14 Feb 2025

Detecting toxic forms of tau before they weave into dense thickets of tangles could pave the way for earlier diagnosis and treatment of tauopathies, including Alzheimer’s disease. In the February 10 Nature Medicine, researchers led by Thomas Karikari, University of Pittsburgh, unveil their characterization of early stage clumps of tau, aka soluble tau assemblies. STAs contain a core stretch of amino acids, featuring two phosphorylated serines, 262 and 356. Antibodies to p-tau262 and p-tau356 bound STAs in AD brains. Further, STAs in CSF correlated with tangles in the brain and with cognitive decline, and distinguished people with AD from those with other tauopathies.

Researchers define the structural regions of soluble tau assemblies.
STA cores could be biomarkers for pretangle tau.
They are found in cerebrospinal fluid.

Others were impressed by the work. “This study represents a significant step forward in understanding early tau pathology and advancing biomarker development for Alzheimer’s disease,” Seong Kang, Emory University School of Medicine, told Alzforum. “The identification of phosphorylation at serine-262 and serine-356 within the soluble tau assembly core provides new insights into prefibrillar tau species that precede neurofibrillary tangles.”

Khalid Iqbal, New York State Institute for Basic Research, Staten Island, called the findings highly significant. (Comments below.)

In related news, scientists also reported this week that soluble protofibrils of amyloid track well with total tau and with markers of neurodegeneration (Feb 2025 news).

Tau tangles don’t appear overnight. Smaller oligomers seed larger fibrillar tangles, fueling the spread of pathology throughout the brain. Indeed, soluble tau species may be more toxic to neurons than larger aggregates (Kopeikina et al., 2013). However, scientists don’t know the exact makeup of these early stage assemblies. First authors Tohidul Islam, Gothenburg University, Sweden, Emily Hill, Warwick University, UK, and Eric Abrahamson, University of Pittsburgh, along with their colleagues, set out to decode the structure of these elusive STAs.

Their key to cracking the code? A tau–fluorescence resonance energy transfer (FRET) assay developed by Revvity, Inc, a biotech company based in Waltham, Massachusetts. In this version, a tau antibody is attached to either a fluorescent donor or acceptor. FRET only occurs when both are close, and because tau monomers can only bind one antibody, they go undetected. Oligomers, however, bind multiple antibodies, sparking energy transfer and a fluorescence signal. The brightness reflects the number and size of STAs (image below).

**No Aggregate? Don’t FRET!** As tau monomers cluster to form STAs, donor and acceptor fluorophores come close in this assay, producing a signal. *[Courtesy of Islam et al., Nature Medicine.]*

Did this assay detect recombinant tau? Check. What about tau in AD brain samples? Ditto.

But was it specific for AD? To find out, the authors compared extracts from AD and control brains with those from non-AD tauopathies, including corticobasal degeneration, progressive supranuclear palsy, and Pick’s disease. The tau-FRET signal in AD extracts was 300 times higher than in controls and 15 to 40 times higher than in other tauopathies.

Next, Islam and colleagues used this FRET assay to probe the structure of the STAs. Immunoprecipitation pulled tau from brain extracts, leading to a drop in FRET signal. Testing nearly two dozen tau antibodies, they found that those targeting the N-terminus, some of the mid region, or C-terminus completely abolished the FRET signal, suggesting these segments are more accessible to antibodies and likely form the outer “fuzzy” coat of insoluble tau previously seen in cryo-EM (Oct 2021 news). In contrast, antibodies against the microtubule-binding region (amino acids 244–368) made only a dent in the FRET signal, indicating that this stretch harbors the STA core.

At either end of that core lie two serine residues, 262 and 356, that previously have been reported to be phosphorylated (Biernat et al., 1993; Jensen et al., 1999). Islam and colleagues wondered if these sites might serve as biomarkers. To find out, they stained hippocampal sections from AD brains with commercially available antibodies to these phospho-serine and other p-tau isoforms. In neurons, antibodies to p-tau262 and p-tau356 clustered within discrete foci, whereas antibodies to p-tau202/205—sites lying outside the STA core—blanketed much of the cell body and dendrites (image below). This pattern of staining echoes those previously reported for p-tau 262 in AD brains (Augustinack et al., 2002).

Additional labeling with the fibril-binding dye X-34 confirmed that neurofibrillary tangles avidly bound p-tau202/205 antibodies, but not p-tau262 or p-tau356 antibodies. The authors propose that Ser262 and Ser356 phosphorylation labels the earliest stages of insoluble tau build-up, before fibrils fully coalesce into tangles.

Tangled Beginnings. Antibodies to p-tau262 detected distinct puncta (green) inside neurons (left and right) while antibodies to p-tau202/205 detected more diffuse tau (middle and right). [Courtesy of Islam et al., Nature Medicine.]

Can these early forms of soluble tau be detected in the cerebrospinal fluid? To test this, Islam and colleagues devised a highly sensitive single-molecule (Simoa) assay using antibodies to the STA core—though not to p-tau262 or p-tau356—and tested CSF collected from a small neuropathology cohort of 67 elderly people, some of whom had been diagnosed with Alzheimer’s. CSF was collected roughly four years before death. Those who had more tangles in their brain had less CSF STA, suggesting that as soluble tau aggregates into fibrils, STA levels declined. They also analyzed a separate cohort of 185 living participants who underwent tau-PET scans. This group included adults who were either tau-negative or tau-positive on PET and who had been diagnosed with mild cognitive impairment or Alzheimer’s disease, and two groups of cognitively normal adults, one in their early 20s and one in their early 70s. Among people with mild cognitive impairment or AD, those who tested positive on PET had lower STA relative to total tau in the CSF than did the tau PET–negative volunteers. The ratio was also higher in cognitively normal adults.

Bernard Hanseeuw, Massachusetts General Hospital, Boston, called the work beautiful, but he wondered if the Simoa assay, which relies on a capture antibody directed against tau truncated at position 368, and a detection antibody recognizing the C-terminal of the microtubule-binding region, captures STAs. “I see no strong evidence that the assay measures tau assemblies,” noted Hanseeuw (comment below). “The assay quantifies all tau fragments 321-368, whether monomeric or oligomeric.”

Iqbal thinks this study helps justify tau immunotherapy. Some have questioned the wisdom of that approach because tangles are mostly intracellular, not parenchymal like amyloid plaques. “This study demonstrated the extracellular presence of full-length or almost full-length hyperphosphorylated tau in AD. This suggests that passive immunization with tau antibodies is a viable therapeutic target and that the CSF level of p-tau 262/356 or of the microtubule binding domain (MTBD) can serve as a useful biomarker of tau pathology to monitor tau therapeutics in clinical trials,” he wrote.

Recent work has also shown that levels of MTBR-tau243, a fragment of tau that accumulates in plasma and CSF, closely track with tangle burden in the brain and cognitive decline (Aug 2024 conference news; Dec 2020 news).

Karikari believes this new assay could stratify individuals enrolled in clinical trials by finding those in the early stages for tau pathology. “You see people who test positive for Aβ, and only some will go on to develop tau tangles,” he told Alzforum. “Our assay can help identify which patients may most benefit from targeted therapies.”

He thinks testing for STAs could also help determine when to treat. “Similar to amyloid-targeting, intervening before tau fully aggregates might offer greater clinical benefits than aiming at insoluble fibrils,” he said. “We might be able to save drug companies a lot of time!”—George Heaton.

George Heaton is a freelance writer in Durham, North Carolina.

Comments

- Kanta Horie
  Washington University School of Medicine
- Posted: 14 Feb 2025
This study is impactful because soluble tau assemblies (STA) containing phospho-tau serine-262 and phospho-tau-serine-356 were identified as components of pre-tangles, which are precursors of insoluble neurofibrillary tangles.

From previous Cryo-EM experiments, it is known that the tau filament core in AD brain is composed of amino acids 302 to 378. In this study, an integrative biochemical approach combining immunoprecipitation/immunodepletion using antibodies with various epitopes, and molecular profiling techniques, was used to identify aa 258-368 as one of the STA species. In vitro experiments were used to show that these have high aggregation competency and neurotoxicity. However, it is important to note that STA is not limited to aa 258-368, and there are various possible forms, including some containing adjacent residues.

This finding is also consistent with the previous report of Lövestam et al. on the effects of protein length on tau filament assembly (Lövestam et al., 2022).They produced recombinant tau of various lengths and reported that microtubule-binding region (MTBR) fragments with cleavage sites before the first repeat (R1) have low aggregation competency or do not form the AD filament structure. On the other hand, the aa 258-378 MTBR fragment, with a cleavage site in the middle of R1, does exhibit both strong aggregation competency and the AD filament structure.

Interestingly, Lövestam also reported that when the cleavage extends to the region spanning R3, such as aa 306 and aa 310, aggregation competency or AD filament structure is lost. This is consistent with Islam et al.’s finding that short MTBR species, e.g., aa 302-368, have relatively low aggregation competency and neurotoxicity.

Islam et al. also used pathologically confirmed, or tau-PET cohorts to test whether STAs can be used as CSF biomarkers to identify AD tau pathology in the brain, particularly at an early, pretangle stage. Their CSF STA sandwich assay used an anti-N368 neo-epitope antibody and an antibody to aa 321–371 as an antigenic epitope. In this method, the CSF STA/t-tau ratio tended to decrease with tau pathology progression, e.g., Braak stages, due to the sequestering effect of STA, which is consistent with the previous report with a similar method using an anti-N368 antibody (Blennow et al., 2020) and our other report on the R4 region peptide MTBR-tau354 (aa 354-369) in CSF using LC/MS (Horie et al., 2021; Horie et al., 2023).

However, there were large individual variations in this CSF STA biomarker, which include the tau filament core residues, and there were also large overlaps between groups. If highly sensitive methods are established for specifically measuring CSF p-tau262 and p-tau356, which are extensively discussed in this paper, then it may be possible to develop highly accurate biomarkers for identifying early stage AD tau pathology.

References:

Lövestam S, Koh FA, van Knippenberg B, Kotecha A, Murzin AG, Goedert M, Scheres SH. Assembly of recombinant tau into filaments identical to those of Alzheimer's disease and chronic traumatic encephalopathy. Elife. 2022 Mar 4;11 PubMed.

Blennow K, Chen C, Cicognola C, Wildsmith KR, Manser PT, Bohorquez SM, Zhang Z, Xie B, Peng J, Hansson O, Kvartsberg H, Portelius E, Zetterberg H, Lashley T, Brinkmalm G, Kerchner GA, Weimer RM, Ye K, Höglund K. Cerebrospinal fluid tau fragment correlates with tau PET: a candidate biomarker for tangle pathology. Brain. 2020 Feb 1;143(2):650-660. PubMed.

Horie K, Barthélemy NR, Sato C, Bateman RJ. CSF tau microtubule binding region identifies tau tangle and clinical stages of Alzheimer's disease. Brain. 2021 Mar 3;144(2):515-527. PubMed. Correction.

Horie K, Li Y, Barthélemy NR, Gordon B, Hassenstab J, Benzinger TL, Fagan AM, Morris JC, Karch CM, Xiong C, Allegri R, Mendez PC, Ikeuchi T, Kasuga K, Noble J, Farlow M, Chhatwal J, Day G, Schofield PR, Masters CL, Levin J, Jucker M, Lee JH, Roh JH, Sato C, Sachdev P, Koyama A, Reyderman L, Bateman RJ, McDade E, and the Dominantly Inherited Alzheimer Network. Change in Cerebrospinal Fluid Tau Microtubule Binding Region Detects Symptom Onset, Cognitive Decline, Tangles, and Atrophy in Dominantly Inherited Alzheimer's Disease. Ann Neurol. 2023 Jun;93(6):1158-1172. Epub 2023 Mar 16 PubMed.

View all comments by Kanta Horie

- Seong Kang
  Emory University
- Posted: 14 Feb 2025
This study represents a significant step forward in understanding early tau pathology and advancing biomarker development for AD. The identification of phosphorylation at serine-262 and serine-356 within the soluble tau assembly (STA) core (~tau258–368) provides new insight into prefibrillar tau species that precede neurofibrillary tangles (NFTs). By developing a CSF assay capable of detecting these STA-specific phospho-epitopes, the authors offer a novel approach to distinguishing AD from other tauopathies while also enabling the tracking of early disease progression.

Unlike p-tau181 and p-tau217, which are strongly linked to amyloid and the emergence of fibrillar tau in NFTs, the STA-specific biomarkers (p-tau262 and p-tau356) highlight an earlier prefibrillar tau state, potentially offering a distinct window into tau pathogenesis before NFT formation. This positions them as potential tools for detecting AD at a stage where therapeutic interventions may still prevent irreversible neurodegeneration.

Beyond early detection, this STA-specific CSF assay holds promise as a pharmacodynamic marker for drug development. Current therapeutic strategies, including anti-tau antibodies and aggregation inhibitors, could benefit from a biomarker that reflects soluble tau levels before NFTs become prominent. Since clinical trials targeting amyloid-b have shown the greatest benefits in individuals with lower NFT pathology, monitoring STA levels may provide a more dynamic measure of treatment efficacy. Furthermore, the STA assay complements tau-PET imaging, which primarily detects fibrillar tau in later Braak stages. By capturing early soluble aggregates, this assay could refine patient stratification in clinical trials, ensuring that those at risk of developing symptomatic AD are identified sooner.

Despite these promising advances, there are challenges to address. Longitudinal studies are necessary to confirm whether STA levels predict disease progression reliably, and efforts to translate this assay into a less-invasive blood-based biomarker would significantly enhance accessibility. Additionally, exploring the functional impact of STA core peptides on synaptic function and network hyperexcitability may provide further mechanistic insight into tau-driven neurotoxicity.

Ultimately, this work represents a shift in focus from targeting fibrillar tau to intercepting its soluble precursors, aligning biomarker strategies with the earliest molecular drivers of AD pathology and opening new avenues for precision medicine in neurodegenerative disease.

View all comments by Seong Kang

- Khalid Iqbal
  New York Institute for Basic Research in Developmental Disabilities
- Posted: 14 Feb 2025
Tau pathology in humans or animals, without fail, is made up of the hyperphosphorylated protein. In AD tau is hyperphosphorylated sub-stoichiometrically at multiple sites by several combinations of protein kinases. Six isoforms of tau in the human brain and hyperphosphorylation at multiple sites in AD generate numerous protein species.

It is for this reason that hyperphosphorylation of tau at any one particular site in CSF or plasma has been insufficient to serve as a diagnostic test of tau pathology. The hyperphosphorylation makes tau from microtubule assembly-promoting to -disrupting molecule and promotes its self-assembly into aggregates leading to neurofibrillary tangles (NFT).

We found that p-tau had the same characteristics in plasma as in the brain from AD patients and thus can serve as a useful screening test for AD. The VeraBIND Tau test, a novel plasma assay for active tau pathology, identifies individuals with positive F18MK6240 tau-PET signal regardless of amyloid status (Feb 2025 conference news).

Although previous studies showed that tau aggregates through the microtubule binding domain repeats (MTBR) and that the abnormal hyperphosphorylation at Ser 262 and Ser 356, the only two sites on the MTBR, promotes its aggregation, the presence of this pathological change in the CSF of AD patients was not previously reported. Thus, this study by Karikari et al on the presence of tau hyperphosphorylated at Ser-262 and Ser-356 both in the brain and the CSF in in Alzheimer’s disease (AD) is highly significant.

Contrary to a previous study which detected only the amino terminal half of tau in the extracellular space in the brain, this study demonstrated the extracellular presence of full length or almost full-length hyperphosphorylated tau in AD. This suggests that passive immunization with tau antibodies is a viable therapeutic target and that the CSF level of p-tau262/356 or of the microtubule binding domain (MTBD) can serve as a useful biomarker of tau pathology to monitor tau therapeutics in clinical trials.

Given that the authors of this study have access to several cohorts of well-characterized AD and control brains and biological fluids, hopefully they will also study and report plasma levels of p-tau 262/356 from these cases.

View all comments by Khalid Iqbal

- Bernard Hanseeuw
  Cliniques Universitaires Saint-Luc and Massachusetts General Hospital
- Posted: 14 Feb 2025
This beautiful work uses human brain samples, particularly the FRET assay showing that there are soluble tau oligomers, named “soluble tau assemblies.” It is puzzling to see that these STAs are AD specific!

For the CSF analyses, I am less convinced of the authors’ interpretation. They used SIMOA with a capture antibody directed against the truncated version of tau at position 368, and then a detection antibody recognizing the C-terminal portion of the microtubule-binding region (321-371). They also named the resulting analytes “STAs,” as in the FRET experiments, when in reality I see no strong evidence that the assay measures tau assemblies. The assay quantifies all tau fragments 321-368, whether monomeric or oligomeric.

Furthermore, the normalization by total tau is questionable. When we look at the non-normalized SIMOA data (Ext. Fig. 6) nothing is significant. In my opinion, this means that increase in CSF total tau is driving the association with tau-PET stages.

A remarkable finding, though, is that they are able to reliably quantify tau MTBR in CSF using Simoa, despite low concentrations and tau fragmentation. Future studies are needed to further characterize MTBR fragments in CSF, including truncated but also post-translationally modified forms. Developing fluid biomarkers accurately reflecting AD and non-AD brain changes in tau protein is critical for including the appropriate patients in clinical trials and monitor biological responses to drug exposure.

View all comments by Bernard Hanseeuw

- Stephen Salloway
  Brown University
- Posted: 14 Feb 2025
This contains encouraging findings about the identification of phosphotau species that may correlate with pretangles and be a marker of risk for progression of AD-related cognitive decline. If replicated and expanded, these findings could provide important insights for biomarker and drug development.

The key to AD treatment will require intervening early in the disease course to arrest or substantially slow disease progression. We will need diagnostic, prognostic and potentially theragnostic tools, like those proposed, to make that possible. The phosphotau biomarkers coming into clinical practice correlate most closely with amyloid plaque burden in the brain and to a lesser degree with aggregated forms of tau.

MBTR-243 is emerging as a good marker of tau PET accumulation but it is not clear how well MBTR-243 correlates with pretangle fragments. Ultimately, it will be most impactful to have plasma tests for early-tau biomarkers.

View all comments by Stephen Salloway

- Dave Morgan
  Michigan State University
- Posted: 14 Feb 2025
This work is a tour de force, with a series of well-planned studies examining soluble tau assemblies (STAs) using a FRET assay in which both components are the same antibody, although the exact antibody used is not clear from the manuscript.

The authors define a core domain in these assemblies that is somewhat longer than that for neurofibrillary tangles, but, like NFTs, also has variable degrees of the tau sequence N-terminal and C-terminal to the core domains. From this, they have developed a CSF biomarker assay for these early stage pretangle assemblies, using the ratio of the STA values divided by the total tau values, which declines significantly as the fibrillar tau pathology accumulates. They demonstrated this relationship with both pathological specimens and by tau PET, providing a new fluid biomarker measure that predicts NFTs.

Curiously, both the STAs and total tau increase in parallel with NFTs, but the total tau denominator increases more, leading the ratio to decline. The apparent advantage of the ratio is to reduce variance found in the STA measure alone. A final important observation is that p-tau 262 is a good marker histologically of pretangles, which should prove useful for further experimental model and neuropathological studies.

View all comments by Dave Morgan