Evidence continues to pile up that filaments of aggregated tau form unique strains in different tauopathies. Why is that? A paper published online in Cell on February 6 suggests that post-translational modifications help decide a filament’s ultimate shape. Investigators led by Anthony Fitzpatrick, Columbia University, New York, paired cryo-electron microscopy images of tau filaments from people with corticobasal degeneration (CBD) and Alzheimer’s disease with mass spectrometry to identify amino acid adducts. They report unique modifications in the protofibril of each structure.

  • CryoEM reveals the structures of CBD tau filaments in high resolution.
  • Mass spectrometry identifies post-translational modifications.
  • Those amino acid changes may govern the formation of unique tau strains.

“The powerful combination of cryoEM with mass spectrometry gives a more complete representation of the aggregated tau protein as it actually exists in the diseased brain,” said Lary Walker, Emory University, Atlanta.

“CryoEM alone yields important insights into the core structure of tau tangles, but precisely localizing post-translational modifications adds flesh to the bones,” he said.

Scientists led by Sjors Scheres and Michel Goedert at the MRC Laboratory of Molecular Biology in Cambridge, England, U.K., also use cryoEM to resolve the structure of CBD fibrils. On February 12 in Nature online, they reported a mysterious molecule hiding in a fold in the protofibril, much like they had found in tau protofibrils from a person with chronic traumatic encephalopathy. (This paper was previously uploaded to bioRxiv on October 21, 2019). Fitzpatrick’s group found a similar molecule. Scheres does not believe this is a post-translational modification, because it does not seem covalently attached. Together, the two papers offer the first high-resolution view of tau fibrils in CBD.

Tau Doublet. CBD tau protofibrils comprise two monomers joined back-to-back. Each C-shaped monomer comprises 107 amino acids (circles), that form 11 β-sheets (solid black lines). Non-tau moieties (pink) lie trapped within the fold or are covalently connected to the outside. [Courtesy of Arakhamia et al., 2020.]

In recent years, Scheres, Goedert, and colleagues have been methodically resolving the structures of tau filaments found in various tauopathies. They found that paired helical filaments and straight fibrils of tau from AD brain contained the same C-shaped protofibril structure (Jul 2017 news). Protofibrils in tau filaments from Pick’s disease and chronic traumatic encephalopathy assumed different J- and C-shaped structures (Aug 2018 news; Mar 2019). No one had yet examined filaments from CBD, which form from a tau isoform containing all four microtubule-binding domains. Tau can be alternatively spliced to have either three (3R) or four (4R) of these repeats. Fibrils in Pick’s disease incorporate only 3R tau, those in both AD and CTE have 3R and 4R forms.

Previous structural analyses of tau fibrils with cryoEM focused on its fibril-forming core. To isolate that, researchers used the enzyme pronase to remove the “fuzzy” outer coat of the fibril, revealing the more stable interior of the filament. However, pronase can strip away post-translational modifications as well, and tau accrues a whole host of them, some being disease-dependent (Jul 2015 news; Sep 2015 news). The group wondered if these alterations explain the unique structures of fibrils found in different tauopathies.

To find out, co-first authors Tamta Arakhamia, Christina Lee, and Yari Carlomagno used cryoEM to examine the undigested tau fibrils taken postmortem from the brain of a person with CBD. As has been reported previously, the sarkosyl-insoluble material was made up of both twisted and straight filaments (Ksiezak-Reding et al., 1996). The former was twice as wide and abundant as the latter.

The two fibrils were made of the same conformer of misfolded tau. While straight fibrils comprised just one column of monomers, each rung of the twisted fibril consisted of a linked pair (see image above). For both, the core-forming protofibril spanned amino acids 274 to 380. It included the last residue of R1, all of R2, R3, and R4, and 12 residues after R4. These regions formed 11 β-sheets—three from R2, three from R3, four from R4, and one formed by the last 13 amino acids. The sheets folded into four layers, forming a C-shaped loop (see image above).

Scheres and Goedert also analyzed undigested CBD tau fibrils using cryoEM. First author Wenjuan Zhang and colleagues found essentially the same β-sheet configuration and fold as did Fitzpatrick and colleagues. Zhang also found a molecule inside the fold of the protofibril. It was not covalently attached to any amino acid. Based on the positively charged amino acids that surround it, Zhang predicted this molecule to have a net negative charge of -3, and be 4 x 9 Ångstroms in size. Scheres had reported a similarly mysterious molecule inside the fold of CTE protofibrils, but that one was hydrophobic.

Arakhamia also found a large density inside the molecule, deep within a hydrophilic cavity formed by amino acids 281–296 and 358–374. It was not covalently bound, and so does not appear to be an amino acid modification. However, they found other large, non-tau densities adorning the outside of the fibrils. On the straight fibrils, these were attached to lysines 321, 343, 353, and 369, and to one histidine, H362. On the twisted form, they linked to K321, K353, and H362.

To identify these non-tau densities, Arakhamia and co-authors analyzed fibrils from several people with CBD by mass spectrometry, then mapped the identified PTMs onto the cryoEM structure (see image below). The authors found numerous phospho, trimethyl, acetyl, and ubiquitin additions. Some amino acids were either acetylated or ubiquitinated. Strikingly, while a few phospho groups attached to the superficial fuzzy outer coat, acetyl and ubiquitin groups predominated in the fibril-forming core.

PTM Map. Mass spectrometry identified modifications on amino acid sidechains of tau monomers from CBD (left) and AD (right). For the most part, acetylation (blue), ubiquitination (orange), and trimethylation (red) modified the fibril-forming cores, while phosphorylation (green) took place outside. [Courtesy of Arakhamia
et al., 2020.]

“I found that to be surprising,” said Li Gan, Weill Cornell Medicine, New York. “I would have assumed that the tau fibrils in the diseased brain would be hyperphosphorylated.”

Do these modifications affect folding of tau fibrils? That acetyl and ubiquitin groups bound to the core suggested to the authors that these were present as tau fibrils formed and played a hand in their aggregation. Acetylation may make tau protein less soluble, as it neutralizes positive charges on side chains and reduces their repulsion, predicted the authors. Ubiquitin may stabilize stacks of β-sheets by providing more surfaces for hydrogen bonding. Likewise, Zhang and colleagues think the mysterious hydrophilic molecule inside the fold might also be important in formation of the filament. That it is buried inside each monomer suggests that it is continuously incorporated during fibrillization and may stabilize the CBD fold during filament assembly, they wrote.

Could modifications of tau dictate which type of fold, and therefore which “strain,” accumulates in the brains of different diseases? Arakhamia and colleagues compared CBD tau PTMs with those on tau fibrils from AD. Again, they mapped mass spectrometry data from many fibrils onto the cryoEM structure. As in the CBD fibril, phosphoryl groups attached mainly beyond the protofibril core of AD tau, while acetyl and ubiquitin groups bound to the core. However, the amino acids modified were different in the different protofibrils and in the filaments they formed. In CBD, ubiquitinated K353 and acetylated K343 were found in twisted fibrils. The reverse, acetylated K353 and ubiquitinated K343, modified straight filaments. Similarly, acetyl groups bound K311 and K317/K321 in AD paired helical filaments, but ubiquitin occupies each of those sidechains in straight filaments. The results hint that PTMs influence the shape of aggregating tau fibrils.

“This finding implies that ubiquitin ligases and acetyltransferases modulate the behavior of tau, potentially tuning the ratio of fibril subtypes in tau inclusions,” Fitzpatrick wrote to Alzforum. “It will be informative to use our approach of combining cryoEM with PTM mapping by mass spectrometry to determine the additional structural role played by PTMs in tau oligomer formation and template-based seeding.”

“This paper illustrates, on a single-molecule level, that the interplay between acetylation and ubiquitination could play a role in tau fibrillization and strain properties,” Gan told Alzforum. If PTMs play such an important role in fibril formation, the recombinant seeds people have been using may not be as biologically relevant, she added. Goedert emphasized this at the Tau2020 meeting held in Washington, D.C., February 12–13. He noted that tau structures formed from recombinant protein using heparin are different from those isolated from brain tissue, particularly with respect to the fourth repeat and the 12 amino acids that come after it. “CryoEM findings cast a lot of doubt on work using recombinant tau structures,” he said.

Gan noted that the physiological consequences of the different tau strains—or whether they are even toxic—is unclear. “Before we develop strain-specific approaches, we need to understand what the strains do.” On that note, Marc Diamond, UT Southwestern Medical Center, Dallas, wondered whether PTMs were causal or incidental. He suggested that researchers find out by removing PTMs from fibrils before seeding. If that does not change the strain output, it would imply that they were not required for strain identity.—Gwyneth Dickey Zakaib


  1. These two papers from Fitzpatrick/Petrucelli and Scheres/Goedert represent a major advance toward a holistic structural understanding of the tau fibril conformers that drive distinct neurodegenerative tauopathies (i.e. "strains", which link specific amyloid conformations to defined biological or pathological phenotypes). Like previous work from the aforementioned authors, the technical achievements of the presented manuscripts are unparalleled and deserve universal attention and acclamation. Nevertheless, one is drawn to a disconnect between the structures presented in the two works. Intriguingly, both structures highlight the presence of tau-independent molecules in fibrils purified from corticobasal degeneration (CBD) brain. In the Scheres/Goedert work on three such samples, a non-covalently attached polyanionic cofactor is implicated, whereas the Fitzpatrick/Petrucelli study (n=2) suggests that post-translational ubiquitin moieties are present. While seemingly pedantic, the relative degree of the pursuant claims of the individual papers makes these discrepancies worthy of additional comment. 

    As two separate groups purified the CBD tau fibrils using slightly different protocols, discrepancies in presence (or absence) of cofactors is not particularly surprising in itself. The provocative claim is that of Fitzpatrick/Petrucelli, who suggest that identified ubiquitin moieties are instrumental in specifying conformer identity and thus strain characteristics. For this to be valid, a simple experiment would suffice: Do such fibrils maintain strain identity in the absence of ubiquitin? A variety of simple cell culture systems have been developed to test this hypothesis and are available to the experimenters. I note this not to be combative, but rather out of deep interest for learning whether non-tau cofactors contribute to strain identity. With strong claims come great interest, and such claims require great evidence.

  2. Arakhamia et al. hypothesize that pronase treatment removes PTMs from the ordered cores of tau filaments, which would lead to their absence in resulting cryo-EM structures. However, in Fitzpatrick et al., 2017, we compared cryo-EM structures of the ordered cores of AD PHFs and SFs prepared with or without pronase. Pronase removed the disordered "fuzzy coat" of the filaments, but did not change the cryo-EM structures of the ordered filament cores. Importantly, this included the additional densities on the outside of the filament cores, hypothesized by Arakhamia et al. to correspond to PTMs.

    Moreover, the cryo-EM structures of tau filaments from a case of PiD (Falcon et al., 2018) were prepared with pronase, whereas pronase was not used in the preparation of tau filaments from the three additional cases of AD in Falcon et al., 2018, from the three cases of CTE in Falcon et al., 2019, or from the three cases of CBD in Zhang et al., mentioned above. Yet, all of these structures show similar additional densities on the outside of the filament cores. These results indicate that pronase treatment does not alter the cryo-EM structures of tau filament cores. 


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

    . Structures of filaments from Pick's disease reveal a novel tau protein fold. Nature. 2018 Sep;561(7721):137-140. Epub 2018 Aug 29 PubMed.

    . Tau filaments from multiple cases of sporadic and inherited Alzheimer's disease adopt a common fold. Acta Neuropathol. 2018 Nov;136(5):699-708. Epub 2018 Oct 1 PubMed.

    . Novel tau filament fold in chronic traumatic encephalopathy encloses hydrophobic molecules. Nature. 2019 Apr;568(7752):420-423. Epub 2019 Mar 20 PubMed.

Make a Comment

To make a comment you must login or register.


News Citations

  1. Tau Filaments from the Alzheimer’s Brain Revealed at Atomic Resolution
  2. Conformers Confirmed: Structure of Pick’s Tau Distinct from AD Tau
  3. Traumatic Tau: Filaments from CTE Share Distinct Structure
  4. Inventory of Tau Modifications Hints at Undiscovered Functions
  5. New Type of Toxic Tau? Acetylated Form Correlates With Memory Defects

Paper Citations

  1. . Ultrastructural instability of paired helical filaments from corticobasal degeneration as examined by scanning transmission electron microscopy. Am J Pathol. 1996 Aug;149(2):639-51. PubMed.

External Citations

  1. bioRxiv

Further Reading


  1. . Elucidating Tau function and dysfunction in the era of cryo-EM. J Biol Chem. 2019 Jun 14;294(24):9316-9325. Epub 2019 May 14 PubMed.
  2. . The elusive tau molecular structures: can we translate the recent breakthroughs into new targets for intervention?. Acta Neuropathol Commun. 2019 Mar 1;7(1):31. PubMed.

Primary Papers

  1. . Posttranslational Modifications Mediate the Structural Diversity of Tauopathy Strains. Cell. 2020 Feb 20;180(4):633-644.e12. Epub 2020 Feb 6 PubMed.
  2. . Novel tau filament fold in corticobasal degeneration. Nature. 2020 Apr;580(7802):283-287. Epub 2020 Feb 12 PubMed.