A third-generation, high-affinity biparatopic anti-tau antibody inhibits intracellular tau aggregation seeded by Alzheimer’s brain extracts Authors: Eric Hatterer, Lucile Broyer, Sara Pecoraro, Fabien Cusin, Naz Simsek, John J Breve, Mathieu Clavel, Kevin L. Batenburg, Bart Klein, Thuur Zuidweg, Kristin Piché, Jeremie Bourguignon, Margot Bonnet-Gamard, Pauline Malinge, Franck Gueneau, Valery Moine, Nadia Anceriz, Krzysztof Masternak, Nicolas Fischer, Torleif Tollefsrud Gjølberg, Jenny Skjermo Telstad, Gaia Calamera, Simone Mester, Wiep Scheper, Wilma D.J. Van de Berg, Jean-Philippe Courade Published: 2026 ABSTRACT BackgroundTau immunotherapy has recently shown clinical promise but required high dosing. We developed NIDB-3101, a novel third-generation, high-affinity anti-tau biparatopic antibody designed for superior tau binding, aggregation inhibition, and extended half-life.MethodsNIDB-3101 binds tau’s microtubule-binding region and C-terminal domains. Various binding and cellular functional assays using recombinants, but more importantly human AD extracts were used to assess NIDB-3101 benefits. Half-life mutations impact was assessed via FcRn binding and cellular assays recycling.ResultsNIDB-3101 exhibited sub-nanomolar affinity, binding a broad spectrum of pathological tau species in AD homogenates, inhibited AD extracts-induced cellular effect compared to benchmark antibodies. Mutations enhanced hFcRn-mediated cellular recycling.ConclusionsNIDB-3101 captures a broad spectrum of pathological tau species leading to strong cellular efficacy using human AD extracts, supporting further clinical development as a potential disease-modifying therapy for AD and related tauopathies.Supplementary InformationThe online version contains supplementary material available at https://doi.org/10.1186/s13195-026-02155-9. Background Tauopathies, including Alzheimer’s disease (AD), are characterized by pathological deposition of tau in neurons leading to neuronal dysfunction and neurodegeneration [1, 2]. The tau protein undergoes pathological changes and transforms from a soluble microtubule-stabilizing protein into misfolded, hyperphosphorylated and aggregated forms of tau species in the diseased brain [1]. The accumulation and spread of tau throughout the brain correlates with cognitive decline and predicts disease progression in AD [3, 4]. As such, therapeutic strategies targeting and reducing pathological tau species hold a promise to slow down disease progression in AD and other tauopathies. Several tau-directed therapies, including clearance of pathological tau species by anti-tau monoclonal antibodies, are in various stages of clinical development. The first generation of tau antibodies, largely directed against N-terminal epitopes, failed to demonstrate efficacy in phase 2 trials for progressive supranuclear palsy (PSP) and AD [5–7]. This is likely due to suboptimal epitope selection, pharmacokinetic properties and insufficient affinity for pathological tau species leading to low target engagement [5, 8–10]. Second-generation antibodies targeting epitopes within or near the MTBR (Microtubule Binding Region) have entered clinical testing [5, 11]. Among these, etalanetug (formerly E2814) has recently demonstrated CSF target engagement in Phase 1 in healthy participants [12]. In a very small study in dominantly inherited Alzheimer’s disease patients, it did reduce the biomarkers ptau217 and MTBR-tau243 [13]. In a larger phase 2 trial, bepranemab formerly (UCB0107) recently showed encouraging data demonstrating significant effects on secondary endpoints such as tau PET and ADAS-Cog14, although there was no benefit on the primary endpoint (CDR-SB) [14, 15]. This was the first clinical proof of concept that a systemically administered tau antibody could slow tau pathology and the progression of cognitive decline. However, the requirement for very high doses limits clinical and commercial application, opening the path to third-generation tau antibodies. Recent structural, biochemical and immunohistochemical studies showed a wide range of pathological tau assemblies in human AD brain tissue underscoring the need for innovative approaches that target a broader range of pathological tau species [16–18]. Biparatopic antibodies (bpAb) - capable of engaging two distinct epitopes on the same antigens - have emerged as promising strategies in oncology and immunology but remain unexplored in AD [19]. Given the structural and biochemical heterogeneity of pathological tau species — including truncated, phosphorylated (or other post-translational modifications, PTM), and aggregated conformers — a biparatopic approach could allow binding to a broad range of tau species, thereby improving functional efficacy against diverse pathological assemblies [6, 16–18]. In an effort to improve efficacy of anti-tau antibodies, we aimed to generate a novel third-generation, high affinity, biparatopic anti-tau antibody, NIDB-3101. We incorporated Fc-engineered half-life extension (YTE mutation) for improved pharmacokinetics to reduce human efficacious dose to reach target occupancy and reduce injection patient burden. The combination of the two tau-binding paratopes was rationally chosen: one targeting the fourth seeding-prone MTBR (R4), a core structural component of tau filaments identified in cryo-electron microscopy studies of AD brain aggregates [16, 20, 21], and the other recognizing a distinct C-terminal domain epitope. C-terminal targeting antibodies have been shown to reduce the internalization of both monomeric tau and tau fibrils [22, 23]. This design enables simultaneous engagement of different tau domains, enhancing avidity and capturing a broad spectrum of pathological tau species. This work demonstrates the therapeutic potential of bpAb’s engineered to have an extended half-life through improved hFcRn (human neonatal Fc receptor)-mediated recycling in achieving detection of pathological tau, inhibition of seeded tau aggregation combined with improved antibody FcRn-mediated cellular recycling. Materials and methods Reagents and cell lines Reagents and Cells used in this study are listed in the online Supplementary Material & Method. Identification and affinity maturation of tau antibodies The isolation of anti-tau sequences was performed using proprietary phage display libraries with scFv from VH dummy naïve lambda and kappa libraries (libraries diversity, selection/ optimization processes, and outcomes of screening are depicted in Fig. S1) [24]. To select MTBR R4 specific candidates, biotinylated recombinant peptides of the fourth domain of MTBR tau (CAT#T08-55NB-50, Tau-K18, SignalChem) were utilized in solution. Individual clones were cultured in a 96-well plate, and scFv production was induced with 0.02 mM IPTG. Following centrifugation, scFvs were collected from the supernatant. Beads coated with various tau proteins, including tau 2N4R (CAT#T08-54B2N, SignalChem), tau 0N3R (CAT#SPR-490, StressMarq), tau 216–391 (CAT#T08-55LNB, SignalChem) and tau K18 (CAT#T08-55NB-50, SignalChem), were prepared in a 384-well CellInsight plate alongside an irrelevant protein. A detection mixture with human anti-c-myc antibody and anti-human Fc AF647 secondary antibody was added for specific clone selection based on tau protein binding, resulting in their reformatted IgG. For identifying Ct-specific antibodies, immunotubes were coated with recombinant tau protein ON3R (CAT#TP313364, ORIGEN-LabForce AG). Clones were amplified, and phage production was initiated through superinfection with M13K07 helper phage. Phage binding on tau protein ON3R (CAT#TP313364, ORIGEN-LabForce AG) as well as on irrelevant protein was assessed via HRP-conjugated anti-M13 antibody and Amplex-Red substrate. Affinity maturation focused on enhancing antibody binding affinities through oligonucleotide-directed mutagenesis of the CDRs in the variable light chain while maintaining the heavy chain. Electrocompetent bacteria or yeast cells were electroporated and grown respectively in 2 x TYAG2% (w/v) and SD-CAA medium to generate a titrated library. Benchmark hIgG1 analogs were generated from patents US2018/0201666Al, US2020/0369755A1 and WO2019077500A1 for bepranemab analog, BMS-986,446 analog and etalanetug analog, respectively. Binding characterization to human recombinant tau proteins The assays for evaluation of binding to monomeric recombinant human tau involved Nunc MaxiSorp 96-well plates coated overnight with goat anti-human Fc (in 1x PBS, pH 7.4). After three washing steps with PBS Buffer/Tween 20 (0.05% v/v) the plates are blocked with PBS Buffer/BSA (2% w/v)/Tween 20 (0.05% v/v) for 1 h followed by three washing steps. Anti-tau mAbs were added at 0.2 to 1 µg/mL (in PBS Buffer/BSA (2% w/v)/Tween 20 (0.05% v/v), incubation 1 h), followed by three washing steps. Biotinylated tau recombinant protein 2N4R (clone 37, in-house) is added (in PBS Buffer/BSA (2% w/v)/Tween 20 (0.05% v/v) for dose-response analysis (incubation 1 h). After washing steps the binding of tau mAbs to the tau 2N4R protein was detected using an HRP-antibody (in PBS Buffer/BSA (2% w/v)/Tween 20 (0.05% v/v), incubation 1 h), followed by five washing steps and Amplex-Red substrate (15–20 min incubation before reading). For evaluating binding to tau preformed fibrils, plates were coated with 0.5–1 µg/mL of tau 2N4R PFF P301S (CAT#SPR-329, StressMarq) or human recombinant alpha-synuclein PFF (CAT#SPR-322, StressMarq) to assess specificity (in 1x PBS). The washing and blocking procedures were identical to the prior method, followed by the addition of tau antibodies targeting either the fourth domain of the MTBR or the C-terminal domain of tau, diluted across a range of concentrations (6.3 × 10-5–133 nM). After additional washes, anti-human Fc HRP (1 h incubation) and Amplex-Red substrate were added to visualize antibody binding. Generation of a biparatopic anti-tau κλ body Lead scFv candidates were reformatted into human IgG1 format and further characterized. Anti-tau arms (anti-tau aa 344–349 and anti-tau aa 426–430) were selected for κλ body generation [24]. The two light chains and the common heavy chain sequences were cloned into a single tricistronic pNovi kHλ vector containing three expression cassettes under the transcriptional control of the human cytomegalovirus promoter. The vector also contains a glutamine synthetase gene (GS) that enables the selection and establishment of stable cell lines. First, second and third promoters drove respectively the expression of the lambda light chain (anti-tau aa 344–349), the common heavy chain, and the kappa light chain (anti-tau aa 426–430). Expi293 cells (RRID: CVCL_D615) were transiently transfected using polyethylenimine (Polysciences) at a final concentration of 5 µg/mL and a DNA concentration of 1.25 µg/mL. Cells were grown in Expi293 culture medium (Gibco) for 7 days at 37 °C, 8% CO2 and 125 rpm in an ISF1 Kuhner incubator. Six to seven days post-transfection, the supernatant was harvested, clarified by filtration after addition of diatomaceous earth (Sartorius) and purified applying a three-step purification process. First, the FcXP affinity matrix (Thermo Fisher Scientific), pre-washed with PBS, was added to the clarified supernatant. After incubation overnight at 4 °C and 15 rpm, sample tubes were centrifuged at 1,000 g for 10 min, supernatant was removed and the resin washed twice with PBS. Then, the resin was transferred to Amicon Pro devices (Millipore) and elution was performed by adding a solution containing 50mM glycine at pH 3.5, incubating for 5 min and centrifugating at 200 g for 5 min. Several elution fractions were generated, pooled and neutralized with Tris HCl pH7.4 (Invitrogen). Two additional affinity purification steps were applied, in the same conditions as described for the first step, using the CaptureSelect KappaXP and then the LambdaXP resins (Thermo Fisher Scientific). After the third purification step, elution fractions were pooled, dialyzed and concentrated on Amicon 50 kDa centrifugal filter units (Millipore). The κλ body was formulated in histidine 25 mM, NaCl 125 mM, pH 6.0. Sample purity and integrity were analyzed by electrophoresis on a Bioanalyzer 2100 with the Protein 80 kit (Agilent Technologies) in denaturing and reducing conditions and by isoelectric focusing (Cambrex pH 7–11 IsoGel agarose plates). Aggregate and fragment levels were determined by size exclusion chromatography – ultra-high-performance chromatography (SEC - UPLC) with an Acquity UPLC BEH SEC column, 200Å (Waters). Consent statement & ethics approval All donors signed informed written consent for brain donation and the use of material and clinical information for research purposes. The study was conducted in accordance to the Declaration of Helsinki and the procedures were approved by the local ethical committee of the Amsterdam UMC, Vrije Universiteit Amsterdam (METC number 2009/148). Human brain tissues Postmortem human brain tissue from pathologically confirmed AD cases (n = 10/group), were acquired from the Netherlands Brain Bank (NBB; Amsterdam, The Netherlands). Human postmortem brain tissue from age-matched non-demented controls (termed control cases; n = 10) were obtained from the Normal Aging Brain Collection Amsterdam (NABCA; Amsterdam, The Netherlands) [25]. Neuropathological diagnosis and pathological characterization was established according to international guidelines of Brain Net Europe II (BNE) consortium and NIA-AA criteria for AD [26]. We selected advanced AD cases with high burden of tau pathology in the MTG (middle temporal gyrus) and limited co-pathology (Braak NFT 5–6, Thal phase 1–3, Braak aSyn 0, pTDP43 negative, CAA negative) for this study (see for details of cohort Van der Gaag et al. [18]). Frozen MTG tissue of selected donors was processed for biochemical analysis or isolation of tau seeds. Formalin-fixed paraffin-embedded MTG tissue was used to evaluate pathological features detected with NIDB-3101 with immunohistochemistry. Brain tissue homogenization and dot-blotting To prepare brain tissue extracts, 500 µL of ice-cold detergent-free lysis buffer (5 mM HEPES, 320 mM sucrose, 1x complete mini EDTA-free protease inhibitor cocktail, Roche #11836170001; 1x PhosSTOP phosphatase inhibitor cocktail, Roche #PHOSS-RO, pH 8.0) was added to 50 mg of frozen MTGa tissue. After thawing the tissues on ice for 10 min, they were homogenized using a TissueLyser LT (CAT#85600, Qiagen) and centrifuged at 1000 x g for 2 min at 4 °C and homogenized again. To remove endogenous human immunoglobulins, 50 µL of settled immobilized Pierce™ Protein A/G (CAT#20421, ThermoScientific) was washed and added to the homogenates, which were incubated for 2 h at 4 °C. The supernatant was collected, and NuPAGE Sample Reducing Agent (1 mM DTT, CAT# NP0004, Invitrogen) was added to prevent oxidation. Antibody binding to human brain tissue extracts was evaluated using a DB (Dot Blot) assay. Samples were prepared with 1 µg of protein diluted in 100 µL of ice-cold 1x TBS. Nitrocellulose membranes (CAT #1620112, LICOR) were hydrated and used to spot samples. Total protein loading was verified using Revert™ 700 Total Protein Stain Kit (CAT#926-11016, Licor). Membranes were blocked, incubated with primary antibodies (overnight at 4 °C), and then secondary antibodies (1 h at room temperature; goat anti-rabbit IRDye 800CW, Licor CAT# 926-32211; donkey anti-mouse IRDye 800CW, CAT#926-32212, Licor) before imaging on a Licor Odyssey SA Imager. As a negative control, an additional blot was processed in parallel with the same brain extracts loaded, but only the secondary antibodies were applied. For semi-quantitative analysis of dot blots (DB), Image Studio Software (version 5.2) was used to obtain background-corrected signal values for each measurement region. These signal values were further corrected by subtracting the corresponding values obtained from the negative control blot (processed with secondary antibodies only). To determine the relative amount of protein of interest, the background- and control-corrected signal detected in the 800-channel was normalized to the total protein signal measured in the 700-channel using the Revert total protein stain [27]. Biochemical fractionation, gel electrophoresis and western blotting To prepare brain tissue extracts, 500 µL of ice-cold detergent-free lysis buffer (5 mM HEPES, 320 mM sucrose, 1x cOmplete mini EDTA-free protease inhibitor cocktail, CAT#11836170001, Roche; 1x PhosSTOP phosphatase inhibitor cocktail, CAT#PHOSS-RO, Roche) was added to 50–70 mg of frozen MTG tissue. After thawing the tissues on ice for 10 min, they were homogenized for 1 min at 50 Hz using a TissueLyser LT (CAT# 85600, Qiagen and centrifuged at 1000 x g for 2 min at 4 °C. Homogenization and centrifugation were repeated at the same settings as above. To remove endogenous human immunoglobulins, 50 µL of settled immobilized Pierce™ Protein A/G (CAT#20421, ThermoScientific) was washed and added to the homogenates, which were incubated for 2 h at 4 °C, with gentle shaking. The supernatant was collected, and NuPAGE Sample Reducing Agent (500 mM DTT, CAT#NP0004, Invitrogen) was added to a final concentration of 1 mM to prevent oxidation. This generated fraction is termed S1, which was used to assess antibody binding to crude human brain tissue homogenates using a DB assay. Remaining supernatant went through ultracentrifugation at 100’000 x g for 60 min at 4 °C. The resulting supernatant was collected without disturbing the pellet and termed the S2 soluble fraction, which was used to evaluate antibody binding characteristics using a WB (Western Blot) assay. The pellet was washed using lysis buffer and incubated for 10 min on ice, which was followed by ultracentrifugation at 100’000 x g for 30 min at 4 °C. SDS/Urea buffer (8 M Urea, 5% SDS, and TBS (pH 7.6) was then added to the pellets in ultracentrifugation tubes. The pellets were then dissolved by sonication (3/7 setting, 30% amplitude, 10 pulses). Hereafter, suspensions were transferred to new tubes and boiled for 10 min at 100 °C. This generated fraction is termed the S3 insoluble fraction, which like the S2 soluble fraction is used within the WB assays. BCA (CAT# 23225, ThermoScientific) was performed according to the manufacturer’s instructions to evaluate protein concentrations of S1, S2, and S3 fractions. Aliquots of all fractions mentioned above were made of which S1 and S2 fractions were flash frozen on dry ice and subsequently stored at − 80 °C, while aliquots of fraction S3 were stored at room temperature. Samples were thawed on ice and diluted to load 5 µg of protein per well with Bolt LDS 4x Sample Loading buffer (CAT# B0007, ThermoScientific) and Bolt Reducing agent 10x (CAT#B0009, ThermoScientific) as recommended by the supplier. Samples were heated at 70 °C for 10 min prior to loading, followed by brief centrifugation. NuPAGE™ MES SDS Running Buffer (20X; CAT# NP0002, ThermoScientific) was diluted 1:20 in MQ and loaded in the outer chamber of the XCell SureLock Mini-Cell Electrophoresis System (CAT#EI0001, ThermoScientific). Bolt™ Antioxidant (CAT# BT0005, ThermoScientific) was added at 0.25% to MES-SDS running buffer and loaded in the inner chamber. Proteins were separated on Bolt Bis-Tris 4–12% gels (CAT#NW04125BOX, Thermoscientific) and gels were run on 200 V for 35 min with MES-SDS as running buffer. Bolt antioxidant was added as per recommendation by the supplier. Precision Plus Protein Kaleidoscope Standard was used as the protein ladder. As a positive control for tau and αSyn a Tau Protein Ladder (CAT# T-1007-4, rPeptide) and 100 ng recombinant aSyn (CAT#S-1001-1, rPeptide) were used, respectively. The gels were equilibrated in 20% ethanol for 5–10 min, then proteins were transferred to a 0.2 μm nitrocellulose membrane using the P0 setting on the iBlot 2 Transfer Device (CAT#IB21001, Thermoscientific). The transfer used a stepped voltage profile: 20 V for 1 min, 23 V for 4 min, and 25 V for the remainder of the run. After the protein transfer membranes were air-dried for 30 min. Total protein loaded per sample was verified after rehydration of the membrane with TBS using Revert™ 700 Total Protein Stain Kit (CAT# 926-11016, Licor). Before imaging membranes were washed with a reverse wash solution (6.7% (v/v) glacial acetic acid, 30% (v/v) methanol, in MQ water). Membranes were imaged on a Licor Odyssey DLx Fluorescent Imager using the following settings: 169 μm resolution, automated intensity level and a 0.0 mm offset. Membranes were blocked with Intercept® (TBS) Blocking Buffer (LiCOr) for 60 min at 20 ˚C − 22 ˚C with gentle rocking. Which was followed by incubation with primary antibodies overnight at 4 ˚C. And subsequently washed with TBS + 0.1% Tween-20 followed by secondary antibody (goat anti-human IRDye 800CW, CAT# 926-32232, Licor) incubation for 60 min at 20 ˚C − 22 ˚C with gentle rocking before imaging on a Licor Odyssey DLx Fluorescent Imager as described above. Immunohistochemistry of human AD brain tissue Consecutive FFPE 6 μm tissue sections from MTG were deparaffinized for 3 × 10 min in 100% xylene substitute and rehydrated through a graded ethanol series of 100%, 96%, 80%, and 70%, with 5-minute incubations at each concentration, followed by rinsing in deionized water for 5 min. After deparaffinization sections were incubated for 5 min in 10 mM Tris-EDTA buffer (pH 9.0), followed by heat induced antigen retrieval with 10 mM Tris-EDTA buffer (pH 9.0) for 30 min. Sections were cooled down to room temperature for 40 min and washed for 5 min in 1x TBS (pH 7.4) and treated with 1% H2O2 in 1x TBS (pH 7.4) for 30 min at room temperature to block endogenous peroxidase activity, followed by three 5-minute washes with 1x TBS (pH 7.4). To minimize nonspecific binding, sections were blocked with 3% normal donkey serum (NDoS; CAT# 017-000-121, Jackson ImmunoResearch) for 30 min in 1x TBS (pH 7.4) with 0.1% Triton X-100. Primary antibodies (NIDB-3101, bepranemab analog) were then incubated at a concentration of 1–25 µg/mL, respectively, at 4 °C overnight in TBS with 0.1% Triton X-100 containing 3% normal donkey serum (NDoS; CAT# 017-000-121, Jackson ImmunoResearch). After three 5-minute washes with 1x TBS (pH 7.4), sections were incubated with a mouse anti-human HRP-conjugated IgG Fc (CH2 domain) secondary antibody (CAT# MA5-16859, Invitrogen) diluted 1:1000. To amplify and detect the signal sections were incubated with mouse EnVision HRP-conjugated (CAT#K4001, DAKO), followed by incubation with 0.05% (w/v) Diaminobenzidine-3,3-tetrahydrochloride (2.34 mM, CAT#D5637, Sigma) with 0.01% H2O2 diluted 1:2000 in Tris-HCL (pH 7.6). After three 5-minute washes with Tris-HCL (pH 7.6) nuclei were stained with hematoxylin for 60 s, followed by a 5-minute incubation in running tap water. Sections were then dehydrated through increasing ethanol concentrations of 70%, 80%, 96%, and 100%, and 100% xylene with 2-minute incubations at each concentration. After, sections were mounted with Entellan. Images of representative sections were captured with a Leica DM5000 microscope with 20x and 63x objectives (1.4 N.A.). Ordinal scores from 0 to 3 (1 = 0–5 structures; 2 = 5–10; 3 > 10 in the 20x objective field) were determined by two investigators (TZ, WvdB) for a variety of pathological morphology features (e.g. granular cytoplasmic staining, neurofibrillary tangles, ghost tangles, coiled bodies, threads, and dystrophic neurites). Isolation of Tau PHFs (Paired Helical Filaments) from postmortem AD brain tissue Tissue was thawed and the grey matter was manually separated from white matter and blood vessels, using a scalpel and tweezers. Tissue was kept on ice during the procedure [28]. Briefly, 5 ± 0.1 gram of pooled material was transferred to a BigPrep Lysing matrix D 50 mL tube (MPBio) kept on dry ice and homogenized (Fraction H1) in six volumes (w/v) High Salt Buffer [HSB; 10mM Tris-HCl (Bio Connect), 0.8M NaCl (L0080, Honeywell) and 10% Sucrose (Serva), supplemented with 2mM dithiothreitol (DTT; Fisher), 0.1% sarcosyl (Sigma), and complete protease and PhoStop phosphatase inhibitor cocktail (both Roche) using the Fastprep-24 5G homogenizer, using two cycles at 6 m/s for 30 seconds, with a 5 minute rest on ice between cycles. The resulting H1 fraction was centrifuged at 10.000 g for 10 minutes at 4°C using the Allegra X-30R centrifuge (Beckman Coulter), after which pellet (P1) and supernatant (S1) fractions were obtained. P1 was re-extracted by homogenization in four volumes (w/v) HSB after which pellet (P2) and supernatant (S2) were obtained. The supernatants of both extractions were transferred to 50 ml tubes and pooled (S1+2), sarkosyl was added to reach 1% final concentration and incubated overnight (o/n) at 4°C in a rotator. Subsequently, S1+2 was ultracentrifuged at 130-150.000 x g for 1 hour at 4°C in a 32 TI rotor (Optima XE90, both Beckman Coulter). After centrifugation the supernatant (S3) was removed, and the sarkosyl-insoluble pellet (P3) was washed once in 1 ml PBS and resuspended in PBS (∼ 200 uL/g tissue) by passing through 23G and subsequently 27G needles (0.5 mm, both BD). P3 was sonicated on ice at 190W 10s on/5s off for 1min10s, after 5 minutes on ice the sonication procedure was repeated (CAT#UP200st, Hielscher). Sonicated P3 was ultracentrifuged at 102.000 x g in Sorvall M120-SE equipped with a fixed-angle rotor for 30 minutes at 4°C. The pellet (P4) was washed once in 1 ml PBS and resuspended in PBS at one pre-centrifugation volume by passing through 23G and subsequently 27G needles. was sonicated on ice at 190W 10s on/5s off for 1min10s, after 5 minutes on ice the sonication procedure was repeated, and S5 was obtained after centrifugation at 10.000 g for 30 minutes at 4°C in an Allegra X-30R Centrifuge. S5 was stored as 200 uL aliquots into autoclaved-sterile 0.6 mL tubes (CAT#MCT-060-C, Corning) and stored at -80ºC for further use. Immunogold transmission electron microscopy Immunogold transmission electron microscopy (Immuno-TEM) was performed on AD brain-derived tau fibrils (see above), using anti-Human IgG (H+L) conjugated to 5 nm gold (CAT#SKU: AC-5-03-05, Cytodiagnostics,) as the secondary. First, 10 uL of sample at 1 mg/mL were applied to a formvar/carbon film nickel grid and then blocked with 1% BSA and 0.1% Tween-20 in PBS for a half-hour. This was followed by incubation with either a negative control antibody or primary anti-tau antibodies (NIDB-3101, bepranemab analog, BMS-986446 analog and etalanetug analog) diluted 1:50 in blocking buffer for 1 hour, washed and then incubated with the conjugated secondary antibody diluted 1:25 in blocking buffer a half-hour. The grids were washed again, then stained with 2% uranyl acetate and visualized with a TEM (CAT#L120C, Thermo Scientific Talos) in the Microscopy Imaging Facility, University Health Network (Toronto, ON, Canada). For AD brain-derived Tau, a minimum of 16 images were taken at varying magnifications (36000 x, 45000 x, 57000 x, 73000 x) depending on resolution. To determine number of AD brain-derived Tau fibrils, each fibril with an image was annotated in Microsoft Paint to avoid accidental duplicating. Averages, standard deviations, and ratios were calculated. Ratios calculated were number of dots divided by number of PFFs/field. This quantification was done manually and in a blinded manner. Cell-based seeding aggregation assay using recombinant or human brain AD seeds Tau RD P301S FRET Biosensor cells (CRL-3275, ATCC, from Washington University and M. Diamond) were detached the day before from the cell culture flask and were plated in a 96-well plate [29]. Before being added to the cells, tau seeds were sonicated to disrupt PFFs and optimize cell membrane diffusion, using a Fisherbrand™ Model 505 Sonic Dismembrator at 20kHZ in a room temperature (RT) water bath to avoid denaturation of the material by the heat generated. To evaluate the ability of our anti-tau Abs in preventing tau propagation and aggregation, recombinant tau441 (2N4R) P301S mutant PFFs were opsonized with anti-tau Abs for 10 min, at RT, before being added to the cells. Tau441 (2N4R) P301S mutant PFFs with heparin was used due to its higher productivity and seeding efficiency compared to unmutated PFFs. In case tau-enriched fraction from AD brain tissues were used as tau seeds, samples were opsonized with anti-tau Abs. Lipofectamine was added and after 20 min of incubation at RT, the preparation was added to the cells for 48 h of incubation at 37 °C. Cells were detached from wells and fixed with IOTest™ 3 Fixative Solution (Beckman Coulter) to be analyzed with CytoFlex™ flow cytometer. CFP and FRET signals were detected by exciting cells at 405 nm; emitted fluorescence was captured with a 450/45 nm filter, and a 525/40 nm filter, respectively. YFP was detected by exciting cells at 488 nm and fluorescence was captured with a 525/40 nm filter. CFP vs. FRET bivariate plots were used to identify FRET-negative and positive cells with results displayed as integrated FRET (percentage of FRET-positive cells multiplied by GeoMean FRET intensity) [29]. Data were expressed as normalized FRET values, with the signal obtained from the hIgG1 isotype control-treated sample, set as 100%. Statistics Raw data were processed in Microsoft Excel, and statistical analyses were performed using GraphPad Prism (version 9.3.1) or R studio (version 4.4.3). Age at onset, age at death, disease duration, and post-mortem delay (PMD) were compared using a one-way ANOVA (two-tailed) with Tukey’s multiple comparisons test. Data distribution was assessed through visual inspection of QQ plots and the Shapiro–Wilk normality test. Values presented as mean +/- SEM. P values are the following * = P < 0.05; ** = P < 0.01; *** = P < 0.001; **** = P < 0.0001. Results Identification and activity of anti-tau epitope-specific antibodies Using phage display technology, we identified anti-tau antibodies that target N-terminal, R4-MTBR and C-terminal regions of the tau protein and improved each of them by performing affinity maturation with human recombinant tau. Best binders (Fig. 1A) were then selected for further analysis. Binding assays with full-length recombinant tau (tau 2N4R) showed similar binding curves among the three selected anti-tau antibodies (Fig. 1B). Further analysis using tau-derived peptides revealed that the first selected antibody specifically bound a N-terminal peptide delineating amino acids 1-108 (Fig. 1C), with its epitope localized on amino acids 1–8 using overlapping peptide binding technique (Fig. 1A and Fig. S2A). The second selected antibody showed specific binding to a tau R4 peptide encompassing amino acids 337–369 (Fig. 1D), with epitope positioning confirmed at amino acids 344–349 using overlapping peptide binding technique (Fig. 1A and Fig. S2B), while the third uniquely bound a C-terminal peptide specifying amino acids 391–441 (Fig. 1E), further refined at amino acids 426–430 using overlapping peptide binding technique (Fig. 1A and Fig. S2C). To evaluate the functional activity of these antibodies, we tested the capacity of tau-antibodies to inhibit human recombinant preformed fibrils (PFF)-induced tau aggregation in tau biosensor cells [17–21, 24–30]. PFF treatment induced a FRET signal in 13% of the cells, determined by FACS analysis (Figs. S3A, B), indicating induction of intracellular tau aggregation. Preincubation with anti-tau aa 1–8 did not affect the integrated FRET signal compared to treatment with a hIgG1 non-tau binding isotype control (Fig. 1F and Fig. S3B). In contrast both anti-tau aa 344–349 and anti-tau aa 426–430 effectively reduced the FRET signal, showing their capacity to inhibit seed-induced tau aggregation (Fig. 1F and Fig. S3B). This indicates a preference to target R4 and C-terminal epitopes rather than N-terminal epitopes within the tau protein for functional activity. Fig. 1Characterization of anti-tau monospecific antibodies. Domain locations of three monospecific human IgG1 mAbs (anti-tau aa 1–8, anti-tau aa 344–349 and anti-tau aa 426–430) targeting either the N-terminal, the fourth domain of the MTBR or the C-terminal domains of tau with their respective epitopes determined by overlapping binding peptides (A). Concentration dependent intensity of binding of different tau mAbs to recombinant human tau 2N4R (B), shortened recombinant tau peptides delimiting the amino acids (aa) sequences of an N-term domain aa 1-108 (C), the fourth domain of the MTBR aa 337–369 (D) and C-terminal domain of tau aa 391–441 (E). To conduct these ELISA binding assays, recombinant proteins were coated and a range of antibody concentrations was applied to assess their binding efficacy to the specified recombinant protein. Each recombinant tau protein used is depicted according to their respective amino acids positioning using tau 2N4R numbering as reference. hIgG1 is a non-tau binding isotype control. Effect of anti-tau human IgG1 mAbs on tau seeded-aggregation cell-based assay using recombinant seeds (F). Data is expressed as normalized FRET values and set at 100% based on the signal obtained in the presence of a isotype control hIgG1 used as a negative control (dashed line). A minimum of 5 independent experiments were run in duplicate. Individual data points depict individual experiments. In each panel, anti-tau aa 1–8 is color coded in green, anti-tau aa 344–349 in blue and anti-tau aa 426–430 in brown. Values presented as mean +/- SEM. The one-way analysis of variance (ANOVA) with Tukey’s post hoc test is used for statistical significance calculation. P values are the following * = P < 0.05; ** = P < 0.01; *** = P < 0.001; **** = P < 0.0001 Characterization of anti-tau monospecific antibodies. Domain locations of three monospecific human IgG1 mAbs (anti-tau aa 1–8, anti-tau aa 344–349 and anti-tau aa 426–430) targeting either the N-terminal, the fourth domain of the MTBR or the C-terminal domains of tau with their respective epitopes determined by overlapping binding peptides (A). Concentration dependent intensity of binding of different tau mAbs to recombinant human tau 2N4R (B), shortened recombinant tau peptides delimiting the amino acids (aa) sequences of an N-term domain aa 1-108 (C), the fourth domain of the MTBR aa 337–369 (D) and C-terminal domain of tau aa 391–441 (E). To conduct these ELISA binding assays, recombinant proteins were coated and a range of antibody concentrations was applied to assess their binding efficacy to the specified recombinant protein. Each recombinant tau protein used is depicted according to their respective amino acids positioning using tau 2N4R numbering as reference. hIgG1 is a non-tau binding isotype control. Effect of anti-tau human IgG1 mAbs on tau seeded-aggregation cell-based assay using recombinant seeds (F). Data is expressed as normalized FRET values and set at 100% based on the signal obtained in the presence of a isotype control hIgG1 used as a negative control (dashed line). A minimum of 5 independent experiments were run in duplicate. Individual data points depict individual experiments. In each panel, anti-tau aa 1–8 is color coded in green, anti-tau aa 344–349 in blue and anti-tau aa 426–430 in brown. Values presented as mean +/- SEM. The one-way analysis of variance (ANOVA) with Tukey’s post hoc test is used for statistical significance calculation. P values are the following * = P < 0.05; ** = P < 0.01; *** = P < 0.001; **** = P < 0.0001 Generation of the biparatopic anti-tau antibody NIDB-3101 and epitope mapping We utilized the kappa/lambda body platform [24] to create the fully human bpAb NIDB-3101, featuring two distinct paratopes, using anti-tau aa 344–349 and anti-tau aa 426–430 parental antibodies described above. This process involved pairing a lambda (λ) light chain from the selected anti-tau arm targeting the fourth domain of the MTBR (anti-tau aa 344–349) with a kappa (k) light chain from the anti-tau arm that binds to the C-terminal domain (anti-tau aa 426–430) of tau (Fig. 2A). The biparatopic nature of NIDB-3101 was confirmed using overlapping peptide arrays, which showed strong binding responses to the epitopes 343KLDFKDR349 and 426ATLAD430, indicating that the bpAb recognizes both epitopes (Fig. 2B and Fig. S4A). To further clarify the key interaction residues within each epitope, we crystallized the individual anti-tau arms as Fab fragments with their respective tau peptides. High-resolution crystallographic analysis validated the epitope locations identified in the peptide array, further detailing the three-dimensional amino acid interactions within the binding sites of each tau epitope for both anti-tau aa 344–349 (KLDFKDRVQS) and anti-tau aa426-430 arms (PQLATLADEVSASL) (Figs. S4B-C, right panels, with highlighted amino acid residues). Fig. 2Characterization of a biparatopic anti-tau antibody, NIDB-3101. Schematic structure of the anti-tau biparatopic antibody pairing anti-tau arms recognizing the fourth domain of the MTBR (anti-tau aa 343–349) and the C-terminal domain of tau (anti-tau aa 426–430) into a fully human and native κλ body format (A). The biparatopic binding domains within tau 2N4R protein are visualized with epitope locations determined by overlapping binding peptides (B). Concentration dependent binding of NIDB-3101 to tau 2N4R preformed fibrils mutated at position P301S and produced in SF9-insect cells expression system (including post-translational modifications). Binding is compared to respective monospecific and bivalent antibodies (anti-tau aa 344–349 and anti-tau aa 426–430) (C). Effect of the anti-tau biparatopic κλ body, NIDB-3101, on tau seeded-aggregation cell-based assay using recombinant seeds from E. coli expression system. Activity compared to respective monospecific and bivalent antibodies (anti-tau aa 344–349 and anti-tau aa 426–430) (D). Data is expressed as normalized FRET values and set at 100% based on the signal obtained in the presence of a non-binder hIgG1 used as a negative control (dashed line). A minimum of 3 independent experiments were run in duplicate. Individual data points depict individual experiments. Values presented as mean +/- SEM. The one-way analysis of variance (ANOVA) with Tukey’s post hoc test is used for statistical significance calculation. P values are the following * = P < 0.05; ** = P < 0.01; *** = P < 0.001; **** = P < 0.0001 Characterization of a biparatopic anti-tau antibody, NIDB-3101. Schematic structure of the anti-tau biparatopic antibody pairing anti-tau arms recognizing the fourth domain of the MTBR (anti-tau aa 343–349) and the C-terminal domain of tau (anti-tau aa 426–430) into a fully human and native κλ body format (A). The biparatopic binding domains within tau 2N4R protein are visualized with epitope locations determined by overlapping binding peptides (B). Concentration dependent binding of NIDB-3101 to tau 2N4R preformed fibrils mutated at position P301S and produced in SF9-insect cells expression system (including post-translational modifications). Binding is compared to respective monospecific and bivalent antibodies (anti-tau aa 344–349 and anti-tau aa 426–430) (C). Effect of the anti-tau biparatopic κλ body, NIDB-3101, on tau seeded-aggregation cell-based assay using recombinant seeds from E. coli expression system. Activity compared to respective monospecific and bivalent antibodies (anti-tau aa 344–349 and anti-tau aa 426–430) (D). Data is expressed as normalized FRET values and set at 100% based on the signal obtained in the presence of a non-binder hIgG1 used as a negative control (dashed line). A minimum of 3 independent experiments were run in duplicate. Individual data points depict individual experiments. Values presented as mean +/- SEM. The one-way analysis of variance (ANOVA) with Tukey’s post hoc test is used for statistical significance calculation. P values are the following * = P < 0.05; ** = P < 0.01; *** = P < 0.001; **** = P < 0.0001 NIDB-3101 binds truncated tau and blocks tau seeded aggregation using recombinant seeds Engagement of two non-overlapping epitopes by NIDB-3101 may allow to recognize a broader range of tau proteoforms than monospecific antibodies. In AD, tau truncation at different sites (residues 391 and 421) leads to neurotoxic fragments [31–38]. ELISA (Enzyme-Linked Immunosorbent Assay) binding assays were conducted using recombinant tau 2N4R (Fig. S5A) and tau peptides covering residues 216–391 (Fig. S5B) and 421–441 (Fig. S5C). Using these assays, there was a slight shift in binding in favor of NIDB-3101 compared to anti-tau aa 344–349 and anti-tau aa 426–430 (Fig. S5A) and bound robustly to all truncated forms, while parental antibodies were selective (anti-tau aa 344–349 and anti-tau aa 426–430, Figs. S5B-D). Importantly, unlike anti-tau aa 426–430, NIDB-3101 recognized aggregated truncated tau species encompassing aa 297–391 domain, which closely resemble the pathological conformers observed via Cryo-EM in AD’s disease brain extracts (Fig. S5D) [39]. We next evaluated the impact of phosphorylation on binding to tau preformed fibrils (PFFs) from the 2N4R variant produced in insect cells-SF9 expression system (Fig. 2C). Mass spectrometry confirmed the presence of AD-relevant phosphorylation sites on tau PFFs, including pS202, pT205, pS396 (Fig. S6A) [40–42]. NIDB-3101 displayed superior maximal binding signal to tau PFFs with various post-translational modifications (PTMs) compared to monospecific mAb variants (anti-tau aa 344–349 and anti-tau aa 426–430) suggesting a binding advantage with a biparatopic engagement compared to a classical mAb (Fig. 2C). Data suggest that the anti-tau aa 426–430 (C-term) arm of NIDB-3101 is a crucial binding component, exhibiting superior binding compared to the anti-tau aa 344–349 (R4-MTBR) arm. However, the combination of both arms results in an overall stronger binding signal than either monospecific antibody used independently. In the seeded aggregation biosensor assay using recombinant seeds, NIDB-3101 significantly reduced the FRET signal relative to a non-tau binding isotype control antibody and outperformed each monospecific parental antibody (anti-tau aa 344–349 and anti-tau aa 426–430) (Fig. 2D). Moreover, NIDB-3101 recapitulated the efficacy of the combination of anti-tau aa 344–349 and anti-tau aa 426–430 (each combined at 166.5 nM, to reach 333 nM final concentration) in a single bpAb format (tested at 333 nM, final concentration) (Fig. 2D). NIDB-3101 exhibited comparable activity when utilizing either recombinant seeds derived from E. coli (Fig. S6B) or those expressed in SF9-insect cells (Fig. S6C), the latter displaying PTMs that are characteristics of tau proteins found in AD brain tissues [40–42]. These findings corroborated our binding data (Fig. 2C) and indicate that NIDB-3101 strongly binds to tau seeds and inhibits seeded tau aggregation, marking it as a promising therapeutic candidate with broad-spectrum activity compared to monospecific parental tau antibodies. NIDB-3101 binding efficiency to AD brain-derived tau species For comparative studies, we confirmed the binding by ELISA of NIDB-3101 and three second-generation anti-tau antibodies analogs (Fig. 3A): one binding adjacent to the MTBR (bepranemab analog – Fig. 3A and Figs. S7A-F) and two others specifically targeting the MTBR (BMS-986446 analog and etalanetug analog – Fig. 3A and Figs. S7A-F). SPR (Surface Plasmon Resonance) analysis revealed that NIDB-3101 has a robust apparent affinity for monomeric 2N4R tau (0.035 nM ± 0.004), significantly surpassing bepranemab analog (2.4 nM ± 0.004), BMS-986,446 analog (10.2 nM ± 0.7) and etalanetug analog (1.3 nM ± 0.2) (Fig. 3B). ELISA tests with recombinant tau PFFs confirmed that NIDB-3101 captures tau aggregates with higher Emax and at least 10 folds lower EC50 (Emax = 52168 fluorescence intensity, EC50 = 0.027 nM) than the three clinical anti-tau analogs (Emax = 46711 fluorescence intensity, EC50 = 0.26 nM, Emax = 38628 fluorescence intensity, EC50 = 15.7 nM, and Emax = 50507 fluorescence intensity, EC50 = 0.11 nM for bepranemab, BMS986446 and etalanetug analogs, respectively) (Fig. 3C). To determine the relevance of NIDB-3101 binding in pathophysiological context, human brain tissue homogenates (Fig. S8) were analyzed by dot-blot. Data showed minimal binding of NIDB-3101 to MTG tissue of control subjects (Fig. 3D). Importantly, NIDB-3101 exhibited significantly stronger binding to human AD MTG tissue compared to BMS-986,446 analog (Fig. 3E). The benefits of engaging two different epitopes for improved binding to human brain samples were further illustrated by comparing NIDB-3101 to its monospecific variants, anti-tau aa 344–349 and anti-tau aa 426–430 (Figs. S9A-B). To further characterize tau species captured by NIDB-3101, we evaluated, by Western Blot, HEPES-soluble and SDS/Urea-insoluble protein fractions from three control and three AD donors (Figs. 4A-B). Bepranemab analog and NIDB-3101 detected tau species in the soluble fraction from AD and control samples (Figs. 4A-B). Western blotting confirmed that bepranemab analog and NIDB-3101 detect a wide range, including high (> 75 kD), mid (50–75 kD) and low (< 50 kD) molecular weight, of tau species in the insoluble fractions of AD brains (Figs. 4A-B). Some labeling in control brains, both in the soluble and insoluble fractions, is attributed to tissue homogenization, which enables the detection of both intracellular and extracellular non-pathological tau species, and is not indicative of nonspecific binding. Furthermore, the differences between dot-blot (Fig. 3D) and Western Blot (Fig. 4B) are attributed to the distinct methodologies used, as dot-blot analyzes crude brain homogenates, while Western Blot employs biochemical fractionation that enriches tau in the brain samples. It is noteworthy that bepranemab, an antibody which was tested and safe clinically in healthy participants and patients with AD and PSP, displays a broader -labelling in control brain homogenates compared to NIDB-3101. While NIDB-3101 identifies only a few bands (50–75 kD) in control brains, it labels a diverse range of tau species (> 75 kD, 50–75 kD, and < 50 kD) in insoluble extracts of AD brain homogenates, suggesting its ability to identify pathogenic or aggregated tau entities. To further establish the binding specificity of NIDB-3101, binding assays were conducted using a set of irrelevant proteins, including hHER2, hCEA, hFRa, hFCRL5, and hPDL-1 (Fig. S10A). The data demonstrate that NIDB-3101 does not exhibit detectable binding to any of these irrelevant proteins (Fig. S10A). Furthermore, there was no binding observed to the surface of human lung cell line A-549 (Fig. S10B), nor the surface of human neuroblastoma cell line SH-SY5Y (Fig. S10C). These results indicate that the dual recognition of distinct domains within tau by NIDB-3101 does not compromise its binding specificity. Overall, these findings highlight the binding characteristics of NIDB-3101 and its ability to bind a broad range of tau species in human AD brain fractions. Fig. 3NIDB-3101 binding efficacy to post-mortem human brain samples from AD pathology compared to second generation anti-tau clinical analogues. Schematic view comparing epitope location within tau 2N4R of two second generation monospecific anti-tau benchmarks, targeting either an adjacent region to the MTBR (bepranemab analog) or specific to the MTBR domain (BMS-986446, etalanetug analogs), to the third-generation bispecific anti-tau, NIDB-3101 (A). Comparative affinity characterization Table of NIDB-3101, bepranemab BMS-986,446 and etalanetug analogues to monomeric tau protein 2N4R (B). SPR results are depicted in the Table and show the affinity values to tau monomeric 2N4R. Affinities were measured by surface plasmon resonance on a Biacore instrument in 3 independent experiments and reported as mean ka (M-1s-1), kd (s-1) and KD (nM) values +/- SD (B). Concentration dependent binding comparing NIDB-3101, bepranemab analog, BMS-986,446 analog and etalanetug analog to tau 2N4R preformed fibrils mutated at position P301S from SF9-insect cells expression system (including post-translational modifications) (C). hIgG1 is a non-binder negative isotype control (C). Binding signal by DB on post-mortem human brain homogenates from non-dementia control (n = 10 donors) (D) or AD (tau and amyloid beta pathology n = 10 donors) pathology (E). Binding data of a hIgG1 non-tau binding isotype control compared to NIDB-3101 and bepranemab analog, BMS-986,446 analog and etalanetug analog (D-E). Individual data points depict one single brain donor per pathology. Values presented as mean +/- SEM. The one-way analysis of variance (ANOVA) with Tukey’s post hoc test is used for statistical significance calculation. P values are the following * = P < 0.05; ** = P < 0.01; *** = P < 0.001; **** = P < 0.0001 NIDB-3101 binding efficacy to post-mortem human brain samples from AD pathology compared to second generation anti-tau clinical analogues. Schematic view comparing epitope location within tau 2N4R of two second generation monospecific anti-tau benchmarks, targeting either an adjacent region to the MTBR (bepranemab analog) or specific to the MTBR domain (BMS-986446, etalanetug analogs), to the third-generation bispecific anti-tau, NIDB-3101 (A). Comparative affinity characterization Table of NIDB-3101, bepranemab BMS-986,446 and etalanetug analogues to monomeric tau protein 2N4R (B). SPR results are depicted in the Table and show the affinity values to tau monomeric 2N4R. Affinities were measured by surface plasmon resonance on a Biacore instrument in 3 independent experiments and reported as mean ka (M-1s-1), kd (s-1) and KD (nM) values +/- SD (B). Concentration dependent binding comparing NIDB-3101, bepranemab analog, BMS-986,446 analog and etalanetug analog to tau 2N4R preformed fibrils mutated at position P301S from SF9-insect cells expression system (including post-translational modifications) (C). hIgG1 is a non-binder negative isotype control (C). Binding signal by DB on post-mortem human brain homogenates from non-dementia control (n = 10 donors) (D) or AD (tau and amyloid beta pathology n = 10 donors) pathology (E). Binding data of a hIgG1 non-tau binding isotype control compared to NIDB-3101 and bepranemab analog, BMS-986,446 analog and etalanetug analog (D-E). Individual data points depict one single brain donor per pathology. Values presented as mean +/- SEM. The one-way analysis of variance (ANOVA) with Tukey’s post hoc test is used for statistical significance calculation. P values are the following * = P < 0.05; ** = P < 0.01; *** = P < 0.001; **** = P < 0.0001 The binding of NIDB-3101 was also assessed by immunohistochemistry using formalin-fixed paraffin-embedded MTG sections obtained from 6 AD patients and compared to AT8 and bepranemab analog anti-tau antibodies (Figs. 4C-N). Specificity of the staining was compared to a non-tau binding isotype control hIgG tested as negative control. NIDB-3101 revealed distinct pathological features in AD MTG brain tissue, including granular cytoplasmic neuronal tau immunoreactivity, neurofibrillary tangles, ghost tangles, neuropil threads, coiled bodies and dystrophic neurites (Figs. 4C-M). No immunostaining was observed with NIDB-3101 in MTG tissue of selected control donors (Fig. S11). Numerous granular cytoplasmic neuronal immunoreactivity and neurofibrillary tangles in the AD MTG were detected with NIDB-3101. The semi-quantitative scoring did not show a difference in neuronal cytoplasmic immunoreactivity across antibodies (AT8, bepranemab and NIDB-3101) (Fig. 4N). In summary, these data confirm that NIDB-3101 binds to tau-immunoreactive pathological features in AD brain tissue. Fig. 4NIDB-3101 binding to multiple tau species in human AD brain tissues. Binding of bepranemab analog and NIDB-3101 to human MTG fractions and brain tissue paraffine-embedded tissue sections. Western blotting confirmed binding of bepranemab analog (A) and NIDB-3101 (B) to a broad range of tau species in hepes soluble and SDS/Urea insoluble fractions of human brain MTG samples from non-neurological controls and AD donors (n = 3/group). Tau species of 37–70 kDa were detected with both antibodies in the soluble MGT control and AD fractions. High (> 75 kDa), mid (50–75 kDa) and low (< 50 kDa) molecular weight tau species were detected in the insoluble fraction with the bepranemab analog and NIDB-3101 antibody of AD insoluble MTG fraction, whereas few bands between 50–70 kDa were detected in the insoluble fraction of controls. Representative images of morphological structures visualized with anti-tau AT-8, NIDB-3101 and bepranemab analog immunohistochemistry (C-N). NIDB-3101 showed a variety of morphological structures: granular cytoplasmic neuronal tau immunoreactivity (C), neurofibrillary tangles (D), neuropil threads (E), coiled bodies (F), dystrophic neurites (G), dendritic grains (H), globular oligodendroglial inclusions (I), and ghost tangles (J). Comparison between AT8 (K), bepranemab analog (L) and NIDB-3101 (M) confirmed that these antibodies bind pathological features in AD MTG tissue: granular cytoplasmic neuronal tau immunoreactivity (by black bold hashtags), neurofibrillary tangles (by black bold arrowheads) and neuropil threads (by black bold asterisks) (K-M). Frequency distributions of semi-quantitative scores of cellular morphologies confirmed strong binding of AT8, bepranemab analog, and NIDB-3101 to granular cytoplasmic neuronal tau and presence of neurofibrillary tangles in AD MTG tissue (n = 10) (N). Scale bars represent 25 μm in each image NIDB-3101 binding to multiple tau species in human AD brain tissues. Binding of bepranemab analog and NIDB-3101 to human MTG fractions and brain tissue paraffine-embedded tissue sections. Western blotting confirmed binding of bepranemab analog (A) and NIDB-3101 (B) to a broad range of tau species in hepes soluble and SDS/Urea insoluble fractions of human brain MTG samples from non-neurological controls and AD donors (n = 3/group). Tau species of 37–70 kDa were detected with both antibodies in the soluble MGT control and AD fractions. High (> 75 kDa), mid (50–75 kDa) and low (< 50 kDa) molecular weight tau species were detected in the insoluble fraction with the bepranemab analog and NIDB-3101 antibody of AD insoluble MTG fraction, whereas few bands between 50–70 kDa were detected in the insoluble fraction of controls. Representative images of morphological structures visualized with anti-tau AT-8, NIDB-3101 and bepranemab analog immunohistochemistry (C-N). NIDB-3101 showed a variety of morphological structures: granular cytoplasmic neuronal tau immunoreactivity (C), neurofibrillary tangles (D), neuropil threads (E), coiled bodies (F), dystrophic neurites (G), dendritic grains (H), globular oligodendroglial inclusions (I), and ghost tangles (J). Comparison between AT8 (K), bepranemab analog (L) and NIDB-3101 (M) confirmed that these antibodies bind pathological features in AD MTG tissue: granular cytoplasmic neuronal tau immunoreactivity (by black bold hashtags), neurofibrillary tangles (by black bold arrowheads) and neuropil threads (by black bold asterisks) (K-M). Frequency distributions of semi-quantitative scores of cellular morphologies confirmed strong binding of AT8, bepranemab analog, and NIDB-3101 to granular cytoplasmic neuronal tau and presence of neurofibrillary tangles in AD MTG tissue (n = 10) (N). Scale bars represent 25 μm in each image Low Fc-effector function and plasma half-life-extension variant of NIDB-3101 LALA amino acid substitutions were incorporated in NIDB-3101 bpAb IgG1 Fc fragment (L234A/L235A) to reduce neuroinflammation risk through decreased binding to Fcγ receptors (FcgRs) [43]. Additionally, YTE IgG1 Fc substitutions (M252Y/S254T/T256E) were introduced to enhance the binding of the IgG1 Fc fragment to hFcRn at the endosomal pH-range (pH 5.5–6.5) without affecting its binding at pH 7.4, increasing receptor-mediated cellular recycling efficiency and thus extending systemic half-life (Fig. 5A) [44]. This variant is expected to increase NIDB-3101 concentrations in circulation at the same systemic dose as a non-engineered variant, thereby improving its target engagement and reducing dosing frequency burden for patients. IgG1 Fc-wild-type, LALA and LALA/YTE-Fc variants of NIDB-3101 were compared in their capacity to bind FcgR and FcRn, as well as for their capacity to block intracellular Tau aggregation induced by extracellular recombinant seeds. Affinity to human CD64 (FcγR-I), CD32a-R167 polymorphism (FcγR-IIA), CD16a-V158 polymorphism (FcγR-IIIA) was determined by Octet and affinity towards hFcRn was determined by SPR. The introduction of the LALA substitutions was confirmed to strongly reduce the binding to various FcγRs (Fig. 5B and Fig. S12), with a hierarchy across the tested FcγRs in line with known traits for IgG1 with LALA substitutions [45]. Surprisingly, the combination of LALA and YTE substitutions resulted in a further reduction in binding hFcγR-I and hFcγR-IIIA compared to LALA substitutions alone (Fig. 5B and Figs S12) [46]. The introduction of the YTE substitutions was confirmed to increase the binding to hFcRn at acidic pH by 14.5-fold compared to the WT and LALA variants (Fig. 5B and Fig. S12). Importantly, all NIDB-3101 exhibited a high degree of pH-dependency in their hFcRn binding, with no binding response measured for WT and LALA and only a very minor response for the YTE variant at pH 7.4 (data not shown). This is in line with the expected effect of the YTE substitutions, where increased binding to hFcRn at acidic pH promotes receptor engagement and only weak binding at neutral pH promotes efficient release at the cell surface following hFcRn-mediated recycling. Notably, the affinities of the LALA and WT variants towards hFcRn were nearly identical, entailing that the LALA substitutions in themselves did not impact NIDB-3101’s binding to hFcRn. To provide additional validation of the benefit of the YTE Fc-engineering, hFcRn affinity chromatography was deployed. A longer elution time demonstrated stronger receptor binding for hIgG1 LALA/YTE variant compared to the WT variant (Fig. S13). The capacity of NIDB-3101 variants to undergo hFcRn-mediated rescue from intracellular degradation was finally evaluated in a cellular context using human endothelial cell-based recycling assay (HERA), which was shown to predict plasma half-life extension in vivo [47–50]. We showed that compared to both the WT and the LALA variants, incorporating YTE amino acid substitutions significantly increased the overall amount of NIDB-3101 within hFcRn-expressing microvascular endothelial cells at the end of an initial uptake phase (Fig. 5C), which reflects the outcome of an ongoing uptake and recycling process over the course of 4 h in a buffer of pH 7.4 where increased recycling efficiency of the YTE-variant enables a higher degree of rescue from intracellular degradation. Throughout a subsequent overnight incubation period, this translated into a significant increase in recycled amounts of the YTE-variant (Fig. 5D). Importantly, the increase in recycling surpassed the concomitant increase in residual intracellular amount of the YTE-variant at end of experiment (Fig. 5E), which signifies increased recycling efficiency and predicts prolonged circulatory half-life as a function of improved hFcRn-mediated cellular transport. All experiments included internal controls for discrimination of enhanced hFcRn-mediated transport, specifically in the form of WT and YTE-variants of well-characterized model, hapten-specific antibodies produced by Authera (Fig. S14), for which the YTE-substitutions have been shown to increase cellular recycling efficiency in HERA in a manner that correlates with prolonged plasma half-life in human FcRn transgenic mice [51]. The model antibodies exhibited the expected hierarchy for all HERA parameters in all experiments, thus validating the viability of used cell cultures and ongoing hFcRn-mediated cellular transport. Notably, the magnitude by which YTE increased the HERA parameters of NIDB-3101 was largely in line with that observed for the model antibodies. Notably, the LALA variant exhibited a HERA profile near identical to the WT variant, signifying that the implemented engineering for abolished immune effector functions in itself does not alter hFcRn-mediated cellular transport. Importantly, the WT variant or the one carrying the LALA or LALA/YTE substitutions effectively and similarly inhibited intracellular tau aggregation induced by recombinant tau seeds, achieving approximately 80% reduction compared to a hIgG1 non-tau binding isotype control (Fig. 5F). Fig. 5Low Fc-effector function and half-life-extension variant of NIDB-3101. Schematic structure of the anti-tau bpAb, NIDB-3101, incorporating LALA and YTE Fc amino acids substitutions (A). Binding affinities of the biparatopic anti-tau NIDB-3101 Fc-variants (wild type, LALA only or LALA/YTE) were determined by Octet on human CD64 (FcγR-I), human CD32A R167 (FcγR-IIA), human CD16A V158 (FcγR-IIIA) (B). Binding affinity values to FcRn at pH 6.0 and 6.4 were determined by SPR and are depicted (B). Amino acid substitutions introduced in the Fc region are described. Each value is provided as the mean measurement of 2 independent experiments (B). Association rate constants (ka, M-1s-1), dissociation rate constant (kd, s-1) and binding affinities (KD, nM) to all the relevant human FcγRs are depicted in Supplementary Fig. 9. Cellular uptake (C), recycling transport (D) and residual accumulation (E) of NIDB-3101 Fc-variants assessed by using Authera’s platform based on hFcRn-expressing cells. Results are obtained from three individual experiments performed with technical triplicates and shown as relative values derived by normalizing absolute values for NIDB-3101 Fc-variants against corresponding intraexperimental absolute values for NIDB-3101 Fc-wild type (dashed lines) (C-E). Activity comparison of NIDB-3101 Fc-variants on tau seeded-aggregation cell-based assay using recombinant seeds (F). Recombinant seeds (tau PFF, 2.5 µg/mL ≈ 55 nM) mixed with a fixed concentration of anti-tau mAbs (50 µg/mL = 333 nM) are applied to tau-biosensor cells for 2 days. Head-to-head comparison of NIDB-3101 as hIgG1, hIgG1-LALA and hIgG1-LALA/YTE (F). Data is expressed as normalized FRET values and set at 100% based on the signal obtained in the presence of a hIgG1 non-tau binding isotype control (dashed line). Individual data points depict individual experiments. Values presented as mean +/- SEM. The one-way ANOVA with Dunnett’s post hoc test is used for statistical significance calculation with Dunnett’s post hoc test. ****P < 0.0001, ***P < 0.001, **P < 0.01, *P < 0.05, ns = non-significant Low Fc-effector function and half-life-extension variant of NIDB-3101. Schematic structure of the anti-tau bpAb, NIDB-3101, incorporating LALA and YTE Fc amino acids substitutions (A). Binding affinities of the biparatopic anti-tau NIDB-3101 Fc-variants (wild type, LALA only or LALA/YTE) were determined by Octet on human CD64 (FcγR-I), human CD32A R167 (FcγR-IIA), human CD16A V158 (FcγR-IIIA) (B). Binding affinity values to FcRn at pH 6.0 and 6.4 were determined by SPR and are depicted (B). Amino acid substitutions introduced in the Fc region are described. Each value is provided as the mean measurement of 2 independent experiments (B). Association rate constants (ka, M-1s-1), dissociation rate constant (kd, s-1) and binding affinities (KD, nM) to all the relevant human FcγRs are depicted in Supplementary Fig. 9. Cellular uptake (C), recycling transport (D) and residual accumulation (E) of NIDB-3101 Fc-variants assessed by using Authera’s platform based on hFcRn-expressing cells. Results are obtained from three individual experiments performed with technical triplicates and shown as relative values derived by normalizing absolute values for NIDB-3101 Fc-variants against corresponding intraexperimental absolute values for NIDB-3101 Fc-wild type (dashed lines) (C-E). Activity comparison of NIDB-3101 Fc-variants on tau seeded-aggregation cell-based assay using recombinant seeds (F). Recombinant seeds (tau PFF, 2.5 µg/mL ≈ 55 nM) mixed with a fixed concentration of anti-tau mAbs (50 µg/mL = 333 nM) are applied to tau-biosensor cells for 2 days. Head-to-head comparison of NIDB-3101 as hIgG1, hIgG1-LALA and hIgG1-LALA/YTE (F). Data is expressed as normalized FRET values and set at 100% based on the signal obtained in the presence of a hIgG1 non-tau binding isotype control (dashed line). Individual data points depict individual experiments. Values presented as mean +/- SEM. The one-way ANOVA with Dunnett’s post hoc test is used for statistical significance calculation with Dunnett’s post hoc test. ****P < 0.0001, ***P < 0.001, **P < 0.01, *P < 0.05, ns = non-significant NIDB-3101 hIgG1-LALA/YTE inhibits human AD brain seeded tau aggregation To further substantiate that the biparatopic engagement of tau by NIDB-3101 hIgG1-LALA/YTE is preserved and accessible in pathological tau fibrils, we tested gold particle-labelled antibody binding to tau fibrils isolated from human AD brain tissues by electron microscopy. Tau fibrils were labeled with either a hIgG non-tau binding isotype control, NIDB-3101 hIgG1-LALA/YTE, bepranemab analog or BMS-986,446 analog (Fig. 6A). The results indicate that NIDB-3101 hIgG1-LALA/YTE had a higher number of immunogold particles bound to AD brain-isolated PHFs than the three second generation clinical anti-tau analogs (Fig. 6B). Subsequently, we evaluated the ability of these antibodies to inhibit seed-induced tau aggregation using tau fibrils isolated from human AD brain. NIDB-3101 significantly inhibited tau aggregation, achieving a mean reduction of approximately 70–80% in comparison to the isotype negative control (Fig. 6C). Notably, NIDB-3101’s capacity to inhibit human AD brain seed-induced tau aggregation is superior to that of bepranemab analog, BMS-986,446 analog and etalanetug analog (all tested at 333 nM, final concentration), which exhibited mean percentage reductions of ~ 11–33% (Fig. 6C). Dose range analysis revealed that NIDB-3101 hIgG1 wild type variant exhibits comparable effectiveness to its hIgG1-LALA/YTE variant (Fig. S15), showing inhibition of intracellular tau aggregation within a concentration range of at 0.02 nM to 333 nM (Fig. 6D). Fig. 6Binding and activity mediated of NIDB-3101 using human AD brain tau seeds. Representative immunogold labelled staining for a hIgG1 non-tau binding isotype control, bepranemab analog, BMS-986,446 analog and NIDB-3101-hIgG1-LALA/YTE on human AD tau seeds (A). Quantification of the ratio between number of dots and number of fibrils per image per Abs. A minimum of 12 images were quantified per condition. Individual data points depict individual images analyzed per condition (B). Head-to-head activity of NIDB-3101 to bepranemab analog, BMS-986,446 analog and etalanetug analog on tau seeded-aggregation cell-based assay using tau-enriched post-mortem human AD brain seeds (C). Data is expressed as normalized FRET values and set at 100% based on the signal obtained in the presence of a hIgG1 non-tau binding isotype control used as a negative control (dashed line). Minimum of 6 independent experiments were run in duplicate. Individual data points depict one unique experiment. Values presented as mean +/- SEM. The one-way analysis of variance (ANOVA) with Tukey’s post hoc test is used for statistical significance calculation. P values are the following * = P < 0.05; ** = P < 0.01; *** = P < 0.001; **** = P < 0.0001. Concentration dependant activity of NIDB-3101 on tau seeded-aggregation cell-based assay using tau-enriched post-mortem human AD brain seeds (n = 2 independent experiments). The signal obtained in the presence of a hIgG1 non-tau binding isotype control used as a negative control is presented with a dashed line (D) Binding and activity mediated of NIDB-3101 using human AD brain tau seeds. Representative immunogold labelled staining for a hIgG1 non-tau binding isotype control, bepranemab analog, BMS-986,446 analog and NIDB-3101-hIgG1-LALA/YTE on human AD tau seeds (A). Quantification of the ratio between number of dots and number of fibrils per image per Abs. A minimum of 12 images were quantified per condition. Individual data points depict individual images analyzed per condition (B). Head-to-head activity of NIDB-3101 to bepranemab analog, BMS-986,446 analog and etalanetug analog on tau seeded-aggregation cell-based assay using tau-enriched post-mortem human AD brain seeds (C). Data is expressed as normalized FRET values and set at 100% based on the signal obtained in the presence of a hIgG1 non-tau binding isotype control used as a negative control (dashed line). Minimum of 6 independent experiments were run in duplicate. Individual data points depict one unique experiment. Values presented as mean +/- SEM. The one-way analysis of variance (ANOVA) with Tukey’s post hoc test is used for statistical significance calculation. P values are the following * = P < 0.05; ** = P < 0.01; *** = P < 0.001; **** = P < 0.0001. Concentration dependant activity of NIDB-3101 on tau seeded-aggregation cell-based assay using tau-enriched post-mortem human AD brain seeds (n = 2 independent experiments). The signal obtained in the presence of a hIgG1 non-tau binding isotype control used as a negative control is presented with a dashed line (D) Discussion There is a strong need for high-affinity anti-tau therapeutic antibodies to slow down or halt disease progression in AD or other tauopathies. Here, we present a novel biparatopic anti-tau antibody engaging both the R4 and C-terminal tau domains with a sub-nanomolar affinity, underscoring its therapeutic potential for AD. Our findings demonstrate that NIDB-3101 recognizes pathological tau species, including truncated and aggregated forms and in the presence or absence of PTMs, prevalent in AD human brain tissue. NIDB-3101 captures a broad range of tau species as evidenced by WB and detects distinct pathological features in AD human brain tissue with high sensitivity using immunohistochemistry. We confirmed that NIDB-3101 binds the critical minimal core-forming R3-R4 fibrils as revealed by cryo-EM in human AD brains and overcomes the narrow specificity of monospecific antibodies [16, 39]. NIDB-3101 other binding site includes C-terminal tau which have been targeted by antibodies previously which reduced the internalization of both monomeric tau and tau fibrils [22, 23]. The biparatopic format enhances avidity and functional potency, as evidenced by robust inhibition of tau seeding in biosensor assays using human AD brain-derived PHFs. This efficacy is thought to be due to its binding to two independent tau epitopes close to MTBR and binding to tau even when truncations occur in human brains. Our results demonstrate that NIDB-3101 outperforms the second generation bepranemab, BMS-986,446 and etalanetug antibody analogs in both binding affinity and functional activity. Binding assays revealed that NIDB-3101 presents a significantly higher apparent affinity for both monomeric and aggregated tau proteins compared to its monospecific counterparts. In summary, we designed a third generation bpAb, NIDB-3101, with a high affinity to a broad range of pathological tau species, as innovative approach for the treatment of AD. NIDB-3101 antibody was designed with multiple advantages, including: (1) its ability to bind to two non-overlapping tau epitopes (aa 344–349 and aa 426–430), providing high apparent affinity and specificity; (2) demonstrated superior binding affinity and efficacy compared to second-generation monospecific anti-tau antibodies, effectively capturing a broad range of tau species in AD brain extracts and seeds; (3) strong binding to various pathological features in human AD brain tissue, including pre-tangles, mature tangles, ghost tangles, and neuropil threads; and (4) significant inhibition of seeded tau aggregation in a cellular assay, achieving suppression levels that surpassed both parental and benchmark antibodies. The expected mechanism of action of NIDB-3101, as suggested by the cellular assay using AD brain tau-enriched seeds, involves the targeted recognition and neutralization of pathologic tau species, effectively inhibiting their transsynaptic spread and aggregation. The biparatopic approach enhances the antibody’s capacity to disrupt tau propagation, ultimately contributing to the prevention of tau-related neurodegeneration. In contrast to other bsAb constructs, NIDB-3101 retains the sequence of a natural hIgG without any engineering. This feature confers several advantages. First, it minimizes immunogenicity risk—an important consideration for chronic, long-term administration in AD. Second, NIDB-3101 was manufactured using a standard process for recombinant IgG production, enabling straightforward purification with commercially available resins that yield robust overall production efficiency. NIDB-3101 incorporates Fc engineering (YTE substitutions) to extend its half-life and thereby reduce the patient burden of dosing requirements, a strategy validated in other therapeutic FDA-approved antibodies such as Beyfortus (nirsevimab-alip) with T1/2 ~ 70 days for RSV prophylaxis and Evusheld (tixagevimab/cilgavimab) with T1/2 ~ 90 days for SARS-CoV-2 prophylaxis [52, 53]. The IgG1 backbone with reduced FcγR binding is anticipated to minimize microglia-mediated inflammation, a critical consideration given neuroinflammatory risks associated with immunotherapy [54]. The combination of biparatopic engagement, high affinity and long half-life is expected to lead to higher target engagement at lower injected doses than other antibodies with the consequent lower predicted human efficacious doses. These findings position NIDB-3101 as a potential low-risk partner for combination therapy with amyloid-targeting antibodies, offering the potential to reduce the risk at neuroinflammation-related ARIA liability while targeting tau-mediated neurodegeneration. Considering the Fc mutations, in addition to its role in blocking transsynaptic tau spread, we anticipate that the removal of the tau-NIDB-3101 complex from the extracellular brain environment will involve minimal contributions from microglial engulfment, due to the decreased binding of NIDB-3101 to FcγR. Consequently, we propose that the immune complex is likely cleared from the brain to the periphery (CSF turn over is 4–5 times a day – [55]), facilitating its subsequent elimination from the systemic circulation. In conclusion, the strong preclinical performance of NIDB-3101 in clinically relevant human extracts-derived models supports its advancement into clinical development as a next generation disease-modifying therapy for AD. Limitations This study used a wide range of techniques for an in-depth characterization of NIDB-3101 using AD human brain tissues. NIDB-3101 exhibited high binding affinity and efficacy against various pathological tau forms. Its ability to recognize different tau species and inhibit tau aggregation suggests a positive outlook for therapeutic use in AD. Despite these strengths, we acknowledge several limitations. A limitation to consider is the fact that NIDB-3101 binding and function was only tested using human AD extracts and different tau structures have been described in other tauopathies (Progressive Supranuclear Palsy, Pick’s disease, etc.), highlighting the need to assess NIDB-3101’s effect with extracts from these other tauopathies to allow translatability to these diseases [56–58]. Another limitation is the lack of in vivo data. A study is currently being conducted in non-human primates (NHP) to evaluate NIDB-3101 pharmacokinetics (PK) and central nervous system (CNS) biodistribution, as well as pharmacodynamic (PD) and safety endpoints. This study will include tissue cross-reactivity assessments to further reinforce the specificity of NIDB-3101 in a relevant model. This data will be published after completion of the study. Furthermore, while tau transgenic mice could provide insight into in vivo efficacy, their use is limited due to their typical overexpression of a single isoform of tau and the absence of human FcRn expression, which does not support the assessment of NIDB-3101’s pharmacology in the context of its LALA Fc mutation affecting human FcγR binding. As previously mentioned, the data from the ongoing NHP study will be published after completion of the study. Tau transgenic mice are not suitable for evaluating NIDB-3101’s pharmacology nor pharmacokinetic profile due to their overexpression of a single tau isoform and the lack of cross-reactivity with human FcRn and FcγR. Therefore, we opted for the ongoing study in non-human primates (NHPs), which is expected to provide more relevant insights into the antibody’s complex pharmacological profile. Conclusions NIDB-3101 is a novel, third-generation anti-tau biparatopic antibody which can simultaneously engage two distinct epitopes: one in the fourth domain of the MTBR (residues 344–349) and another in the C-terminal domain (residues 426–430). This ‘dual-engagement’ strategy is designed to increase avidity and ensure that even if one epitope is masked or truncated, the antibody remains effective. Furthermore, we’ve incorporated YTE and LALA Fc-mutations to extend the half-life and reduce effector-function-related toxicity. Importantly, binding and function was tested on disease relevant human post-mortem AD brain extracts. Key findings include: (i) NIDB-3101 targets two epitopes located in the fourth MTBR domain and C-terminal region of tau, with sub-nanomolar affinity for both monomeric and aggregated tau, (ii) The biparatopic anti-tau κλ-bodies show an improvement in apparent binding potency, allowing the capture of broader range of pathological tau species compared to benchmark monospecific mAbs, (iii) NIDB-3101 displays an enhanced cellular efficacy by blocking tau seeding induced by both recombinant tau seeds and post-mortem human AD brain extracts compared to benchmark anti-tau antibodies, (iv) NIDB-3101 Fc mutations YTE increased FcRn binding at acidic pH, leading to better recycling in cellular assay. Its LALA mutations decreased binding to FcγgR, reducing the neuroinflammation risk. Supplementary Information Supplementary Material 1. Supplementary Material 1. Supplementary Material 2. Supplementary Material 2.