Tau amyloid corona-shelled nanoparticle-based drug screening platform for discovering tau oligomer-degrading drugs

Tau amyloid corona-shelled nanoparticle-based drug screening platform for discovering tau oligomer-degrading drugs | Authorea try { document.documentElement.classList.add('js'); } catch (e) { } var _gaq = _gaq || []; _gaq.push(['_setAccount', 'G-8VDV14Y67G']); _gaq.push(['_trackPageview']); (function() { var ga = document.createElement('script'); ga.type = 'text/javascript'; ga.async = true; ga.src = ('https:' == document.location.protocol ? 'https://ssl' : 'http://www') + '.google-analytics.com/ga.js'; var s = document.getElementsByTagName('script')[0]; s.parentNode.insertBefore(ga, s); })(); Skip to main content Preprints Collections Wiley Open Research IET Open Research Ecological Society of Japan All Collections About About Authorea FAQs Contact Us Quick Search anywhere Search for preprint articles, keywords, etc. Search Search ADVANCED SEARCH SCROLL Aggregate This is a preprint and has not been peer reviewed. Data may be preliminary. 19 March 2025 V1 Latest version Share on Tau amyloid corona-shelled nanoparticle-based drug screening platform for discovering tau oligomer-degrading drugs Authors : Hyo Gi Jung , Dongsung Park 0000-0001-5235-7114 , Junho Bang 0009-0000-9749-6513 , Yeon Ho Kim , Jae Won Jang , Yonghwan Kim , Hyunji Kim , … Show All … , Seungmin Lee , Wonbin Moon , Kyo Seon Hwang , Jeong Hoon Lee , Dongtak Lee , and Dae Sung Yoon [email protected] Show Fewer Authors Info & Affiliations https://doi.org/10.22541/au.174235163.36542009/v1 Published Aggregate Version of record Peer review timeline 424 views 235 downloads Contents Abstract Supplementary Material Information & Authors Metrics & Citations View Options References Figures Tables Media Share Abstract Tau oligomers are recognized for their critical role in causing neuronal toxicity and synaptic dysfunction in a diverse array of neurodegenerative diseases collectively referred to as tauopathies. However, the discovery of drugs that specifically target tau oligomers has been impeded by the absence of appropriate screening methods. Here, we suggest a drug screening platform based on tau amyloid corona-shelled nanoparticles (TACONs) to assess the efficacy of tau oligomer-degrading compounds through aggregation-induced colorimetric responses of TACONs. TACONs were engineered via the encapsulation of gold nanoparticles (AuNPs) with homogeneous tau oligomers by leveraging heparin as a co-factor. Our TACON-based strategy harnesses two primary functions of AuNPs: (i) catalytic activators for the selective isolation of tau oligomers and (ii) optical reporters for quantifying colorimetric responses induced by tau oligomer-degrading agents. To validate this approach, we employed proteases that can degrade tau oligomers (protease XIV and plasmin) along with various small molecules known to aid in the treatment of tauopathies. Furthermore, we significantly enhanced screening efficiency by integrating a time-series deep learning architecture, enabling rapid identification of effective compounds within one hour. These results highlight the great potential of deep learning-assisted TACON-based drug screening platform as a powerful strategy for streamlining drug discovery in tauopathies. Tau amyloid corona-shelled nanoparticle-based drug screening platform for discovering tau oligomer-degrading drugs Hyo Gi Jung a,b,# , Dongsung Park a,h,# , Junho Bang a,b,# , Yeon Ho Kim a,b , Jae Won Jang a,b , Yonghwan Kim a,b , Hyunji Kim a,b , Seungmin Lee a,d,e , Wonbin Moon a,b , Kyo Seon Hwang c , Jeong Hoon Lee d,e, *, Dongtak Lee f,g, *, and Dae Sung Yoon a,b,i, * a School of Biomedical Engineering, Korea University, Seoul 02841, South Korea b Interdisciplinary Program in Precision Public Health, Korea University, Seoul 02841, South Korea c Department of Clinical Pharmacology and Therapeutics, College of Medicine, Kyung Hee University, Seoul 02447, South Korea d KU-KIST Graduate School of Converging Science and Technology, Korea University, 145 Anam-ro, Seongbuk-gu, Seoul 02841, Korea e Department of Integrative Energy Engineering, College of Engineering, Korea University, 145 Anam-ro, Seongbuk-gu, Seoul 02841, Korea f Division of Bioengineering, College of Life Sciences and Bioengineering, Incheon National University, Incheon, 22012, Republic of Korea g Division of Bioengineering, Incheon National University, Incheon, 22012, Republic of Korea h Center for Functional Nanomaterials, Brookhaven National Laboratory, Upton, NY 11973, USA i Astrion Inc, Seoul 02841, South Korea * Corresponding authors: D.S.Y. ( [email protected] ); D.L. ( [email protected] ); J.H.L. ( [email protected] ) # These authors contributed equally to this work Tau oligomers are recognized for their critical role in causing neuronal toxicity and synaptic dysfunction in a diverse array of neurodegenerative diseases collectively referred to as tauopathies. However, the discovery of drugs that specifically target tau oligomers has been impeded by the absence of appropriate screening methods. Here, we suggest a drug screening platform based on tau amyloid corona-shelled nanoparticles (TACONs) to assess the efficacy of tau oligomer-degrading compounds through aggregation-induced colorimetric responses of TACONs. TACONs were engineered via the encapsulation of gold nanoparticles (AuNPs) with homogeneous tau oligomers by leveraging heparin as a co-factor. Our TACON-based strategy harnesses two primary functions of AuNPs: (i) catalytic activators for the selective isolation of tau oligomers and (ii) optical reporters for quantifying colorimetric responses induced by tau oligomer-degrading agents. To validate this approach, we employed proteases that can degrade tau oligomers (protease XIV and plasmin) along with various small molecules known to aid in the treatment of tauopathies. Furthermore, we significantly enhanced screening efficiency by integrating a time-series deep learning architecture, enabling rapid identification of effective compounds within one hour. These results highlight the great potential of deep learning-assisted TACON-based drug screening platform as a powerful strategy for streamlining drug discovery in tauopathies. Keywords: Tau oligomers, Tauopathies, Colorimetric drug screening platform, Aggregation, Tau-oligomer degrading agents, Deep learning . 1. Introduction Tau, a non-catalytic structural protein, is a microtubule-associated protein (MAP) predominantly expressed in neurons that stabilizes axonal microtubules and facilitates tubulin assembly [1] . However, under pathological conditions, tau undergoes hyperphosphorylation, thereby disassociating from microtubules and facilitating the accumulation of misfolded tau in the cytoplasm [2] . Misfolded tau self-assembles into insoluble aggregates known as neurofibrillary tangles (NFTs), which are the hallmarks of several neurodegenerative diseases, including Alzheimer’s disease (AD) [3] , chronic traumatic encephalopathy [4] , and frontotemporal dementia [5] , collectively known as tauopathies. Recently, growing evidence has implicated small soluble tau oligomers as key players, rather than insoluble tau aggregates such as NFTs or paired helical filaments (PHFs), as primary culprits in the pathogenesis of tauopathies [6] . Consequently, the inhibition of tau oligomerization using peptides [7] , small molecules [8] , and antibodies [9] has been proposed as a therapeutic strategy for tauopathies. However, the inter- and intracellular transmission of tau oligomers, which further propagate pathological aggregation, hinders drug development [10] . Therefore, anti-tau drugs should specifically target and eliminate existing tau oligomers in patients suffering from tauopathies. Numerous studies have explored the identification of drugs that degrade tau aggregates through conventional approaches such as fluorescence spectroscopy with chemical dyes [11] , fluorescence resonance energy transfer (FRET) in cell-based assays [12] , and molecular dynamics simulations [13] . However, these methods have faced significant challenges in achieving precise and rapid drug screening owing to their low reproducibility [14] , need for additional preprocessing steps [15] , and limited accuracy [16] . In particular, the identification of hit compounds that specifically target tau oligomers has been more difficult because of the absence of an adequate tau oligomer-specific screening model. The kinetic instability of toxic tau oligomers frequently results in their rapid transformation into larger tau aggregates and fibrils, rendering the arrest of tau proteins in their oligomeric states challenging [17] . The transient nature of tau oligomers significantly complicates their isolation, making it difficult to assess the efficacy of tau oligomer-targeting drugs. Herein, we developed tau amyloid corona-shelled nanoparticles (TACONs) comprising a plasmonic gold nanoparticle (AuNP) uniformly encapsulated by 4~5 nm size heparin-induced tau oligomers (Figure 1A). The conformational and physiochemical characteristics of the amyloid corona on TACONs were analyzed by graphene transistor-based immunoassay, transmission electron microscopy (TEM), X-ray photoelectron spectroscopy (XPS), and dynamic light scattering (DLS). Effective reagents degrade the tau amyloid corona on the TACONs, exposing the AuNP surface and making the AuNP unstable. This process induces particle aggregation, causing aggregation-induced colorimetric changes in the TACON solution from red to purple (Figure 1B). Conversely, ineffective reagents fail to alter the amyloid corona on TACONs, resulting in no colorimetric response. This strategy was confirmed using tau-degrading proteases (Protease XIV and plasmin) and their inactivated forms. Using TACONs, we evaluated the effectiveness of the tau oligomer-degrading agents (epigallocatechin gallate (EGCG), 1, 4-naphthoquinone, rosmarinic acid, and fulvic acid) and derived their half-maximal effective concentration (EC 50 ) as 0.937, 1.896, 1.266, and 1.981 mM respectively. Furthermore, by integrating a time-series deep learning architecture into our TACON-based platform, we substantially reduced the overall screening time from 3 hours to 1 hour with 91% classification accuracy at low drug concentrations (Figure 1C). We believe that our TACON-based drug screening platform has significant potential for the kinetic analysis of tau oligomer-degrading drugs and holds promise for facilitating drug discovery for tauopathies. 2. Results 2.1. Fabrication of tau amyloid corona-shelled nanoparticle (TACON) We synthesized TACONs employing AuNPs with heparin-induced misfolded tau monomers. Heparin, an anionic cofactor, facilitates tau protein misfolding, thereby forming tau aggregates and used various tau studies [18] . Specifically, negatively charged heparin induces the misfolding of tau proteins by neutralizing highly positively charged regions within the repeated domain (R). This promotes changes in the local and global conformations of the tau polypeptides, facilitating tau aggregation [19] . In our approach, TACONs are engineered to arrest the progression of tau aggregation in an oligomeric state, leveraging the interactions between AuNPs and tau monomers. On the surface of AuNP, the negatively charged citrates attract the positively charged N-terminal and R domains in the tau monomers via electrostatic and solvation forces, thereby inducing concentration gradients of tau near the AuNP surface. These interactions facilitate the primary nucleation of tau proteins and their subsequent oligomerization on the AuNP surface [20] . The formation of sterically stable TACONs is strongly correlated with the initial concentration of amyloid proteins [20b, c] . Our results align with those of previous reports, demonstrating that the steric stability of the tau amyloid corona relies on the initial concentration of tau monomers. At lower concentrations of tau monomers (49.6 nM), the UV-vis signals decreased with concentration owing to the accumulation of tau aggregates on the surface of the AuNPs, causing TACON coagulation (Figure S1). Therefore, we determined 49.6 nM as the optimal tau concentration for TACON fabrication. Further, the uniformity of tau amyloid corona was confirmed using HRTEM (Figure 2A). HRTEM images of TACONs and bare AuNPs revealed the complete coating of TACONs with a uniform amyloid corona, with thicknesses ranging as 4–5 nm. Furthermore, this configuration exhibited good steric stability for up to seven days (Figure 2B and Figure S2). FIGURE 1. Schematic illustration of tau amyloid corona-shelled nanoparticle (TACON)-based drug screening workflow. (A) Schematic of the TACON fabrication process, starting from seeding with AuNPs and tau monomers to the oligomerization facilitated by heparin, culminating in the synthesis of TACONs. The inset focuses on the heparin-binding domain critical for the oligomerization process. (B) Schematic of the drug screening assay where the interaction of TACONs with an effective drug results in aggregation, while an ineffective drug maintains dispersion, detectable via colorimetric changes. It demonstrates the high-throughput screening capability of the TACON platform across multiple wells, correlating drug concentration with tau oligomer degradation as indicated by the peak shifts in the UV-vis spectrum. (C) Schematic illustration of the time-series analysis framework of deep learning model. 2.2. Physicochemical characteristics of the TACONs To analyze the tau amyloid corona on TACONs, we employed XPS (Figure 2C). The XPS spectrum of the TACONs exhibited distinct N 1 s peaks at approximately 400 eV, along with reduced Au 4 f and Au 4 d peaks, compared to those of the bare AuNPs. These results imply the presence of a tau amyloid corona on the AuNPs, characterized by their amine groups. Further, the hydrodynamic diameter of the TACONs (~38 nm) was approximately 9 nm larger than the bare AuNPs (~29 nm) (Figure 2D), which are as per the estimated thickness (4–5 nm) of the tau amyloid corona from the HRTEM images (Figure 2A). These corroborate the result of the zeta potential of TACONs, in which tau oligomers slightly neutralized the negative charge of bare AuNPs [21] . The uniformity of the TACONs was also confirmed via agarose gel electrophoresis in TAE buffer, indicating a homogeneous size and high reproducibility of TACON synthesis (Figure S3). Finally, the freeze–thaw stability of the TACONs was enhanced relative to that of bare AuNPs. Following freeze–thawing, the bare AuNPs aggregated and their UV-vis spectra were not observed. In contrast, the TACONs remained intact in the UV-vis spectra after the freeze–thaw test, indicating the irreversible and stable coating of tau amyloid corona on the surface of AuNPs [22] (Figure S4). This cryostability was quantified using the relative absorbance ratio A 525 /A 650 (the absorbance at 525 nm divided by that at 650 nm), which are indicators of TACON dispersion (A 525 ) and aggregation (A 650 ), respectively (Figure 2F). Thus, TACONs exhibited superior cryopreservation storage capacity, preventing the denaturation of tau proteins. We used gFETs (Figure S5) functionalized with tau conformation-specific antibodies (monoclonal Tau-5 and polyclonal T22) to examine the physical properties of the tau amyloid corona on TACONs (Figure 2G). We covalently immobilized Tau-5 and T22 antibodies to the gFETs using carbodiimide chemistry, and then treated them with ethanolamine to mitigate the nonspecific protein adsorption. The monoclonal Tau-5 antibody served as a positive control for capturing all conformational species of tau proteins, including monomers and oligomers. The T22 antibody (specific to oligomeric tau species) was adopted to check whether the tau amyloid corona on TACONs retain oligomeric conformation. Prior to the conformational characterization of TACONs, we evaluated the binding affinities of specific antibodies to various in vitro synthesized tau species, including monomers and heparin-induced oligomers. We confirmed the topological characteristics of tau monomers and heparin-induced oligomers by AFM (Figure S6). Both tau monomers and heparin-induced tau oligomers were used to evaluate the sensing performances of our gFETs. These gFET sensors monitor conductivity changes within graphene channels triggered by the gating effect, as antibodies capture their specific targets [23] . We measured the real-time changes in the ratio of drain current (| ΔI ds /I 0 |) for Tau-5- and T22-immobilized gFETs upon the injection of each tau species (20 nM). The results demonstrate that our gFETs could precisely identify the specific conformations of various tau species (Figure S7). Finally, we investigated the conformation of the tau amyloid corona on TACONs employing Tau-5- and T22-immobilized gFET sensors. As a negative control, we injected bare AuNPs into Tau-5- and T22-immobilized gFETs, yielding negligible responses (0.019 ± 0.004% and 0.078 ± 0.029%, respectively) of |Δ I ds /I 0 |. In contrast, TACONs induced significant changes (0.507 ± 0.019% and 0.346 ± 0.04%, respectively) in | ΔI ds /I 0 | of Tau-5- and T22-immobilized gFETs (Figure 2H and 2I). Figure 2H and 2I summarize the electrical responses of the Tau-5- and T22- immobilized gFETs, respectively; the inset graphs represent the corresponding real-time responses following the injection of bare AuNPs or TACONs. We double-checked these results by dot blot assay (Figure S8) and concluded that the tau amyloid corona on TACONs predominantly comprises tau oligomers. FIGURE 2. Physicochemical and conformational characteristics of TACON. (A) Transmission electron microscopy (TEM) image of bare AuNP (upper) and TACON (lower). The white scale bar is 25 nm in size. (B) UV-vis spectra of TACON reacted in PBS at 0, 1, 2, and 24 hours (C) X-ray photoelectron spectroscopy (XPS) data of bare AuNP (blue) and TACON (pink). (D) Dynamic light scattering (DLS) data of bare AuNP (blue) and TACON (pink). The hydrodynamic diameter of bare AuNP and TACON are 28.74 ± 0.78 and 38.46 ± 1.58 nm, respectively. (E) The ζ-potential data of bare AuNP (blue), TACON (pink), and heparin-induced tau oligomer (green). The ζ-potential of bare AuNP, TACON, and heparin-induced tau oligomer are -65.41 ± 0.64, -57.76 ± 0.75 mV, and -9.83 ± 5.46 mV respectively. (F) A 525 /A 650 data of bare AuNP (blue) and TACON (pink) before and after the 1 freeze-thaw cycle and their images (upper inset of data). (G) Schematic of conformational validation of TACONs using gFET sensor with Tau-5 (red, tau-specific), and T22 (green, tau oligomer-specific) antibodies. The green check mark represents a significant sensor response, while the red circle with a white bar represents a negligible response of the gFET sensor. (H, I) Sensing responses of bare AuNPs (blue) and TACONs (pink) using gFET sensors immobilized with Tau-5 antibody (H) and T22 antibody (I) , respectively. Here, I ds represents the source-drain currents, and I 0 denotes the calibrated baseline signal for each gFET during continuous buffer-flow injection. All of the measurements were conducted in triplicate, and the data presented in (D, E, H, I) are mean±standard deviation. 2.3. Proteolytic activity analysis of tau oligomer-degrading proteases In order to validate our drug screening platform, we employed a protease XIV capable of degrading all species of tau protein into small peptide fragments [24] . The extent of TACON degradation was evaluated by measuring the A 650 /A 525 ratio. A higher ratio indicates increased particle aggregation, reflecting more extensive degradation of tau oligomers (amyloid corona on TACONs). Why this behavior happens is quite clear. When the amyloid coronas on AuNPs are degraded, the AuNPs become unstable, leading to their aggregation. The aggregation of nanoparticles dramatically affects UV-vis spectra of the TACON solutions, enabling the quantification of TACON degradation. As the exposure time to protease XIV increased, the LSPR peak of the TACON solution significantly changed, demonstrating active degradation of tau oligomers by proteolytic activity (Figure 3A and S9A). In contrast, denatured protease XIV elicited no change in UV-vis spectra, owing to the loss of proteolytic activity caused by thermal denaturation (Figure 3B). The A 650 /A 525 ratio of the TACON solution increased progressively with increasing concentrations of protease XIV, whereas no change was observed with denatured protease XIV. We utilized a sigmoidal dose-response model to analyze the A 650 /A 525 ratio based on varying protease XIV concentrations (Equation (1), Figure 3C). A 650 /A 525 =0.2155 +\(\frac{0.6172}{1+10^{1.838-[Protease\ XIV]}}\), R 2 = 0.98 (1) where [ Protease XIV ] is the concentration of protease XIV. Based on this model, the EC 50 of protease XIV was 1.838 mg·mL −1 , and maximal efficacy was 0.6172 arbitrary units (a.u.). Denatured protease XIV exhibited no significant changes in the A 650 /A 525 ratio. The aggregation of TACONs caused by protease XIV was confirmed via TEM and gel electrophoresis (Figure S9C and S10B). These results demonstrate the capability of our TACON-based platform for the quantitative analysis of the proteolytic activity of the tau oligomer-degrading enzyme. To further validate the capabilities of our platform, we used plasmin, a broad-specific serine protease [25] that influences the pathogenesis of AD by reducing tau accumulation and facilitating neuronal recovery [26] . Serine, a constituent of the tau protein, enables plasmin to degrade various tau species, thereby rendering it compatible with our platform. The UV-vis spectra of the TACON solution treated with 15 μg·mL -1 of plasmin exhibited significant peak shifts with increased A 650 /A 525 ratios over time, indicating that plasmin easily degraded the tau oligomers and induced particle aggregation (Figure 3D and S9D). This trend aligns with both the results from TEM analysis and gel electrophoresis (Figure S9F and S10C). A concentration-dependent response was evident as TACON solutions containing less than 10 μg·mL -1 of plasmin remained stable over time, whereas those exceeding 10 μg·mL -1 exhibited significant changes in the A 650 /A 525 ratio (Figure S9E). However, plasmin pre-incubated with 52 μg·mL -1 of α2-macroglobulin, an inhibitor of plasmin, showed no significant changes in the UV-vis spectra of the TACON solution as time progressed (Figure 3E). Using Equation (2), the A 650 /A 525 ratios of TACON solutions were analyzed with a sigmoidal dose-response model in response to varying concentrations (1-20 μg·mL -1 ) of both active and inhibited plasmin (Figure 3F). A 650 /A 525 =0.2669 +\(\frac{0.6001}{1+10^{11.39-[Plasmin]}}\), R 2 = 0.99 (2) where [ Plasmin ] is the plasmin concentration. The EC 50 was 11.39 μg·mL -1 , and the maximal efficacy was 0.6001 (a.u.). Negligible changes were observed in the A 650 /A 525 ratio during incubation with the inhibited plasmin. Collectively, these results indicate that the TACON-based platform can quantitatively assess both the proteolytic activity and inhibition of tau oligomer-degrading enzymes. FIGURE 3. Kinetic analysis of tau-degrading enzymes using TACON. (A, B) Time-resolved UV-vis spectra of TACON solution treated with ( A) active protease XIV (2 mg·mL −1 ) and ( B) thermally denatured protease XIV (exposed to 90 °C for 4 h, 2 mg·mL −1 ). The results were measured for 90 min at 10-min intervals. (C) Graphs fitted with the sigmoidal dose-response curve based on the concentrations of protease XIV and denatured protease XIV. ( D, E) Time-resolved UV-vis spectra of TACON solution treated with ( D) plasmin (15 μg·mL -1 ) and (E) inhibited plasmin (15 μg·mL -1 ) treated by the α-2-macroglobulin. The results were measured for 90 min at 10-min intervals. ( F) Dose-response curve of plasmin and denatured plasmin. Data points in c and f represent the mean ± S.D, with each experiment performed in triplicate (n=3) (The S.D. values smaller than 0.01 could not displayed on the graph). 2.4. TACON platform for evaluating tau oligomer-degrading agents and integration with Artificial Intelligence To assess the ability of our TACON-based platform to evaluate tau oligomer-degrading agents, we examined eight reagents (Figure S11). We incubated the TACON solutions with these reagents across a concentration range (0–10 mM) and assessed their efficacy using UV-visible spectroscopy (Figure S12). The EGCG, 1, 4-NQ, RA, and FA, exhibited substantial shifts in the LSPR peak of the TACON solutions depending on their concentration (Figure S12A–D). In contrast, the control compounds (EPPS, ibuprofen, sinomenine, and puerarin) exhibited no changes, regardless of their concentrations (Figure S12E–H). Time-dependent assays also confirmed that only the tau-degrading agents induced substantial peak shifts over time (Figure S13). We further validated the tau oligomer degradation by EGCG, NQ, RA, and FA using TEM imaging, AFM analysis, and gel electrophoresis (Figure S14, S15, and S16). TEM images showed that the reagents removed the tau amyloid corona from the TACONs, inducing their massive aggregation. AFM analysis also revealed that the degree of particle aggregation varied with EGCG concentration. These were consistent with the results of gel electrophoresis. The images of gel electrophoresis revealed broader and more diffuse bands for the reagent-treated TACONs, compared to those of intact TACONs. The broader bands were due to various sized TACON clusters formed by massive aggregation. In addition, we conducted selectivity test of TACON with various interfering biomolecules and heparin antagonist including glucose, bovine serum albumin (BSA), human serum albumin (HSA), immunoglobulin G (IgG), and protamine (Figure S17). The result showed that the TACONs are not aggregated in the presence of other interfering molecules and heparin antagonist, indicating that the TACONs selectively react with oligomer-degrading agents. To further evaluate the practicality and versatility of our platform, we analyzed the A 650 /A 525 ratios of the TACON solutions in a dose-dependent manner. As shown in Figure 4A–D, the A 650 /A 525 ratio increased significantly with the increasing concentration of tau oligomer-degrading agents (0.1–10 mM). Conversely, treatment with EPPS, ibuprofen, sinomenine, and puerarin showed no significant changes in the A 650 /A 525 ratios at various concentrations (Figure 4E–H). We then constructed sigmoidal dose-response curves based on the A 650 /A 525 ratio, for each agent with different concentrations (Table S1). Using these equations, we extracted the key pharmacokinetic parameters for each compound, including EC 50 , maximal efficacy, hillslope, and correlation coefficient R 2 (Figure 4I). Specifically, EGCG emerged as a superior tau oligomer-degrading agent, exhibiting the highest maximum efficacy and lowest EC 50 value. RA produced the steepest hillslope, highlighting its rapid action. To directly compare the efficacy of each agent, we introduced a scoring index calculated by dividing the maximal efficacy by the EC 50 value; a higher index shows greater efficacy. Figure 4J shows the A 650 /A 525 ratios and scoring indices of the TACON solutions with each reagent as comprehensive metrics reflecting their efficacies. The index metrics revealed that EGCG was the most potent, with a scoring index of 1.189, followed by RA (0.660), 1,4-NQ (0.539), and FA (0.359). Overall, these results underscore the significant potential of our TACON-based drug screening platform to accelerate the development of effective drugs targeting tau oligomers in pharmaceutical research. FIGURE 4. Evaluated the efficacies of tau oligomer-degrading compounds using a TACON-based platform. (A–H) Dose-dependent A 650 /A 525 curves of TACON solutions treated with each compound. ( I) Summary of the EC 50 and maximal efficacy values derived from the sigmoidal dose-response curves of each compound. ( J) Comprehensive metrics visualizing the efficacies derived from the dose-response curves of each compound. The A 650 /A 525 ratios of the TACON solutions at different concentrations of each compound were illustrated in a blue heatmap. The scoring index is displayed in a red heatmap, which is defined as the ratio of maximal efficacy to EC 50 for the assorted small molecules. All measurements were conducted three times, and the data presented in (A–H) are mean ± S.D (The S.D. values smaller than 0.01 could not be displayed on the graph). 2.5. Integrating deep learning algorithm to TACON-based platform to reduce screening time To further enhance the efficiency and accuracy of our drug screening platform, we integrated a deep learning-based time-series analysis algorithm into the TACON-based platform. This algorithm is based on our previously reported method [27] . Specifically, we employed a Long Short-Term Memory (LSTM)-based deep learning architecture optimized for spectral analysis to classify the efficacy of tau oligomer-degrading compounds using sequential colorimetric data obtained from TACON solutions. We first obtained time-dependent UV-vis absorbance spectra from TACON solutions incubated with EGCG at three distinct concentrations: low (1 mM), medium (2 mM), and high (3 mM) in 96-well plates (Figure 5A). These spectra were then sequentially processed by our LSTM-based algorithm to extract relevant temporal features for accurate classification of compound efficacy based on distinctive spectral patterns. As illustrated in Figure 5B, the algorithmic workflow involves the sequential spectral data acquisition, feature extraction using an LSTM model, and subsequent compound classification. The classification accuracy significantly improved from 60% to 91% within 60 minutes as additional time-series data were incorporated into the model (Figure 5C). These results demonstrate that our algorithm effectively captures intricate temporal patterns inherent in the colorimetric aggregation responses of TACON solutions, with the model’s accuracy progressively increasing as additional time-series data were integrated. Figure 5D shows the ROC curve analysis of the drug screening results generated by our LSTM-based algorithm, which was trained on time-series spectral data from TACON solutions collected at 10-minute intervals over one hour. The ROC analysis demonstrated an AUC of 0.99, indicating exceptional classification accuracy, particularly at lower drug concentrations. Figure 5E presents the corresponding confusion matrix, which demonstrates the robustness and consistent performance of our deep learning model across various concentration ranges. Collectively, the integration of LSTM-based deep learning algorithms significantly enhanced spectral analysis of our colorimetric TACON-based platform, providing a powerful tool for rapid and accurate identification of tau oligomer-degrading compounds. FIGURE 5. Deep learning algorithm for accelerating drug screening using EGCG. (A) Time-dependent UV-vis spectra of TACONs according to EGCG concentration (1mM, 2mM, and 3mM). (B) Schematic illustration of the time series analysis framework using an LSTM-based architecture for feature extraction and classification of EGCG concentrations into low (1mM), medium (2mM), and high (3mM). (C) Improved accuracy with increasing signal accumulation times, achieving a 91% at 60 minutes. (D) Receiver Operating Characteristic (ROC) curve demonstrating excellent performance with an Area Under the Curve (AUC) of 0.99. (E) Confusion matrix demonstrating robust classification performance. 3. Discussion Despite increasing evidence implicating oligomeric tau species in AD and other tauopathies [28] , the development of drugs specifically targeting tau oligomers remains limited due to the lack of proper in vitro drug screening methods. These limitations stem primarily from challenges in the reproducible synthesis and isolation of tau oligomers. Researchers have employed various approaches to synthesize tau aggregates, including the use of anionic cofactors [29] and seed amplification assays (SAAs), such as real-time quaking-induced conversion (RT-QuiC) [30] . However, the kinetic instability of tau oligomers impedes the reliable arrest of tau aggregates in their preformed oligomeric state during the aggregation process. Therefore, a method that enables the precise and reliable synthesis of tau oligomers is required to facilitate the development of drug screening platforms. Amyloid coronas, which encapsulates nanoparticles with a thickness of a few nanometers, provide remarkable insights into their potential for exploring the kinetics of amyloid aggregation, including its formation [31] , degradation [20a] , and inhibition [32] . The concept of amyloid corona has been developed for assessing the effectiveness of small molecules targeting the oligomeric species of α-synuclein [20b] and Aβ [20c] . Additionally, amyloid coronas composed of synthetically engineered amyloid peptides have been utilized to screen the efficacy of inhibitors against the main protease (M pro ) of SARS-CoV-2 [20a] . This study broadened the application of the amyloid corona to develop a drug screening platform for tauopathies. We engineered TACONs, wherein heparin-induced tau oligomers were uniformly coated on AuNPs, forming a hard protein corona to identify drugs that specifically degraded tau oligomers. After TACON synthesis, we effectively eliminated free-floating tau aggregates via centrifugation, leveraging the robust nature of the tau amyloid corona of TACONs and ensuring their high stability after centrifugation. The processed TACONs preserved exceptional stability, withstanding both long-term durability and freeze-thaw tests. This underscores the potential of our TACON-based platform for the synthesis and isolation of tau oligomers with high stability and reproducibility. In our TACON-based drug screening platform, AuNPs facilitate the monitoring of the effectiveness of tau oligomer-degrading agents. Degradation of the tau oligomer destabilizes the TACONs in solution, inducing the agglomeration of AuNPs. This changes the characteristic LSPR spectrum of the AuNPs, thereby enabling the quantitative colorimetric analysis of tau oligomer degradation by each agent. Based on this detection principle, we validated our platform using two tau oligomer-degrading enzymes and eight small molecules, demonstrating its capability for quantitative measurement of tau oligomer degradation. While our platform may not elucidate the precise degradation mechanisms of each agent, it provides label-free, rapid (<1 h), and real-time monitoring of oligomer-degrading agents, regardless of their degradation mechanism. Moreover, the integration of deep learning models further enhances the platform’s potential by significantly reducing overall screening time with high classification accuracy. Despite these advantages, our platform has limitations in that it does not fully mimic in vivo pathological hallmarks of tauopathies, such as hyperphosphorylated tau oligomers or tau-Aβ coaggregates. Our future research will focus on incorporating these clinically relevant biomarkers into the platform and performing broad compound library screening to discover new drug candidates. We believe that our TACON-based platform, combined with its deep learning-driven spectral analysis, will significantly streamline drug discovery for the therapeutics of tauopathies. 4. Conclusions In this study, we introduced a TACON-based platform utilizing AuNPs encapsulated by tau amyloid coronas for the rapid colorimetric screening of tau oligomer-degrading agents. This approach efficiently and specifically identifies compounds that degrade tau oligomers, which have been implicated in the onset and transmission of AD and other tauopathies. By leveraging the plasmonic properties of AuNPs and a uniform tau oligomer-based hard corona, our method can provide a precise and quantitative description of the degradation of the tau oligomers without relying on chemical stains or antibodies. To verify that our platform can reliably quantify the degradation of tau oligomers, we conducted in vitro kinetic analyses using both active and deactivated proteases. We also evaluated the efficacy of various tau oligomer-degrading compounds by monitoring the colorimetric responses of TACONs under time- and dose-dependent conditions. Through these analyses, we demonstrated that our TACON-based strategy effectively facilitates accurate comparisons of the tau oligomer-degrading efficacy of various compounds. Furthermore, by leveraging the AI-driven algorithm with our TACON platform, we were able to significantly reduce the overall assay time in tau oligomer-targeting drug screening. We anticipate that integrating LSTM-based deep learning model with 384-well plate formats will further accelerate high-throughput screening of compound libraries, paving the way for streamlining the development of therapeutics for tauopathies and other neurodegenerative diseases. Although our current TACON platform does not fully mimic in vivo pathological hyperphosphorylated tau oligomers, we plan to broaden its applicability by incorporating hyperphosphorylated tau oligomers or tau-Aβ aggregates, thereby enhancing its pathophysiological relevance. We believe these advances in quantification and high-throughput capability demonstrate the potential of our TACON-based strategy to significantly contribute to the discovery of novel therapeutics for tauopathies. 5. Materials and methods 5.1. Materials Tau (ab211410, Lot# GR3433942-1) was purchased from Abcam (UK). Gold chloride solution (HAuCl 4 ), trisodium citrate, heparin sodium salt, protease type XIV, plasmin, α 2 -macroglobulin, fulvic acid (FA), rosmarinic acid (RA), EGCG, 1, 4-Naphthoquionone (1, 4-NQ), humic acid, thrombin, puerarin, protease K, ibuprofen, sinomenine, and EPPS were purchased from Sigma-Aldrich (USA). Agarose gel powder (Lot# 1A20019O20Y) was purchased from BioSesang (South Korea). 5.2. Gold nanoparticle (AuNP) synthesis Before synthesis, all glassware was dipped in aqua regia (a 1:3 mixture of concentrated HNO 3 and HCl) for 30 min and washed with Millipore water. After washing, the conventional Turkevich method was used to synthesize the AuNPs. UV-vis spectrophotometer, DLS, and high-resolution TEM (HRTEM) were used to analyze the properties of the synthesized AuNPs. 5.3. Tau monomer solution preparation Lyophilized tau peptide was dissolved in 0.1× BES solution with 1 mg·mL -1 concentration. Following peptide dissolution, the solution was aliquoted into a 1.7 mL low-binding microcentrifuge tube (Sorenson TM , USA) with a 50 μL volume. Each tube was stored at -20℃ before use. 5.4. TACON synthesis To prepare TACON, a tau monomer solution was mixed with a concentrated AuNP solution. Specifically, 1 mL of AuNP was concentrated via centrifugation at 10500 rpm for 16 min and 30 s; subsequently, 960 μL of supernatant was removed. Simultaneously, 600 μL of DW was added to one of the aliquoted tau monomer solutions and 2 mL of heparin solution (0.12 mg·mL -1 ) was added to the tau solution. Thereafter, 160 μL of tau-heparin solution was added to the AuNP solution. A thermomixer (Eppendorf, Germany) was used to incubate the solution for 20 hours, set to 37℃ and shaken at 800 rpm. 5.5. Fabrication, functionalization, and antibody immobilization of graphene-based field-effect transistors (gFETs) sensors GFET sensors were fabricated using a previously described method [20b] . First, SiO 2 substrates were activated with a piranha solution (4:1 H 2 SO 4 /H 2 O 2 ), followed by immersion in a 1% 3-aminopropyldimethylethoxysilane solution in EtOH for 2 h. Then, the graphene oxide solution was spin-coated (3000 rpm for 60 s) onto the substrates and chemically reduced using hydroiodic acid vapor (80°C for 3 h). The reduced graphene oxide (rGO) thin films were patterned into the gFET channels using reactive-ion etching (150 W, 3 min), and Ti/Au (5/65 nm) was deposited via electron-beam (e-beam) evaporation to form the source and drain electrodes. Finally, the devices were passivated with SU-8 2002 negative photoresist, exposing only the necessary areas of the graphene channels. For antibody immobilization on the gFET, the device was first incubated for 2 h at room temperature (RT) with a 20 mM solution of N-Hydroxysuccinimidyl pyrenebutanoate (PBSE; Sigma-Aldrich) in dimethyl sulfoxide (DMSO). Afterward, it was rinsed sequentially with DMSO and distilled water. Afterward, a 10 μg·mL -1 antibody solution in 0.1× phosphate-buffered saline (PBS) was incubated with the functionalized rGO surface for 1 h to form covalent bonds with PBSE. Unreacted PBSE residues were quenched by incubating the surface with 100 mM ethanolamine for 30 min. 5.6. Electrical Characterization of gFET sensor response The electrical response of the gFET sensor was continuously measured via a semiconductor parameter analyzer (KEITHLEY 4200A-SCS) in conjunction with a custom-made setup. This setup included a top jig with an Ag/AgCl reference electrode holder and a bottom jig for device loading. To facilitate real-time monitoring, the gate voltage (V g ) and drain-source voltage (V ds ) were maintained at 0.1 and 0.2 V, respectively. The injection of samples was conducted using a syringe pump at a steady flow rate of 0.1 mL/min. 5.7. XPS analysis To prepare samples for XPS analysis, each 1 mL of bare AuNP and TACON particles were concentrated via centrifugation at 10500 rpm for 16 min and 30 s; subsequently, 980 μL of supernatant was removed. Pellets of these solutions were spotted onto a silicon wafer and dried at RT. X-ray (Al Kα line: 1486.6 eV) at 4.8 × 10 −9 mb controlled by a K-alpha instrument (Thermo VG, UK) was using for performing XPS analysis. A survey scan was conducted in the binding energy range of 0–700 eV. Elemental scans of Au 4p, N 1s, and Au 4f were obtained at 4 points on each sample, with a passing energy of 40 eV. 5.8. Freeze-thaw performance To confirm the dispersion stability of TACON, we performed a freeze-thaw process. We stored TACON and AuNP solutions at -80℃ for 20 min, using AuNPs as control. After freezing, the samples were incubated at RT for 30 min to allow defrosting. The UV-vis spectra of the TACON and bare AuNPs were measured before / after freeze-thawing to confirm their stability. Images of the samples were captured using a Galaxy S22 Ultra microscope (Samsung, South Korea). 5.9. Spectroscopic analysis All UV-vis spectra within the range of 800-400 nm were obtained using a LAMBDA 365 UV-vis spectrophotometer (PerkinElmer, USA) at a scan rate of 600 nm·min -1 . SpectraMax® ABS Plus microplate reader (Molecular Devices, USA) was used to measure the kinetic assay of relative absorbance (A 650 /A 525 ). 5.10. Protease XIV kinetic analysis For kinetic analysis, protease XIV was dissolved in 1× PBS (1.54 mM KH 2 PO 4 , 155.17 mM NaCl, 2.70 mM Na 2 HPO 4 -7H 2 O) at various concentrations (0.125 to 3.125 mg·mL -1 ). The solutions with various concentrations of protease XIV were incubated at 37℃ for 30 min for complete dissolution and activation. To prepare as a negative control, the protease XIV solutions were heated up to 90℃ for 4 h for denatured. Various concentrations of protease XIV and denatured protease XIV were added to the TACON solution (the total volume being 1 mL), and the degree of particle aggregation was measured using a UV-vis spectrophotometer. The kinetics of protease XIV are represented by the relative absorbance (A 650 /A 525 ) of the solution. 5.11. Plasmin kinetic analysis For kinetic analysis, plasmin was dissolved in 1× PBS (1.54 mM KH 2 PO 4 , 155.17 mM NaCl, 2.70 mM Na 2 HPO 4 -7H 2 O) at various concentrations (1 to 100 μg·mL -1 ). The solutions with various concentrations were incubated at 37℃ for 30 min to completely dissolve and activate plasmin. For negative control of plasmin, the plasmin solutions were incubated with 150 μL of 1 mg·mL -1 plasmin inhibitor (α-2-macroglobulin) for 1 h. A fixed concentration of plasmin or inhibited plasmin was added to the TACON solution (total volume of 1 mL). 5.12. Evaluating the efficacies of small molecules The efficacy of small molecules was evaluated with the colorimetric responses of TACON solution. In detail, each of the compounds was dissolved in 1× PBS (1.54 mM KH 2 PO 4 , 155.17 mM NaCl, 2.70 mM Na 2 HPO 4 -7H 2 O) at various concentrations (0.1–10 mM). Each of the solutions was mixed with the TACON solution and incubated for 3 h at 37 ℃ for a full reaction. Subsequently, the TACON solutions were analyzed by microplate reader and UV-vis spectrophotometer to evaluate the efficacy of small molecules. 5.13. TEM image analysis The TACON and small molecule-treated TACON solutions were centrifuged and the supernatant of each solution was removed. Subsequently, 200 μL DW was added to each pellet, and 10 μL of final solutions were loaded on copper grids. Gravity-driven plunger apparatus was used to quench-frozen the samples in liquid ethane, followed by negative staining with 2% uranyl acetate. Field-emission gun transmission electron microscopy was used to obtain the image of samples at -170℃. 5.14. Electrophoresis of TACO Herein, 1500 mg of agarose gel powder (Biosesang, South Korea) was dissolved in 150 mL of 1× Tris-acetate-EDTA (TAE) buffer solution (20 mM CH 3 COOH, 1.37 mM C 10 H 16 N 2 O 8 , 39.62 mM C 4 H 11 NO 3) . The agarose solution was boiled to 80℃ and poured into the mold. Subsequently, the solution was left at RT for 1 h to lower the temperature, and an agarose gel was obtained. 40 μL of each sample were loaded on an agarose gel template. 5 min after sample loading, the sample-loaded agarose gel was completely sunk in 1× TAE buffer, and an electric voltage was applied for 30 min to draw the samples through the agarose gel electrophoresis. 5.15. Preparation of tau oligomers Tau oligomers were prepared by agitating monomers with heparin. 10 μL of heparin solution was added to 50 μL of the tau monomer solution at a concentration ratio of 5:1. The heparin-tau mixture was incubated for 15 h at 37 ℃ to prepare heparin-induced tau oligomer. The resulting oligomers were confirmed using atomic force microscopy (AFM). 5.16. Deep learning algorithm analysis A time series deep learning algorithm was applied to predict drug concentrations and significantly reduce drug screening time. Data acquisition was performed using a 96-well plate with measurements taken at 10-minute intervals. A volume of 20 μL of TACON was added to varying concentrations (1.25 mM, 2.5 mM, 3.75 mM) of EGCG prepared in PBS, resulting in a total solution volume of 100 μL per well. The final concentrations of EGCG in the wells were 1, 2, or 3 mM. The solutions were incubated in a 96-well plate, and the efficacy of EGCG at these concentrations was assessed using a microplate reader (SpectraMax® ABS Plus Microplate Reader, Molecular Devices, USA). Absorbance readings were taken at 10-minute intervals for 1 hour. The input data consisted of feature vectors representing wavelength values from 450 nm to 750 nm, measured at 10 nm intervals. The output labels were categorized as 1 (low concentration), 2 (medium concentration), and 3 (high concentration). The deep learning model was trained for 1,000 epochs, with cross-entropy employed as the loss function. A total of 2,592 datasets were generated (96 wells × 9 time points × 3 classes). The dataset was split into training, validation, and testing sets in a ratio of 6:2:2. CRediT authorship contribution statement Hyo Gi Jung : Investigation, Conceptualization, Methodology, Data curation, Formal analysis, Writing – original draft, Writing – review & editing, Dongsung Park : Investigation, Conceptualization, Formal analysis, Methodology, Visualization, Writing – original draft, Writing – review & editing, Junho Bang : Investigation, Methodology, Data curation, Visualization, Writing – original draft, Writing – review & editing, Yeon Ho Kim : Data curation, Resources, Jae Won Jang : Data curation, Yonghwan Kim : Data curation, Hyunji Kim : Data curation, Seumgmin Lee : Validation, Visualization, Wonbin Moon : Investigation, Resources, Kyo Seon Hwang : Resources, Jeong Hoon Lee : Resource s , validation, Dongtak Lee : Project administration, Validation, Writing – review & editing, Dae sung Yoon : Supervision, Writing – review & editing. Declaration of competing interest There are no conflicts of interest to declare. Acknowledgments This work was supported by National Research Foundation of Korea (NRF) grants funded by the Korean Government (MSIP) (grant numbers: RS-2025-00553786, NRF-2022R1A2C1091756, NRF-2022R1A6A3A03066467, RS-2024-00400563, RS-2024-00409958, and RS-2024-00414209), the BK21 FOUR Institute of Precision Public Health, the Science and Technology Commercialization Agency (grant number: RS-2024-00423580), and Soseon Foundation. This work was also supported by grants of the Korea Health Technology R&D Project through the Korea Health Industry Development Institute (KHIDI), funded by the Ministry of Health and Welfare (grant numbers: RS-2023-00274152 and RS-2023-00265159). References [1] a) T. Guo, W. Noble and D. P. Hanger, Acta neuropathologica 2017 , 133 , 665-704; b) L. Qiang, X. Sun, T. O. Austin, H. Muralidharan, D. C. Jean, M. Liu, W. Yu and P. W. Baas, Current Biology 2018 , 28 , 2181-2189. e2184.[2] a) A. D. Alonso, L. S. Cohen, C. Corbo, V. Morozova, A. ElIdrissi, G. Phillips and F. E. Kleiman, Frontiers in cellular neuroscience 2018 , 12 , 338; b) S. M. Alavi Naini and N. Soussi-Yanicostas, Oxidative medicine and cellular longevity 2015 , 2015 .[3] K. S. KoSIK, C. L. Joachim and D. J. Selkoe, Proceedings of the National Academy of Sciences 1986 , 83 , 4044-4048.[4] A. C. McKee, T. D. Stein, P. T. Kiernan and V. E. Alvarez, Brain pathology 2015 , 25 , 350-364.[5] M. Goedert, M. G. Spillantini, B. Falcon, W. Zhang, K. L. Newell, M. Hasegawa, S. H. Scheres and B. Ghetti, Frontotemporal Dementias: Emerging Milestones of the 21st Century 2021 , 177-199.[6] a) G. Ghag, N. Bhatt, D. V. Cantu, M. J. Guerrero‐Munoz, A. Ellsworth, U. Sengupta and R. Kayed, Protein Science 2018 , 27 , 1901-1909; b) S. A. Hussong, A. Q. Banh, C. E. Van Skike, A. O. Dorigatti, S. F. Hernandez, M. J. Hart, B. Ferran, H. Makhlouf, M. Gaczynska and P. A. Osmulski, Nature communications 2023 , 14 , 2367; c) S. Shin, D. Kim, J. Y. Song, H. Jeong, S. J. Hyeon, N. W. Kowall, H. Ryu, A. N. Pae, S. Lim and Y. K. Kim, Progress in neurobiology 2020 , 187 , 101782.[7] a) L. Diomede, E. R. Zanier, F. Moro, G. Vegliante, L. Colombo, L. Russo, A. Cagnotto, C. Natale, F. M. Xodo and A. De Luigi, Molecular Psychiatry 2023 , 28 , 2433-2444; b) M. Malhis, S. Kaniyappan, I. Aillaud, R. R. Chandupatla, L. M. Ramirez, M. Zweckstetter, A. H. Horn, E. Mandelkow, H. Sticht and S. A. Funke, ChemBioChem 2021 , 22 , 3049-3059.[8] a) C. H. Lo, C. K.-W. Lim, Z. Ding, S. P. Wickramasinghe, A. R. Braun, K. H. Ashe, E. Rhoades, D. D. Thomas and J. N. Sachs, Alzheimer’s & Dementia 2019 , 15 , 1489-1502; b) L. Wang, R. Kumar, P. F. Pavlov and B. Winblad, European journal of medicinal chemistry 2021 , 209 , 112915.[9] a) R. Abskharon, P. M. Seidler, M. R. Sawaya, D. Cascio, T. P. Yang, S. Philipp, C. K. Williams, K. L. Newell, B. Ghetti and M. A. DeTure, Journal of Biological Chemistry 2020 , 295 , 10662-10676; b) R. R. Chandupatla, A. Flatley, R. Feederle, E. M. Mandelkow and S. Kaniyappan, Alzheimer’s & Dementia: Translational Research & Clinical Interventions 2020 , 6 , e12097.[10] a) J. E. Gerson and R. Kayed, Frontiers in neurology 2013 , 4 , 93; b) I. Leake, Nature Reviews Neuroscience 2023 , 24 , 393-393.[11] S. Chinnathambi and R. Das, Cell & Bioscience 2023 , 13 , 95.[12] S. Jiang and K. Bhaskar, Frontiers in Molecular Neuroscience 2020 , 13 , 586731.[13] J. Wan, Y. Gong, Z. Xu, X. Dong, G. Wei and Q. Zhang, Biophysical chemistry 2021 , 271 , 106541.[14] C. Berney and G. Danuser, Biophysical journal 2003 , 84 , 3992-4010.[15] B. L. Scott and A. D. Hoppe, Scientific reports 2015 , 5 , 10270.[16] E. Lerner, A. Barth, J. Hendrix, B. Ambrose, V. Birkedal, S. C. Blanchard, R. Börner, H. Sung Chung, T. Cordes and T. D. Craggs, Elife 2021 , 10 , e60416.[17] a) S. S. Shafiei, M. J. Guerrero-Muñoz and D. L. Castillo-Carranza, Frontiers in aging neuroscience 2017 , 9 , 83; b) M. Kjaergaard, A. J. Dear, F. Kundel, S. Qamar, G. Meisl, T. P. Knowles and D. Klenerman, ACS chemical neuroscience 2018 , 9 , 3060-3071.[18] a) N. Sibille, A. Sillen, A. Leroy, J.-M. Wieruszeski, B. Mulloy, I. Landrieu and G. Lippens, Biochemistry 2006 , 45 , 12560-12572; b) D. Townsend, N. J. Fullwood, E. A. Yates and D. A. Middleton, Biochemistry 2020 , 59 , 4003-4014; c) A. T. Marvian, T. Strauss, Q. Tang, B. J. Tuck, S. Keeling, D. Rüdiger, N. Mirzazadeh Dizaji, H. Mohammad-Beigi, B. Nuscher and P. Chakraborty, Molecular Neurodegeneration 2024 , 19 , 100; d) H.-L. Song, M.-S. Kim, W.-Y. Cho, Y.-S. Yoo, J.-Y. Kim, T.-W. Kim, H. Kim, D.-H. Kim and S.-Y. Yoon, Molecular Neurodegeneration 2024 , 19 , 76.[19] G. Limorenko, M. Tatli, R. Kolla, S. Nazarov, M.-T. Weil, D. C. Schöndorf, D. Geist, P. Reinhardt, D. E. Ehrnhoefer and H. Stahlberg, Nature Communications 2023 , 14 , 3939.[20] a) D. Lee, H. G. Jung, D. Park, J. Bang, D. Y. Cheong, J. W. Jang, Y. Kim, S. Lee, S. W. Lee and G. Lee, Nature Communications 2024 , 15 , 2108; b) D. Lee, H. G. Jung, D. Park, J. Bang, J. H. Hong, S. W. Lee, S. Roh, J. W. Jang, Y. Kim and K. S. Hwang, ACS applied materials & interfaces 2022 , 15 , 2538-2551; c) D. Lee, D. Park, I. Kim, S. W. Lee, W. Lee, K. S. Hwang, J. H. Lee, G. Lee and D. S. Yoon, Nature communications 2021 , 12 , 639.[21] E. Casals, T. Pfaller, A. Duschl, G. J. Oostingh and V. Puntes, ACS nano 2010 , 4 , 3623-3632.[22] a) E. Trenkenschuh and W. Friess, International Journal of Pharmaceutics 2021 , 606 , 120932; b) A. S. Picco, G. B. Mondo, L. F. Ferreira, E. E. De Souza, L. A. Peroni and M. B. Cardoso, Nanoscale 2021 , 13 , 753-762.[23] D. Park, J. H. Kim, H. J. Kim, D. Lee, D. S. Lee, D. S. Yoon and K. S. Hwang, Biosensors and Bioelectronics 2020 , 167 , 112505.[24] a) J. Brown, C.-L. Lu, J. Coburn and D. L. Kaplan, Acta biomaterialia 2015 , 11 , 212-221; b) M. Li, M. Ogiso and N. Minoura, Biomaterials 2003 , 24 , 357-365.[25] T. Syrovets and T. Simmet, Cellular and Molecular Life Sciences CMLS 2004 , 61 , 873-885.[26] C. Guo, T. Wang, D. Zhang, X. Ge and J. Li, Biochemical and Biophysical Research Communications 2023 , 654 , 102-111.[27] S. Lee, J. S. Park, H. Woo, Y. K. Yoo, D. Lee, S. Chung, D. S. Yoon, K.-B. Lee and J. H. Lee, Nature communications 2024 , 15 , 1695.[28] N. Puangmalai, N. Bhatt, M. Montalbano, U. Sengupta, S. Gaikwad, F. Ventura, S. McAllen, A. Ellsworth, S. Garcia and R. Kayed, Cell death & disease 2020 , 11 , 314.[29] Y. Fichou, Z. R. Oberholtzer, H. Ngo, C.-Y. Cheng, T. J. Keller, N. A. Eschmann and S. Han, Frontiers in neuroscience 2019 , 13 , 1339.[30] S. Vascellari, C. D. Orrù and B. Caughey, Frontiers in Aging Neuroscience 2022 , 14 , 853050.[31] T. John, L. L. Martin and B. Abel, Macromolecular Bioscience 2023 , 23 , 2200576.[32] P. Chen, F. Ding, R. Cai, I. Javed, W. Yang, Z. Zhang, Y. Li, T. P. Davis, P. C. Ke and C. Chen, Nano Today 2020 , 35 , 100937. Supplementary Material File (table s1.docx) Download 13.96 KB Information & Authors Information Version history V1 Version 1 19 March 2025 Peer review timeline Published Aggregate Version of Record 29 Aug 2025 Published Copyright This work is licensed under a Non Exclusive No Reuse License. Collection Aggregate Keywords aggregation colorimetric drug screening platform deep learning tau oligomers tau-oligomer degrading agents tauopathies Authors Affiliations Hyo Gi Jung Korea University View all articles by this author Dongsung Park 0000-0001-5235-7114 Korea University View all articles by this author Junho Bang 0009-0000-9749-6513 Korea University View all articles by this author Yeon Ho Kim Korea University View all articles by this author Jae Won Jang Korea University View all articles by this author Yonghwan Kim Korea University View all articles by this author Hyunji Kim Korea University View all articles by this author Seungmin Lee Korea University View all articles by this author Wonbin Moon Korea University View all articles by this author Kyo Seon Hwang Kyung Hee University College of Medicine View all articles by this author Jeong Hoon Lee KU-KIST Graduate School of Converging Science and Technology View all articles by this author Dongtak Lee Incheon National University College of Life Science and Technology View all articles by this author Dae Sung Yoon [email protected] Korea University View all articles by this author Metrics & Citations Metrics Article Usage 424 views 235 downloads .FvxKWukQNSOunydq8rnd { width: 100px; } Citations Download citation Hyo Gi Jung, Dongsung Park, Junho Bang, et al. Tau amyloid corona-shelled nanoparticle-based drug screening platform for discovering tau oligomer-degrading drugs. Authorea . 19 March 2025. DOI: https://doi.org/10.22541/au.174235163.36542009/v1 If you have the appropriate software installed, you can download article citation data to the citation manager of your choice. Simply select your manager software from the list below and click Download. For more information or tips please see 'Downloading to a citation manager' in the Help menu . Format Please select one from the list RIS (ProCite, Reference Manager) EndNote BibTex Medlars RefWorks Direct import Tips for downloading citations document.getElementById('citMgrHelpLink').addEventListener('click', function() { popupHelp(this.href); return false; }); $(".js__slcInclude").on("change", function(e){ if ($(this).val() == 'refworks') $('#direct').prop("checked", false); $('#direct').prop("disabled", ($(this).val() == 'refworks')); }); View Options View options PDF View PDF Figures Tables Media Share Share Share article link Copy Link Copied! Copying failed. Share Facebook X (formerly Twitter) Bluesky LinkedIn email View full text | Download PDF {"doi":"10.22541/au.174235163.36542009/v1","type":"Article"} Now Reading: Share Figures Tables Close figure viewer Back to article Figure title goes here Change zoom level Go to figure location within the article Download figure Toggle share panel Toggle share panel Share Toggle information panel Toggle information panel Go to previous graphic Go to next graphic Go to previous table Go to next table All figures All tables View all material View all material xrefBack.goTo xrefBack.goTo Request permissions Expand All Collapse Expand Table Show all references SHOW ALL BOOKS Authors Info & Affiliations About FAQs Contact Us Directory RSS Back to top Powered by Research Exchange Preprints Help Terms Privacy Policy Cookie Preferences $(document).ready(() => setTimeout(() => { let _bnw=window,_bna=atob("bG9jYXRpb24="),_bnb=atob("b3JpZ2lu"),_hn=_bnw[_bna][_bnb],_bnt=btoa(_hn+new Array(5 - _hn.length % 4).join(" ")); $.get("/resource/lodash?t="+_bnt); },4000)); (function(){function c(){var b=a.contentDocument||a.contentWindow.document;if(b){var d=b.createElement('script');d.innerHTML="window.__CF$cv$params={r:'a024c13e4c2c300f',t:'MTc3OTg4MjEzMg=='};var a=document.createElement('script');a.src='/cdn-cgi/challenge-platform/scripts/jsd/main.js';document.getElementsByTagName('head')[0].appendChild(a);";b.getElementsByTagName('head')[0].appendChild(d)}}if(document.body){var a=document.createElement('iframe');a.height=1;a.width=1;a.style.position='absolute';a.style.top=0;a.style.left=0;a.style.border='none';a.style.visibility='hidden';document.body.appendChild(a);if('loading'!==document.readyState)c();else if(window.addEventListener)document.addEventListener('DOMContentLoaded',c);else{var e=document.onreadystatechange||function(){};document.onreadystatechange=function(b){e(b);'loading'!==document.readyState&&(document.onreadystatechange=e,c())}}}})();