A small molecule modulating heterogeneous amyloid-β isoforms improves cognitive and pathological outcomes in acute and 5XFAD Alzheimer’s disease models

Abstract

Background

Alzheimer’s disease (AD) is characterized by the pathological aggregation of amyloid-β (Aβ) peptides into neurotoxic assemblies. While recent antibody therapies targeting Aβ have shown clinical efficacy, they face limitations including specificity for particular Aβ species and accessibility challenges. The structural heterogeneity of Aβ isoforms, including Aβ₄₀, Aβ₄₂, and pyroglutamate-modified Aβ_pE3−42, necessitates therapeutic strategies with broad-spectrum activity.

Methods

The effects of YIAD-1003, a dihydropyrrolo[1,2-a]pyrazine derivative, were evaluated in vitro using Thioflavin T fluorescence assays and A11 dot blot to assess fibrillization and dissociation of preformed aggregates across Aβ₄₀, Aβ₄₂, and Aβ_pE3−42. In vivo efficacy was examined in an acute Aβ₄₂-induced AD model and in 5XFAD transgenic mice. Cognitive performance was assessed using spatial and associative memory tests. Amyloid pathology, including soluble Aβ and fibrillar deposition, and neuroinflammation markers were quantified by biochemical and histological analyses.

Results

YIAD-1003 inhibited fibril formation and promoted dissociation of preformed aggregates across multiple Aβ isoforms in vitro. In the acute model, co-administration of YIAD-1003 ameliorated Aβ₄₂-induced memory impairment. In 5XFAD mice, oral administration improved cognitive performance, reduced plaque burden and soluble Aβ levels, and attenuated pathological alterations associated with neuroinflammation.

Conclusions

YIAD-1003, a dihydropyrrolo[1,2-a]pyrazine derivative, exhibits broad-spectrum intervention in Aβ assembly across multiple isoforms and confers functional and pathological benefits in AD mouse models. These findings support the development of multi-isoform small-molecule modulators as a possible AD therapeutic strategy.

Supplementary Information

The online version contains supplementary material available at 10.1186/s13195-026-02113-5.

Introduction

Alzheimer’s disease (AD) is a progressive neurodegenerative disorder pathologically characterized by the abnormal accumulation of amyloid-β (Aβ) peptides [1]. These peptides, generated through sequential proteolytic cleavage of the amyloid precursor protein, undergo aberrant self-assembly into a spectrum of neurotoxic species, including soluble oligomers, protofibrils, and insoluble fibrillar aggregates. Accumulating evidence indicates that soluble Aβ oligomers exert potent acute synaptic toxicity, while fibrillar aggregates drive chronic inflammation and long-term amyloid deposition [2, 3]. Importantly, cerebral Aβ burden and its aggregation state correlate with synaptic loss, cognitive decline, and disease progression, establishing Aβ aggregation as a central molecular hallmark and a validated therapeutic target in AD [4].

Given the central role of Aβ in AD pathogenesis, substantial clinical research has focused on reducing cerebral amyloid burden as a disease-modifying strategy [5]. Monoclonal antibodies targeting aggregated Aβ species, including aducanumab, lecanemab, and donanemab, have demonstrated measurable efficacy in lowering brain amyloid levels and modestly slowing cognitive decline in recent Phase III clinical trials [6–8]. These outcomes provide validation of Aβ as a therapeutic target and support amyloid reduction as a viable disease-modifying approach. However, antibody-based therapies face translational challenges, requiring intravenous infusion at specialized centers, limiting accessibility, high manufacturing costs and limited brain penetration, and selective targeting of specific Aβ conformations or epitopes, potentially missing pathogenic species [9, 10]. These limitations underscore the need for complementary therapeutic modalities that can engage diverse pathogenic Aβ assemblies with improved brain bioavailability and broader isoform coverage.

A key challenge in achieving durable therapeutic efficacy is the structural and biochemical heterogeneity of pathogenic Aβ species. The two most abundant isoforms, Aβ₄₀ and Aβ₄₂, differ by only two C-terminal residues yet display markedly distinct biophysical properties. Aβ₄₂ exhibits 10-fold faster aggregation kinetics, enhanced β-sheet propensity, and greater neurotoxicity compared to Aβ₄₀ [11, 12]. Beyond these full-length peptides, N-terminally truncated and pyroglutamate-modified species, particularly pyroglutamate-modified Aβ_pE3−42 (Aβ_pE3) constitutes up to 50% of total amyloid burden in AD brains. Aβ_pE3 species display increased hydrophobicity, resistance to aminopeptidase degradation, accelerated aggregation kinetics, and potent seeding activity that can nucleate aggregation of other Aβ isoforms [13, 14]. These biochemical differences generate structurally heterogeneous aggregate populations with distinct cellular toxicity profiles, regional deposition patterns, and temporal dynamics throughout disease progression. Consequently, therapeutic strategies relying exclusively on isoform-selective or conformation-specific targeting may fail to address the full complexity of amyloid pathology. Approaches capable of engaging and modulating multiple pathogenic Aβ assemblies simultaneously may therefore be required to achieve comprehensive and sustained therapeutic benefit.

Small-molecule therapeutics represent a promising complementary modality for targeting amyloid pathology, offering distinct advantages including: (i) chemical tunability for multi-target optimization; (ii) oral bioavailability and scalable manufacturing; (iii) capacity for blood-brain barrier (BBB) penetration via passive diffusion or active transport; and (iv) potential for continuous target engagement at lower cost [15–17].

In this study, we screened a library of dihydropyrrolo[1,2-a]pyrazine derivatives, and identified YIAD-1003 as a lead compound capable of modulating pathogenic Aβ aggregation. Through biophysical characterization, we demonstrate that YIAD-1003 acts across multiple Aβ isoforms through multifaceted anti-amyloid activity, inhibition and dissociation. Using complementary structural and kinetic approaches including Thioflavin T (ThT) fluorescence, we assessed its effects on Aβ assembly pathways in vitro. Furthermore, we evaluated the behavioral and pathological effects of YIAD-1003 treatment in vivo through both acute (single Aβ injection) and chronic (5XFAD transgenic) mouse models of AD, assessing cognitive performance, amyloid burden, and neuroinflammation. Together, these findings establish YIAD-1003 as a versatile chemical scaffold for modulating heterogeneous Aβ assemblies and provide mechanistic rationale for multi-target small-molecule strategies in AD therapeutics.

Methods

Study design

YIAD-1003, a dihydropyrrolo[1,2-a]pyrazine derivative (molecular weight 452.45 g/mol) identified through ThT screening, was evaluated using complementary in vitro and in vivo approaches to investigate its anti-amyloidogenic effects across multiple Aβ isoforms. In vitro assays utilizing Aβ₄₂, Aβ₄₀ and Aβ_pE3 examined the inhibitory and dissociative effects of YIAD-1003 on Aβ aggregation and oligomer formation. Mechanistic studies, including fragment-based mapping and molecular docking, were performed to characterize the interaction between YIAD-1003 and Aβ aggregates. In vivo efficacy was assessed in an acute Aβ₄₂-induced mouse model and chronic 5XFAD transgenic model, with cognitive function measured by behavioral tests such as the Y-maze. Subsequent histological and biochemical analyses were conducted to evaluate amyloid burden and soluble Aβ levels.

Animals

The 5XFAD transgenic (strain name: B6SJL-Tg(APPSwFlLon, PSEN1*M156L*L286V)6799Vas/-Mmjax) mice and wild-type (C57BL/6 x SJL) mice were initially obtained from the Jackson Laboratory (Bar Harbor, ME, USA). 6-week-old male ICR mice were acquired from Orient Bio Inc. (Republic of Korea). All mice were bred in a humidity and temperature-controlled facility in Yonsei University (Seoul, Korea) with 12:12 h light-dark cycle. Additionally, the mice were given ad libitum access to food and water. All animal experiments were executed in accordance with the National Institutes of Health guide for the care and use of laboratory animals (NIH Publications). All animal experiments involving transgenic mice were approved by Institutional Animal Care and Use Committee (IACUC) of the Yonsei Laboratory Animal Research Center (Approval No. IACUC-202411-1948-01). Details regarding administration are written in later sections.

YIAD compounds synthesis

The twenty YIAD compounds (YIAD-1000 to -1019) were synthesized through a one-pot three-component coupling reaction as previously reported [18].

Aβ peptides synthesis

The three peptides utilized in this paper, Aβ₄₂ (DAEFRHDSGY EVHHQKLVFF AEDVGSNKGA IIGLMVGGVV IA), Aβ₄₀ (DAEFRHDSGY EVHHQKLVFF AEDVGSNKGA IIGLMVGGVV), and Aβ_pE3 (pyroEFRHDSGYEV HHQKLVFFAE DVGSNKGAII LMVGGVVIA), were synthesized in-house by Fmoc solid-phase peptide synthesis as previously reported [19]. Peptides were dissolved in 100% dimethyl sulfoxide (DMSO) prior to experiments and diluted accordingly with deionized water (DW) or phosphate-buffered saline (PBS) to needed concentrations.

ThT fluorescence assays

To assess the ability of compounds to inhibit and dissociate Aβ aggregation, ThT was used to quantify Aβ fibril content. For the inhibition assay during the screening phase, Aβ₄₂ monomers (25 µM) were incubated with each YIAD compound (50, 100, or 200 µM) for 24 h at 37 °C. For the dissociation assay of the screening process, Aβ₄₂ monomers (25 µM) were pre-incubated for 24 h at 37 °C to form pre-aggregated Aβ₄₂. After a day, each YIAD compound (50, 100, or 200 µM) was added and incubated for another day at 37 °C. For the dose-dependent inhibition and dissociation assays, the same protocol as mentioned above was followed with different incubation times for each Aβ isoforms and YIAD-1003 was applied at 8 different concentrations (2-fold dilution from 500 µM). After the incubation period, 25 µL of each sample was loaded in triplicates to a 96-well half-area black plate with 75 µL of ThT solution (5 µM in 50 mM glycine buffer, pH 8.5). After a 5-min incubation period, fluorescence intensity was measured at Ex 450/Em 485 nm using a microplate reader (TECAN Infinite 200 PRO).

Oligomer dissociation assay

To prepare the oligomer dissociation assay plate, Aβ₄₂-Cys (10 µM) was first immobilized into each well of a 15,153 maleimide plate followed by blocking using L-Cys (10 µg/mL). Flamma552-labeled Aβ₄₂ peptides (10 µM) were then added to each well containing the immobilized Aβ₄₂-Cys peptides and incubated for 8 h to form oligomers. After the incubation period, each YIAD compound was added (50 or 100 µM) and incubated for 24 h at 37 °C. After washing out detached Flamma552-labeled Aβ₄₂ peptides and YIAD compounds with a wash buffer, 100 µL of binding buffer was added to each well. Fluorescence intensity was measured at Ex 550/Em 564 nm using a microplate reader (TECAN Infinite 200 PRO).

Dot blot assays

To analyze oligomeric levels in the tested samples, 2 µL of each sample was blotted onto nitrocellulose membranes, dried for 30 min, and blocked with 5% skim milk in TBS-T (tris-buffered saline with 0.1% Tween-20) for 1 h at room temperature (RT) to prevent non-specific binding. Blocked membranes were then incubated with anti-oligomer A11 antibody (ThermoFisher, AHB0052, 1:1,000 in TBS-T) overnight at 4 °C on a rocker. After three 5-min washes with TBS-T, membranes were incubated with horseradish peroxidase (HRP)-conjugated anti-rabbit IgG antibody (Jackson ImmunoResearch, 111-035-144, 1:10,000 in TBS-T) for 1 h at RT. A11-positive signals were detected using SuperSignal™ West Pico PLUS Chemiluminescent Substrate (ThermoFisher, 34580). Densitometry of the signals was conducted using ImageJ.

Mapping Amyloid Plate (MAP) assay

Consecutive hexamer fragments of Aβ₄₂ and full-length Aβ₄₂ were immobilized in corresponding wells of a 15,153 maleimide-activated microplate and incubated with FITC-labeled Aβ₄₂ (10 µM) for 6 h at 37 °C to form pre-formed aggregates, as previously described [20]. Afterwards, YIAD-1003 (250 µM, 5.5% DMSO in binding buffer) was treated and incubated with the pre-formed aggregates for 24 h at 37 °C. After washing the wells three times with wash buffer, 100 µL of binding buffer was added to each well. Fluorescence intensity was measured at Ex 550/Em 564 nm using a microplate reader (TECAN Infinite 200 PRO).

Molecular docking

To obtain the three-dimensional structure of YIAD-1003, Avogadro 2, a chemical editor and visualization application, was utilized with geometry optimization, which included energy minimization [21]. Each structural model of Aβ₄₂ and Aβ₄₀ was derived from the Protein Data Bank (PDB) structure (each PDB ID: 2NAO and 2M4J) [22, 23]. The docking simulation with YIAD-1003 was performed through GNINA (version 1.3) [24]. The result of the docking simulation was visualized and analyzed using PyMOL software (version 2.1), a molecular visualization system [25]. The interaction between Aβ and YIAD-1003 was investigated by utilizing the Protein-Ligand Interaction Profiler (PLIP) server [26].

Parallel artificial membrane permeability assay (PAMPA) assay

BBB permeability of YIAD-1003 was assessed through PAMPA-BBB using the PAMPA kit (Bioassay systems PAMPA-096). The experiment was conducted according to the manufacturer’s manual. The acceptor plate was prepared by coating the membrane with 5 µL of lecithin solution (4% in dodecane) and each well was filled with 300 µL of PBS. The donor plate wells were treated with 200 µL of each sample. 200 µL of YIAD-1003 (125 µM, 10% DMSO in PBS) and the high/low permeability controls provided in the kit were filled into the wells of the donor plate. Then, the donor plate was placed gently on the acceptor plate to form a sandwich and incubated for 18 h at RT. After the incubation process, 100 µL of controls, acceptor, and donor samples were moved to a UV plate to scan the UV-vis spectra of the solutions using a microplate reader (SpectraMax M2e). Permeability of the samples was calculated using the analysis equation provided in the manufacturer’s manual.

Intracerebroventricular (ICV) injection

To prepare the samples for ICV, purified Aβ₄₂ (10 µM) was pre-incubated for 6 h at 37 °C and treated with YIAD-1003 (400 µM) for an additional 12 h. Vehicle (5.5% DMSO in PBS) and Aβ₄₂-only (10 µM, 5.5% DMSO in PBS) samples were incubated in equal time periods for the control groups. After the samples were prepared, 5 µL of each sample was injected into the ICV region of 7-week-old male ICR mice (n = 15/group) according to the previously reported protocol [27].

5XFAD administration

Female 5XFAD mice were administered YIAD-1003 (2.5% DMSO and 2.5% Tween80 in PBS) daily via oral gavage starting at 6 months of age for 4 weeks at 30 mg/kg. Age- and gender-matched vehicle-treated groups (transgenic and wild-type) were prepared as the control. The dosing volume was adjusted according to individual body weight, which was recorded weekly throughout the treatment period.

Y-maze

Following a 5-day handling period to acclimate mice to the experimenter, the Y-maze test was conducted to evaluate short-term spatial working memory and exploratory behavior. Spontaneous alternation behavior was used as an index of spatial memory, while total arm entries were used to assess locomotor activity. Each mouse was placed at the end of one arm (usually labeled A) of an opaque Y-shaped maze consisting of three identical arms (40 cm length, 10 cm width, 12 cm height) and allowed to explore freely for 10 min. The maze was cleaned with 70% ethanol and deionized water between trials to eliminate olfactory cues. An arm entry was defined as the placement of all four paws beyond the midpoint of an arm. The sequence and total number of arm entries were recorded by an experimenter blinded to the groups, and spontaneous alternation was calculated as the ratio of actual alternations (triads of consecutive entries into three different arms) to the maximum possible alternations, expressed as a percentage.

Contextual fear conditioning (CFC)

CFC was performed for 3 days (habituation, training, and test day). For the habituation day, each mouse was placed in a chamber for 5 min to become familiarized with the chamber. On the training day, mice received two 0.86 mA electric foot shocks for 2 s each through the platform. On the test day, freezing responses of the mice were examined for 5 min and measured by the Packwin software. A breathing filter was applied, and the freezing threshold was set to 1 s.

Morris water maze (MWM)

The MWM was used to assess whether YIAD-1003 treatment improved spatial learning and memory in 5XFAD mice. The apparatus consisted of a circular pool (50 cm height, 90 cm radius) divided into four equal quadrants (A-D) and filled with water maintained at 22 ± 1 °C. Nontoxic white opaque paint was added to render a circular escape platform (9 cm radius) invisible beneath the water surface. Mice underwent acquisition training for five consecutive days. During each trial, mice were allowed a maximum of 60 s to locate the hidden platform. Animals that failed to find the platform within this time were gently guided to it and required to remain there for 10 s to facilitate spatial learning.

A probe trial was conducted on day 6, during which the platform was removed. Mice were released into the pool and allowed to swim freely for 60 s. Swimming trajectories were recorded and analyzed using SMART Video Tracking software (Panlab). Spatial memory performance was evaluated by quantifying time spent in the target quadrant, time spent in the opposite quadrant, and the number of target platform crossings. These parameters were compared across experimental groups to assess YIAD-1003-mediated cognitive improvement. Following completion of the behavioral testing, all mice were sacrificed.

Immunohistochemical (IHC) analysis

To obtain the brain samples, mice were anesthetized with 4% avertin via intraperitoneal injection, analgesia confirmed with a toe pinch, and cardiac perfusion was performed with 0.9% saline solution. The right hemisphere was fixed with 4% paraformaldehyde for 24 h at 4 °C and moved to a 30% sucrose solution until the brain sank at 4 °C. Prepared mice brain hemispheres were frozen in an optimal cutting temperature compound and cut in 25 μm sections using Cryostat (CM1860, Leica). Brain sections were mounted on glass slides and dried completely for IHC staining. Antigen retrieval was done with 1% sodium dodecyl sulfate diluted in PBS for 10 min and blocked with 20% goat serum for 1 h. Primary antibody solution containing 6E10 (BioLegend, SIG-39320, 1:200) and GFAP (Millipore, AB5541, 1:200) were treated overnight at 4 °C. Following primary antibody incubation, fluorescent dye-conjugated secondary antibodies were treated for 1 h at 4 °C. Nucleic acid staining was done with Hoechst 33,342 (Sigma-Aldrich, 10 µg/mL) for 3 min. After staining, the brain slides were imaged with a fluorescence microscope (DM2500, Leica) with the LAS X software program. All obtained images were analyzed using ImageJ software.

Western blot

While the right-brain hemisphere was stored as a whole and used for IHC assays, the left-brain hemisphere was dissected into cortex, hippocampus, and other sections upon sacrifice. The hippocampal tissues were homogenized in radioimmunoprecipitation assay (RIPA) buffer (R0278, Sigma-Aldrich) containing protease inhibitor (Roche, 11836170001) and PhosStop (Roche, 04906845001) and incubated for 30 min on ice. Then the homogenized tissues were centrifuged at 14,000 rpm for 3 min to collect the RIPA-soluble fraction. Protein concentrations of the RIPA-soluble fraction were measured using the Pierce™ BCA Protein Assay Kit (ThermoFisher, 23225) and 15 µg of protein of each sample were separated through SDS-PAGE. Separated proteins were then transferred to a nitrocellulose membrane and checked with Ponceau staining. After blocking the membrane with 5% skim milk in TBS-T to prevent nonspecific binding, primary antibodies were treated overnight at 4 °C, followed by HRP-conjugated secondary antibodies for 1 h at RT. The primary antibodies used for western blot are anti-GFAP antibody (Millipore, AB5541, 1:4,000), anti-Iba1 antibody (Cell Signaling, 17198 S, 1:1,000), and anti-β-actin antibody (Millipore, MAB1501R, 1:10,000). Positive signals were detected with SuperSignal West Pico PLUS Chemiluminescent Substrate (ThermoFisher, 34580) using the FUSION Solo S software program. All obtained images were analyzed using ImageJ software.

Enzyme-linked immunosorbent assay (ELISA)

Human Aβ₄₀ and Aβ₄₂ levels were quantified in the hippocampal brain lysates using a human Aβ₄₀ (Invitrogen, KHB3481) and Aβ₄₂ ELISA kit (Invitrogen, KHB3441) according to the manufacturer’s instructions. Protein concentrations of the brain lysates were measured using the Pierce™ BCA Protein Assay Kit (ThermoFisher, 23225) and ELISA samples were prepared to contain 5 µg of hippocampal protein in each well. After loading 50 µL of each sample, 50 µL of detection antibody was added and incubated overnight at 4 °C. Then, HRP-conjugated anti-rabbit IgG antibody was treated and incubated for 30 min at RT. Following this step, a stabilized chromogen solution was added and incubated in the dark for 30 min and the reactions were concluded with a stop solution. The absorbance was measured at 450 nm using a microplate reader (SpectraMax M2e). Final Aβ₄₀ and Aβ₄₂ concentrations were calculated according to the manufacturer’s instructions.

Pharmacokinetic study of YIAD-1003

Pharmacokinetic profile and brain distribution of YIAD-1003 were investigated in male ICR mice following a single intravenous administration of 30 mg/kg via the tail vein. Brain and plasma samples were collected at 5, 15, and 30 min post-administration (n = 3 per time point). Blood was collected in heparinized tubes and whole brains were rapidly removed. Whole brains were rinsed with PBS, cut into hemispheres, weighed, and snap-frozen on dry ice. Plasma was obtained after centrifuging the blood samples at 3,000 g for 15 min at 4 °C and stored at -80 °C until analysis. Brain hemispheres were homogenized in PBS at three times the tissue weight, centrifuged at 14,000 g for 14 min at 4 °C, and the resulting supernatants from each hemisphere were combined into a single tube. Before analysis, plasma and brain homogenates were transferred to microcentrifuge tubes, and a fixed volume of internal standard solution was added, followed by the addition of ice-cold acetonitrile for protein precipitation. The mixtures were vortexed thoroughly and incubated on ice for 10 min, then centrifuged at 14,000 g for 14 min at 4 °C. The resulting supernatants (about 80 µL) were collected, and an aliquot of the samples was transferred to liquid chromatography (LC) vials for LC-quadrupole time-of-flight mass spectrometry (LC-QTOF-MS) analysis.

Quantification of YIAD-1003 in plasma and brain samples was performed using an Agilent 1290 LC system coupled to an Agilent 6533 QTOF (Agilent Technologies) equipped with an electrospray ionization source operating in positive ion mode. Chromatographic separation was achieved on a Purospher STAR RP-18 endcapped column (150 × 4.6 mm, 3 μm; Merck). The mobile phases consisted of water with 0.1% formic acid (solution A) and acetonitrile with 0.1% formic acid (solution B). YIAD-1003 was eluted using a linear gradient from 50% to 95% solution B over 7 min, held at 95% B for 1 min, and then returned to 40% B at 8.1 min with re-equilibration until 10 min, at a flow rate of 0.8 mL/min. The injection volume was 20 µL. The MS source parameters were as follows: ion source, dual Agilent Jet Stream ESI (positive); fragmentor voltage, 125 V; MS scan range m/z 100–3000; collision energy, 25 eV. Targeted MS/MS acquisition was performed with the transition m/z 451.17 → 395.11 for YIAD-1003. Data were acquired and processed using MassHunter Workstation software (Agilent Technologies).

Statistical analysis

All data were analyzed with one-way ANOVA followed by Bonferroni’s post hoc multiple-comparison test using GraphPad Prism 10 software (*P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001). Data are presented as mean ± SEM.

Results

YIAD-1003 inhibits Aβ fibrillization and dissociates preformed Aβ₄₂ aggregates

To identify small-molecule modulators of Aβ aggregation, we synthesized a library of 20 dihydropyrrolo[1,2-a]pyrazine derivatives (Fig. 1A) [18]. This scaffold has been reported to exhibit anti-amyloid and neuroprotective activities, although the precise molecular mechanism underlying Aβ modulation remains to be fully elucidated [28, 29]. The chemical structures of all derivatives (YIAD-1000 through YIAD-1019) are shown in Fig. 1B.

Fig. 1.

Synthesis scheme and chemical structures of dihydropyrrolo[1,2-a]pyrazine derivatives. A General synthetic procedure for the preparation of YIAD-1000 through YIAD-1019. B Chemical structure of all 20 derivatives

To evaluate the therapeutic potential of the synthesized library against Aβ₄₂ aggregation, we followed a workflow encompassing compound synthesis, in vitro screening, and in vivo validation in both acute and chronic AD mouse models (Fig. 2A). Primary screening of the library was performed using ThT, a fluorescent dye that quantifies fibrillar content by binding to β-sheet structures in Aβ aggregates [30, 31]. We focused on Aβ₄₂ as the primary screening target, as it is the most aggregation-prone and neurotoxic isoform of Aβ [32]. To assess inhibition of Aβ₄₂ aggregation, monomeric Aβ₄₂ (25 µM) was co-incubated with each compound (50, 100, or 200 µM) for 24 h at 37 °C (Fig. 2B). Compound concentrations were adjusted based on solubility constraints, with some derivatives tested only at 50 and 100 µM. All compounds reduced Aβ₄₂ fibril formation compared to Aβ₄₂-only controls at different levels. Among the 20 derivatives, YIAD-1003 exhibited the most significant dose-dependent inhibition of fibril formation.

Fig. 2.

Experimental workflow and screening of dihydropyrrolo[1,2-a]pyrazine derivatives to evaluate chemically driven inhibition and dissociation of Aβ42. A Schematic overview of the experimental workflow from in vitro compound screening to in vivo validation studies. B Inhibitory effects of synthesized compounds in three concentrations (50, 100, or 200 μM) against Aβ42 (25 μM) aggregation evaluated through ThT fluorescence assay. Inhibitory effects were presented in percentage (%) compared to non-compound treated Aβ42 aggregate. C Dissociative effects of synthesized compounds (50, 100, or 200 μM) against Aβ42 fibrils (25 μM) evaluated through ThT fluorescence assay. Dissociative effects were presented in percentage (%) compared to non-compound treated Aβ42 aggregate. D Schematic illustration and (E) results of the A3 plate assay evaluating dissociative efficacy of selected compounds (50 or 100 μM) from the screening process. All statistical analyses were done using one-way ANOVA followed by Bonferroni’s post hoc multiple-comparisons test (*P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001 vs Aβ-only aggregates). Data are presented as mean ± SEM. FITC, fluorescein isothiocyanate. Schematic icons (mice, plate, and behavioral analyses icons) in panel A were created with AI image generation assistance

To evaluate dissociation capacity, Aβ₄₂ was pre-aggregated for 24 h prior to compound treatment. Compounds were then added to the pre-formed aggregates and incubated for an additional 24 h at 37 °C (Fig. 2C). While most compounds reduced ThT fluorescence by approximately 50% relative to the pre-aggregate control (P < 0.0001), YIAD-1003 treatment reduced fluorescence to 34.2% at 200 µM, indicating pronounced dissociation efficacy. Based on these fibril-based screening results, YIAD-1003 was identified as the most effective compound for both inhibiting Aβ₄₂ aggregation and dissociating preformed aggregates.

Given that Aβ oligomers represent the most neurotoxic aggregation state, we next evaluated the oligomer-dissociating activity of YIAD-1003 using the amyloid aggregate (A3) assay. This assay quantifies disruption of pre-formed oligomeric FITC-labeled Aβ₄₂ by measuring fluorescence after compound treatment (Fig. 2D) [20]. Compounds (50 or 100 µM) were incubated with pre-formed oligomers for 24 h. Among all tested compounds, YIAD-1003 exhibited the highest dissociation rate of 49.5% at 100 µM (Fig. 2E). Based on both the fibril and oligomer-based screening results, YIAD-1003 was selected for comprehensive characterization.

To further characterize the dose-dependent activity of YIAD-1003, we performed concentration-response assays at eight concentrations (4-500 µM; two-fold serial dilutions). For inhibition assays, monomeric Aβ₄₂ (25 µM) was co-incubated with YIAD-1003 for 24 h. YIAD-1003 produced a concentration-dependent reduction in ThT fluorescence, with a maximal inhibition of 88.2% at 500 µM (P < 0.0001; Fig. 3A). For dissociation assays, pre-formed Aβ₄₂ aggregates were treated with YIAD-1003 for 24 h. YIAD-1003 likewise exhibited dose-dependent dissociation of pre-formed aggregates, with a maximal reduction of 77% in ThT fluorescence at 500 µM (P < 0.0001; Fig. 3C).

Fig. 3.

Dose-dependent inhibition and dissociation of Aβ isoforms by YIAD-1003. A, E, I Dose-dependent inhibitory effect and IC₅₀ values of YIAD-1003 on Aβ₄₂, Aβ₄₀ and Aβ_pE3, respectively. YIAD-1003 (4-500 μM, two-fold serial dilutions) was incubated with the different isoforms at 37 °C. B, F, J Aβ oligomers were detected by an A11 antibody through dot blot. C, G, K Dose-dependent dissociative effect and EC₅₀ values of YIAD-1003 on Aβ₄₂, Aβ₄₀, and Aβ_pE3, respectively. The same concentrations of YIAD-1003 were treated on pre-aggregated Aβ samples to evaluate how effectively YIAD-1003 can dissociate pre-formed fibrils. D, H, L Aβ oligomers in all samples were detected by an A11 antibody through dot blot. All statistical analyses were done using one-way ANOVA followed by Bonferroni’s post hoc multiple-comparisons test (*P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001 vs Aβ-only aggregates). Data are presented as mean ± SEM. Agg, aggregation; EC₅₀, half-maximal effective concentration; IC₅₀, half-maximal inhibitory concentration; 1003, YIAD-1003

To determine whether the observed reductions in fibrillar content were accompanied by changes in oligomeric species, we performed A11 immunoblotting on samples from both inhibition and dissociation assays. YIAD-1003 treatment significantly reduced A11 immunoreactivity in both assay formats compared to Aβ-only controls (Fig. 3B, D), indicating suppression of oligomer formation during aggregation and dissociation of oligomers from pre-formed aggregates. Together, these data demonstrate that YIAD-1003 exhibits significant dose-dependent inhibition and dissociation of both fibrillar and oligomeric Aβ₄₂ species.

YIAD-1003 exhibits isoform-agnostic inhibitory and dissociative activity against Aβ₄₀, Aβ₄₂, and Aβ_pE3−42

To determine whether the inhibitory and dissociative activity of YIAD-1003 extended beyond Aβ₄₂, we evaluated its effects on Aβ₄₀ and Aβ_pE3 aggregation under the same concentration-response conditions. YIAD-1003 exhibited comparable dose-dependent inhibition of both isoforms (Fig. 3E, I), with maximal ThT fluorescence reductions of 81.1% for Aβ₄₀ and 89.4% for Aβ_pE3 (P < 0.0001). A11 immunoblotting analysis confirmed significant suppression of oligomeric species for both Aβ₄₀ (53.5% reduction) and Aβ_pE3 (88.7% reduction) compared to the Aβ-only controls (P < 0.0001; Fig. 3F, J), indicating effective inhibition across distinct Aβ isoforms. From concentration-response curves, the half-maximal inhibitory concentration (IC₅₀) values for inhibition of Aβ₄₂, Aβ₄₀, and Aβ_pE3 fibrillization were approximately 90.36, 36.48, 38.82 µM, respectively (Fig. S2A).

We next examined the ability of YIAD-1003 to dissociate preformed aggregates of Aβ₄₀ and Aβ_pE3. YIAD-1003 reduced ThT fluorescence in a dose-dependent manner for both isoforms, with maximal reduction of 78.9% for Aβ₄₀ and 72.6% for Aβ_pE3 at 500 µM (P < 0.0001; Fig. 3G, K). Consistent with fibril dissociation, A11 immunoblot analysis demonstrated corresponding reduction in oligomeric species (50.6% for Aβ₄₀, 69.8% for Aβ_pE3; P < 0.0001; Fig. 3H, L). The half-maximal effective concentration (EC₅₀) values for dissociation of preformed Aβ₄₂, Aβ₄₀, and Aβ_pE3 fibrils were approximately 62.52, 25.70, and 138.0 µM, respectively (Fig. S2B). Together, these data demonstrate that YIAD-1003 exhibits robust inhibitory and dissociative activity across multiple Aβ isoforms, supporting isoform-agnostic modulation of pathogenic Aβ assemblies.

YIAD-1003 preferentially disrupts aggregation-prone regions of Aβ₄₂

To gain insight into the sequence-dependent determinants underlying YIAD-1003-mediated aggregate destabilization, we employed a fragment-based aggregate dissociation assay using a MAP, as previously established [20]. In this assay, immobilized Aβ₄₂-derived hexamer fragments serve as sequence-defined interaction interfaces that support the retention of aggregated full-length Aβ, enabling resolution of compound effects at the level of individual aggregation-prone motifs. Following an 8 h aggregation period, YIAD-1003 (250 µM) was applied, and residual fragment-associated aggregate signal was quantified at 16 h post-treatment. Reduction in signal reflects disruption of stabilizing interactions between immobilized fragment sequences and aggregated Aβ, rather than direct binding to monomeric peptides (Fig. 4A).

Fig. 4.

YIAD-1003 binding analysis to Aβ via MAP assay and molecular docking simulations. A Schematic illustration of the fragment-based aggregate dissociation Mapping Amyloid Plate (MAP) assay experiment. B Aggregate dissociation rates of YIAD-1003 on each hexamer fragment-based fAβ₄₂ aggregates through the decreased fluorescence intensities normalized to those of each well before YIAD-1003 treatment. Data are presented as mean ± SEM. C Amino acid-level heat map summarizing sequence-dependent dissociation efficiencies. Molecular docking simulation results of YIAD-1003 interaction with Aβ₄₂ (D) and Aβ₄₀ (E)

YIAD-1003 induced differential dissociation across Aβ₄₂ fragments, with pronounced disruption observed for fragments encompassing residues 5–10 and 15–20, corresponding to aggregation-prone regions of the peptide (Fig. 4B). In contrast, fragments derived from the N-terminal region exhibited comparatively modest or negligible responses to YIAD-1003, consistent with the lower contribution of these sequences to aggregate core formation and their solvent-exposed nature in aggregated assemblies.

To visualize sequence-resolved dissociation behavior, dissociation efficiencies were summarized in a heat map (Fig. 4C). Regions corresponding to the central hydrophobic core and C-terminal β-sheet-forming segments showed consistently elevated sensitivity, whereas N-terminal residues remained largely unaffected. Together, these data indicate that YIAD-1003 preferentially destabilizes aggregates by targeting sequence elements that contribute directly to aggregation nucleation and structural reinforcement, rather than indiscriminately disrupting peptide-surface interactions.

Preferential association of YIAD-1003 with aggregation-prone Aβ regions revealed through computational docking

To further examine whether the sequence-dependent dissociation patterns observed experimentally were consistent with predicted interaction preferences, computational docking analyses were performed using Aβ₄₂ and Aβ₄₀ structural models. YIAD-1003 was docked onto representative conformations of aggregated Aβ using Avogadro 2, and predicted interaction poses were evaluated based on docking scores and spatial localization (Fig. 4D, E).

According to molecular docking simulation results, the binding affinity of YIAD-1003 was calculated as − 5.11 kcal/mol with Aβ₄₂ and − 4.99 kcal/mol with Aβ₄₀. In the convolutional neural network docking score, which demonstrated the confidence level of the docking pose, Aβ₄₂ was calculated with a higher score (0.9185) compared to Aβ₄₀ (0.8669) [24]. The docking simulation of YIAD-1003 with Aβ₄₂ showed that YIAD-1003 formed hydrogen bonds with Phe20 and Lys28, hydrophobic interactions with Phe19, Phe20, and Val24, and van der Waals interactions with Val18 and Asn27, whereas Phe19 established π-π stacking with the aromatic ring of YIAD-1003. The residues His6, Gly9, Tyr10, Glu11, and Asn27 of Aβ₄₀ showed hydrogen bonding with YIAD-1003, whereas Ser8 and Tyr10 engaged in van der Waals interaction and hydrophobic interaction, respectively.

Comparison of docking-derived interaction patterns with fragment dissociation profiles revealed qualitative correspondence between regions of predicted association and experimentally observed aggregate disruption (Fig. 4E). Although these computational analyses do not establish direct binding or a specific molecular mechanism, they provide supportive evidence that YIAD-1003 preferentially associates with aggregation-prone regions of Aβ that are critical for fibril stability.

YIAD-1003 improves cognitive performance in acute and transgenic AD mouse models

The functional effects of YIAD-1003 were first evaluated in an acute Aβ₄₂-induced mouse model. Aβ₄₂ (10 µM) was co-administered with YIAD-1003 (400 µM) via ICV injection in a total volume of 5 µL (5.5% DMSO in PBS). The experiment scheme is illustrated in Fig. 5A. Spatial working memory was assessed using the Y-maze spontaneous alternation task 5 days following ICV injection. Representative heat maps of mouse trajectories during the sessions show Aβ₄₂-injected mice to have a stronger preference for only two arms (Fig. 5B). Relative to Aβ₄₂-injected mice, animals co-injected with YIAD-1003 exhibited a significant increase in spontaneous alternation behavior (62.4% vs. 68.9%, P = 0.0182), while total arm entries were not significantly altered, indicating that changes in task performance were not attributable to altered locomotor activity (Fig. 5C). Additionally, CFC was performed to evaluate hippocampus-dependent associative memory. Consistent with the Y-maze results, Aβ42-injected mice showed impaired contextual memory, as evidenced by reduced freezing behavior, whereas mice co-injected with YIAD-1003 exhibited restored freezing responses (Fig. 5D).

Fig. 5.

Therapeutic efficacy of YIAD-1003 in AD models following different administration routes. A Experimental scheme of in vivo studies utilizing an acute AD mouse model. B Representative heat map of mice paths during Y-maze session. C Spontaneous alternations (%) and total number of entries (n). D Freezing duration (%) of each mouse during the third day of CFC. E BBB permeability of YIAD-1003 assessed by BBB-PAMPA. F Pharmacokinetic profile of YIAD-1003 in plasma and brain after intravenous administration to ICR mice. G Experimental scheme of in vivo studies utilizing 5XFAD mouse model. H Representative heat map of mice paths during Y-maze session. I Spontaneous alternations (%) and total number of arm entries (n) of 5XFAD mice. J Learning curve of 5XFAD mice over the course of the training period (5 days). K Representative path tracking of each group during the probe day. Time in zone (%) mice spent in target (L) and opposite (M) quadrants during the probe day test. All statistical analyses were done using one-way ANOVA followed by Bonferroni’s post hoc multiple-comparisons test (*P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001 vs Aβ-treated group and TG vehicle group, respectively). Data are presented as mean ± SEM. ICV, intracerebroventricular; conc., concentration; TG, transgenic; Veh, vehicle; WT, wild-type; 1003, YIAD-1003

To evaluate the brain permeability of YIAD-1003, passive BBB permeability was assessed using the BBB-PAMPA assay. Under the assay conditions, YIAD-1003 exhibited an effective permeability coefficient of 2.34 × 10⁻⁶ cm/s, intermediate between the kit’s high- (5.56 × 10⁻⁶ cm/s) and low-permeability (1.64 × 10⁻⁶ cm/s) reference compounds (Fig. 5E), indicating moderate passive BBB permeability. Consistent with the PAMPA results, a short-term in vivo pharmacokinetic study further confirmed that YIAD-1003 reaches the brain following systemic administration. Following an intravenous dose of 30 mg/kg in ICR mice, YIAD-1003 was detected in plasma and brain samples. The brain-to-plasma area-under-the-curve (AUC) ratio (K_p, 5–30 min) was approximately 0.16 (Fig. 5F), indicating that brain exposure was approximately 16% of the corresponding plasma exposure over the sampled interval. Together, these findings support the suitability of YIAD-1003 for subsequent in vivo evaluation following oral gavage administration in transgenic AD mice.

The effects of orally administered YIAD-1003 were next examined in 5XFAD mice following the experimental scheme (Fig. 5G). Animals received oral gavage of YIAD-1003 at 30 mg/kg/day for 4 weeks, followed by behavioral assessment. Biosafety of YIAD-1003 was monitored through the body weight changes of TG (vehicle) and TG (YIAD-1003) mice throughout the administration. Body weight of examined mice was not statistically different from the start to the end (Fig. S3). In the Y-maze test, YIAD-1003-treated 5XFAD mice exhibited increased spontaneous alternation compared with vehicle-treated transgenic controls (49.8% vs. 65.2%, P = 0.0014; Fig. 5H, I). MWM was conducted in which mice were trained to find a hidden platform for 5 days then examined to find the removed platform for 1 min on the final day. YIAD-1003-treated mice showed similar learning curves to the wild-type group indicating ameliorated long-term memory compared to the vehicle-treated 5XFAD group (Fig. 5J). During the probe sessions, YIAD-1003-treated mice showed comparable tracks to the wild-type mice as shown in the representative tracking (Fig. 5K). Furthermore, YIAD-1003-treated mice spent significantly less time in the opposite zone of the target (Fig. 5L) and more time in the target zone (Fig. 5L and M) indicating improved long-term memory.

Together, these findings demonstrate improved cognitive performance of YIAD-1003-treated mice in two different AD mouse models.

Chronic YIAD-1003 administration reduces amyloid plaque burden and soluble Aβ levels in 5XFAD mouse brains

To determine whether the behavioral improvements observed following oral administration of YIAD-1003 were accompanied by changes in amyloid pathology, IHC analyses of 5XFAD mouse brain samples were performed. Amyloid plaque burden was assessed by histological staining of brain sections encompassing the cortex and hippocampus.

Quantitative analysis following the scheme (Fig. 6B) revealed a significant reduction in plaque burden in YIAD-1003-treated 5XFAD mice compared with vehicle-treated controls. In the cortex, plaque count was reduced by 30.7% (P = 0.0245), and hippocampal plaque count was decreased by 85.4% (P = 0.0145, Fig. 6C). Representative images demonstrated fewer and smaller plaque deposits in YIAD-1003-treated 5XFAD mice relative to the vehicle-treated transgenic mice (Fig. 6A).

Fig. 6.

YIAD-1003 reduces amyloid burden and Aβ peptide levels in 5XFAD mouse brains. A Right hemisphere immunohistochemical (IHC) images of Aβ plaques stained with 6E10 antibody in two different magnifications. Scale bars represent 2 mm and 500 μm. B Brain schematic diagram of areas imaged and used for each quantification data. C Quantification of the area and number of 6E10-stained Aβ plaques of total, cortical, and hippocampal regions. D Amount of Aβ₄₀ and Aβ₄₂ in soluble hippocampal brain fractions detected through ELISA. All statistical analyses were done using one-way ANOVA followed by Bonferroni’s post hoc multiple-comparisons test (*P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001 vs post-administration TG vehicle group and TG vehicle group, respectively). Data are presented as mean ± SEM. Post, post-administration; Pre, pre-administration; Veh, vehicle; 1003, YIAD-1003

To further assess the effects of YIAD-1003 on soluble Aβ levels, ELISA was conducted. Chronic oral administration of YIAD-1003 resulted in a significant reduction in soluble Aβ₄₀ and Aβ₄₂ levels, with decreases of 24.4% (P = 0.0091) and 12.6% (P = 0.04), respectively, compared with vehicle-treated 5XFAD mice (Fig. 6D). These findings demonstrate that chronic oral administration of YIAD-1003 is associated with reduced amyloid plaque burden and decreased levels of soluble Aβ species in the brains of 5XFAD mice.

YIAD-1003 attenuates astrocytic and microglial activation in 5XFAD mice

To evaluate whether the reduction in amyloid pathology following chronic oral administration of YIAD-1003 was accompanied by changes in inflammation markers, brain sections from treated 5XFAD mice were analyzed for markers of astrocytic and microglial reactivity. Hippocampal sections were immunostained for GFAP and Iba-1 to assess astrocytic and microglial activation.

Quantitative analysis revealed a significant reduction in GFAP immunoreactivity in YIAD-1003-treated 5XFAD mice compared with vehicle-treated controls. GFAP-positive area was decreased by 24.1% in the hippocampus (P = 0.0006; Fig. 7B, D), indicating attenuation of astrocytic activation. Immunostained brain slide images demonstrated reduced GFAP signal in YIAD-1003-treated animals relative to controls (Fig. 7A). Microglial activation was similarly affected by YIAD-1003 treatment. Iba-1-positive area was reduced by 51.5% in the hippocampus compared with vehicle-treated 5XFAD mice (P = 0.0053; Fig. 7C, D). To confirm the consistency of these effects across animals, all data points for GFAP and Iba-1 quantification are shown in Fig. S6. Across all analyzed regions and markers, YIAD-1003 treatment was associated with reduced glial reactivity relative to vehicle-treated 5XFAD controls indicating amelioration of neuroinflammation.

Fig. 7.

YIAD-1003 attenuates astrocytic and microglial activation in 5XFAD mouse brains. A Magnified immunohistochemical (IHC) images of astrocytes (red) in the hippocampal regions of 5XFAD mice. Scale bars represent 500 μm. B Imaging area of the hippocampus indicated with a box. C Expression levels of GFAP and Iba-1 in the RIPA-soluble hippocampal fractions of 5XFAD mice. D Relative intensities were normalized to β-actin levels of each mouse. All statistical analyses were done using one-way ANOVA followed by Bonferroni’s post hoc multiple-comparisons test (*P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001). Data are presented as mean ± SEM. TG, transgenic; Veh, vehicle; WT, wild-type; 1003, YIAD-1003

Discussion

Aberrant aggregation of Aβ into structurally heterogeneous assemblies is a central pathological feature of AD and a persistent therapeutic challenge. Here, we demonstrated that YIAD-1003, a small-molecule modulator, effectively targets pathogenic Aβ aggregation through both inhibitory and dissociative activities across multiple isoforms. Through in vitro and in vivo studies, YIAD-1003 suppressed fibril formation, dissociated preformed aggregates, improved cognitive performance, and reduced amyloid burden and glial activation in AD mouse models.

The most noteworthy property of YIAD-1003 is its robust activity across multiple Aβ isoforms, including Aβ₄₂, Aβ₄₀, and Aβ_pE3, which differ markedly in aggregation behavior and pathological relevance. Given the coexistence of diverse Aβ species and assembly states in the AD brain, isoform- or conformation-selective strategies are inherently limited in scope. The ability of YIAD-1003 to modulate aggregation and dissociate assemblies across structurally distinct isoforms demonstrates a mechanism that is not restricted to a single aggregation pathway, addressing a critical limitation of more selective therapeutic approaches.

A key distinguishing feature of YIAD-1003 is potent dissociative activity against preformed aggregates, complementing its inhibitory effects on de novo fibrillization. The MAP assay revealed preferential disruption of aggregation-prone regions within Aβ₄₂, particularly the central hydrophobic core and C-terminal segment. Computational docking analyses confirmed qualitative correspondence with these experimentally sensitive regions, providing mechanistic support for selective interaction with structural elements critical for fibril stability.

Importantly, YIAD-1003 demonstrated robust therapeutic efficacy in both acute and transgenic AD mouse models. In an acute Aβ₄₂ injection model, ICV co-administration of YIAD-1003 improved memory performance without affecting locomotion. Leveraging its brain-penetrant properties, oral administration in 5XFAD mice reduced amyloid plaque burden, soluble Aβ levels, and glial activation, accompanied by improved cognitive function. These findings establish that aggregation modulation by YIAD-1003 translates into measurable behavioral and pathological benefits, validating its therapeutic potential.

Although YIAD-1003 promotes dissociation of Aβ aggregates in vitro, chronic treatment reduced soluble Aβ₄₀ and Aβ₄₂ levels in the hippocampus in vivo. This observation is not necessarily contradictory, because aggregate disassembly in the brain does not occur in a closed system. Dissociated Aβ species may undergo rapid clearance through multiple pathways, including BBB-mediated efflux, CSF/glymphatic drainage, and enzymatic degradation, thereby preventing sustained accumulation of soluble Aβ in brain tissue [33]. The ELISA results in the present study reflect endpoint Aβ levels following prolonged administration rather than transient increases in soluble intermediates immediately after aggregate dissociation. Consistent with this interpretation, EPPS, a previously reported Aβ-dissociating small-molecule, reduces cerebral Aβ oligomer and plaque burden while promoting brain-to-blood efflux of dissociated Aβ species in an AD mouse model, resulting in an overall reduction of brain Aβ rather than accumulation of soluble Aβ [34, 35]. Therefore, the observed reduction in plaque burden and soluble Aβ after YIAD-1003 treatment likely reflects net depletion of Aβ burden through coupled dissociation and clearance mechanisms, rather than evidence against its aggregate-dissociating activity.

Taken together, our findings identify YIAD-1003 as a small-molecule modulator of heterogeneous Aβ isoforms that improves cognitive and pathological outcomes in both acute and 5XFAD mouse models. Nonetheless, several limitations should be acknowledged. First, our study focuses on Aβ pathology and does not examine the effects of YIAD-1003 on tau aggregation or tau-related neurodegeneration, even though hyperphosphorylated tau and neurofibrillary tangles represent another major hallmark of AD. Future studies will therefore need to determine whether YIAD-1003 influences tau pathology and to evaluate the compound in tau-driven or Aβ/tau models. Second, although we have confirmed brain exposure of YIAD-1003, comprehensive pharmacokinetic/pharmacodynamic profiling and target-engagement studies remain necessary to fully assess the translational potential of this scaffold. Despite these limitations, YIAD-1003 represents a meaningful advance in addressing the challenge of Aβ heterogeneity. By combining multifaceted anti-amyloid activity, inhibition and dissociation, with multi-isoform targeting capability, YIAD-1003 establishes a promising chemical scaffold for further development.

Conclusion

In summary, YIAD-1003 is a small-molecule capable of modulating heterogeneous Aβ assemblies through both inhibitory and dissociative activities across multiple pathogenic isoforms, including Aβ₄₂, Aβ₄₀, and Aβ_pE3. By preferentially destabilizing aggregation-prone regions of Aβ, YIAD-1003 attenuated fibril formation and disrupted preformed aggregates in vitro, with corresponding reductions in oligomeric species. These properties were translated into functional and pathological improvement in vivo. Both the acute Aβ₄₂-induced model and 5XFAD mice demonstrated improved cognitive performance following YIAD-1003 administration. 5XFAD mice showed reduced amyloid plaque burden, decreased soluble Aβ levels, and attenuation of astrocytic and microglial activation. Given the structural and biochemical heterogeneity of Aβ species, therapeutic strategies that target multiple aggregation states may offer advantages over selective approaches. While further pharmacokinetic and mechanistic studies are warranted, these findings establish YIAD-1003 as a potential multi-targeting small-molecule therapeutic candidate for AD.

Abbreviations

Aβ

Amyloid-β

Aβ₄₀

Amyloid-β (1–40)

Aβ₄₂

Amyloid-β (1–42)

Aβ_pE3

Pyroglutamate-modified amyloid-β (3–42)

AD

Alzheimer’s disease

ANOVA

Analysis of variance

BBB

Blood-brain barrier

BCA

Bicinchoninic acid

CFC

Contextual fear conditioning

DMSO

Dimethyl sulfoxide

DW

Deionized water

ELISA

Enzyme-linked immunosorbent assay

FITC

Fluorescein isothiocyanate

Fmoc

9-Fluorenylmethoxycarbonyl

GFAP

Glial fibrillary acidic protein

HRP

Horseradish peroxidase

Iba-1

Ionized calcium-binding adapter molecule 1

ICV

Intracerebroventricular

IHC

Immunohistochemical

LC

Liquid chromatography

MAP

Mapping amyloid plate

MWM

Morris water maze

NIH

National Institutes of Health

PAMPA

Parallel artificial membrane permeability assay

PDB

Protein Data Bank

PLIP

Protein-Ligand interaction profiler

QTOF-MS

Quadrupole time-of-flight mass spectrometry

RIPA

Radioimmunoprecipitation assay

SEM

Standard error of the mean

SDS-PAGE

Sodium dodecyl sulfate-polyacrylamide gel electrophoresis

TBS-T

Tris-buffered saline with Tween-20

TG

Transgenic

ThT

Thioflavin T

WT

Wild-type

Authors’ contributions

Y.S.K. and I.K. conceived and designed the study. J.K. and D.K. performed the experiments, conducted data analysis, and prepared the figures. J.K. and H.Y.K. drafted the manuscript. I.W.P. and S.Y. synthesized and prepared the Aβ peptides. I.C. and M.P. performed the MAP and A3 assays. W.S. conducted the molecular docking analyses. S.L. and D.K. prepared the compound library and YIAD-1003 for administration. J.K., M.P., and I.W.P. conducted the PK experiment. Y.S.K., I.K., and H.Y.K. supervised the project and secured funding. All authors read and approved the final manuscript.

Funding

This research was supported by a grant of the Korea Dementia Research Project through the Korea Dementia Research Center (KDRC), funded by the Ministry of Health & Welfare and Ministry of Science and ICT, Republic of Korea (Grant Number: RS-2024-00349158), and Mid-Career Researcher Program (Grant Number: RS-2025-00523607, I.K.; RS-2021-NR059653, Y.S.K.), and Basic Science Research Program (Grant Number: RS-2018-NR031048, H.Y.K., I.K., and Y.S.K.) through the National Research Foundation of Korea (NRF), funded by the Ministry of Health & Welfare and Ministry of Science and ICT, Republic of Korea. This research was also supported by Korea Institute for Advancement of Technology (KIAT) funded by the Ministry of Trade, Industry and Energy in 2024 (Grant Number: RS-2024-00418203) and Amyloid Solution Inc.

Data availability

The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.

Declarations

Ethics approval and consent to participate

All animal experiments involving transgenic mice were approved by IACUC of the Yonsei Laboratory Animal Research Center (Approval No. IACUC-202411-1948-01) and were conducted in accordance with institutional and national guidelines for the care and use of laboratory animals.

Consent for publication

Not applicable.

Competing interests

Y.S.K. is an employee of Amyloid Solution and holds stock options and equity interests in the company. All other authors declare no competing interests.