Phospho-tau serine-262 and serine-356 as biomarkers of pre-tangle soluble tau assemblies in Alzheimer’s disease
Tau-FRET aggregation assay
We used the homogeneous time-resolved fluorescence energy transfer tau aggregation assay (originally from Cisbio, now Revvity) which contains a tau-specific antibody conjugated to either Tb cryptate or d2, generating a FRET signal when the labeled antibodies are in proximity. The resulting signal, which is proportional to the number and complexity of aggregates in the sample, was read at the 665 nm and 615 nm wavelengths on a VICTORX4 plate reader (PerkinElmer). A negative control consisting of the two labeled antibodies in diluent (without sample) was used to calculate the ΔF percentage, a value that reflects the signal to background of the assay.
Initially, a dilution linearity test was performed to identify the most suitable fold dilution to use for the brain samples. The TBS-soluble fraction from the AD and control brain samples was used in a test run, first brought to the same total protein concentration of 1.317 mg ml−1, before a dilution series. The samples were serially diluted 1:100, 1:50, 1:10 and 1:5 in TBS, incubated overnight with the labeled antibodies added according to the manufacturer’s protocol in 96-wells low-volume white microplates. The ΔF percentage values were calculated using the ratio between the wavelength and the negative control provided with the kit. The observed signals were multiplied by the fold dilution and compared with the expected signals to determine the linearity of dilution.
Expression and purification of recombinant α-synuclein and tau constructs
The DNA sequence for full-length α-synuclein and Tau441 (UniProt ID: P10636-8) and the tau peptides representing the STA (~tau258–368) and fibril (tau302–368) core peptides, as well as the N-terminal and C-terminal ends (tau1–224 and tau368–441, respectively) were amplified using PCR with primers representing the 5′ and 3′ sequence of each fragment, respectively, with the complementary DNA for full-length tau441 (cat. no. RC213312, Origene) as the template. The PCR fragments were cloned directly into the pET_SUMO vector (cohort 3: the Shiley-Marcos Alzheimer’s Disease Research Center (ADRC), University of California, San Diego (UCSD)), an expression vector with a 6× His-SUMO tag N-terminally fused to the protein or peptide of interest. Constructs containing the 6× His-SUMO-tau fusion protein were sequenced and transformed into the Escherichia coli BL21 (DE3) strain for protein expression.
To express the SUMO fusion proteins and peptides, E. coli BL21 (DE3) cells harboring the construct of interest were inoculated into 20 ml lysogeny broth (LB) medium supplemented with kanamycin at a concentration of 50 µg ml−1 and incubated overnight to obtain a starter culture. The overnight culture was used to inoculate 1 l of LB medium with kanamycin (50 µg ml−1) and incubated at 37 °C. When the optical density (OD)600 reached 0.5–0.7, protein expression was induced with 1.0 mM isopropyl ß-d-1-thiogalactopyranoside and grown overnight at 27 °C. The next morning, the culture was centrifuged at 7,000 rpm for 20 min at 4 °C and the dry weight was calculated. The pellet was stored at −20 °C until purification.
The pellet was gently thawed at room temperature and resuspended in 1× native buffer containing 50 mM sodium phosphate, pH 8.0, and 0.5 M sodium chloride (Invitrogen) added at a ratio of 8 ml buffer to 1 g of dry weight of pellet. Lysozyme solution (Thermo Fisher Scientific) was added and the lysate was incubated on ice for 30 min, followed by sonication and then centrifugation at 12,000 rpm for 20 min at 4 °C, after which the supernatant was collected. The protein extract was added to Ni-NTA agarose columns (Novex) equilibrated with 10 mM imidazole in 1× native buffer and incubated with gentle rotation at 4 °C for 1 h. The Ni-NTA agarose column was washed with 1× native buffer containing 20 mM imidazole, and the 6× His-SUMO-tau fusion protein of interest eluted with 250 mM imidazole in 1× native buffer. The eluted protein was dialyzed against 50 mM Tris-HCl, 150 mM NaCl, pH 7.8, for 1 h; the buffer was replenished with fresh supply and the process was repeated for another hour. Dithiothreitol (DTT) (1 mM) and a SUMO protease containing a 6× His tag was added and cleavage continued overnight at 4 °C. To remove the SUMO tag and the protease, an Ni-NTA column equilibrated with 20 mM imidazole in 1× native buffer was added and incubation proceeded for 1 h at 4 °C. The flowthrough was collected and the remaining bound protein was eluted separately after adding 20 mM imidazole in 1× native buffer. Eluate fractions were dialyzed against 1× PBS; aliquots of these samples were examined using gel electrophoresis on 4–12% NuPAGE SDS gel (Invitrogen) and stained with Imperial Protein Stain (Thermo Fisher Scientific). Fractions that showed high immunoreactivity for the constructs of interest were pooled and stored at −80 °C until use, in line with previously described protocols61,87. Where necessary, the protein or peptide constructs were further polished using size exclusion chromatography according to published methods61.
Preparation of recombinant tau aggregates
The monomeric forms of the recombinant tau peptides prepared according to the procedures described above and frozen at −80 °C in 1× PBS were used to generate the aggregates. Each tau protein variant was diluted to a final concentration of 46 µM in 1× PBS supplemented with 2 mM EDTA and incubated for 72 h on a shaking incubator (Thermomixer comfort, Eppendorf) at 350 rpm at 37 °C.
Human postmortem tissue and CSF studies (cohorts 1–3)
Human brain tissue and CSF specimens were obtained under permission and used in accordance with the Declaration of Helsinki 2013 and the relevant ethical boards at the respective institutions. The samples were from the following sources: cohort 1, the Queen Square Brain Bank for Neurological Disorders, Department of Clinical and Movement Neurosciences, Institute of Neurology, University College London, London, UK; cohort 2, the Netherlands Brain Bank, Amsterdam, the Netherlands; and cohort 3, the Shiley-Marcos ADRC, UCSD. Ethical approval for these studies was provided by the institutional review boards (IRBs) at the participating institutions, with written consent sought for and provided by the participants or their close family members if deemed to be incapable of making such decisions at that time in accordance with IRB requirements. The Queen Square Brain Bank for Neurological Disorders has generic ethical approval from a London multicenter research ethics committee under a license from the Human Tissue Authority. The Netherlands Brain Bank cohort was approved by the ethics committee of the Vrije Universiteit Medical Center, Amsterdam. The research protocol for the UCSD cohort was reviewed and approved by the human subject review board at UCSD, while informed consent was obtained from all patients or their caregivers as consistent with California State law.
For cohort 1, we used frontal gray matter tissue samples from n = 50 patients, including n = 10 each from AD, PSP, CBD, PiD and controls to enable pathological comparison across tauopathies. Neuropathological diagnosis followed established guidelines50,88.
For cohort 2, the samples were taken from the superior parietal gyrus. Participants with AD were at Braak stages V and VI while controls were at Braak 0, fulfilling the criteria of Braak staging1,2. Complete demographic information has been published previously29. Briefly, age at death (~64 years), sex distribution (25% males) and the postmortem interval (~6–7 h) were similar between the two groups.
Detailed methodological description of cohort 3 has been provided previously46,56. Briefly, the individuals included received clinical assessment for cognitive changes annually until death and signed to allow for neuropathological examination after their death. The evaluation results were carefully assessed at a consensus conference of experts to give a research diagnosis and determine the overall evaluation of cognition (normal, MCI; diagnosed after standard criteria, or dementia)89.
Biofluids, including the CSF, were periodically collected from the participants who consented. In this study, we measured CSF samples from individuals with both neuropathological examination and antemortem CSF samples within 5 years of death. We included individuals with sporadic disease, excluding those with a family history of autosomal dominant AD, dominantly inherited mutations (such as PSEN1, PSEN2 and APP mutations) or early-onset disease (under 50 years).
The autopsy procedures followed established protocols90. For pathological diagnosis of AD, neuritic plaques, diffuse plaques and NFT were identified either with 1% thioflavin S staining viewed with ultraviolet illumination and a 440 µm bandpass wavelength excitation filter, or with immunohistochemical staining using antibodies to Aβ (antibody 69D, rabbit polyclonal from E. Koo, 1:1,200 dilution) and PHF1 tau (from P. Davies, 1:600 dilution). Neuritic plaque density and NFT pathology were assessed according to CERAD91 and Braak staging2, respectively. For more recent cases, pathological diagnosis of AD was made using the National Institute on Aging (NIA)-Alzheimer’s Association (AA) consensus criteria50,92. The National Alzheimer’s Coordinating Center Neuropathology Working Group50 recommendations were followed to stage the severity of cerebral amyloid angiopathy, grading from 0 (absent) to 3 (severe).
Lewy body pathology was evaluated using hematoxylin and eosin staining in addition to immunostaining with antibodies against α-synuclein (p-synuclein 81A from V. Lee, 1:15,000 dilution). Disease staging was performed in accordance with consensus LBD guidelines93. TDP-43 pathology was identified using immunohistochemical staining (polyclonal, 1:12,000 dilution, cat. no. 10782-2-AP, Proteintech).
Homogenization and characterization of brain tissue isolates
The procedure used for cohorts 1 and 2 has been described previously29. Briefly, frozen brain tissue from each autopsy-verified case was dissected from the indicated region and 100 mg were dissolved in 0.5 ml TBS buffer (20 mM Tris-HCl, 137 mM NaCl, pH 7.6) containing cOmplete protease inhibitor cocktail (Roche Diagnostic). Tissue homogenization was performed on ice with TissueLyser II (QIAGEN) under the following conditions: 200-Hz frequency for 2 min. The homogenate was transferred to a new 0.5 ml TBS buffer and centrifuged at 27,000g for 20 min at 4 °C. The supernatant (referred to as the TBS-soluble fraction) was aliquoted and stored frozen at −80 °C. The total protein concentration in the various TBS extracts was determined using the DC Protein Assay (Bio-Rad Laboratories).
The brain tissue homogenization protocol used for the MS and immunoblotting experiments followed a protocol described in Islam et al.30. Both this method and the one used for cohorts 1 and 2 (which involved centrifugation of brain extracts at 135,000g and 27,000g, respectively) led to the separation of tau oligomers and tangle-free filaments (sedimentable at 235,000g) from monomers6.
IP and depletion of brain tau
The antibodies used in these experiments are described in Supplementary Table 5. For precipitation and depletion of tau from TBS-soluble fractions of AD brain isolates, the indicated anti-tau antibodies were conjugated to Dynabeads M-280 sheep anti-mouse or anti-rabbit IgG (Thermo Fisher Scientific), respectively, depending on the origin of the antibody, and according to the manufacturer’s recommended protocol. Briefly, 10 µg of total protein from the brain extract was incubated with the Dynabeads–antibody complex (that is, 4 µg antibody added to 50 µl beads in 1× PBS) and incubated overnight at 4 °C with gentle rocking to enable even mixing. The next morning, the Dynabeads–antibody complex was recovered by using a magnetized rack, the supernatant was reincubated in new 50 µl antibody–Dynabeads conjugate and the immunoprecipation or depletion process repeated for 2 h at room temperature. Afterwards, the Dynabeads–antibody complex (the precipitate fraction) was recovered and the remaining sample (the depleted fraction) was used in the tau-homogeneous time-resolved fluorescence energy transfer assay, where 10 µl of each sample was analyzed and untreated TBS-soluble brain extract from the same patient was used as control for the depleted samples.
Tau forms precipitated on the Dynabeads captured in the Dynabeads–antibody complex were eluted from the Dynabeads with 50 µl of 0.1 M citrate buffer, pH 2.75, into tubes containing 15 µl of 1 M Tris buffer, pH 9.0, for neutralization. The Dynabead-free tau samples were analyzed using immunoblotting, a capillary-based protein separation and immunodetection assay34, to detect the different tau fragments present according to the manufacturer’s recommendations.
IP–MS
The IP–MS experiments were performed at the MS facility in the Biofluid Biomarker Laboratory, Department of Psychiatry, University of Pittsburgh. Briefly, 30 µl of a pooled TBS-soluble fraction (7.1 mg ml−1) from the mid-temporal regions of postmortem brains was diluted to 1 ml with binding buffer (100 mM Tris-HCl, pH 7.4, 300 mM NaCl, 0.2% w/v n-dodecyl-ß-d-maltoside, 0.2% w/v n-Nonyl-β-d-thiomaltoside (cat. no. N373, Dojindo Laboratories)), supplemented with 10% v/v Neurology Panel 4-PLEX E CSF sample diluent (cat. no. 103727, Quanterix) to minimize nonspecific binding. Tau protein forms were immunoprecipitated using 50 µl of Dynabeads (M-270 Epoxy, cat. no. 14301; cohort 3: Shiley-Marcos ADRC, UCSD) conjugated with 1.25 μg of the specified antibodies (tau12, HT7, BT2, tau5, 77G7 and tau46), incubated overnight at 4 °C with rotation. The supernatant was then removed and the beads were washed twice with 0.5 ml PBS. After the removal of all residual liquid, proteins bound to the beads were eluted twice with 100 µl glycine buffer (50 mM glycine, pH 2.8, 0.1% n-dodecyl-ß-d-maltoside). The combined eluates were then neutralized with 5.5 µl 2N NaOH. Two replicate IPs were performed for each tau antibody.
Proteins were digested using SP3-based trypsin digestion94, similarly as described in ref. 95. Briefly, 50 µl of nondepleted and depleted fractions and 160 µl of precipitated fractions were brought up to 200 µl with 100 mM Tris, pH 8.0, and 2% SDS. The samples were then reduced with 10 mM DTT at 56 °C for 10 min and alkylated with 20 mM iodoacetamide at room temperature in the dark for 1 h. Subsequently, 1 ml of 100% ethanol and 30 µl of 20 mg ml−1 Sera-Mag SpeedBeads Carboxylate-Modified Magnetic Particles (equal mix of hydrophobic and hydrophilic beads; cat. no. 65152105050250 and cat. no. 45152105050250, GE Healthcare) were added to each sample. The mixtures were incubated at room temperature with shaking at 1,400 rpm for 20 min, followed by three washes with 80% ethanol. After removing all liquid, 1 µg of trypsin (Sequencing Grade Modified Trypsin; cat. no. V5111, Promega Corporation) in 100 µl of 50 mM ammonium bicarbonate with 1 mM CaCl2 was used to digest the proteins bound to the magnetic particles. After digestion, the samples were desalted with C18 spin cartridges (cat. no. SMM SS18V, The Nest Group), dried using a SpeedVac and then reconstituted in 0.1% formic acid (20 µl for nondepleted and depleted fractions and 16 µl for the precipitate).
The reconstituted peptides were analyzed using reverse-phase LC–tandem MS (MS/MS) using a nanoflow LC (a Dionex UltiMate 3000 RSLCnano System) coupled to an Orbitrap Exploris 480 mass spectrometer (Thermo Fisher Scientific). The Xcalibur software (v.2.2 SP1.48, Thermo Fisher Scientific) was used to operate the LC–MS/MS system. For each analysis, 1 µl peptides were directly injected onto a 5-cm Aurora series electrospray ionization column with a 150 µm ID filled with 1.6 μm reversed-phase C-18 packing material (120-Å pore size) (IonOpticks). Peptides were eluted using a linear gradient of 3–34% mobile phase B (0.1% formic acid in acetonitrile) in 5.5 min, then to 90% B for an additional 1 min, all at a constant flow rate of 1 µl min−1. Data acquisition parameters included a full MS scan from 350 to 1,600 m/z at a 30,000 resolution and an automatic gain control (AGC) target of 300%, followed by four data-dependent MS/MS scans at a 15,000 resolution and a standard AGC target, and a retention-time-scheduled PRM analysis of 18 tau peptides. The PRM parameters included an Orbitrap resolution of 15,000, a standard AGC target, an automatic injection time, an isolation window of 2 m/z and a higher-energy C-trap dissociation-normalized collision energy of 30. Supplementary Table 1 shows the targeted inclusion list with the retention-time-scheduled PRM scans. Each peptide sample was analyzed twice using LC–MS/MS.
Skyline (v.21.2.0.568) was used to facilitate the extraction of peptide quantification data from the PRM scans. To ensure accuracy, the chromatogram peak selection for each PRM assay was based on the presence of at least ten coeluting fragment ions. The final quantification of each peptide was based on the total area of the top three high-quality fragment ions (Supplementary Table 1). Replicate injections were averaged before further data analysis. The normalized intensity area ratio of peptides from the precipitate to the nondepleted fractions was used to compare the relative enrichment efficiency of tau peptides from different regions. This was achieved by dividing each ratio by the sample’s mean ratio.
MS characterization of recombinant tau441 phosphorylation
Recombinant tau441 proteins, phosphorylated by various kinases, underwent in-solution trypsin digestion as detailed below: 1 μg of each protein was brought up to a final volume of 90 μl using 50 mM ammonium bicarbonate. This was followed by a reduction with 10 mM DTT at 56 °C for 10 min and alkylation with 20 mM iodoacetamide at room temperature in the dark for 1 h. Then, 0.25 μg of trypsin was added to each sample and the mixture was incubated overnight at 37 °C. After digestion, peptides were desalted using C18 spin cartridges, dried using SpeedVac, and reconstituted in 36 μl of 0.1% formic acid. MS analysis proceeded in a manner similar to the IP–MS experiment, with the exception that different tau peptides, as listed in Supplementary Table 2, were targeted for the PRM analysis. The Skyline software was used to extract quantitative data, similar to the IP–MS experiments.
Biochemical characterization of the p-tau262 and p-tau356 antibodies
Sandwich ELISAs were used to validate the specificity of the p-tau262 and p-tau356 antibodies. For each measurement, 80 μl of the antibodies at 2 μg ml−1 in PBS, pH 7.2, was added to the well and incubated overnight at 4 °C. The well was then blocked with 200 μl PBS/0.1% BSA (cat. no. 81-053-3, Merck Millipore) for 1 h at room temperature. After blocking, the well was washed twice with 300 μl PBS with 0.05% Tween 20 (PBST). Subsequently, 50 μl of recombinant tau441 at concentrations ranging from 0 to 400 ng ml−1 were added, followed by the addition of 50 μl of PBST with 2% milk and 50 μl biotinylated tau12 (specific to tau441 aa 6–18) at 1 μg ml−1 in PBST. Immunocomplex formation proceeded for 1 h with gentle shaking at 300 rpm at room temperature. After incubation, the wells were washed with 300 μl PBST five times. Pierce High Sensitivity Streptavidin-HRP (cat. no. 21130, Thermo Fisher Scientific) was added and incubated for 1 h at room temperature, followed by five washes with 300 μl PBST. The 3,3′,5,5′-tetramethylbenzidine substrate (cat. no. 34022, Thermo Fisher Scientific) was added and allowed to incubate for 30 min before stopping the reaction with 100 μl of Stop solution (cat. no. N600, Thermo Fisher Scientific). The OD at 450 nm with subtraction of the background OD at 550 nm was used to determine color development. The indicated synthetic peptides at a 0.1 μg ml−1 concentration were added to each well for competitive ELISA during the immunocomplex formation step.
Immunoblotting
Samples (5 μg for untreated and depleted, or precipitate from 5 μg tissue lysates after the IP procedures described above, using either PBS or the IP–MS buffer with 10% Neurology Plex 4E CSF sample diluent as the IP buffer) complemented with 1× Laemmli sample buffer (cat. no. 161-0747, Bio-Rad Laboratories) were loaded on 4–12% NuPAGE Bis-Tris gradient gels (cat. no. NP0322B0X, Thermo Fisher Scientific) and separated for 3 h at room temperature in 1× NuPAGE MOPS Running Buffer (cat. no. NP0001, Thermo Fisher Scientific) at 100 V. Separated proteins were then transferred to nitrocellulose membranes (Invitrogen iBlot 2 Transfer Stacks, cat. no. IB23002, Thermo Fisher Scientific) using the iBlot2 Western Blot Transfer System (cat. no. IB21001, Thermo Fisher Scientific) at 20 V for 5 min at room temperature. Membranes were incubated for 1 h in Intercept (PBS) Blocking Buffer (cat. no. 927-70001, LI-COR Biosciences), then overnight in the presence of the anti-tau antibody tau12 (1:1,000 dilution, cat. no. 806501, BioLegend) in Intercept (PBS) Blocking Buffer with 0.2% Tween 20. Membranes were then incubated with Cy3 goat anti-mouse antibody (cat. no. 115-165-166, Jackson ImmunoResearch), diluted 700× in Intercept (PBS) Blocking Buffer with 0.2% Tween 20. Immunoblots were then dried and scanned using the ChemiDoc MP system (Bio-Rad Laboratories).
TEM
The recombinant tau aggregate preparations (5 µl) were pipetted onto copper grids that had been glow-discharged and carbon-coated, allowed 1 min to adhere onto the grid surface and then rinsed with ultrapure distilled water. Next, the grids were treated with 0.75% uranyl formate (Electron Microscopy Sciences) for 30 s to enable negative staining. TEM micrographs were taken on a Talos L120C 120 kV TEM microscope (Thermo Fisher Scientific) fitted with a BM-CETA camera-4.096 × 4.096, 14-µm pixel complementary metal–oxide–semiconductor. Microscopic imaging was performed at the Centre for Cellular Imaging at the University of Gothenburg.
For the AD brain samples, TBS-soluble homogenates were first immunoprecipitated (following the procedures described above) with the tau12 antibody (which has been shown to enrich for tau forms that stretch into the MTBR; Extended Data Fig. 1 and Supplementary Fig. 4) to enrich for STAs. The resulting precipitated fractions were negatively stained by treating with 1% uranyl acetate, allowed to dry and analyzed on a JEM-1400Flash TEM Microscope (JEOL) at ×25,000 direct magnification, at the Center for Biological Imaging, University of Pittsburgh.
SPR
The SPR experiments were performed using a Biacore T100 biosensor (GE Healthcare). Immobilization of the CT19.1 antibody (epitope: aa 331–361 of tau441) ligand on the surface of a CM5 chip was performed at a 5 µl min−1 flow rate to a level of 4,000 response units using standard amino coupling reagents (Cytiva). Thereafter, the analytes (that is, the truncated tau peptides) were injected at a flow rate of 20 µl min−1, with the experiments being performed in PBS at 25 °C. The BIAevaluation and Prism 9 (GraphPad Software) software programs were used for data processing and presentation, respectively.
Generation and characterization of CT antibodies
The new library of anti-tau mAbs was generated by immunizing 8-week-old BALB/c mice with 100 µg of recombinant tau241–441 peptide in complete Freund’s adjuvant (Sigma-Aldrich). After 2–3 further dosages of the immunogen (100 µg per mouse) in incomplete Freund’s adjuvant (Sigma-Aldrich), mice were euthanized, the spleen was removed and B cells were fused with the SP2/0 myeloma cell line according to standard protocols. Approximately 10 days after fusion, direct ELISA experiments were performed to screen the cell medium for antibodies that react with full-length recombinant tau441 (2N4R) or tau241–441. Positive clones were grown further, subcloned and subsequently frozen in liquid nitrogen. Antibody specificity was verified and the isotype determined using the Pierce Rapid Isotyping Kit-Mouse. Thereafter, mAbs were purified using a Hitrap Protein G column (Cytiva) according to the manufacturer’s instructions. Epitope mapping for each mAb was performed using direct ELISA against five custom-designed overlapping peptides spanning the tau241–441 sequence (Caslo ApS), specifically tau241–291, tau281–331, tau321–371, tau361–411 and tau401–441.
Generation and characterization of polyclonal antibody specific for truncation at aa 368
A new polyclonal antibody specific against tau truncated at aa 368 was generated by immunizing rabbits with 200 µg of a peptide containing the tau360–368 sequence (Caslo ApS) in complete Freund’s adjuvant. After one more dose of the immunogen (200 µg per mouse) in incomplete Freund’s adjuvant, and two more doses of the immunogen (100 µg per mouse) in incomplete Freund’s adjuvant, the rabbits were euthanized and standard antibody generation procedures followed. Antibody validation using MS is shown in Supplementary Information.
IHC and immunofluorescence studies
Cases and brain tissue samples
Studies were approved by the University of Pittsburgh Committee for Oversight of Research and Clinical Training Involving Decedents. Hippocampal tissue samples were obtained from autopsy cases at the University of Pittsburgh ADRC brain bank, including cases with NFT stage B1 or B2 and those with severe NFT stage B3 (ref. 96). At autopsy, samples of the hippocampus were dissected at the level of lateral geniculate nucleus and placed in cold (4 °C), 4% paraformaldehyde (cat. no. 158127-5006, Sigma-Aldrich) made in 0.01 M sodium phosphate buffer (pH 7.2) (sodium phosphate monobasic and dibasic, cat. nos. S374 and S1319, respectively, Thermo Fisher Scientific) for 48 h. After fixation, samples were sequentially immersed for 48 h in each of 15% and 30% sucrose (cat. no. S512, Thermo Fisher Scientific) solutions made in sodium phosphate buffer. Samples were sliced on a freezing, sliding microtome (model 860, American Optical Corporation) into 40-μm-thick sections that were stored at −20 °C in cryoprotectant solution97.
IHC and immunofluorescence
Tissue sections were removed from the cryoprotectant solution and rinsed three times in 0.1 M Trizma-buffered saline containing 0.25% Triton X-100 (Trizma, pH 7.4, cat. no. T7693, Sigma-Aldrich; Triton X-100, cat. no. T9284, Sigma-Aldrich). Chromogen-based IHC was performed as described previously98 using the VECTASTAIN Elite ABC-HRP Kit (cat. no. PK-6100, Vector Laboratories) with Ni-enhanced 3,3′-diaminobenzidine tetrahydrochloride (cat. no. D8001, Sigma-Aldrich) as the chromogen. The details of the primary antibodies, including primary antibodies targeting an epitope at p-tau202/205 (clone AT8, 1:3,000 dilution for chromogen-based IHC and 1:500 dilution for the multi-immunofluorescence experiments, cat. no. MN1020, Thermo Fisher Scientific), an epitope at p-tau231 (clone AT180, 1:1,000 dilution for chromogen-based IHC, cat. no. MN1040, Thermo Fisher Scientific), an epitope at p-tau262 (1:500 dilution for chromogen-based IHC and 1:200 dilution for the multi-immunofluorescence experiments, cat. no. 44-750G, Thermo Fisher Scientific) and an epitope at p-tau356 (1:750 dilution for chromogen-based IHC and 1:250 dilution for the multi-immunofluorescence experiments, cat no. 44-751G, Thermo Fisher Scientific) are given in Supplementary Table 6. Secondary antibodies (all used at 1:250 dilution) included a biotinylated goat anti-mouse IgG (cat. no. 115-065-146, Jackson ImmunoResearch), a biotinylated goat anti-rabbit IgG (cat. no. 111-065-045, Jackson ImmunoResearch), an Alexa Fluor 594-conjugated goat anti-mouse IgG (cat. no. 115-585-146, Jackson ImmunoResearch) and an Alexa Fluor 488-conjugated goat anti-rabbit IgG (cat. no. 111-585-144, Jackson ImmunoResearch). X-34 staining was performed as described previously99. Light microscopy analysis was performed using an Olympus BX53 microscope. The immunofluorescence analysis was performed as described previously100, using the Olympus BX53 microscope connected to a fluorescence illuminator (X-Cite 120Q). The microscope was equipped with an Olympus DP72 digital camera connected to a Dell Precision T5500 Desktop Workstation running the Olympus cellSens Standard v.1.12 imaging software, and with a U PLAN S-APO ×4 objective (numerical aperture (NA) 0.16), a U PLAN S-APO ×10 objective (NA 0.4) and a U PLAN S-APO ×20 objective (NA 0.75). The fluorescence of the Alexa Fluor 488 fluorophore was visualized using a fluorescein isothiocyanate-compatible filter (excitation peak = 480 nm, beam splitter = 505 nm, emission peak = 535 nm; cat. no. 41001, Chroma). The fluorescence of the Alexa594 fluorophore was visualized using a Texas red isothiocyanate-compatible filter (excitation peak 535 nm, beam splitter 565 nm, emission peak 610 nm; #41002, Chroma). The fluorescence of X-34 was visualized using a violet filter set (excitation peak = 405 nm, dichroic mirror DM440, emission peak = 455 nm; cat. no. 11005, Chroma).
Electrophysiology experiments
Preparation of mouse brain slices
All animal care and experimental procedures were reviewed and approved by the institutional animal welfare and ethical review body at the University of Warwick. Animals were kept in standard housing with littermates, provided with food and water ad libitum and maintained on a 12:12 (light–dark) cycle. Male and female 3–4-week-old C57BL/6 mice were euthanized using cervical dislocation and decapitated in accordance with the UK Animals (Scientific Procedures) Act 1986. The brain was rapidly removed and acute parasagittal or horizontal brain slices (350–400 μM) were cut with a Microm HM 650V microslicer in cold (2–4 °C) high Mg2+, low Ca2+ aCSF, consisting of the following: 127 mM NaCl, 1.9 mM KCl, 8 mM MgCl2, 0.5 mM CaCl2, 1.2 mM KH2PO4, 26 mM NaHCO3 and 10 mM d-glucose (pH 7.4 when bubbled with 95% O2 and 5% CO2, 300 mOsm). Slices were stored at 34 °C in standard aCSF (1 mM Mg2+ and 2 mM Ca2+) for 1–8 h.
Incubation of acute brain slices with recombinant tau truncations
After at least 1 h of recovery, slices were either incubated in aCSF (control) or in 444 nM recombinant tau (1–224; N terminus, 258–368; STA core, 302–368; fibril core or 368–441; C terminus) in aCSF for 1 h, in bespoke incubation chambers at room temperature. The incubation chambers consisted of small, raised grids (to allow perfusion of slices from above and below) placed in the wells of a 24-well plate (Falcon), which were bubbled with 95% O2, 5% CO2 using microloaders (Eppendorf). Slices were placed into the incubation chambers one at a time (minimum volume 1.5 ml to cover the raised grid). Individual slices were then placed on the recording rig and perfused with regular aCSF throughout the recording period, so the recombinant tau was only present for the 1-h incubation, as shown previously101.
Whole-cell patch clamp recording from single hippocampal CA1 pyramidal neurons
A slice was transferred to the recording chamber, submerged and perfused (2–3 ml min−1) with aCSF at 30 °C. Slices were visualized using infrared IR differential interference contrast optics with an Olympus BX151W microscope (Scientifica) and a charge-coupled device camera (Hitachi). Whole-cell current clamp recordings were made from pyramidal cells in area CA1 of the hippocampus using patch pipettes (5–10 mΩ) manufactured from thick-walled glass (Harvard Apparatus). Pyramidal cells were identified by their position in the slice, morphology (from fluorescence imaging) and characteristics of the standard step current–voltage relationship. Voltage recordings were made using an Axon Multiclamp 700B amplifier (Molecular Devices) and digitized at 20 kHz. Data acquisition and analysis were performed using pClamp 10 (Molecular Devices). Recordings from neurons that had an RMP of between −55 and −75 mV at whole-cell breakthrough were accepted for analysis. Bridge balance was monitored throughout the experiments; any recordings where it changed by more than 20% were discarded.
Stimulation protocols
To extract the electrophysiological properties of recorded neurons, both step and naturalistic, fluctuating currents were injected.
Standard IV protocol
A standard current–voltage relationship was constructed by injecting standard (step) currents from −200 pA, incrementing by either 50 or 100 pA (1-s duration) until a regular firing pattern was induced. A plot of step current against voltage response around the resting potential was used to measure the infrared (from the gradient of the fitted line).
Dynamic IV protocol
A dynamic-I-V curve, defined by the average transmembrane current as a function of voltage during naturalistic activity, can be used to efficiently parameterize neurons. The method has been described previously. Briefly, a current waveform, designed to provoke naturalistic fluctuating voltages, was constructed using the summed numerical output of two Ornstein–Uhlenbeck processes86,102,103 with time constants tfast = 3 ms and tslow = 10 ms. This current waveform, which mimics the stochastic actions of α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid and gamma-aminobutyric acid receptor channel activation, is injected into cells and the resulting voltage recorded (a fluctuating, naturalistic trace). The voltage trace was used to measure the frequency of action potential firing and to construct a dynamic-I-V curve. The FR was measured from voltage traces evoked by injecting a current waveform of the same gain for all recordings (to give an FR of ∼2–3 Hz). Action potentials were detected by a manually set threshold and the interval between action potentials was measured. All analyses were completed using either the MATLAB or Julia (v.1.7.3) software platforms104.
Extracellular recording of synaptic transmission
A 400-µM parasagittal slice was transferred to the submerged recording chamber and perfused with aCSF at 4–6 ml min−1 (32 °C). The slice was placed on a grid allowing perfusion above and below the tissue; all tubing (Tygon) was gas-tight to prevent loss of oxygen. To record field excitatory postsynaptic potentials (fEPSPs), an aCSF-filled microelectrode was placed on the surface of the stratum radiatum in CA1. A bipolar concentric stimulating electrode (FHC) controlled by an isolated pulse stimulator model 2100 (AM Systems) was used to evoke fEPSPs at the Schaffer collateral–commissural pathway. fEPSPs were evoked every 30 s (0.03 Hz). Stimulus input and output curves for the fEPSPs were generated using a stimulus strength of 2–80 mA for all slices (stimulus duration 200 µs). Signals were filtered at 3 kHz and digitized online (10 kHz) with a Micro CED (mark 2) interface controlled by the Spike software (v.6.1) (Cambridge Electronic Design). The fEPSP slope was measured from a 1-ms linear region following the fiber volley.
Development and analytical validation of the CSF STA assay
To capture STAs, a rabbit polyclonal antibody targeting an end-specific truncation at aa 368 was coupled to paramagnetic beads (cat. no. 103207, Quanterix), while the detection antibody CT23.1 (epitope: aa 321–371) was conjugated to biotin (cat. no. A3959, Thermo Fisher Scientific) according to the manufacturer’s recommendations. The resulting method was a three-step Simoa assay that combined the assay beads (that is, beads conjugated with the capture antibody) and the helper beads in a 70% to 30% ratio to give 20,000 beads per µl and 1 µg ml−1 of biotin-conjugated detection antibody with 100 µl of undiluted CSF. The average number of enzymes per bead signal for each sample was plotted against the concentration of the inputted biospecimen.
The specificity of the capture antibody to the truncation at tau368 was confirmed with MALDI MS. In this experiment, 8 μg of the polyclonal 368 antibody was added to 50 μl M-280 Dynabeads (sheep anti-rabbit IgG, Invitrogen) per sample according to the manufacturer’s product description. The 368-coated beads were used for IP of either the positive control or antigen (aa ITHVPGGGN equivalent to aa 359–368 with truncation at aa 368) or the negative control (aa GSLDNITHVPGGGNKKIETHKLTFRE 355–380 lacking truncation at aa 368) in PBS. Beads and samples were transferred to a KingFisher magnetic particle processor (polypropylene tubes, Thermo Fisher Scientific) for automatic washing and elution of full-length and truncated peptides. Eluted samples were collected and dried in a vacuum centrifuge and redissolved in 5 μl 0.1% formic acid in 20% acetonitrile and subsequently analyzed using a Bruker Daltonics UltrafleXtreme MALDI/ionization time-of-flight/time-of-flight mass spectrometer (Bruker Daltonics).
Clinical validation was performed using CSF samples from the CSF-to-autopsy and tau-PET studies (cohorts 3 and 4).
Tau-PET cohort
Study participants
Participants (n = 185) included in the study were selected from the Translational Biomarkers of Aging and Dementia (TRIAD) cohort, McGill University, Canada43,105. These participants had undergone Aβ and tau-PET imaging, the core CSF biomarker (Aβ42/40, p-tau181 and t-tau) analyses and had a CSF sample available for tau aggregate quantification.
In the TRIAD cohort, CN participants were defined as having an MMSE score greater than 24 and a CDR score of 0. This group included both young individuals (younger than 30 years) and older adults (older than 55 years). Participants with MCI had a CDR score of 0.5, with subjective and objective impairments in cognition, while their activities of daily living were preserved. Patients with AD dementia met the diagnostic criteria of the NIA and AA, and had a CDR score greater than or equal to 0.5 (ref. 106).
All participants provided written informed consent; the research protocol was approved by the Montreal Neurological Institute PET working committee and the Douglas Mental Health University Institute Research Ethics Board.
Brain imaging
[18F]AZD4694 and [18F]MK6240 PET were used to assess brain Aβ and tau pathologies, respectively. Imaging was acquired at two time points, that is, 40–70 min and 90–110 min after injection. PET scans were conducted using a Siemens High Resolution Research Tomograph (Siemens Medical Solutions).
To process the imaging data, PET scans from each participant were combined with their magnetic resonance imaging data. The cerebellar gray matter and the inferior cerebellar gray matter were used as reference regions for calculating the standard uptake value ratio (SUVR) for amyloid-β and tau-PET, respectively.
Aβ positivity was determined as a global [18F]AZD4694 SUVR equal to or greater than 1.55 (ref. 107). For tau-PET, a global index of tau pathology was obtained by calculating the average SUVR in the temporal meta-region of interest. Tau positivity was then defined as an SUVR equal to or greater than 1.24 (ref. 108). Moreover, participants were categorized into PET-based Braak stages based on the topography of tau-PET abnormality, as described in previous studies12.
Statistical methods, data analysis and software
All statistical tests were two-sided.
Tau-FRET studies
Data were presented as the mean ± s.e.m. Two groups were compared with the Mann–Whitney U-test, whereas a Kruskal–Wallis analysis of variance with Dunn’s multiple comparison test was used to examine three or more groups in Prism 9.
Electrophysiology studies
Prism 9 was used. Because of the small sample sizes (n < 15), statistical analysis was performed using nonparametric methods, that is, Kruskal–Wallis analysis of variance, Mann–Whitney U-test and Wilcoxon signed-rank test as required. All data are presented as the mean ± s.e.m. with individual experiments represented by single data points. For all experiments, significance was set at P ≤ 0.05. Data points for each experimental condition were derived from a minimum of four individual animals.
CSF-to-autopsy study (cohort 3)
Statistical analyses were performed using R v.4.3.1. The nonparametric Kruskal–Wallis rank-sum test was used for comparisons between group categories in demographic tables and figures alike, with significant results followed by post hoc pairwise Mann–Whitney U-tests with Benjamini–Hochberg FDR adjustment for multiple comparisons. Categorical variables were compared using a chi-squared test, with significant results followed by pairwise chi-squared tests with FDR adjustment for multiple comparisons.
Tau-PET cohort studies (cohort 4)
Python v.3.11.2 was used to perform nonimaging statistical analyses. For several data processing and statistical tasks, several additional packages were used. Pandas (v.1.5.3) was used as a powerful data analysis tool, providing data structures like DataFrames and Series, which allowed for efficient data handling and transformation. NumPy (v.1.24.2) was used for numerical computations, enabling the manipulation of multidimensional arrays and matrices. Scikit-learn (v.1.2.2) was used for regression, clustering and model evaluation. Statsmodels (v.0.13.5) was used for statistical modeling and hypothesis testing.
To assess the overall differences among the multiple diagnostic categories, a Kruskal–Wallis test was performed. If the Kruskal–Wallis test yielded a significant result, indicating that there were overall differences among the groups, post hoc pairwise comparisons were conducted using a Mann–Whitney U-test. To control the increased risk of type I error because of multiple pairwise comparisons, a Bonferroni correction was applied to adjust the significance threshold (alpha) for each comparison and maintain an overall alpha level for the entire set of tests.
Correlation analyses were performed to examine the relationships between a set of selected variables. Specifically, the Pearson correlation coefficient was calculated to measure the linear associations between continuous variables. To control the risk of type I errors arising from multiple comparisons, FDR correction was applied to adjust the P values. Only correlations with an FDR-corrected P < 0.05 were considered statistically significant and included in the final correlation matrix. Significant correlations were visualized using a heatmap, with color intensity representing the strength and direction of the associations between variables.
Neuroimaging analyses were carried out using the VoxelStats toolbox (https://github.com/sulantha2006/VoxelStats), a MATLAB‐based analytical framework that allows for the execution of multimodal voxel‐wise neuroimaging analyses109. Using this toolbox, a voxel-wise linear model was constructed to assess the relationship between the CSF STA and t-tau ratio and [18F]MK6240 PET SUVR while correcting for age, sex, the presence of APOE ε4 and [18F]AZD4694 PET SUVR, resulting in a t-map to display the strength of this relationship in a voxel-wise manner across different brain regions.
Reporting summary
Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.