others, and all sites are recognized by the antibody 12E8 generated by Peter Seubert at Elan. This type of phosphorylation, especially at S262, reduces tau-microtubule interactions as well as tau aggregation (Schneider et al., 1999). The new map shows only residues S324 (in R3) and S356 (in R4) in an unphosphorylated form. S324 faces outside of the C-shaped structure, consistent with its minor role in inhibiting PHF assembly. S356 lies on the inside of the C-shape and appears to be tightly squeezed between charged residues K354 and D358, suggesting that phosphorylation could disrupt the conformation and thus PHF assembly. The environment of S262 remains to be elucidated.

    Non-tau components of PHFs: Recombinant tau protein is soluble at concentrations >200 µM, even in a highly phosphorylated state (Tepper et al., 2014). On the other hand, neuronal tau is in the low µM range, ~10-fold lower than tubulin (Drubin et al., 1984). Therefore, tau by itself would not assemble in cells nor in vitro in physiological buffer conditions unless it interacts with a cellular partner that compensates its excess positive charges, especially in the repeat domain that harbors the elements for β-structure. In vitro, filament aggregation can be achieved by polyanions such as heparin (Goedert et al., 1996), RNA, or acidic peptides (Friedhoff et al., 1998). The acidic C-terminal tail of tubulin would qualify as a partner for tau aggregation, except that microtubules act as chaperones of tau. Erickson and Voter likened the tubulin-tau interaction to the complex coacervation of polyelectrolytes, where assembly of (acidic) tubulin is induced by (basic) polycations, e.g. tau (Erickson and Voter, 1976). Reversing the argument, the assembly of (basic) tau can be induced by (acidic) polyanions. In vitro, heparin serves as an inducer, but in the cytoplasm there must be alternatives. Examples are the abundant RNAs (rRNA, mRNA, tRNA). Indeed, stress granules containing RNA and proteins including tau can be viewed as one type of complex coacervate (Vanderweyde et al., 2016). The question is: Where are the extra components in the PHF map? The ordered elements reflect only a part of tau itself. It is possible that other components are hidden in the disordered background, or that they were removed by proteolysis during aging or preparation. In any case, the secret to understanding pathological tau aggregation might well lie in the polyanionic binding partner(s), rather than just in tau.

    In 1903, when Alois Alzheimer moved to the University Clinic in Munich headed by Emil Kraepelin, he declined to become a regular professor with teaching duties, but instead asked for a room full of microscopes to pursue the research for which he became famous. One hundred and fourteen years and a huge progress in microscopy methods later, the new structure of the protein that forms neurofibrillary tangles is a fitting follow-up that will propel future directions of research in neurodegenerative diseases.


    . Molecular structure determination by electron microscopy of unstained crystalline specimens. J Mol Biol. 1975 May 25;94(3):425-40. PubMed.

    . A Bayesian view on cryo-EM structure determination. J Mol Biol. 2012 Jan 13;415(2):406-18. Epub 2011 Nov 12 PubMed.

    . Microtubule-associated protein tau. A component of Alzheimer paired helical filaments. J Biol Chem. 1986 May 5;261(13):6084-9. PubMed.

    . A protein factor essential for microtubule assembly. Proc Natl Acad Sci U S A. 1975 May;72(5):1858-62. PubMed.

    . The primary structure and heterogeneity of tau protein from mouse brain. Science. 1988 Jan 15;239(4837):285-8. PubMed.

    . Neuropathological stageing of Alzheimer-related changes. Acta Neuropathol. 1991;82(4):239-59. PubMed.

    . Assembly of tau protein into Alzheimer paired helical filaments depends on a local sequence motif ((306)VQIVYK(311)) forming beta structure. Proc Natl Acad Sci U S A. 2000 May 9;97(10):5129-34. PubMed.

    . Structural polymorphism of 441-residue tau at single residue resolution. PLoS Biol. 2009 Feb 17;7(2):e34. PubMed.

    . Atomic structures of amyloid cross-beta spines reveal varied steric zippers. Nature. 2007 May 24;447(7143):453-7. PubMed.

    . Side chain-dependent stacking modulates tau filament structure. J Biol Chem. 2006 Dec 8;281(49):37820-7. PubMed.

    . Domains of tau protein and interactions with microtubules. Biochemistry. 1994 Aug 16;33(32):9511-22. PubMed.

    . Folding of the Tau Protein on Microtubules. Angew Chem Int Ed Engl. 2015 Aug 24;54(35):10347-51. Epub 2015 Jun 19 PubMed.

    . Straight and paired helical filaments in Alzheimer disease have a common structural unit. Proc Natl Acad Sci U S A. 1991 Mar 15;88(6):2288-92. PubMed.

    . Paired helical filaments are twisted ribbons composed of two parallel and aligned components: image reconstruction and modeling of filament structure using atomic force microscopy. J Neuropathol Exp Neurol. 1997 Jan;56(1):79-85. PubMed.

    . Human Tau isoforms assemble into ribbon-like fibrils that display polymorphic structure and stability. J Biol Chem. 2010 Aug 27;285(35):27302-13. PubMed.

    . Alzheimer-like paired helical filaments and antiparallel dimers formed from microtubule-associated protein tau in vitro. J Cell Biol. 1992 Aug;118(3):573-84. PubMed.

    . Assembly of microtubule-associated protein tau into Alzheimer-like filaments induced by sulphated glycosaminoglycans. Nature. 1996 Oct 10;383(6600):550-3. PubMed.

    . The core of tau-paired helical filaments studied by scanning transmission electron microscopy and limited proteolysis. Biochemistry. 2006 May 23;45(20):6446-57. PubMed.

    . Oxidation of cysteine-322 in the repeat domain of microtubule-associated protein tau controls the in vitro assembly of paired helical filaments. Proc Natl Acad Sci U S A. 1995 Aug 29;92(18):8463-7. PubMed.

    . β-Sheet core of tau paired helical filaments revealed by solid-state NMR. J Am Chem Soc. 2012 Aug 29;134(34):13982-9. PubMed.

    . Mechanistic basis of phenothiazine-driven inhibition of tau aggregation. Angew Chem Int Ed Engl. 2013 Mar 18;52(12):3511-5. PubMed.

    . Oligomer formation of tau protein hyperphosphorylated in cells. J Biol Chem. 2014 Dec 5;289(49):34389-407. Epub 2014 Oct 22 PubMed.

    . Studies on the expression of the microtubule-associated protein, tau, during mouse brain development, with newly isolated complementary DNA probes. J Cell Biol. 1984 Mar;98(3):1090-7. PubMed.

    . Assembly of microtubule-associated protein tau into Alzheimer-like filaments induced by sulphated glycosaminoglycans. Nature. 1996 Oct 10;383(6600):550-3. PubMed.

    . Rapid assembly of Alzheimer-like paired helical filaments from microtubule-associated protein tau monitored by fluorescence in solution. Biochemistry. 1998 Jul 14;37(28):10223-30. PubMed.

    . Polycation-induced assembly of purified tubulin. Proc Natl Acad Sci U S A. 1976 Aug;73(8):2813-7. PubMed.

    . Interaction of tau with the RNA-Binding Protein TIA1 Regulates tau Pathophysiology and Toxicity. Cell Rep. 2016 May 17;15(7):1455-1466. Epub 2016 May 6 PubMed.

    View all comments by Eckhard Mandelkow
  3. This study represents a breakthrough in our understanding of the pathological conformation of the Alzheimer-related protein tau. In a unique manner it combines preparations of insoluble protein deposits from a patient with Alzheimer’s disease, which were first discovered more than 30 years ago, with technical advances in structural biology, in particular in single-particle cryo-EM. The study provides high-resolution insights into the conformation of amyloid fibrils of tau. In agreement with previous studies, it shows that the pseudo-repeat R3, which is the most hydrophobic region of the tau sequence, is part of the core of tau fibrils.

    Rather unexpectedly, however, the β-structure-containing region, which is well-defined in the electron density maps, includes the complete R4 region and even extends to F378 downstream of R4. In liquid-state NMR measurements of fibrils of 2N4R tau prepared in vitro (Bibow et al., 2011; Sillen et al., 2005), residues starting at ~400 become observable, suggesting that the structured region of tau amyloid fibrils might even extend beyond F378. In addition, residues in R2 and even R1 might contribute to the fibril core, although the definition of electron density in these regions was not sufficient to make that determination—potentially due to the presence of multiple isoforms, which either have or don’t have R2, in the patient-derived material.

    Another interesting question is how much variation in the local β-structure can occur in different preparations of tau fibrils purified from patients, given the observation that a significant amount of heterogeneity was observed in heparin-induced tau fibrils of K18/K19 by solid-state NMR spectroscopy (Andronesi et al., 2008; Daebel et al., 2012; Xiang et al., 2017). The observation of additional electron density, which cannot unambiguously be assigned to specific amino acids in tau, further stresses the need to integrate cryo-EM studies of patient-derived material with NMR studies of in vitro aggregated amyloid fibrils. This is, for example, highlighted by the observation of additional electron density, which might originate from the very N-terminus of tau consistent with an interaction of this region with the amyloid core suggested by liquid-state NMR studies (Bibow et al., 2011). In addition, in vitro-prepared amyloid fibrils of tau, will enable the study of the influence of site-directed mutations and post-translational modifications on the structure of aggregated states of the protein. The study forms the basis for characterization of the structure of insoluble deposits from other patients with AD, or those in different stages of AD potentially including presymptomatic AD, as well as from other forms of tauopathy.


    . The dynamic structure of filamentous tau. Angew Chem Int Ed Engl. 2011 Nov 25;50(48):11520-4. PubMed.

    . Regions of tau implicated in the paired helical fragment core as defined by NMR. Chembiochem. 2005 Oct;6(10):1849-56. PubMed.

    . Characterization of Alzheimer's-like paired helical filaments from the core domain of tau protein using solid-state NMR spectroscopy. J Am Chem Soc. 2008 May 7;130(18):5922-8. PubMed.

    . β-Sheet core of tau paired helical filaments revealed by solid-state NMR. J Am Chem Soc. 2012 Aug 29;134(34):13982-9. PubMed.

    . A Two-Component Adhesive: Tau Fibrils Arise from a Combination of a Well-Defined Motif and Conformationally Flexible Interactions. J Am Chem Soc. 2017 Feb 22;139(7):2639-2646. Epub 2017 Feb 8 PubMed.

    View all comments by Markus Zweckstetter

Make a Comment

To make a comment you must login or register.