Tau seeds occur before earliest Alzheimer's changes and are prevalent across neurodegenerative diseases

The study investigates when and where Alzheimer’s disease (AD)-specific 3R/4R tau seeds appear relative to classical histopathologic tau changes and how they relate to comorbid proteinopathies (notably α-synuclein) across neurodegenerative diseases. Prior work suggests tau seeds may precede overt pathology, but the extent, disease specificity, and correlation with staging and histological burden remain unclear. The authors use ultrasensitive and selective RT-QuIC assays to quantify AD-specific 3R/4R tau seeds, contrasting them with 4R tau seeds and α-syn seeds, to define the spatiotemporal relationship between tau seeding activity, Braak stage, amyloid pathology, and disease diagnosis. They also explore the prevalence of tau seeds in clinically and neuropathologically normal individuals, potential co-pathology effects with α-synuclein, and sex-related differences in tau seeding activity. The purpose is to determine the predictive value of tau RT-QuIC for AD neuropathologic change and to refine understanding of early tau seeding relative to classical histopathology and comorbid pathologies.

The introduction summarizes evidence that tau seeds are implicated in tauopathies and may be detectable before significant neuropathology. Biosensor cell assays have shown early tau seeding and progressive accumulation with Braak stage, but they may not discriminate between distinct seed conformers (3R/4R vs 4R) within mixed samples. RT-QuIC assays provide greater dynamic range and conformer selectivity via substrate choice (e.g., τ306-378, K12 for AD 3R/4R; K18/K19 for 4R/3R) and can distinguish disease-specific tau and α-syn conformers. Literature also indicates frequent tau pathology in synucleinopathies and potential synergistic interactions between α-synuclein and tau aggregates in models, though direct, sensitive measures of co-occurring seed-competent species in human tissues have been lacking. The authors build upon these studies to directly and selectively quantify AD 3R/4R tau seeds and evaluate their relationships with histology, amyloid pathology, co-pathologies, and clinical variables such as sex.

• Cases and neuropathologic assessment: 67 autopsy cases from UCSD plus additional younger controls and one AD case from Case Western Reserve University were classified by expert neuropathologists for Thal phase, Braak tau stage, CERAD neuritic plaques, AD neuropathologic change (ADNC), α-synuclein pathology, TDP-43, and ARTAG, using standard criteria.
• Digital histology: Middle frontal cortex (MFC) formalin-fixed, paraffin-embedded sections were immunostained with phospho-tau (AT8) and an AD-specific conformational tau antibody (GT38). Whole slide imaging and QuPath were used to calculate percent area occupied (%AO) by tau in grey and white matter using validated sampling and thresholding methods.
• Recombinant protein preparation: K23Q α-synuclein, K12CFh (AD-selective), τ306, K18CFh, and K19CFh tau substrates were expressed in E. coli and purified via Ni-NTA and ion-exchange chromatography with quality control across batches.
• Tissue processing: Frozen mid-frontal lobe tissue was homogenized (10% w/v) in PBS with protease inhibitors. For some analyses, homogenates were treated with proteinase K (with sarkosyl) to assess protease resistance; efficacy verified by SDS-PAGE. Sarkosyl-insoluble tau was prepared by high-salt/sucrose extraction, sarkosyl treatment, ultracentrifugation, and resuspension in PBS; concentrations estimated by SDS-PAGE densitometry.
• Mass spectrometry: PK-treated samples underwent reduction/alkylation, trypsin digestion, and LC-MS/MS on an Orbitrap Exploris 480 for peptide identification (Mascot, FDR < 2%).
• RT-QuIC assays: Endpoint dilution RT-QuIC measured seeding doses (SD50) for 3R/4R tau (K12 and AD RT-QuIC), 4R tau (K18+K19), and α-syn (K23Q). Reactions included appropriate recombinant substrates, buffers (HEPES or phosphate), salts (NaCl/NaF/citrate), heparin or poly-Glu where appropriate, ThT, and beads, run at assay-specific temperatures with shaking/rest cycles. Fluorescence thresholds for positivity were set based on negative-control baselines (mouse tau KO brain homogenate for tau; defined z-score exclusion and SD-based thresholds). Immunoprecipitation with AT8 assessed depletion of tau seed activity prior to K12 RT-QuIC. Mouse tau KO BH served as negative controls.
• Quantitation and statistics: SD50 values were determined via Spearman–Kärber analysis from endpoint dilutions. Correlations between seeding doses and digital histology (%AO), Thal phase, CERAD, and ADNC were analyzed (Pearson). Group comparisons used one-way ANOVA. Sex effects were analyzed using median quantile regression controlling for age (≤70 vs ≥70), Braak stage (0–III vs IV–VI), and the interaction of sex with Braak stage. Data availability upon request.

• 3R/4R tau seeding is widespread: Detectable in all diagnostic categories including controls (p < 0.0001 vs tau KO negative control). AD cases showed ~100–1000-fold higher 3R/4R tau seeding than synucleinopathies, 4R tauopathies, and controls. Controls exhibited up to ~10^6 seeding doses.
• Relation to Braak stage: 3R/4R tau seeding in the mid-frontal cortex is significantly related to Braak stage; tau seeds are present even at Braak 0–II (regions without histologically visible tau), with a significant increase at higher stages (median quantile regression p = 0.003). Largest quantitative increases observed after Braak III/IV.
• Correlation with histological tau burden: In AD, seeding doses correlated with digital pathology measures in grey matter for GT38 (r = 0.69, p = 0.03) and with AT8 in both grey (r = 0.71, p = 0.02) and white matter (r = 0.62, p = 0.04). LBD showed modest correlation in GM with AT8 (r = 0.55, p = 0.01) and weak/non-significant correlations otherwise; LBD %AO was much lower than AD for comparable seeding doses.
• Relation to amyloid and ADNC: 3R/4R tau seeding correlated with Thal Aβ phase (r = 0.63, p < 0.0001), CERAD neuritic plaque score (r = 0.68, p < 0.0001), and strongly with overall ADNC (r = 0.81, p < 0.0001).
• Seed properties across stages: Early (Braak 0–I) and late (Braak VI) 3R/4R seeds are largely protease resistant; PK treatment did not substantially reduce SD50 values. AT8 immunodepletion reduced seeding by ~0.78 log (~83%) in AD cases. Sarkosyl-insoluble tau was detectable even at Braak I (lower levels than AD), and sarkosyl-insoluble fractions showed seeding activities approaching those in total homogenate.
• Assay selectivity and 4R seeds: Seeding quantitation was consistent across two independent AD-selective RT-QuIC assays (K12 and AD RT-QuIC). 4R RT-QuIC detected high 4R tau seeding in PSP (10^5–10^7 SD50) and low/absent 4R seeding in most normals; some AD cases had low-level 4R seeding, confirming assay isoform selectivity and indicating normal-case seeding predominantly reflects 3R/4R tau.
• α-synuclein co-seeding: α-syn seeds were detected in 100% of LBD, 50% of MSA, 63% of AD, and 50% of PSP cases; none in CBD. Tau seeding doses were higher in neocortical and amygdala-predominant Lewy body stages than in brainstem-predominant stages.
• Sex differences: At Braak ≥ IV, females exhibited higher 3R/4R tau seeding doses than males; the interaction between sex and Braak stage was significant (p = 0.028). Female cases showed a median increase of ~2.875 logs (≥IV vs ≤III), males ~1.375 logs, with median seeding doses ~1.625 logs higher in ≥IV females compared to males.

The findings demonstrate that AD-specific 3R/4R tau seeds are prevalent well before histologically detectable tau pathology, including in Braak 0–II stages and even in neuropathologically normal and some younger individuals. Quantitatively, mid-frontal cortex 3R/4R seeding predicts overall Braak stage and ADNC, supporting RT-QuIC as a sensitive, selective biomarker of disease burden that surpasses reliance on lag-phase kinetics by using end-point dilution SD50 estimation. Early-stage seeds share protease resistance and sarkosyl insolubility characteristics with late-stage seeds, suggesting early pre-fibrillar or sub-fibrillar species with similar structural order to those forming NFTs. The strong correlation with amyloid measures and ADNC underscores the coupling of tau seeding with broader AD pathology. Widespread co-occurrence of α-syn and tau seeds across diseases aligns with mixed proteinopathies in neurodegeneration; higher tau seeding in neocortical/amygdala-predominant Lewy body stages suggests stage-related associations in synucleinopathies. Sex-related analyses indicate that females at higher Braak stages have greater tau seed loads than males, consistent with reports of higher tau PET burden and faster accumulation in women, potentially reflecting biological vulnerability. Together, these results refine the temporal and quantitative landscape of tau seeding, emphasize its predictive value for AD pathology, and highlight the need to consider co-pathologies and sex in disease characterization.

This study shows that 3R/4R tau seeds are widespread in the human brain prior to histologically visible tau pathology, are quantitatively associated with Braak stage and AD neuropathologic change, and frequently co-occur with α-syn and, in some cases, 4R tau seeds across neurodegenerative diseases. Early and late seeds share protease resistance and sarkosyl-insolubility, indicating qualitatively similar, structured species present before NFT formation. RT-QuIC provides selective, ultrasensitive, and quantitative assessment of tau seed burden, offering potential utility for disease staging and subtype definition. Females at higher Braak stages exhibit higher 3R/4R tau seeding than males, warranting further sex-stratified studies. Future research should include larger cohorts, longitudinal and region-specific mapping, comprehensive evaluation of tau post-translational modifications, examination of interactions with co-occurring protein seeds, and cross-method validation with equally selective, sensitive assays.

• Methodological cross-validation: Lack of confirmation with alternative seed-selective methods (e.g., biosensor cells do not differentiate 3R/4R vs 4R within mixed samples and often require immunocapture) limits cross-platform validation.
• PTM assessment: Limited evaluation of tau PTMs; while AT8 depletion reduced seeding ~1 log, comprehensive mapping of PTM impacts at early vs late stages was not performed.
• Quantitation and kinetics: Tenfold dilution-based SD50 quantitation and RT-QuIC’s broad dynamic range may miss subtler kinetic effects of co-occurring seeds on aggregation parameters.
• Spatial resolution: Homogenate-based analyses lack spatial information on seed distribution; contralateral hemisphere sampling may not perfectly match histopathologically assessed tissue, and some early pathology might be below detection of chosen antibodies.
• In vitro conditions: Assay conditions may not fully recapitulate in vivo filament structures and interactions.
• Sample composition: Relatively small and unevenly sex-distributed diagnostic groups (e.g., LBD vs AD) limit sex-specific inferences within disease categories; regional specificity of sex differences cannot be excluded.