Baicalein inhibits amyloid beta42 aggregation through disruption of the Asp23-Lys28 salt bridge

Abstract

The aggregation of amyloid beta-42 (Aβ42) into β-sheet-rich fibrillar structures is a critical pathogenic feature of Alzheimer’s disease (AD). Baicalein (BCN), a natural flavonoid, has been shown to inhibit aggregation in amyloidogenic proteins, including human islet amyloid polypeptide (hIAPP), which shares structural similarities with Aβ42. This study investigates the inhibitory and disaggregation effects of BCN on Aβ42 using biophysical assays and atomistic molecular dynamics (AT-MD) simulations. Thioflavin-T (ThT) fluorescence, circular dichroism (CD) spectroscopy, and fluorescence microscopy reveal that BCN significantly reduces fibril formation and induces disaggregation of preformed Aβ42 fibrils in a concentration-dependent manner. Dynamic light scattering (DLS) analysis further confirms that BCN stabilizes Aβ42 in its monomeric form, preventing the formation of larger aggregates. AT-MD simulations show that BCN interacts with the aggregation-prone region of Aβ42, specifically disrupting the Asp23-Lys28 salt bridge, which is crucial for β-sheet formation. The simulations also reveal that BCN promotes the formation of α-helical structures, reducing β-sheet content and hindering aggregation. Secondary structure analysis via DSSP plots confirms that BCN shifts Aβ42 towards less aggregation-prone conformational states. These results highlight BCN’s dual function in inhibiting both the formation and disaggregation of Aβ42 fibrils. This study provides mechanistic insights into BCN’s therapeutic potential for amyloid-related diseases, suggesting that it can effectively target the β-sheet spine structures common to multiple amyloidogenic proteins, offering a promising approach for mitigating AD progression.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-025-21991-7.

Introduction

The accumulation of amyloids is a hallmark of several progressive and often fatal diseases, including Alzheimer’s, Parkinson’s, Huntington’s diseases, and type II diabetes^1–5. Among these, AD is a complex neurodegenerative disorder characterized by the aggregation of Aβ proteins into insoluble fibrils, which form amyloid plaques in regions of the brain associated with memory and learning^6–8. The aggregation of Aβ, particularly the Aβ42 isoform, is strongly associated with neurotoxicity, inflammation, and neuronal death, making it a critical therapeutic target in AD⁹. Despite significant research, the molecular mechanisms governing Aβ42 aggregation and associated toxicity remain incompletely understood.

The aggregation pathway of Aβ begins with monomeric forms assembling into oligomers, which are the primary neurotoxic specie^10–13. These oligomers disrupt neuronal function through membrane destabilization, dysregulated calcium influx, oxidative stress, and apoptosis^14–16. Although fibrils are less toxic, they can dissociate into smaller toxic species or interact with cell membranes directly, exacerbating damage¹⁷. Emerging hypotheses, such as the lipid-chaperone model, emphasize the role of lipid-protein interactions in amyloid-mediated toxicity, underscoring the complexity of the aggregation process^18,19. These findings emphasize the need for therapeutic strategies that inhibit Aβ aggregation, disaggregate existing fibrils, and alleviate neurotoxic effects.

Current therapeutic options for AD, such as acetylcholinesterase inhibitors and NMDA receptor antagonists, provide only symptomatic relief without halting disease progression²⁰. Consequently, there is significant interest in developing agents that target the underlying mechanisms of amyloid toxicity. Strategies under investigation include immunotherapies, peptides, small molecules, and natural compounds that inhibit fibril formation or promote fibril clearance^21–25. Among these, natural compounds are particularly attractive due to their diverse bioactivities and low toxicity.

BCN, a natural flavonoid derived from Scutellaria baicalensis, has garnered attention for its broad pharmacological properties, including anti-inflammatory, antioxidant, and anti-amyloidogenic activities^26,27. BCN has demonstrated efficacy in inhibiting the aggregation of other amyloidogenic proteins, such as hIAPP, whose β-sheet spine structure closely resembles that of Aβ42^28,29. Despite these promising findings, the molecular mechanisms underlying BCN’s ability to modulate Aβ42 aggregation and disaggregation remain poorly understood.

In this study, we investigate the potential of baicalein to inhibit and reverse Aβ42 aggregation, leveraging a combination of experimental and computational approaches. Biophysical assays, including ThT fluorescence, CD spectroscopy, and fluorescence microscopy, are employed to evaluate BCN’s effects on Aβ42 fibrillation and oligomerization. Complementary AT-MD simulations elucidate the molecular interactions between baicalein and Aβ42, revealing its capacity to destabilize β-sheet-rich aggregation-prone conformations. Our findings demonstrate that baicalein disrupts the Asp23-Lys28 salt bridge, a critical structural motif in Aβ42 fibril formation, thereby inhibiting aggregation and promoting disaggregation of preformed fibrils. This work provides a detailed understanding of baicalein’s dual inhibitory and disaggregative effects on Aβ42 and highlights its therapeutic potential as a natural compound capable of modulating amyloid aggregation. By targeting the molecular underpinnings of Aβ42 pathology, baicalein offers a promising avenue for the development of effective AD therapeutics, addressing both early and late stages of disease progression.

Materials and methods

Reagents

Aβ42 was synthesized and procured from Biochain Incorporated, India. BCN and ThT were sourced from Sigma-Aldrich Inc. (St. Louis, MO). All other chemicals used were of analytical grade. Mass spectrometry and HPLC data for Aβ42 is provided in Figure S1 and S2.

Stock preparation

Monomeric Aβ42 was prepared by dissolving the peptide in 0.1 N sodium hydroxide to achieve a concentration of 1 mg/mL. The resulting solution was filtered through a 0.22 μm Millex filter, aliquoted, and stored at − 80 °C. The stock solution of BCN was prepared by dissolving the flavonoid to a final concentration of 10 mM. ThT was also dissolved in water, filtered, and its concentration was determined using the molecular extinction coefficient ε = 36,000 M¹ cm¹, measured at an optical density of 412 nm.

Aβ42 aggregate preparation for inhibition and disaggregation reactions

Aβ42 fibrils were prepared by diluting the stock solution in PBS (pH 7.4) to a final concentration of 30 µM and incubating at 37 °C for 48 h. Similarly, Aβ42 samples were also incubated with different ratios of BCN under similar conditions. To study disaggregation of Aβ42 fibrils, previously described protocol was used to prepare Aβ42 fibrils and BCN was added in increasing ratios after fibril formation and samples were kept then for 24 h. All samples were prepared in low-binding Eppendorf tubes, kept on ice, and handled with careful pipetting to prevent the introduction of air bubbles.

Fluorescence light scattering

Light scattering intensity measurements for Aβ42 samples, with and without BCN, were conducted using a Shimadzu RF-6000 PC fluorescence spectrophotometer, equipped with a Julabo Eyela water circulator to maintain temperature control. Measurements were acquired in a quartz cuvette with a 1 cm path length. Samples were excited at 300 nm, and emission spectra were recorded over a range of 300–400 nm. The slit widths for both excitation and emission channels were fixed at 1.5 nm. All samples, except monomeric Aβ42 without BCN, underwent a 48-hour incubation period prior to light scattering analysis.

Thioflavin-T binding

ThT dye was added to all samples, with or without BCN, to a final concentration of 50 µM. Samples were incubated in the dark for 20 min before measurement. The measurements were conducted using the previously described instrument, with samples placed in a quartz cuvette of 1 cm path length. Excitation was set to 440 nm, and emission spectra were recorded over a range of 450–600 nm, using slit bandwidths of 5 nm for excitation and 10 nm for emission. Background signals from blank samples containing only BCN were subtracted.

The aggregation kinetics of Aβ42 were assessed by recording ThT fluorescence intensities at specific time intervals for each sample. A global fitting equation was used to fit the kinetic data³⁰.

where f, a, and x_o represent fluorescence intensity, intercept, and half-life of the elongation phase, respectively. The term 1/b denotes K_app, the apparent rate constant associated with fibril growth, while x_o-2b indicates the lag time. The values for x_o and a were obtained using a three-parameter sigmoid fitting equation in GraphPad software.

Circular dichroism (CD) measurements

Far-UV CD spectra of the samples were recorded on a JASCO J-815 spectropolarimeter with a Peltier temperature controller. Spectra were collected in the 190–250 nm range, with a scan speed of 200 nm/min, a bandwidth of 1 nm, and a response time of 2 s. Blank spectra were subtracted from the experimental data. Samples were loaded into a quartz cuvette with a 1 mm path length, and measurements were taken over three cycles. For improved visual clarity, a Savitzky-Golay smoothing filter with a width of 13 was applied to the raw data.

DLS measurements

DLS measurements were performed using a Zetasizer Nano ZS (Malvern Instruments) under ambient conditions, following a 120-second equilibration period. The instrument was equipped with a He-Ne laser at a wavelength of 633 nm, and data analysis utilized Mark-Houwink parameters. Measurements were collected at a backscatter angle of 173°, with each set comprising three runs of 30 s each. The Zetasizer’s built-in software tools were used to process and analyze the recorded average values.

Fluorescence microscopic imaging

Staining of Aβ42 samples, both with and without BCN, was performed by adding ThT at a protein-to-ThT ratio of 1:10. After incubation, the samples were immediately transferred to a glass slide for imaging. Visualization was carried out using a fluorescence microscope (Zeiss M2 Imager; Zeiss, Göttingen, Germany) equipped with a 20X objective and a mercury discharge lamp.

Molecular docking studies

The 3D structure coordinates for the protein were obtained from the Protein Data Bank (https://www.rcsb.org/). The solution NMR-resolved PDB entry 1IYT, with model 10 was used as the starting structure for the Aβ42 monomer³¹. The PDB entry 5OQV, which refers to Aβ42 fibrils resolved by cryo-EM, represents the homodimer protofibril³². To reduce computational costs, a pentamer structure was selected for the study. Ligand coordinates were sourced from the PubChem database in .sdf format. The PyRx package was employed for the preparation of both ligands and proteins, as well as for molecular docking³³. The ligand was minimized using the MMFF94 force field and converted to .pdbqt format via the OpenBabel module of PyRx. Protein structures were prepared using default parameters. A grid box was generated to map the aggregation-prone regions (residues 16–22)³⁴. An exhaustiveness of 100 was applied for more precise ligand localization on the protein structure. A total of nine conformations were generated, which were analyzed using PyMOL and Schrodinger Maestro Visualizer. The best pose, based on appropriate scoring functions, was selected for further dynamics studies.

Molecular dynamics studies

The GROMACS 2018.1 package was employed for AT-MD simulations using the amber99sb-ILDN force field as previously used for Aβ42 simulations^35,36. The Aβ42 monomer and BCN complex were centered in a cubic box with a 10 Å gap between the complex and the box boundary in the x, y, and z directions, resulting in a box dimension of 34 × 31 × 14 Å. Periodic boundary conditions were applied. The system was solvated using the TIP3P water model, resulting in 4173 water molecules. Neutralization was achieved by adding 3 Cl⁻ ions, and physiological conditions were mimicked by adding Na + and Cl − ions to achieve a 150 mM NaCl concentration, as used in experimental studies. The LINCS algorithm was used to constrain all bond lengths, including hydrogen bonds. Energy minimization was performed using the steepest descent method for 50,000 steps. The system was equilibrated in an isochoric-isothermal (NVT) ensemble for 200 ps with a 2 fs time step, using the modified Berendsen thermostat to maintain a temperature of 310 K. The system was further equilibrated under isothermal-isobaric (NPT) conditions for 1 ns using the modified Berendsen thermostat and the Parrinello-Rahman barostat to stabilize both temperature and pressure at 310 K and 1 bar, respectively³⁷. During this equilibration, the Aβ42 peptide and BCN complex were position-restrained with a spring constant of 1 kcal/mol/Ų. The Verlet cutoff scheme was used for neighbor searching, and the smooth particle mesh Ewald (PME) method with cubic interpolation and a Fourier grid spacing of 16 Å was employed for long-range electrostatic calculations³⁸. All nonbonded interactions were evaluated with a cutoff of 1 nm, and GPU acceleration was used for these computations. Position restraints were removed, and the system was further equilibrated for 2 ns before a 1 µs production run was performed. Three independent runs were performed starting from different initial velocities in absence and presence of BCN for enhancing statistical efficiency.

For disaggregation studies of the Aβ42 fibril, the BCN-Aβ42 pentamer structure was placed in a cubic box with a 10 Å edge and periodic boundary conditions applied in all directions. Solvation was achieved using the TIP3P water model, and 18 Cl⁻ and 15 Na⁺ ions were added to neutralize the system and mimic the 150 mM NaCl experimental conditions. Minimization, equilibration, and MD production phases were conducted similarly to those for the monomer, with the fibril MD production run lasting 100 ns for data extraction. 3D atomic coordinates were recorded at 10 ps intervals. Three independent simulations were conducted with distinct initial velocity seeds, both in the absence and presence of BCN, to improve statistical robustness. MD trajectory analysis was performed using GROMACS 2018.1 utilities, including the gmx do_dssp module for secondary structure calculations and the gmx hbond module to calculate hydrogen bonds between peptide chains in the fibril throughout the simulation. The gmx distance module was used to generate average distance profiles between peptides in the pentamer fibril. Snapshots of velocities were visualized using the render option in Visual Molecular Dynamics (VMD)³⁹. GraphPad Prism version 8.0.0 was used to generate graphs.

Results and discussion

Potential of Baicalein to inhibit amyloid formation

The self-association and fibrillation of Aβ42 have been extensively investigated under various experimental conditions. In this study, the effect of BCN on Aβ42 aggregation was assessed using RLS at 350 nm, a widely employed method for monitoring protein aggregation by detecting light scattering from proteins lacking chromophore regions⁴⁰. As shown in Fig. 1A, Aβ42 was incubated under amyloid-inducing conditions (37 °C, pH 7.4, 150 mM NaCl) both in the absence and presence of BCN. The results revealed a concentration-dependent decrease in fluorescence intensity, with significant reductions observed at only 1:2 ratios of Aβ42: BCN after 48 h of incubation. Since increased scattering intensity correlates with the presence of higher molecular weight species²¹, the reduced intensity in the presence of BCN suggests its anti-aggregation effect.

ThT assay validates the inhibitory effect of BCN on amyloid formation

Since scattering intensity measurements alone do not provide detailed insights into the types of aggregates formed, a ThT assay was employed to confirm amyloid fibril formation. Incubation of Aβ42 at 37 °C for 48 h at pH 7.4 in the absence of BCN resulted in a pronounced increase in ThT fluorescence intensity, indicative of amyloid fibril formation (Fig. 1B). In contrast, the presence of BCN significantly reduced ThT fluorescence in a ratio-dependent manner, confirming its inhibitory effect on amyloid fibril formation. Even at a low concentration of 1:2, BCN inhibited amyloid formation by approximately 70%. Further with increased ratios of BCN there is seen to be 79, 83, 87, 91% of inhibition for 1:4, 1:6, 1:8 and 1:10 ratios of Aβ42:BCN respectively.

Fig. 1.

Inhibition of monomeric Aβ42 aggregation experimentally in presence of BCN. (A) RLS fluorescence intensity for Aβ42 read at 350 nm in absence and presence of BCN. (B) ThT fluorescence spectra of Aβ42 (30 µM) before and after aggregation reaction, in presence of different ratios of BCN. (C) Dose dependent inhibition of fibrillation kinetics of Aβ42 (30 µM) measured by ThT assay at 485 nm for 48 h. Error bars = mean ± SD, n = 3, for all samples. (D) far-UV CD spectra of Aβ42 alone and Aβ42 co-incubated BCN (1:10). Fluorescence micrographs of ThT stained Aβ42 after 48 h (E) without BCN and (F) in presence of ten fold BCN. Hydrodynamic radii measurement by DLS method under various conditions (G) monomeric Aβ42, (H) Aβ42 fibrils, (I) Aβ42 in presence of ten folds BCN.

Amyloid fibril formation, a common self-assembly process of normally soluble proteins, exhibits three characteristic kinetic phases: an initial lag phase, a rapid growth phase, and a plateau phase⁴¹. Consistent with this model, the ThT fluorescence profile of fibrillar Aβ42 demonstrated a lag phase followed by a sigmoidal elongation phase and saturation (Fig. 1C). Kinetic analysis revealed that, in the absence of BCN, Aβ42 exhibited a lag phase of approximately 3.31 h. The presence of BCN extended the lag phase in a ratio-dependent manner, increasing to 4.87, and 6.9 h at 1:2 and 1:10 ratio of Aβ42:BCN, respectively. The kinetic parameters are summarized in Table 1. These findings demonstrate that BCN not only inhibits the formation of amyloid fibrils but also delays the aggregation process by prolonging the lag phase and reducing the rate of aggregation. These dual inhibitory effects highlight BCN as a promising candidate for therapeutic intervention in amyloid-related pathologies.

Table 1.

Thioflavin-T kinetic parameters of Aβ42 with or without BCN.

S. no.	Samples	Kapp (h− 1)	Lag time (h)
1	Aβ42 (monomeric)	0.112	3.31
2	Aβ42:BCN (1:2)	0.093	4.87
3	Aβ42:BCN (1:10)	0.06991	6.9

Retention of Aβ42 native structure in the presence of BCN

A hallmark feature of amyloid formation is the structural transition of proteins from α-helix to a β-sheet conformation or, more broadly, an increase in β-sheet content⁴². Secondary structure perturbations in proteins are commonly studied using far-UV CD spectroscopy. The Aβ42 peptide, in its native state, exhibits intrinsically disordered behavior in solution⁴³. This is evident from its far-UV CD spectrum, which, at 25 °C and 0 h in the absence of BCN, displays a prominent single minimum around 198 nm a characteristic feature of intrinsically disordered proteins. As aggregation progresses, the CD spectrum of Aβ42 shows a gradual decrease in negative ellipticity and the emergence of a single minimum at approximately 218 nm, indicating the formation of β-sheet structures. In the absence of BCN, Aβ42 readily transitions into this β-sheet-rich conformation. However, when incubated with BCN, this structural transition is markedly inhibited. At the highest Aβ42:BCN ratio (1:10), the CD spectrum closely resembles that of the native structure, with minimal β-sheet signature. This suggests that BCN may stabilize the native state of Aβ42 and prevent its structural transition to the amyloid-prone β-sheet form. The CD spectra of Aβ42, both with and without BCN, are presented in Fig. 1D, further highlighting BCN’s stabilizing effect on the protein’s native structure.

BCN maintains Aβ42 stability in its monomeric form

Protein aggregation leads to the formation of various heterogeneous species with differing particle sizes⁴⁴. The particle size distribution provides valuable insights into the nature of the species generated during Aβ42 fibrillation. DLS can be used to analyze the presence of monomeric, oligomeric, and mature fibrillar species by examining their diffusion behavior in the reaction solution. At 0 h, monomeric Aβ42 exhibits a hydrodynamic radius of 0.313 nm, consistent with previous studies (Fig. 1G)⁴. However, after 48 h of incubation at 37 °C, Aβ42 forms larger, mature fibrillar aggregates with a particle size of around 89.4 nm (Fig. 1H). It is important to note that since particle scattering intensity correlates with molecular weight, the true abundance of these larger particles is likely much lower than what is represented in the size distribution, which is influenced by scattering intensity⁴⁵. When Aβ42 was incubated with 1:10 ratios of Aβ42:BCN, multiple peaks of size 0.433 and 58.71 nm were observed in the particle size distribution (Fig. 1I). The first peak closely resembles the monomeric form of Aβ42 alongwith a second peak which might describe formation of some low order aggregates. The corresponding hydrodynamic radii values in the absence and presence of BCN are summarized in Table 2. These results are consistent with previous findings, further suggesting that BCN stabilizes the monomeric form of Aβ42 and prevents its aggregation.

Table 2.

Hydrodynamic radii of Aβ42 in different aggregation conditions^* (inhibition studies).

S. no.	Samples	Hydrodynamic radii (nm)
1	Aβ42 (monomeric)	0.313 ± 0.51804
2	Aβ42 (Aggregated)	89.4 ± 15.23
3	Aβ42:BCN (1:10)	0.433 ± 0.7837 (89.91%) 58,71 ± 11.68 (10.09%)

Impact of BCN on the morphological characteristics of Aβ42 amyloid fibrils

Following DLS analysis, Fluorescence Microscopy was employed to examine the morphology of Aβ42 aggregates in this study. Fluorescence Microscopy imaging was conducted for Aβ42 incubated at 37 °C for 48 h to induce amyloid formation, both in the absence and presence of BCN. Under native conditions, Aβ42 do not form any aggregates. However, when incubated at 37 °C for 48 h, long filamentous amyloid structures were observed (Fig. 1E). In the presence of 1:10 ratio of Aβ42:BCN, very few low order aggregates were visible (Fig. 1F). These imaging results support the hypothesis that BCN inhibits amyloid formation in Aβ42. Thus, it can be concluded that BCN significantly prevents the aggregation of Aβ42, which is implicated in AD.

Mechanism of Inhibition of Aβ42 aggregation by BCN

In order to understand the mechanism of inhibition of Aβ42 at atomistic level computational modelling methods have been utilized. Evidences indicates that the aggregation-prone region (APR) of Aβ42 consists of a spine of residues spanning from 16 to 22³⁴. Therefore, we docked BCN specifically to this APR region using AutoDock Vina. The binding energy calculation showed a spontaneous, favorable interaction with a binding energy of − 6.3 kcal/mol. The interacting residues and their corresponding interactions are visualized in Fig. 2. Previous studies reported the formation of a semi-cyclic structure during the early misfolding events of Aβ42 which is stabilized by Asp23-Lys28 salt-bridge interactions⁴⁶. Here we observed, BCN forms two H-Bonds with the Asp23 residue of Aβ42 which in turn destabilize this Asp23-Lys28 salt bridge thus preventing its aggregation.

Fig. 2.

Molecular docking geometry of the Aβ42 monomer with BCN, highlighting various interacting residues. (A) Docked pose and interacting residues of the Aβ42 monomer with BCN in 3D ribbon representation while ligand is represented in ball and stick model. (B) 2D interaction diagram describing various interactions and their types.

Further to understand time dependant evolution of Aβ42 monomer in the absence and presence of BCN, AT-MD was performed. Simulations involving monomeric Aβ42, both with and without BCN in explicit water, were conducted to investigate how BCN influences the conformation of the peptide as it unfolds in solution. From these simulations, we describe the interactions between different regions of the Aβ42 peptide sequence in the presence of BCN. These findings provide valuable insights into the structural changes and the potential mechanisms by which BCN modulates the conformation of Aβ42, potentially influencing its aggregation behavior. To address how BCN inhibits the aggregation of Aβ42 and alters the structure of Aβ42 monomers in solution, we examined the secondary structure of Aβ42 over 1 µs⁴⁷. While enhanced sampling methods, such as replica exchange MD, would be more suitable for capturing the diverse conformations of Aβ42 that are likely to be more prevalent in cellular environments and contribute to aggregation, we believe that 1 µs simulations were sufficient to detect notable changes in the secondary structure⁴⁸. These changes provide insight into how the conformational shifts may influence aggregation processes over longer timescales. Figure 3 illustrates the conformational changes observed in the Aβ42 monomer during our simulations. We used the free Aβ42 in solution as a control to compare with BCN bound Aβ42. While free Aβ42 undergoes various secondary structure changes as it unfolds over time, the BCN bound Aβ42 shows minimal transformation in its secondary structure over the same timescale. In the absence of the BCN, Aβ42 forms β-sheet structures between residues 27 (Asn) − 31 (Ile) and 36 (Val) − 41 (Ile) around 200 − 500 ns, similar to those observed in previous studies^49,50. These β-sheets declined between 500 − 600 ns but increased again for the remainder of the simulation. Additional β-sheets are also observed between the N- and C-termini after 600 ns. The most stable structure that forms after 200 ns resembles an extended β-hairpin with a bend around residues 32 − 24. Over the course of 1 µs simulations, 38% of the residues adopted a random coil conformation,13% formed β-sheets, 4% β-bridges, 21% were in a bend, 20% were turns, and 10% exhibited helical character. These findings are consistent with previous computational studies, which reported 5 − 20% helical character and 0 − 15% β-character for free Aβ42 in solution⁵¹.

Fig. 3.

Secondary structure analysis (DSSP) and conformations over a 1 µs simulation time for (a) the Aβ42 monomer alone and (b) the Aβ42 monomer with bound BCN. In the visualizations, secondary structures of Aβ42 are represented by the following colors: α-helix (purple), 310-helix (blue), extended-β (yellow), bridge-β (tan), turn (cyan), and random coil (white). BCN is depicted in green. For clarity, the side chains of Aβ42, hydrogens, water molecules, and ions are omitted.

In contrast, when BCN is bound to Aβ42, the α-helical structures (residues 13 − 17, 30 − 36) are more persistent, and peptide do not adopts β-sheets. The average secondary structure of BCN bound Aβ42 consists of 37% random coil, 2% β-bridge, 20% bend, 22% turn, and 19% helical character across the simulations. These results indicate that BCN binding stabilizes the helical conformation of Aβ42 and reduces β-sheet formation.

Previous computational studies have demonstrated the formation of antiparallel β-hairpin structures between the central hydrophobic core residues (16 − 22) and residues 29 − 36⁴⁹. However, when BCN binds to Aβ42, we observe that these residues are too distant to form the typical antiparallel β-sheets. The coordination of BCN between APR region (16–22) enables it to interact with the hydrophobic core of Aβ42, thereby preventing the formation of these β-hairpin structures. We did performed three independent runs and in every case similar results and found. The data for run 2 and run 3 in absence of BCN and in presence of BCN can be found in Figure S3 and S4 respectively.

Contact maps for the Aβ42 monomer without BCN (Fig. 4A) reveal interactions between the N- and C-termini as the peptide adopts an extended hairpin conformation, leading to the formation of β-sheets. In contrast, when BCN is present (Fig. 4B), the interaction between residues becomes more dispersed, with BCN forming strong interactions with the hydrophobic core. These interactions effectively disrupt the contact between residues 16 − 22 and 29 − 36. The calculated short-range interaction energies (Lennard-Jones and Coulomb) between BCN and residues 16 − 22 and 29 − 36 are − 7.91 ± 0.82 kJ/mol and − 11.37 ± 1.32 kJ/mol, respectively.

Fig. 4.

Contact maps for monomer simulations of (a) Aβ42 alone and (b) Aβ42 with bound BCN. In (b), the last residue (residue 43) corresponds to the BCN complex.

It is crucial to note that β-sheet formation in the monomer does not directly correlate with β-sheet formation in higher-order aggregates or fibrils. In fibrils, β-sheets form between individual Aβ42 peptides that aggregate in solution. For the monomer to aggregate, it must first form favorable interactions with other monomers. In our simulations of free Aβ42, the peptide adopts a folded conformation similar to the typical β-hairpin structure observed in Aβ fibrils⁵². The presence of β-hairpins plays a key role in fibril growth, with monomers aggregating to form β-sheets in solution. This extended β-hairpin conformation remains stable for around 700 ns in solution suggesting that the peptide is readily capable of forming aggregates under appropriate conditions. However, when BCN binds to Aβ42, the peptide does not adopt the folded β-hairpin conformation. Instead, it stabilizes into helix-like structures, which are less prone to aggregation and fibril formation. Thus, BCN binding appears to hinder Aβ42 aggregation by stabilizing a conformation that is unfavorable for fibril formation.

BCN perturbs the Asp23–Lys28 salt bridge by increasing the inter-residue distance

To quantitatively assess how BCN binding influences the structural ensemble of Aβ42, we performed clustering analysis of the 1 µs monomeric simulations in the absence and presence of BCN. In the Aβ42 peptide alone, the conformational landscape was dominated by a single β-sheet-rich cluster, which accounted for 46% of the total population. This cluster was characterized by a reduced average Asp23-Lys28 side-chain distance of 5.6 Å, compared to the initial 8.45 Å, indicating the formation and persistence of the salt bridge throughout much of the trajectory. Such close Asp23–Lys28 contacts are known to stabilize β-hairpin motifs and promote aggregation-prone conformations. In stark contrast, the BCN-bound system displayed a markedly altered population distribution, with the most populated cluster representing only ~ 16% of the total conformations. This cluster maintained an average Asp23–Lys28 distance of 8.74 Å, consistent with a fully disrupted salt bridge. The reduced prevalence of β-sheet-rich states in the BCN-bound ensemble suggests that ligand binding shifts the peptide toward more disordered and helical conformations that are less competent for nucleating fibril growth. These findings, together with the DSSP-based secondary structure analysis and CD experimental data, provide strong evidence that BCN exerts its inhibitory effect by destabilizing the Asp23-Lys28 salt bridge and suppressing the structural transitions required for β-sheet propagation (Fig. 5).

Fig. 5.

Clustering analysis of monomeric Aβ42 trajectories (1 µs) in the absence and presence of BCN, showing population distributions of dominant clusters and corresponding average Asp23–Lys28 distances. BCN binding reduces the occupancy of β-sheet-rich states and maintains an expanded Asp23–Lys28 separation, consistent with salt bridge disruption.

BCN also disrupts the pre-formed fibrils of Aβ42 in ratio-dependent manner

To evaluate the effect of BCN on pre-formed Aβ42 fibrils, we employed RLS, the ThT binding assay, DLS, and Fluorescence microscopy TEM. Aβ42 fibrils formed after incubation at 37 °C.

for 48 h were confirmed by fluorescence imaging, showing characteristic fibrillar structures. Upon further incubation for 24 h, the Aβ42 fibrils demonstrated strong light scattering at 350 nm, indicating the persistence of fibrillar aggregates. However, in the presence of BCN at 1:5 and 1:10 ratios of Aβ42 fibrils: BCN, a reduction in light scattering was observed, suggesting a decrease in the size or number of aggregates (Fig. 6A). The ThT assay corroborated these findings, revealing a ratio-dependent decline in ThT fluorescence intensity, indicative of fibril disassembly (Fig. 6B). DLS further substantiated these observations by showing distinct size distributions of fibrillar Aβ42. At 1:10 ratio of Aβ42 fibrils: BCN, four hydrodynamic radii were detected approximately 0.626, 4.39, 23.44, 55.19 nm suggesting the formation of trimeric and smaller aggregates (Fig. 6E-F). Fluorescence imaging provided additional confirmation, with typical ordered fibrils observed in untreated samples. In the presence of BCN, fewer fibrils of reduced length and size were observed, consistent with DLS results (Fig. 6C-D). A summary of different hydrodynamic radii have been provided in Table 3.

Fig. 6.

Disaggregation of Aβ42 fibrils in the presence of BCN at different stoichiometric ratios. (A) RLS intensity of Aβ42 fibrils (30 µM) in the absence and presence of fivefold and tenfold BCN, measured at 350 nm. (B) ThT emission spectra showing Aβ42 fibril disaggregation by BCN. (C) Fluorescence micrographs of Aβ42 fibrils. (D) Fluorescence micrographs of Aβ42 fibrils in the presence of tenfold BCN. Hydrodynamic radius measurements for (E) Aβ42 fibrils alone, (F) Aβ42 fibrils with tenfold BCN.

Table 3.

Hydrodynamic radii of Aβ42 in different aggregation conditions^* (disaggregation studies).

S. no.	Samples	Hydrodynamic radii (nm)
1	Aβ42 (Aggregated)	87.66 ± 14.41
2	Aβ42-fibrils: BCN (1:10)	0.626 ± 0.619 (21.81%)
		4.39 ± 0.12 (8.74%)
		23.44 ± 3.39 (13.37%)
		55.19 ± 11.35 (56.08%)

Collectively, these results demonstrate that BCN disrupts Aβ42 fibrils into monomeric and smaller aggregate forms, highlighting its dual functionality as both an inhibitor of fibril formation and a disruptor of pre-formed amyloid fibrils.

Molecular mechanism of disintegration property of BCN for pre-formed Aβ42 fibrils

To evaluate the ability of BCN to disrupt pre-formed Aβ42 fibrils, we utilized a cryo-EM structure (PDB: 5OQV) of the homodimer. For computational efficiency, a single pentameric unit was selected for analysis, reducing computational cost while retaining sufficient structural complexity to assess BCN’s disaggregation potential. Atomistic MD simulations were conducted for 200 ns, although only the initial 100 ns of data were analyzed due to a lack of significant convergence beyond this timescale. The do_dssp module of GROMACS was used to calculate secondary structural changes⁴⁷, revealing a marked reduction in β-sheet content in the presence of BCN (Fig. 7A, B). Data for run 2 and run 3 both in the absence and presence of BCN can be found in Figure S5 and S6.

Fig. 7.

Secondary structure information (DSSP) of the Aβ42 fibrils and conformations over 100 ns simulation time. (A) Aβ42 pentamer alone and (B) Aβ42 pentamer with bound BCN. In the snapshots, the secondary structures of Aβ42 are represented as follows: α-helix (purple), 3₁₀-helix (blue), extended-β (yellow), bridge-β (tan), turn (cyan), and random coil (white). BCN is depicted in red. For clarity, side chains of Aβ42, hydrogens, water molecules, and ions are omitted.

BCN disrupted the fibrillar stability by interfering with the hydrogen bond framework that stabilizes the β-sheet structures. The hydrogen bond distribution over the 100 ns simulation (Fig. 8A) showed a decrease in the number of intermolecular hydrogen bonds for the BCN-bound system compared to the Aβ42 pentamer alone. This reduction highlights BCN’s ability to destabilize the fibrillar assembly by disrupting key interactions. The disruption of hydrogen bonds resulted in increased β-strand spacing, assessed as the distance between the centers of mass of adjacent chains^53,54. Average β-strand spacing increased from 2.07 nm in the control system to 1.68 nm in the presence of BCN (Fig. 8B), indicating steric clashes between chains that impede the maintenance of the fibrillar structure.

Fig. 8.

(A) Average number of hydrogen bonds between neighboring peptide chains in the Aβ42 pentamer without and with bound BCN, calculated using the center of mass of Cα atoms. (B) Average distance between neighboring chains in the Aβ42 pentamer alone and the Aβ42 pentamer with bound BCN.

Conformational stability was evaluated using radius of gyration Rg) and solvent-accessible surface area (SASA) (Fig. 9). An increase in Rg values in the BCN bound system corresponded to the disruption of the hydrogen bond framework and greater β-strand spacing. Similarly, increased SASA values indicated the exposure of hydrophilic regions previously buried within the fibril, further supporting the disaggregation effect of BCN. Overall, our analyses suggest that BCN disrupts Aβ42 fibrils by destabilizing intermolecular hydrogen bonds, increasing β-strand spacing, and inducing significant secondary structural perturbations, leading to reduced β-sheet content and increased solvent exposure.

Fig. 9.

(A) Solvent accessible surface area (SASA) and (B) radius of gyration for Aβ42 fibrils in the absence and presence of BCN.

Conclusion

This study highlights the dual functionality of BCN in inhibiting the aggregation of Aβ42 monomers and disassembling pre-formed amyloid fibrils, making it a promising candidate for therapeutic intervention in AD. Experimental and computational results collectively demonstrate BCN’s ability to modulate amyloid aggregation through structural stabilization, disruption of β-sheet interactions, and prevention of fibrillar assembly. BCN effectively inhibits Aβ42 aggregation by prolonging the lag phase of fibril formation and reducing the rate of aggregation in a ratio-dependent manner, as shown by RLS and ThT assays. It prevents the transition of Aβ42 from a random coil or α-helical structure to a β-sheet-rich conformation, which is critical for amyloid fibril formation. CD spectroscopy revealed that in the presence of BCN, the native-like structure of Aβ42 is retained. This stabilization inhibits the conformational shifts necessary for aggregation. DLS and fluorescence microscopy further confirm that BCN reduces the size and number of Aβ42 aggregates. At higher BCN concentrations, the monomeric and low-order aggregate forms are preserved, supporting BCN’s role in stabilizing Aβ42 in its monomeric state. Computational docking and MD simulations revealed that BCN binds with high affinity to key residues in the hydrophobic core of Aβ42, destabilizing the critical Asp23-Lys28 salt bridge and preventing the formation of β-hairpin structures necessary for fibril propagation. BCN binding also stabilizes α-helical regions, reducing the formation of aggregation-prone intermediates. BCN also disrupts pre-formed Aβ42 fibrils, as demonstrated by RLS, ThT assays, DLS, and fluorescence imaging. Fibrillar structures disassembled into smaller aggregates and monomeric forms in a dose-dependent manner, with higher BCN concentrations resulting in fewer and shorter fibrils. Molecular dynamics simulations provided insights into the underlying mechanism, showing that BCN binds to the aggregation-prone region of Aβ42, disrupting the hydrogen bond network and β-sheet integrity. This disruption increases β-strand spacing, destabilizes fibrillar assemblies, and enhances the solvent exposure of hydrophilic residues. These results position BCN as a potential therapeutic agent for AD disorder, warranting further exploration through in vivo studies and enhanced sampling techniques to refine its therapeutic applicability.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

F.N.: Conceptualization, data curation, formal analysis, investigation, methodology, validation, writing of original draft. O.A.: Figures and data curation. M.R.A.: resources. M.A.: Formal analysis. I.P.: Investigation. F.M.A.: resources. R.H.K.: resources, supervision, and review and editing of manuscript.

Data availability

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

Declarations

Competing interests

The authors declare no competing interests.