Immunotherapy Against Acetylated Tau at K280 Ameliorates Behavioral Impairments and Pathologies of Tau Transgenic Mouse

Background : Tau is gaining attention as a target of therapeutic antibodies for Alzheimer's disease (AD) and other tauopathies but it was recently of debate which epitope on tau has therapeutic potency. While the hexapeptide 275 VQIINK 280 is a critical region for tau aggregation and K280 is acetylated in various tauopathies including AD, there has been no immunotherapy targeting acetylated K280. Methods : Acetylated K280 was rst chosen as therapeutic target from small pilot studies comparing in vivo behavioral ecacy of tau transgenic mice by immunizing with several putatively pathogenic epitopes such as acetylated or phosphorylated or cleaved tau. Results : Therapeutic ecacy in tau transgenic mice was further evaluated with behavior, pathology and biochemical experiments after either active immunization with tau peptide acetylated at K280 or passive immunization with anti-tau antibody targeting K280 acetylation. Vaccination with tau peptide acetylated at K280 or antibody (Y01) targeting it ameliorated memory impairments and pathology in tau transgenic mice via intracerebroventricular or intraperitoneal route. Conclusions : This study showed that acetylated K280 tau is a good therapeutic target and Y01 could be a novel therapeutic antibody for AD and other tauopathies.


Abstract
Background : Tau is gaining attention as a target of therapeutic antibodies for Alzheimer's disease (AD) and other tauopathies but it was recently of debate which epitope on tau has therapeutic potency. While the hexapeptide 275 VQIINK 280 is a critical region for tau aggregation and K280 is acetylated in various tauopathies including AD, there has been no immunotherapy targeting acetylated K280.
Methods : Acetylated K280 was rst chosen as therapeutic target from small pilot studies comparing in vivo behavioral e cacy of tau transgenic mice by immunizing with several putatively pathogenic epitopes such as acetylated or phosphorylated or cleaved tau.
Results : Therapeutic e cacy in tau transgenic mice was further evaluated with behavior, pathology and biochemical experiments after either active immunization with tau peptide acetylated at K280 or passive immunization with anti-tau antibody targeting K280 acetylation. Vaccination with tau peptide acetylated at K280 or antibody (Y01) targeting it ameliorated memory impairments and pathology in tau transgenic mice via intracerebroventricular or intraperitoneal route.
Conclusions : This study showed that acetylated K280 tau is a good therapeutic target and Y01 could be a novel therapeutic antibody for AD and other tauopathies.

Background
Tau protein aggregates, such as neuro brillary tangles (NFTs), are a key pathological nding in Alzheimer's disease (AD) and other neurodegenerative disorders, collectively referred to as tauopathies. In AD, tau pathology has a hierarchical pattern of accumulation that rst appears in the transentorhinal cortex and subsequently spreads to the hippocampus and neocortex (Braak and Braak, 1991). The spread of tau pathology is underscored by cell-to-cell transmission of pathological tau species (de Calignon et al., 2012;Michel et al., 2014;Wu et al., 2016;Clavaguera et al., 2009), which may act as targets of therapeutic antibodies. However, it is unclear which species and distinct features of tau are critical for pathological tau propagation. Tau is expressed as six isoforms with zero, one, or two N-terminal acidic inserts (0N, 1N, or 2N, respectively) and either three or four microtubule-binding repeats (MTBR) (3R or 4R) (Goedert et al., 1989). Tau (2N4R) is a long protein consisting of 441 amino acids and can be modi ed by posttranslational modi cations such as phosphorylation, acetylation, and cleavage at various epitopes. This makes it more challenging to select the most effective epitope for tau in contrast to short β-amyloid (Aβ).
Hence, the roles and relationships of these various modi ed epitopes in AD pathogenesis should be carefully considered.
Tau is acetylated at lysine residues in the MTBR, hence providing a novel regulatory modi cation other  (Cohen et al., 2011) but is rarely observed in control brain tissues (Irwin et al., 2012), which delineates the disease relevant nature of K280 acetylation. Deletion of lysine-280 (ΔK280) of tau is reported in frontotemporal dementia with Parkinsonism (FTDP)-17 and AD (Momeni et al., 2009;Rizzu et al., 1999), and ΔK280 facilitates β-structure formation and accelerates tau aggregation in in vitro and in vivo models (Mocanu et al., 2008), hence the hexapeptide 275 VQIINK 280 , encompassing K280, is well-known as a critical region for tau aggregation (Mukrasch et al., 2005).
Various tau species and modi cations have been reported to be related to AD, but it is unclear which forms are true therapeutic targets rather than merely being pathologically correlated ndings. Here, we addressed this with small pilot studies comparing in vivo behavior e cacy of tau transgenic mice by immunizing with several putatively pathogenic epitopes such as acetylated or phosphorylated or cleaved tau, then chose acK280 as a target. We validated therapeutic e cacy targeting tau-acK280 with active immunization or antibody in tau transgenic mice.
Tau-P301L mice were generated from JNPL3 (Tau-0N4R, P301L) mice purchased from Taconic, Inc. (NY, USA) and backcrossed to C57BL/6J over ve generations. Tau-P301S mice were generated from PS19 (Tau-1N4R, P301S) mice purchased from Jackson Laboratories, Inc. (Bar Harbor, ME, USA) and backcrossed to C57BL/6J over ve generations. All mouse genotyping was con rmed by PCR. The mice were caged (3 to 5 mice per cage) and kept on a 12h light/dark cycle at a constant ambient temperature (22±1 °C) with 40-60% humidity. Food and water were available ad libitum. All animal protocols were approved by Asan Institute for Life Science Animal Experimentation Committee.

Generation of hybridoma cells
Mouse hybridoma cells were manufactured by Young-In Frontier Co., Ltd. (Seoul, Korea). Brie y, the antigen was mixed with Complete Freund adjuvant (Sigma-Aldrich, St. Louis, MO, USA) and intraperitoneally injected into female BALB/c mice. After 2 weeks, serum was collected from mice and antibody titer was measured using ELISA. The antigen was mixed with PBS and injected into mice with a second injection for boosting. After 3 days, the spleen was extracted. After washing the tissue with culture medium, cells were separated. Myeloma cells (Sp2/0Ag14) were fused with cells separated from mice using polyethylene glycol (PEG) (Roche Life Science, Mannheim, Germany). Fused cells were cultured using 1XHAT culture medium (Sigma-Aldrich, St. Louis, MO, USA). Using a HT culture medium (Gibco Life Technologies, NY, USA), cells were pipetted into a 96-well plate. Hybridoma cells were cultured for 7-10 days. For positive clone screening, hybridoma supernatants were screened by ELISA for reactivity with K280-ac peptides. ELISA screens and the cloning process were repeated until the nal clone was veri ed. The best clone was selected and termed ADEL-Y01 antibody.

Preparation of ADEL-Y01 antibody
Hybridoma cells were grown at 37°C in CD Hybridoma Medium (Gibco Life Technologies, NY, USA) supplemented with 8% glutamax at 1:25 dilution, 2% ultra-low IgG fetal bovine serum (FBS), and 1% penicillin streptomycin (Gibco). After 1 week, supernatant was collected and centrifuged at 13,000 rpm for 15 min at 4°C. Supernatant was ltered with 0.4 mm lter unit and 1 M Tris-pH 9.0 was added for neutralization. All experiments were performed on AKTA-start chromatography system (GE Healthcare Life Sciences, Piscataway, NJ, USA). The equipment was washed with distilled water, and the column was connected. The column was equilibrated with 1X Phosphate-Buffered Saline (PBS), and the sample was allowed to ow. Once the sample was exhausted, a wash with 1XPBS was performed, and the antibody was eluted using 0.1 M glycine (pH 2.7). The puri ed antibody was neutralized with 1 M Tris pH 9.0. The antibody was dialyzed with 1XPBS. The band was checked using Coomassie blue staining. The Y01 antibody was stored at -20°C. Active immunization At 3 months of age, male transgenic P301L mice and C57BL/6 mice were immunized with either peptide or Adju-Phos adjuvant (InvivoGen, San Diego, USA) only. Mice were intraperitoneally injected with 50 mg peptide mixed 1:1 (v/v) with Adju-Phos adjuvant (25 mg per mouse). After the rst immunization, treatment was administered at intervals of 2 weeks. Injections were performed once a month. After the last injection, mice were subjected to behavioral tests and diffusion tensor imaging (DTI). All mice were sacri ced at 8 months.
Passive immunization Intracerebroventricular infusion At 8 months of age, C57BL/6 and P301L mice were administered intracerebroventricular (icv) injections of immunoglobulin G (IgG) control or Y01 antibody (1.9 mg/mL). The pump implantation surgery was performed according to the manufacturer's instructions. Brie y, before the surgery, an L-shaped infusion cannula was attached to catheter tubing (Alzet, California, USA). A brain infusion kit was attached to a micro-osmotic pump (Alzet, California, USA). The assembly pump was implanted using a stereotactic apparatus (Harvard Apparatus, Massachusetts, USA) into the right lateral ventricle at 0.58 mm posterior to bregma, 1 mm lateral to the midline, and 2 mm from the skull surface. An osmotic pump was subcutaneously implanted into the back of each mouse. Each pump was lled with IgG control or Y01 antibody (1.9 mg/mL). The osmotic pump delivered antibodies continuously at 0.11 μL/h for 28 days. Reservoir volume was 100 μL. Behavioral analysis was performed during the last 3 weeks of infusion. At 9 months of age, all mice were sacri ced.

Intraperitoneal injection
At 7 months of age, C57BL/6 and P301L mice were injected intraperitoneally (IP) with IgG control or Y01 antibody (50 mg/kg). Mice were subjected to behavioral tests. At 10 months of age, all mice were sacri ced. At 14 months of age, C57BL/6 and P301L mice were injected IP with IgG control or Y01 antibody (50 mg/kg). The mice were subjected to behavioral testing. At 17 months of age, all mice were sacri ced. At 5 months of age, C57BL/6 and P301S mice were injected IP with two different doses of Y01 antibody (5 and 50 mg/kg) or IgG control (50 mg/kg). At 8 months of age, all mice were sacri ced. All mice received injections weekly for 3 months. Behavioral analysis was performed during the last 4 weeks of injection.

Behavioral tests
Nest-building Mice were isolated in home cages a day before testing. Cotton was placed in each cage for the scoring of nesting behavior. The next day, nesting behavior was scored according to the following guidelines: score 1, cotton rarely touching and >90% intact cotton; score 2, cotton partially ripped with 50-90% intact; score 3, cotton ripped with <50% intact and spread around the cage; score 4, cotton mostly ripped with <10% intact and gathered into side of the cage; score 5, cotton almost ripped and perfect nest shape.

Grip strength test
To measure strength, a weights test was performed. Animals were held by the base of the tail and lowered to the apparatus comprising a coiled wire ball for gripping. Apparatus weight was less than 3g.
Time taken to drop the apparatus was measured. All tests were repeated three times.

Vertical grid test
All groups were subjected to a vertical test. Mice were placed on a wire net (horizontal, 18 cm; vertical, 30 cm). The wire net was turned over, and the time taken for mice to fall was measured. All tests were repeated three times.

Y-maze
Mice were subjected to a Y-maze test to assess short-term spatial memory. The apparatus was a Yshaped plastic maze with three arms. Mice were habituated for 5 min. One arm (Y-shaped left arm) was open and the other arm (Y-shaped right arm) was blocked during a 5-min test. After 1 h, the other arm (Yshaped right arm) was unblocked and behavior was analyzed for 5 min. Time in novel arm is the time to stay on the opened Y-shaped right arm.

Rotarod
Mice were placed on an accelerating rotarod, and exercise capacity was measured. The speed was slowly increased from 4 to 40 rpm. The time (sec) and speed (rpm) at which mice fell were measured. Mice were trained for 4 days before the test. On the fth day, motor test was performed. All tests were repeated 3 times.

Water maze
Mice were subjected to a water maze (Vorhees, 2006). The water pool was divided into four quadrants, NE, North East; SE, South East; SW, South West; NW, North West quadrants. Brie y, all mice were trained to locate a visible platform for 4 days. Training was conducted three times a day and start positions were selected as NE, SE and SW quadrants. On the 5day, the platform was removed, and movement was analyzed, such as time taken to reach the target.

Contextual fear conditioning
Mice were placed in a context in which they received a foot shock (Hirotaka Shoji1, 2014). On day 1, all mice were habituated to the chamber for 2 min. The sound was presented for 30 seconds paired with a shock up to the last 2 s. This process was repeated three times. On day 2, after su cient rest, mice were habituated to the context without any stimulation for 3 min. Then, the sound was presented without shock for 3 min, and behavior was analyzed.

Diffusion tensor imaging
Diffusion tensor imaging was performed in mouse after acquiring T2 weighted (T2w). All mice were anesthetized with 1-1.5% iso urane in a mixture of oxygen and room air gases (1:2) delivered via a nose cone. Respiratory rate, electrocardiogram, and rectal temperature were monitored. Parameters and data processing were performed as previously described (H-J Kim, 2017). All mouse using a 4 shot spin-echo based echo planar imaging (EPI) sequence with TR/TE = 3750/46.22 ms, 96 × 96 matrix and an encoding scheme of 30 gradient directions at b-value = 1000 s/mm 2 (H-J Kim, 2017). The hippocampus was selected as the ROI. Data were analyzed by fractional anisotropy (FA) and mean diffusivity (MD) values.

Blood sampling
Blood was collected from the facial vein 1 week before the start of immunization (T0), 1 week after the third injection (T1), 1 week after the fourth injection (T2), and at sacri ce (TF). Blood plasma was harvested by centrifugation, and antibody titer was analyzed.

Enzyme-linked immunosorbent assay
For measuring plasma titers of antibodies, indirect ELISA was performed. Brie y, 96-half well plates (Costar, Corning Life Sciences, Massachusetts, USA) were coated with either 250 ng/well immunizing peptides diluted in capture solution overnight at 4°C. After blocking for 1 h, plasma samples were diluted 1:500 in blocking solution and incubated for 2 h. After 2 h of incubation at room temperature, horseradish peroxidase (HRP)-conjugated secondary antibody was added to each well for 2 h. All steps were followed by three washes with wash solution. Then, samples were processed with 3,3′,5,5′-tetramethylbenzidine (TMB) and a microtiter plate reader MAX190 (Molecular Devices, California, USA) calibrated to 450 nm. mM NaCl, 20 mM NaF, 1 mM Na3VO4, protease/phosphatase inhibitors) and centrifuged at 50,000 ´ g for 20 min. The RAB insoluble pellet was solubilized with RIPA buffer (0.15 M NaCl, 50 mM Tris, 0.5% deoxycholic acid, 1% Triton X-100, 0.5% SDS, 25 mM EDTA, pH 8.0, protease/phosphatase inhibitor) and centrifuged at 50,000 ´ g for 20 min. The RIPA insoluble pellet was solubilized with 70% formic acid.

Immunoprecipitation
Brain lysates of immunized mice were incubated with 100 mL of protein G-agarose bead (GE Healthcare Life Sciences, Piscataway, NJ, USA) at 4°C overnight. The samples were collected by centrifuging at 6,000 rpm for 1 min at 4°C and washed three times with 0.1% Triton X-100 in PBS. The samples were boiled in SDS sample buffer and processed by western blot.

Immunohistochemistry
Frozen mouse brain blocks were sectioned and attached on coated glass slides. Sectioned brain tissues were washed with 1XPBS and tissues were permeabilized with 1XPBS containing 0.1% Triton X-100 for 10 min at RT. The tissues were washed and blocked with PBS containing 3% bovine serum albumin for 1 h at RT. The tissues were then washed with 1XPBS and incubated with primary antibody at 4°C overnight. After washing with 1XPBS, for DAB staining, a biotinylated secondary antibody (Vector Laboratories, California, USA) was added for 1 h at RT. Tissues were washed and incubated in an avidin-biotinperoxidase complex (Vector Laboratories, California, USA). For visualization, tissues were incubated for 10 min in diaminobenzidine (DAB) solution. Tissues were washed and mounted with Canada balsam (Sigma-Aldrich, St. Louis, MO, USA). DAB-reacted images were obtained using a Zeiss microscope (Carl Zeiss, Oberkochen, Germany) and processed using the AxioVision Imaging System.

Statistical analysis
All data were analyzed with GraphPad Prism v.5. Data were analyzed using one-way ANOVA (Tukey's post hoc test) or Student's t-test. A p-value < 0.05 was considered statistically signi cant.

Results
Active immunization with Tau-acK280 is the most effective at reducing behavioral impairments in tau transgenic mice Since phosphorylation, acetylation and cleavage of tau are associated with AD, we rst did a small pilotstudy to test which target we will proceed further for prevention of tau-related impairments, including phosphorylation at T231 (pT231), acetylation at K280 (acK280), acetylation at K311 (acK311), and cleavage at E391 (ΔE391), as suggested in AD pathogenesis (Additional le 1: Figure S1a). To this end, we immunized tau-P301L transgenic (tau-Tg) mice from 3 to 6 months of age with each keyhole limpet hemocyanine (KLH)-conjugated peptides in aluminum adjuvant and compared their effects on memory and motor impairments (Additional le 1: Figure S1b). Wild-type and control tau-Tg mice received adjuvant alone. When mice were immunized with each type of peptide, they exhibited a robust antibody response against each immunogen, whereas minimal responses were observed in control and tau-acK311-immunized mice (Additional le 2: Figure S2). After immunization, we performed behavioral tests including grip strength and vertical tests for motor performance (Additional le 1: Figure S1c, d) as well as Morris water maze tests and fear conditioning (Additional le 1: Figure S1e, f). Only mice immunized with tau-acK280 demonstrated consistent improvement in cognitive and motor performance (Fig. 1, Additional le 3: Figure S3), suggesting that tau-acK280 is the effective therapeutic target for AD and tauopathy relatively than other tau epitopes tested. To further con rm the protective e cacy of tau-acK280 immunization, we immunized mice (n = 9) with tau-acK280 in an independent experiment (Fig. 1, Additional le 3: Figure S3). Nest building test (Fig. 1A, B), learning and memory tests including the Ymaze (Fig. 1C), Morris water maze (Fig. 1D), and motor tests (Additional le 3: Figure S3a, b) revealed impaired cognition and motor performance in tau-Tg mice, but immunization with tau-acK280 attenuated these impairments. In the rotarod test, none of the tau-Tg mice were able to reach 30 rpm, but tau-acK280 vaccination fully recovered rotarod performance (64% of mice) comparable to wild type performance (67%) (Additional le 3: Figure S3c, d).
Active immunization with Tau-acK280 ameliorates pathological changes in tau-Tg mice After immunization for about 4 months, diffusion tensor imaging (DTI) was performed to analyze changes in brain pathology in vivo. DTI analysis showed decreased fractional anisotropy (FA) and mean diffusivity (MD) in the hippocampus of tau-Tg mice (Additional le 4: Figure S4), implying microstructural anomalies and decreased white matter integrity. However, these effects were normalized by tau-acK280 immunization (Additional le 4: Figure S4), implying recovery of microstructural anomalies and functional connectivity by immunization. Western blot data revealed an increase in abnormally phosphorylated tau in tau-Tg mice, but this effect was decreased after tau-acK280 immunization ( Fig. 2A,   B). Acetylated tau was increased in tau-Tg mice, but decreased after tau-acK280 immunization ( Fig. 2A,  B). The levels of synaptic proteins such as PSD95, NMDA receptors, synaptophysin, and synapsin-1 were decreased in the brains of tau-Tg mice (Fig. 2C, D), indicating impaired synapse structure and function. However, these changes were mostly recovered by tau-acK280 immunization (Fig. 2C, D), implying the protective effects of immunization. Increased accumulation of phosphorylated tau in tau-Tg mice and its normalization by tau-acK280 immunization was also observed using immunohistochemistry (Fig. 2E, F). Intracerebroventricular infusion of anti-tau-acK280 antibody, Y01, ameliorates cognitive and pathology in tau-Tg mice To test whether the anti-tau-acK280 antibody, Y01, had therapeutic potential similar to that of tau-acK280 active immunization (Fig. 1, 2), continuous intracerebroventricular (icv) infusion of Y01 antibody into the lateral ventricle of 8-month-old tau-Tg mice was performed for a month (Fig. 3A). As per the results of active immunization, passive icv immunization with Y01 antibody produced improvements in behavioral tests including nest building test, Y-maze, water maze test, and grip strength test (Fig. 3B-D, Additional le 5: Figure S5a). To con rm characterize the aggregate status of tau in the brain, sequential biochemical extraction was performed with RAB (aqueous buffer), radio immunoprecipitation assay (RIPA) (detergent buffer), and 70% formic acid (FA) to solubilize the nal pellet. Insoluble tau was increased in tau-Tg mouse brains and was reduced by Y01 antibody infusion (Fig. 3E). Western blots with cortex lysates revealed that total tau (tau-5), tau-pS396, and tau-acK280 levels were decreased in antibody-infused brains (Fig. 3F, G). Immunohistochemistry revealed that increased phosphorylated tau in tau-Tg mice was normalized by Y01 antibody (Fig. 3H, I). The levels of synaptic proteins such as PSD95 and synapsin-1 were decreased in the brains of tau-Tg mice, which showed a trend to be reversed by Y01 antibody, although this did not reach statistical signi cance (Additional le 5: Figure S5b, c), implying protective effects of immunization.
Systemic administration of Y01 antibody ameliorates behavioral de cits and pathological changes in tau-Tg mice To assess whether peripheral administration of the antibody produced similar effects to those of icv infusion, Y01 antibody was injected intraperitoneally (i.p.) (Fig. 4A). Similar to the results of icv infusion (Fig. 3), improvements in behavioral tests including nest building test, Y-maze, and Morris water maze test (Fig. 4B-D) were observed following weekly i.p. injection with Y01 antibody for 3 months starting from 7 or 14 months of age (Fig. 4). To investigate whether the antibody entered the brain and bound to antigens, brain samples were perfused systemically with cold PBS at sacri ce to remove any blood from the brain and were then incubated with protein-G-sepharose (PGS) to pull down the antibodies in brain parenchyma. Western blots with anti-mouse IgG antibody revealed that both control IgG and Y01 antibody were present in the brain (Fig. 4E), con rming that antibodies entered the brain even though the blood-to-brain ratio may have been low. Tau protein acetylated at K280 was also pulled down alongside PGS only in Y01 antibody-injected mouse brains (Fig. 4E), indicating that peripherally injected Y01 antibody entered the brain and bound to the antigen, tau-acK280. Semi-denatured western blots with cortex lysates either from mice of 10 or 17 months of age revealed that total tau (tau-5), tau-pT231, and tau-acK280 aggregate levels were decreased in antibody-injected brains (Fig. 4F, G). The levels of synaptic proteins such as PSD95 and synapsin-1 were decreased in the brains of tau-Tg mice, which showed a trend to be reversed by Y01 antibody, although this did not reach statistical signi cance (Additional le 6: Figure S6a, b), implying protective effects by immunization. To further evaluate the e cacy of Y01 in a different tau-Tg strain, Y01 antibody was injected i.p. to tau-P301S mice at two different doses (5 mg/kg, 50 mg/kg) for 3 months starting from 5 months of age (Fig. 5A). Improvements in the water maze test were observed in the group that was administered 50 mg/kg of Y01 (Fig. 5B).
Semi-denatured western blots with hippocampus lysates which were fractionated into formic acid from 8 months of age revealed that levels of total tau (tau-5) and tau-acK280 monomers and oligomers/aggregates were decreased in the brains of mice injected with 50 mg/kg of Y01 (Fig. 5C, D).

Discussion
There has been several anti-tau antibodies developed for AD and most of them target phosphorylated or non-modi ed tau (Sigurdsson, 2018). We here for the rst time provide anti-acetylated tau antibody as a potential AD therapeutic pipeline, which tau acetylation has been widely suggested to be a critical pathogenic modi cation in AD (Kontaxi et al., 2017). p300 acetyltransferase acetylates tau (Min et al., 2010), and p300 acetyltransferase activity is increased in AD brains according to severity (Wong et al., 2013;Aubry et al., 2015) and in tau-Tg mice brains (Min et al., 2015). Phosphorylated p300, the active form of p300, was observed in neurites and cytoplasm of neurons in AD hippocampus and frontal cortex (Wong et al., 2013;Aubry et al., 2015). Pathological activation of p300 inhibits autophagic ux leading to increased tau secretion (Chen et al., 2018). In AD brains, this may aggravate the pathological secretion and propagation of K280-acetylated tau. The pilot results of in vivo epitope screening (Additional le 1: Figure S1) underscore the importance of selecting the right epitopes to achieve therapeutic e cacy. Among our tested epitopes, pT231, ΔE391, acK280, and acK311, acK280 was the most therapeutically e cient epitope (Additional le 1: Figure S1). Since the e ciency of antibody formation against acK311 was too low (Additional le 2: Figure S2), we were unable to conclude that this epitope was a poor therapeutic target. However, this result may re ect the practical di culties in developing vaccines or antibodies targeting this epitope. Although we tried to select the therapeutically best epitope, our in vivo screening experiments have limitations that it is a small number pilot study and only covered a subset of AD-related pathological modi cations of tau. Hence, other more therapeutically effective tau epitopes may exist. Nevertheless, acK280 is an attractive target because it is an AD-related modi cation and promotes tau aggregation (Cohen et al., 2011); further, the hexapeptide 275 VQIINK 280 encompassing K280 residue is critical for tau aggregation (Mukrasch et al., 2005).
Tau protein amount in total cell lysates or tau aggregates were decreased by acK280 immunization or Y01 administration (Fig. 2-5), accompanied by decrease of acetylated or phosphorylated tau. This reduction of total tau or tau aggregates may contribute to the neuroprotective effect of targeting Tau-acK280. We think that potential mechanisms of tau reduction by Y01 include 1) inhibiting tau aggregation and propagation and 2) enhancing microglial tau clearance since we observed these effects in cell cultures (data not shown). Total tau amount in tau-Tg mice is a steady-state level of tau production, degradation and accumulation. Increased production of pathogenically mutated tau, P301L or P301S, in tau-Tg may hamper normal degradation of tau, augmenting a pathogenic release of aggregated tau and propagation to next neurons. This subsequently aggravates tau aggregation in next neurons increasing the half-life of tau, which may be a vicious cycle of tau accumulation. Anti-tau antibody like Y01 may stop this vicious cycle leading to decrease of total tau amount or tau aggregates by inhibiting propagation or aggregation or by enhancing tau clearance by microglia.
The structure of tau brils extracted from the AD brain at atomic resolution has recently been described (Fitzpatrick et al., 2017). Paired helical laments (PHFs) and straight laments (SFs) share a common C-shaped lament core comprising a series of β-sheets from R3 and R4 domains. K280 residue and the rest of R2 reside in the fuzzy coat and not in the bril's core. Hence, the K280 residue does not seem to be hindered from access by the ADEL-Y01 antibody. The nding that the R2 domain is not in the bril's core is surprising because the 275 VQIINK 280 segment is well-known to be important for tau aggregation (Mukrasch et al., 2005). However, lament structures in the AD brain only represent the end product of tau aggregation, which does not exclude the involvement of other parts of the protein in the initial formation and progression of laments. Eisenberg's team recently reported the structure of the 275 VQIINK 280 segment forming a tight steric zipper with strong intermeshed teeth to confer 275 VQIINK 280 with greater potency for tau aggregation than that of 306 VQIVYK 311 (Seidler et al., 2018). Inhibitors based on the structure of the 306 VQIVYK 311 segment only partially inhibit full-length tau aggregation and are ineffective at inhibiting seeding by full-length brils. However, structurally designed 275 VQIINK 280 inhibitor more potently inhibits tau aggregation and seeding induced by tau brils (Seidler et al., 2018). Hence, they suggest that the bril core itself is not the primary driver of seeding; rather, it serves as a scaffold that clusters 275 VQIINK 280 segments together in the fuzzy coat, thereby poising the 275 VQIINK 280 steric zippers for seeding. They also suggest that 275 VQIINK 280 localization to the fuzzy coat would make it more accessible to protein monomers in the cellular milieu, which in turn explains its prominent role in seeding and exquisite sensitivity to inhibition. Thus, ADEL-Y01 antibody could access the K280 residue which is critical for aggregation and seeding in the fuzzy coat and inhibit these pathogenic mechanisms.

Conclusions
Several therapeutic anti-tau antibodies are tested in clinical trials and some controversial issues have been raised about which epitope should be targeted in AD or other tauopathies (Courade et al., 2018). We anticipate Y01, the rst antibody targeting tau-acK280 of the key aggregation site 275