Abstract
The transmission of tau pathology has been proposed as one of the major mechanisms for the spatiotemporal spreading of tau pathology in neurodegenerative diseases. Over the last decade, studies have demonstrated that targeting total or pathological tau using tau antibodies can mitigate the development of tau pathology in tauopathy or Alzheimer's disease (AD) mouse models, and multiple tau immunotherapy agents have progressed to clinical trials. Tau antibodies are believed to inhibit the internalization of pathologic seeds and/or block seed elongation after seed internalization. To further address the mechanism of tau antibody inhibition of pathological spread, we conducted immunotherapy studies in mouse primary neurons and wild-type mice (females) seeded with AD patient-derived tau to induce the formation and spreading of tau pathology. Notably, we evaluated the effect of a mouse tau-specific antibody (mTau8) which does not interact with AD–tau seeds in these models. Our results show that mTau8 crosses the blood–brain barrier at levels similar to other antibodies and effectively decreases AD–tau-seeded tau pathology in vitro and in vivo. Importantly, our data suggest that mTau8 binds to endogenous intraneuronal mouse tau, thereby inhibiting the elongation of internalized tau seeds. These findings provide valuable insights into the possible mechanism underlying antibody-based therapies for treating tauopathies.
Significance Statement
The transmission of tau pathology plays key role in the pathoclinical progression of tauopathy. Studies have shown that tau antibody treatment can mitigate tau pathology in transgenic and spreading models of tauopathy. To explore the mechanisms involved in this procedure, we conducted immunotherapy studies on human tau seeds induced tau spreading models using a mouse tau-specific antibody (mTau8), which does not interact with human tau seeds. Our findings in the study enhance our understanding of antibody-based therapies for tauopathies.
Introduction
Aggregation of the microtubule-associated protein tau (tau) is associated with about 20 different neurodegenerative tauopathies, including Alzheimer's disease (AD; Kurihara et al., 2019). Notably, the progression of dementia and neurodegeneration in AD corresponds closely with the brain pathological tau burden (Barthelemy et al., 2020), suggesting the critical role of tau in disease pathogenesis. Emerging evidence shows that misfolded tau species act as pathogenic templates (seeds) that induce further tau aggregation under pathological conditions (Huseby et al., 2019). More specifically, the pathogenic tau seeds incorporate intracellular monomeric tau through an elongation process, resulting in the growth and extension of tau filaments (Friedhoff et al., 1998; Guo and Lee, 2014; Guo et al., 2016; Gibbons et al., 2019), which is believed to be facilitated by the posttranslational modifications of tau and diminishes normal tau interaction with microtubules (Huseby et al., 2019). The release of pathological tau seeds from inclusion-bearing neurons may occur via one or more mechanisms, leading to transsynaptic propagation of the tau pathology (Gibbons et al., 2019). Based on these findings, tau spreading models have been described in which treatment with human patient brain-derived pathogenic tau seeds induces tau pathology in neuron cultures and in animals expressing wild-type (WT) tau, subsequently simulating sporadic disease progression (Cornblath et al., 2021). In these tau spreading models, tau pathology is initiated after the internalization of tau seeds and the pathology gradually spread through different brain regions in vivo (Cornblath et al., 2021). The tau seeding models thus provide unique opportunities to investigate tau pathogenesis by quantitatively analyzing the spatiotemporal transmission of tau pathology.
Previous studies on transgenic mouse models demonstrated treatment with anti-tau antibodies can effectively reduce tau pathology in mice overexpressing human mutant tau (Asuni et al., 2007; Sankaranarayanan et al., 2015; Weisova et al., 2019; Corsetti et al., 2020; Samudra et al., 2023). In these studies, antibodies targeted different tau epitopes, including those recognizing linear epitopes that might interact with both soluble and misfolded tau, and others that were more selective for misfolded tau or tau with specific posttranslational modifications. Based on the results of these studies, proposed mechanisms of antibody-mediated reduction of tau pathology include decreasing the level of total tau, neutralizing the pathogenic tau seeds in the interstitial fluid, or inhibiting the intracellular elongation of tau seeds (Sandusky-Beltran and Sigurdsson, 2020). Tau-directed antibodies have also been tested in a mouse model with both Aβ and tau pathology, and targeting pathogenic tau seeds with confirmation-selective antibodies inhibited the spreading of tau pathology (Gibbons et al., 2020). It remains unclear if tau antibodies that bind to linear epitopes reduce tau pathology in mouse models by inhibiting tau seed uptake and/or through intracellular mechanisms, such as blocking the elongation of internalized seeds. To address this uncertainty, we designed a study utilizing the mouse tau-selective monoclonal antibody (Mab) mTau8 to inhibit tau spreading in neuron culture, WT mouse models, and a cell-free system, where initial seeding is with AD patient-derived insoluble tau seeds (AD–tau) or PSP patient-derived tau seeds (PSP–tau). Specifically, the epitope of mTau8, comprising amino acids 133–142 (133DRTGNDEKKA142) of mouse tau (Rodrigues Martins et al., 2023), does not interact with the human pathogenic tau seeds. Therefore, any potential effects on tau pathology would result from events downstream of initial AD–tau seeding. Our results show that treatment with mTau8 can reduce the AD–tau-seeded tau pathology in vivo and in vitro. Moreover, using live imaging microscopy with fluorescently labeled mTau8 antibody, we show that mTau8 enters neurons via clathrin-mediated endolysosomal pathways and binds to endogenous mouse tau at early time points before the formation of tau aggregates. These findings suggest that one mechanism of mTau8-mediated reduction of tau pathology is through binding to endogenous intracellular tau, with a slowing of elongation from internalized AD–tau seeds. These findings provide further information on the potential mechanisms of tau-directed antibodies that may apply to human tau immunotherapy.
Materials and Methods
Experimental design
To understand tau-based immunotherapy, we designed a study utilizing mouse tau-specific antibody mTau8 for an immunotherapy-like treatment to inhibit human tau seed-induced tau spreading in preclinical models of tauopathy. The antibodies were tested in mouse primary neuron culture, WT mouse model, and seed amplification systems seeded with patient-derived insoluble tau seeds in which the treatment antibody does not interact with the initial tau seeds. Analysis of the resulting pathology would be completed with immunocytochemistry, immunoblot, immunohistochemistry, and seed amplification assay. To measure the distribution of internalized mTau8, we also performed biochemistry, live images, and proximal ligation assay.
Purification of tau seeds
AD–tau was extracted from regions of postmortem patient brains as described in prior literature (Xu et al., 2021). Briefly, the human postmortem brain was homogenized in a glass douncer and sequentially fractionated with 1% Triton X-100 and 1% sarkosyl. The insoluble tau fraction was recovered and quantified for seeding experiments using immunoblots.
Tau spreading mouse model
C57BL6 WT mice were purchased at 2.5 months of age and used for these experiments. Animal care and experimental protocols were approved by the University of Pennsylvania's Institutional Animal Care and Use Committee. Twenty mice were anesthetized with ketamine/xylazine/acepromazine and fixed on a stereotaxic frame (David Kopf Instruments). The mice were unilaterally injected with AD–tau fibrils at the following stereotaxic coordinates: bregma, −2.5 mm; lateral, +2 mm; depth, −2.4 mm from the skull at the right dorsal hippocampus. Immunization with mTau8 or IgG2a control (n = 10) began 1 week after AD–tau injection and continued once a week for 4 months at a dose of 30 mg/kg. All the mice survived treatment and were killed at 4.5 months postinjection (4.5 mpi, 7-month-old) for immunohistochemical staining.
Immunohistochemistry
Immunohistochemical staining was completed according to a previously described protocol (Xu et al., 2021). Brains collected from the mice were embedded and sectioned following perfusion with phosphate-buffered saline (PBS) and immersion fixed 10% neutral-buffered formalin according to the protocol. Six micrometer sections were selected at intervals of 20 and were stained with the antibodies following treatments found in Table 1. Analysis of the immunoreactivity of glial markers, AD–tau-seeded tau pathology, and tau spreading in different brain areas was completed in the QuPath software (Bankhead et al., 2017). Analyses of the AT8-p-tau pathology and glial markers utilized 5 consecutive 6 µm brain sections at an interval of 20 sections for n = 10 mice per treatment group. Further analysis of AT8-p-tau pathology was mapped according to the Allen Brain Atlas connectome for a more detailed mapping of the connectome using the QUINT workflow for spatial analysis of features in histological images (Yates et al., 2019). For connectome mapping of tau pathology, mice were chosen based on the availability and intactness of the mouse brain sections (n = 3 per group).
Cell culture and seeding in vitro
Mouse neuronal cell cultures were generated following previously described protocol (Xu et al., 2021). CD1 mouse cortices and hippocampi were dissected at Embryo Day 16–18 and dissociated with papain (Worthington Biochemical). Neurons were resuspended in neural basal medium (Invitrogen, 21103) with 2% B27 (Invitrogen), 1× GlutaMAX (Invitrogen), and 1× Penicillin/Streptomycin (Invitrogen). Coverslips or plates were coated with poly-d-lysine (0.1 mg/ml, Sigma-Aldrich) overnight in borate buffer (0.05 M boric acid), pH 8.5, at room temperature. Cells were plated at a density of 50,000 cells/cm2 surface area. Transduction of AD–tau occurred at DIV 7 in all experiment at a concentration of 50 ng tau (0.5 μg tau/106 cells). Antibody treatments used include mTau8, IgG2a, PBS control, and negative control with no AD–tau lysate. Antibody treatments were administered at DIV 6 or DIV 8 with 10 µg antibody in 10 µl neuro basal medium (Invitrogen).
Immunoblots
At DIV 21, cells were harvested in PBS containing 1% sarkosyl and spun at 30 min at 100,000 × g in an ultracentrifuge. Sarkosyl-soluble protein of 15 µg and the same proportion of sarkosyl-insoluble fractions were loaded on 10% SDS–PAGE gel, transferred to 0.4 µm PVDF membrane, and probed with primary antibodies for immunoblots analysis. Secondary antibodies were incubated for 2 h at room temperature.
For dot blots, cells were washed with PBS and harvested in lysis buffer containing 1% Triton X-100 and proteinase inhibitor cocktail described in Xu et al. (2021). Samples were diluted to the desired concentrations in Tris-buffered saline and loaded on 0.45 mm PVDF membranes and blocked with 5% milk. Primary (R2295M, rabbit polyclonal mouse tau-specific antibody) and secondary antibodies (LI-COR Biosciences, IRDye 680RD goat anti-rabbit IgG secondary antibody) were incubated following the same protocol as for immunoblots. Photomicrographs were taken using an immunoblot imager (LI-COR Biosciences). Optical densities were measured using the Imagestudio software. The quantification standard curve was generated using purified recombinant mouse tau (2N4R).
Immunocytochemistry
At DIV 21, cell cultures (in 96-well plates) were washed with PBS, extracted with 1% hexadecyltrimethylammonium bromide (Sigma-Aldrich), and fixed with 4% paraformaldehyde and sucrose following previous protocol (Xu et al., 2021). Cell cultures were then stained with primary antibodies overnight at 4°C and with secondary antibodies for 2 h at room temperature. Images were taken with InCell Analyzer 2200 and analyzed with InCell Developer Toolbox 1.9.3. and Fiji software 2.9.0. For immunofluorescence costaining with proximity ligation assay (PLA)-labeled antibodies, EEA1 (Santa Cruz Biotechnology, sc-137130 AF647) and LAMP1 (BioLegend, 121608) antibodies were used.
Treatment antibodies
mTau8 and IgG2a control antibodies were developed and provided by Janssen Pharmaceuticals. The antibodies can be acquired upon request following the request policy of Janssen Pharmaceuticals.
Live imaging
mTau8 and IgG2a antibodies were fluorescently labeled with AnaTag HiLyte Flour 488 or 555 labels kit (AnaTag, AnaSpec). Antibodies were transduced at DIV 6 at a concentration of 10 μg/μl. Cells were quenched with trypan blue before imaging at DIV 8, 10, or 14. A live imaging microscope was used to take images at 37°C and 5% CO2 conditions within a closed, humidified chamber (Leica Microsystems).
Direct ELISA
Three hundred eighty-four MaxiSorp plates were coated overnight at 4°C with 30 μl of 2.5 µg/ml or 0.0625 µg/ml recombinant mouse tau (mT40) or recombinant human tau (hT40) and coating (Takeda Pharmaceuticals) buffer (350 ml 0.1 M NaHCO3, 650 ml 0.1 M Na2CO3 solution). Plates were washed with five washes of 100 µl PBS with Tween 20 (PBST) before incubation with Block Ace blocking solution (1% Block Ace powder, 0.05% NaN3 1× PBST 400 ml 0.2% BSA in PBS) at room temperature overnight. Thirty microliter of diluted CSF, plasma, and reference samples were added after removing Block Ace buffer for overnight incubation at 4°C. Plates were then washed with five washes of 100 µl PBST before incubation with 30 µl of 1:3,000 diluted rabbit anti-mouse horseradish peroxidase (HRP) for 1 h at 37°C. Following an additional five washes of 100 µl PBST, 30 µl of TMB substrate was added for a 10 min incubation before 30 µl of 10% phosphoric acid was added to stop the reaction. Plates were then read at 450 nm wavelength for absorbance on a plate reader (SpectraMax M5).
Tau sandwich ELISAs
The tau sandwich ELISAs were done using a previously published protocol (Crowe et al., 2020). Briefly, 384-well MaxiSorp plates (Thermo Fisher Scientific) were coated with 2.5 μg/ml of mTau8 and were blocked with 1% Block Ace solution in PBS. Cell extracts were diluted fivefold in a 0.2% BSA and PBS solution and incubated overnight. Plates were then washed in PBS and incubated with either biotinylated mTau8 for tau preformed fibrils (PFFs) or biotinylated mouse tau-specific Mab T49 or human tau-specific Mab T14 for the detection of different monomeric tau. Detection was done with a streptavidin–HRP-conjugated streptavidin.
CSF and blood plasma collection
C57BL6 WT mice were purchased at 2 months of age and were intraperitoneally injected with mTau8 or IgG2a antibodies at a dose of 30 mg/kg at 2.5 months of age. One week after injection, CSF and blood plasma samples were collected. These samples were then diluted for analysis in an ELISA. CSF samples were diluted to 1:50 and 1:100 for the measurement, and plasma samples were diluted to 1:4k, 1:8k, 1:16k, 1:32k, 1:64k, and 1:128k for the measurement. mTau8 dilution series at 25, 125, 625, 3,125, 15,625, 78,125, 390,625, and 1,953,125 pg/ml were used as references.
PLA
PLA was performed using the Duolink PLA kit. Neurons were washed with PBS and fixed with 4% PFA. The interaction of endogenous mouse tau with mTau8 antibody was probed with mouse tau-specific R2295M polyclonal antibody (rabbit) followed by the incubation of secondary antibodies in the kit. Images were taken using a fluorescent microscope (Leica DM6000).
Seed amplification assay (SAA)
The SAA was performed using a modified protocol (Metrick et al., 2019). Each reaction contained 10 mM HEPES, 300 mM sodium citrate, pH 7.4, 40 µM heparin, 10 µM ThT, 7.5 µM human K18 tau (4R repeated domain), and 3.75 µM full-length mouse tau (mT40, 2N4R). Cysteines on the monomers were blocked with N-ethylmaleimide before use. The reaction mix was preincubated for 48 h with mTau8, IgG2a, or BSA at a 1:10 stoichiometric ratio to mT40. Sarkosyl-insoluble tau of 20 pg isolated from a PSP patient was diluted in 0.5% tau WT mouse brain lysate containing HEPES buffer. Reactions were performed in a beaded 384-well plate at 37°C shaking at 500 rpm, and ThT fluorescence was measured every 15 min for 60 h. As the insoluble PSP tau comprises ∼1% of the total sarkosyl-insoluble protein preparation, 2 ng of nondisease control brain sarkosyl-insoluble lysate was used as the “no seeds” control to match the amount of total brain protein.
Data analysis and statistics
Unpaired t tests were performed for comparisons of two experimental groups. For comparisons involving more than two groups, one-way ANOVA tests were performed, followed by Tukey post hoc multiple-comparison testing if not specified. The GraphPad Prism software (GraphPad Software) was used to perform all statistical testing. Statistical significance was determined as p < 0.05. Unless specified otherwise, data were presented in the format of mean ± standard deviation.
Results
mTau8 is sensitive and specific to mouse tau
To test the sensitivity of the mTau8 antibody, we first tested the binding of varying concentrations of the mTau8 antibody to full-length mouse tau 2N4R (mT40) in an ELISA in which mT40 was immobilized on wells at two concentrations. At 2.5 µg/ml of mT40 coating concentration, the EC50 value for mTau8 antibody binding was 11.1 ± 2.3 ng/ml. At a lower coating concentration of mT40 (0.61 µg/ml), the direct ELISA yielded a similar mTau8 binding EC50 of 13.2 ± 2.5 ng/ml (Fig. 1A). To evaluate the specificity mTau8 antibody for mouse tau, we compared the binding of mTau8 and HT7, a well-established anti-human tau Mab, with mouse and human tau. ELISA plates were coated with identical amounts of mT40 or human 2N4R tau (hT40), and varying concentrations of mTau8 or HT7 antibodies were added to evaluate binding. These results showed that mTau8 had high specificity for mT40 (EC50 = 20.5 ± 3.6 ng/ml), with minimal affinity for hT40 as the response of the direct ELISA did not exceed background levels. In contrast, HT7 exhibited a high affinity for hT40 (EC50 = 12.5 ± 1.79 ng/ml) and reduced affinity for mT40, as demonstrated by an incomplete binding curve (Fig. 1B,C).
To assess mTau8 specificity toward tau fibrils, we modified a previously established sandwich ELISA system (Crowe et al., 2020) in which mTau8 was used as the capture antibody and incubated with either monomeric or PFFs of hT40 or mT40. Captured tau species were then probed with either HRP-conjugated T49 anti-mouse tau Mab or T14 anti-human tau Mab, which have different epitopes than mTau8. Consistently, our results show that mTau8 can capture both monomeric and oligomeric forms of mT40 with no detectable capture of either hT40 monomers or PFFs (Fig. 1D,E). These results reveal that mTau8 can specifically bind monomeric and fibrillar forms of mouse tau with no binding to human tau seeds.
Relative mTau8 antibody exposure in the central nervous system after peripheral administration
To test the relative brain exposure of mTau8 after peripheral administration, we intraperitoneally injected C57BL6 WT mice with mTau8 or mouse IgG2a as the control, collected CSF and blood plasma 1 week postinjection, and assessed the amount of mTau8 in these biofluids (Fig. 2A). We utilized the mT40 direct ELISA for the measurement of mTau8 antibody in the injected mice (Fig. 2A) and detected a level of 3.74 μg/ml of mTau8 in the plasma of mTau8-injected mice by average. As expected, no ELISA signal was detected in the IgG2a-injected mice (Fig. 2B). There was an average of 0.015 μg/ml of mTau8 detected in the CSF of the mTau8-injected mice, whereas no signal was observed in the IgG2a-injected mice (Fig. 2C). The concentrations of mTau8 in the CSF corresponded to 0.4% of the plasma level, which is in the range typically observed for IgG distribution into the brain.
mTau8 reduces tau spreading in AD–tau-seeded WT mice
We next investigated if targeting mouse tau with mTau8 would inhibit the transmission of tau pathology in vivo after seeding with pathogenic AD–tau. To achieve this, we treated 10 WT mice per group with IgG2a or mTau8 antibody after initiation of tau pathology through intracerebral injection of AD–tau. We biochemically enriched AD–tau seeds from the postmortem brain of AD patients and stereotaxically injected the AD–tau (unilaterally in the hippocampus) into the WT mouse brain. Prior research showed that the injected human tau seeds are no longer detectable with anti-human tau antibody (HT7) 7 d after the injection (Xu et al., 2021), which suggests that the initial endocytosis of tau seeds is largely completed by 7 d postinjection time. Therefore, to allow the seeding of tau pathology to start before the antibody treatment, mTau8 antibody or mouse isotype-matched IgG2a control was administrated weekly at a dose of 30 mg/kg beginning 7 d after the AD–tau injections for a total of 16 weeks (Fig. 3A).
The amount and distribution of tau pathology were evaluated in the AD–tau-injected mice through staining of brain sections with the AT8 phospho-tau antibody. In mouse brain regions with highest amount of tau pathology, including the ipsilateral hippocampus, contralateral hippocampus, and ipsilateral entorhinal cortex, mice treated with IgG2a were found with a higher amount of AT8-positive tau pathology than the mTau8-treated mice (Fig. 3B,C). The same trend was also confirmed using the MC1 antibody that binds misfolded tau (Fig. 3D,E) and the AT180 phospho-tau antibody (Fig. 3F,G), with reduced staining in the ipsilateral hippocampus and the entorhinal cortex regions of the mTau8-treated mice compared with those treated with control IgG2a. Altogether, these results suggest that peripheral administration of mTau8 antibody can effectively inhibit the AD–tau-seeded tau pathology in the WT mouse brain.
The spread of tau pathology appears to be primarily driven by the connectivity of neurons (Cornblath et al., 2021), and we have previously shown that AD–tau-induced tau pathology increases linearly for up to 12 months of postinjection time (Xu et al., 2024). To further explore the connectome-dependent changes of tau pathology, we analyzed with published data set of connectivity strength of brain regions to the hippocampal injection site of AD–tau seeds (Oh et al., 2014). Ninety-two of these regions were identified as being in the first-order anterograde connectome relative to the injection site at the hippocampus. The remaining 102 regions are included as the second-order connectome without direct connection (Extended Data Fig. 4-1). Nearly all regions in mTau8-treated mice were found to have decreased AT8-positive tau pathology regardless of connectivity strength (Fig. 4A–C), suggesting a global reduction of tau pathology (Fig. 4C). On average, there was an ∼87.2% overall decrease in pathology in mTau8- versus IgG2a-treated mice. The reduction of tau pathology can be further confirmed by comparing the distribution of tau pathology (mTau8/IgG2a treatment ratio) in 10 regions with the highest and lowest connectivity strength in the first-order connectome, where no difference was observed between areas with high and low connectivity strength (Fig. 4D,E). Interestingly, we also observed a significant reduction of tau pathology in the second-order connectome, suggesting that mTau8 reduces tau spread through both the primary and secondary connectome (Extended Data Fig. 4-1).
Figure 4-1
A. Heatmap shows Quint workflow percentage of AT8 positive areas in the 2nd order connectome in mice treated with IgG2a and mTau8. Each lane represents one individual mouse. B. Ratio of change in AT8 tau pathology in mTau8- and IgG2a-treated mice in the 2nd order connectome. Download Figure 4-1, TIF file.
The presence of tau pathology is associated with increased gliosis and neuroinflammation (Leyns and Holtzman, 2017). To test if the treatment of mTau8 affects gliosis, we immunohistochemically stained mouse brains for astrocytes and microglia. A 23.4% decrease in GFAP-positive astrocyte staining was found in the mTau8-treated mice compared with those receiving control IgG2a (Fig. 5A,B). Similarly, IBA1-positive microglial immunoreactivity also decreased by 19.8% in mTau8-treated mice (Fig. 5C,D), suggesting a decrease in gliosis and neuroinflammation in the mouse brain that likely corresponds to the reduction of tau pathology in the mTau8-treated mice.
mTau8 reduces tau spreading in vitro
To explore the mechanism of mTau8 inhibition of seeded tau pathology, we transitioned to in vitro studies using primary neuron cultures. Mouse primary neurons were treated with mTau8 or IgG2a 24 h before (DIV 6) or 24 h after (DIV 8) transduction of the cultures with AD–tau seeds (DIV 7; Fig. 6A). In both conditions, mTau8 treatment significantly decreased induced mouse tau pathology detected via immunocytochemistry when compared with IgG2a control treatment (Fig. 6B,C). mTau8 treatment caused a 23.2% mean reduction in the amount of tau pathology when added at DIV 6 and a 22.1% mean reduction when treatment was started at DIV 8 (Fig. 6C). Meanwhile, DAPI counts showed no difference between treatment conditions, indicating an absence of cytotoxicity associated with the mTau8 and IgG2a treatments (Fig. 6D; Extended Data Fig. 9-1A). To further confirm a reduction of tau pathology, cell lysates were separated into sarkosyl-soluble and sarkosyl-insoluble fractions, and the amount of tau species in each of them was measured by immunoblotting with R2295M, a rabbit polyclonal anti-mouse tau antibody. Insoluble mouse tau was significantly reduced in the mTau8-treated neurons compared with PBS or IgG2a-treated controls (Fig. 6E). No difference was observed in the amount of sarkosyl-soluble tau across all treatments (Fig. 6F), suggesting the antibody treatment did not cause changes in the total mouse tau level. Altogether, our results show that mTau8 treatment inhibits the AD–tau-seeded tau pathology in primary neurons in vitro.
mTau8 enters neurons via the endocytic pathway and engages with endogenous tau in vitro
To study the distribution of the mTau8 antibody in the primary neuron culture, we fluorescently labeled mTau8 and IgG2a antibodies with amine-reactive dyes (Hylite 555 or 488) and treated mouse primary neurons with the labeled antibodies at DIV 6. Live imaging was performed at DIV 8, 10, and 14 in the presence of 500 μM trypan blue, which acts as a quenching agent for extracellular fluorescence (Fig. 7A; Karpowicz et al., 2017). Our results show that both mTau8 and IgG2a can be visualized within neurons at the DIV 8–10 time points (1 and 3 d after antibody addition), with a time-dependent increase in the amount of internalized antibodies (Fig. 7B).
Prior studies indicated that antibodies may enter neurons via clathrin-dependent pathways (Congdon et al., 2013). To test whether the internalization of antibodies we observed was by this mechanism, we cotreated the neurons at DIV 7 with fluorescently labeled mTau8 (Hylite 488) and transferrin (Alexa Fluor 594), which is known to enter cells via the clathrin-dependent pathway. Live imaging of the neuron cultures on DIV 7 (1 h after antibody addition) and DIV 10 (3 d after antibody addition) showed significant colocalization of antibodies with transferrin (Fig. 8A,B). These results suggest that mTau8 enters neurons through the clathrin-dependent pathway. We postulated that some fraction of the internalized antibodies might ultimately escape from endocytic vesicles or downstream vesicular compartments such as lysosomes and gain access to endogenous tau in the cytoplasm. To test this hypothesis, we used PLA to assess whether mouse tau in neurons might interact with internalized mTau8 antibody. In this experiment, neurons were treated with mTau8, and the cultures were then fixed at different time points. A rabbit polyclonal anti-mouse tau antibody (R2295M) was then used to probe for endogenous mouse tau, where positive PLA staining would indicate the tau antibody and internalized mTau8 were in proximity (Fig. 9A). The PLA assay shows that mTau8 engages with mouse tau as early as 1 h posttreatment with the antibody (Fig. 9B,D) and their interaction increases with time up to 3 d after antibody addition (Fig. 9B,D). TRIM21 was previously reported to be involved in degrading tau in the presence of anti-tau antibodies (McEwan et al., 2017; Mukadam et al., 2023). To test if the mTau8 antibody might inhibit tau seeding via interaction with TRIM21, we also did PLA with a rabbit polyclonal anti-TRIM21 antibody. In general, the PLA revealed an increase of mTau8–TRIM21 interactions, especially at the 1 h post-transduction time point (Fig. 9C,E), which further confirmed the internalization of mTau8 antibody in the neurons. However, the mTau8–TRIM21 interaction did not further increase with time, unlike the mTau8–tau interaction, which might suggest that TRIM21 is not facilitating tau degradation in the mTau8 treatment. More generally, these results suggest that mTau8 enters neurons via the endocytic pathway and ultimately gains access to endogenous tau in the neurons.
Figure 9-1
A. Mouse primary neurons were treated with PBS, mTau8, or IgG2a for 3 days from DIV 7 to DIV 10 and analyzed for DAPI counts. One-way ANOVA with a Tukey post hoc test revealed no significant differences between the treatment groups. B. Representative images of mTau8/R2295 M proximity ligation assay (PLA) co-stained with LAMP1 and EEA1. Mouse primary neurons were treated with IgG2a or mTau8 antibodies at DIV 7. Following fixation at DIV10 (3-days post-treatment), staining was completed with R2295 M anti-mouse tau antibody (rabbit polyclonal) for PLA, LAMP1(lysosomal) antibodies, and EEA1 (early endosomal) antibodies. Arrowhead highlights the colocalized signals of PLA and LAMP1. Arrow highlights the colocalized signals of the PLA and EEA1. C. Quantification of PLA-positive staining in mouse primary neurons treated with mTau8, IgG2a, or PBS. P values were determined with one-way ANOVA and a Tukey post hoc test n = 3, 6, and 4 for PBS, IgG2a, and mTau8 treatments. D. Measurements of PLA and LAMP1/EEA1 colocalization in B in mTau8 treated conditions. Directionality of Mander’s Coefficient is listed under each column. P values were determined with one-way ANOVA and a Tukey post hoc test. n = 6 and 4 for IgG2a and mTau8. Download Figure 9-1, TIF file.
Figure 9-2
A. Schematic of the experimental paradigm in which mouse primary hippocampal neurons were treated with IgG2a or mTau8 antibodies at DIV 7, followed by washing with PBS on DIV 10 to remove extracellular tau and mTau8. Cells were harvested for measurement of intracellular mouse tau and mTau8 with dot blot or mTau8 direct ELISA. B. Dot blot measurements of endogenous mouse tau in mouse primary neurons. Immunoreactivity was revealed with R2295 M antibodies and fluorescently labeled goat anti rabbit secondary antibodies. P values were calculated by one-way ANOVA with a Tukey post hoc test. n = 4 - 5 per group. C. mTau8 direct ELISA measurements of intracellular mTau8 in primary neuron cultures. P values were calculated by one-way ANOVA followed by Tukey post hoc test. n = 4 - 5 per group. D. Seeding amplification assays (SAA) were performed with tauopathy patient derived tau seeds, K18 (4R repeated domain), and recombinant mouse tau (2N4R, mT40). mT40 samples were pre-incubated with either mTau8, IgG2a, (Bovine Serum Albumin) BSA or HEPES buffer for 48 hours. CBL = non-disease control brain lysate. E. Quantifications of the area under curve generated by the SAA in D. Values were normalized to HEPES treated samples as a control. P values were calculated by one-way ANOVA followed by Tukey post hoc test between the indicated groups. n = 4 per group. Download Figure 9-2, TIF file.
To determine the cellular compartment where the mTau8/mouse tau interaction may occur, primary neurons treated with PBS, mTau8, or IgG2a were fixed and subjected to tau–mTau8 PLA with immunofluorescence costaining using LAMP 1 and EEA1 antibodies (Extended Data Fig. 9-1B). As expected, PLA signals are predominately found in the mTau8-treated cells (Extended Data Fig. 9-1C). Colocalization analysis via Mander's coefficient (EEA1/LAMP1 to PLA) shows that PLA signals are better colocalized with EEA1 staining than with LAMP1 in the cells, with the majority of EEA1-positive vesicles having PLA colocalization (Extended Data Fig. 9-1D). However, the majority of the PLA staining does not overlap with either EEA1 or LAMP1 staining (PLA àEEA1/LAMP1), indicating that most tau–mTau8 interactions occur outside of the endolysosomal system (Extended Data Fig. 9-1D), which is consistent with the previous observations (Congdon et al., 2013).
Intracellular mTau8 inhibits tau seeding in vitro
To determine if the intracellular mTau8 can affect tau seeding in the primary neuron, we pretreated neurons at DIV 3 with mTau8 or IgG2a antibodies and then depleted the extracellular antibodies 3 d later (DIV 6) by changing the culture medium (Fig. 9F). A parallel neuronal culture had mTau8 or IgG2a antibodies added at DIV 6 without subsequent antibody removal. The cells were then treated with AD–tau at DIV 7 and analyzed for mouse tau aggregates 14 d later (DIV 21). As expected, the neurons in which mTau8 was added at DIV 6 without antibody removal had a significant reduction of tau pathology (Fig. 9G). Notably, mouse tau pathology was also significantly reduced in neurons pretreated with mTau8 followed by antibody removal so that mTau8 antibody are only present intracellularly (Intra mTau8; Fig. 9G), suggesting that the endocytosed mTau8 can inhibit the seeding of AD–tau in the primary neuron model.
To determine the stoichiometric ratio of endogenous mouse tau to intracellular mTau8 antibody, we treated mouse primary neurons with mTau8, IgG2a, or PBS. The cells were washed with PBS 3 d later, and cell lysates were prepared to measure the amount of mouse tau using dot blot and mTau8 using direct ELISA (Extended Data Fig. 9-2A). Dot blot measurement of mouse tau showed no significant difference between cells treated with mTau8, IgG2a, and PBS (Extended Data Fig. 9-2B). On average, mTau8-treated samples had a 2.8 nM concentration of mouse tau in the cell lysate (note that this does not represent the intracellular tau concentration due to dilution of lysate preparation). Direct ELISA measurement of mTau8 in the same cell lysates demonstrated the presence of intracellular mTau8 with an average 0.31 nM concentration (Extended Data Fig. 9-2C). When taken together these results estimate a roughly 1:9.1 ratio of mTau8 to mouse tau harvested from the cell culture. Assuming an intracellular mechanism of mTau8 inhibition of tau seeding, these data suggest that a substoichiometric amount of mTau8 is sufficient to reduce the formation of tau aggregates. However, the determination of the mTau8:mouse tau ratio in the cellular lysates does not distinguish between free tau from other tau which might not be accessible to AD–tau seeds or mTau8 antibody due to compartmentalization (e.g., bound to microtubules). Thus, the actual ratio of mTau8 to accessible tau substrate could be higher than that suggested by total tau measurements.
Utilizing a similar stoichiometric ratio of mTau8 to mouse tau, we performed a SAA with tauopathy patient-derived tau seeds and full-length mouse 2N4R tau (mT40). The mT40 was preincubated with mTau8, IgG2a, BSA, or HEPES buffer (reaction buffer of SAA) for 48 h at a 1:10 ratio. The SAA showed that the preincubation of mTau8 can partially inhibit the fibrillization of mT40 when compared with other conditions (Extended Data Fig. 9-2D).
Discussion
In the study, we report mTau8 Mab that specifically targets mouse tau on reducing tau pathology in models of seeded tau aggregation both in WT mice and neuron cultures. Our biochemical results confirm that mTau8 is highly specific to mouse tau with little to no affinity toward human tau, and administration of mTau8 to WT mice revealed brain exposure (0.4% of plasma levels) that is typical of IgG molecules (Gibbons et al., 2020). When compared with other studies that tested antibody treatment as a method to reduce tau pathology in preclinical models, our studies are unique in exclusively targeting endogenous tau as a possible therapeutic strategy. These prior studies demonstrated a reduction in pathology in mouse models overexpressing mutant human tau or with the use of antibodies that do not distinguish between the tau seeds and other forms of tau (d'Abramo et al., 2013; Schroeder et al., 2017; Gibbons et al., 2020). Our experiments leveraged the use of a tau spreading model in which pathology was induced by intracerebral seeding of WT mouse brains with human AD–tau seeds. In our experimental design, the mTau8 antibody specifically targets endogenous tau under a “post pathology onset” paradigm, wherein the seeding of tau pathology by AD–tau begins before antibody treatment and with the knowledge that mTau8 cannot directly interact with the exogenous human tau seeds. Thus, any beneficial effect of mTau8 treatment would be due to events that occur after AD–tau internalization into neurons. Utilizing the AD–tau-seeded WT mouse model, we found that treatment with the mTau8 antibody led to an 87% reduction in tau pathology. Interestingly, the magnitude of tau pathology reduction observed in this study is relatively robust when compared with the efficacy of other tau-targeting antibodies in vivo (Sankaranarayanan et al., 2015; Schroeder et al., 2017; Gibbons et al., 2020). This may relate to slower and more physiological seeding dynamics of tau pathology in WT mice compared with transgenic mouse models with tau overexpression (He et al., 2018; Xu et al., 2024).
On the connectome level, an effect of mTau8 on the early steps of seeding might be expected to reduce all subsequent spread of tau pathology. Conversely, if the effect of antibody treatment was primarily on the inhibition of secondary seeding, one would expect more distant connections that depend on at least one synaptic transmission to be more affected by mTau8 treatment. An analysis of the tau spread after mTau8 treatment demonstrates a global decrease in tau pathology, suggesting a mechanism in which the antibody interferes with the early seeding of endogenous tau such that all subsequent primary and secondary retrograde and anterograde seeding is reduced, as well as any connectome-independent secondary seeding.
Further characterization of mTau8 action was conducted with a primary neuronal cell culture model, which revealed that mTau8 could also inhibit the formation of seeded tau pathology in this in vitro model. Live imaging studies revealed that labeled mTau8 was internalized into the same neuronal compartments as labeled transferrin within 24 h of treatment, indicating a clathrin-dependent uptake mechanism. This observation is consistent with previous literature suggesting that antibodies can be readily internalized into the cells (Shamir et al., 2020; Congdon et al., 2022). One possible pathway involves IgG antibody binding to FcγR receptors that are then internalized via clathrin-coated vesicles (Fuller et al., 2014; Andersson et al., 2019). FcγR receptors are expressed in various cell types in the brain, including microglia, neurons, astrocytes, oligodendrocytes, and endothelial cells (Fuller et al., 2014; Zhao et al., 2023). Notably, previous studies have reported FcyR-dependent uptake of antibodies in rodent neuron models (Masliah et al., 2005; Congdon et al., 2013; Kondo et al., 2015), including both anti-tau and anti-α-synuclein antibodies. Moreover, the blockage of the FcyR receptor prevents the endocytosis of treatment antibodies (Kondo et al., 2015). Interestingly, our neuron culture PLA studies revealed that the mTau8 antibody binds to endogenous mouse tau within 24 h after antibody addition. Since mTau8 does not bind to the exogenous AD–tau used to seed pathology in the neuron cultures, the internalization of the antibody and its subsequent interaction with intracellular mouse tau may prevent the recruitment of mouse tau by the AD–tau seeds, thereby blocking the formation of neuronal inclusions. In this regard, our estimation of a 1:9 mTau8-to-tau intraneuronal ratio in treated neuronal cultures and the relatively low brain exposure of mTau8 (0.4% of plasma levels) suggest a substoichiometric amount of mTau8-to-total tau is sufficient to reduce seeded tau aggregation. However, tau in neurons is largely bound to microtubules, and thus only a fraction of the total neuronal tau is likely ad libitum available to engage with AD–tau seeds or internalized mTau8 antibody. It is unclear which compartments within neurons AD–tau might first seed fibril elongation by endogenous mouse tau, but it is possible that the effective tau concentration at these locales is relatively low, such that sufficient mTau8 antibody is available to impact seeded fibrillization.
We also cannot fully exclude the possibility that mTau8 may also bind to released mouse tau aggregates at later time points and contribute to the reduction in tau pathology via neutralizing secondary seeding in the mouse models. Notably, TRIM21 receptor-mediated degradation of tau was recently proposed as one of the major mechanisms by which tau antibodies neutralize their targets (McEwan et al., 2017; Mukadam et al., 2023). We observed an initial mTau8/TRIM21 interaction in the PLA assay that then decreased over time in culture (Fig. 9E). When compared with the mTau8/mouse tau interactions, the number of mTau8/TRIM21 interactions appeared to be much less frequent, suggesting that TRIM21-mediated degradation may not play a prominent role in the mTau8 inhibition of tau inclusion formation. These results are in line with the observation that soluble tau was not significantly reduced in primary neurons treated with mTau8, as would be the case if TRIM21 promoted degradation of mTau8–tau complexes. The difference in results between the current study and the prior publication on TRIM21 may relate to differences in the experimental models or antibody mechanisms.
Altogether, the results presented here utilizing the mTau8 antibody suggest that endogenous intracellular tau may be a viable immunotherapy target that can be engaged to potentially slow the spread of tau pathology and disease progression. If this proposed mechanism is correct, it could mitigate concerns related to the targeting of released misfolded tau species that are thought to propagate the spread of tau pathology. For example, there is some evidence that the spread of tau pathology may occur through the release of exosomes or other extracellular vesicles that harbor misfolded tau (Ruan et al., 2021; Zhao et al., 2023). If true, antibodies might not be able to access tau captured within these extracellular vesicles. In contrast, an intraneuronal mechanism of tau antibody action would not depend on binding to released tau species. Future studies are required to further substantiate a proposed intracellular mechanism of tau antibody action, and further confirmation of this mode of action would have important implications for tau-based immunotherapy.
Footnotes
We thank J. Robinson, T. Schuck, and H. for their technical assistance. This work was supported by National Institute on Aging (NIA) U19 AG062418 and R01 AG076434 and CurePSP Venture Grant 677-2021-12.
The authors declare no competing financial interests.
- Correspondence should be addressed to Hong Xu at hongxu{at}upenn.edu.