γ-secretase is an intramembrane proteolytic enzyme that is mainly involved in the cleavage and hydrolysis of the amyloid precursor (APP). The catalytic subunit presenilin 1 (PS1) is the catalytic subunit of γ-secretase. Since it was found that PS1 is responsible for Aβ-producing proteolytic activity, which is involved in Alzheimer’s disease, it is believed that reducing the activity of PS1 and preventing or delaying the production of Aβ could help treat Alzheimer’s disease. Consequently, in recent years, researchers have begun investigating the potential clinical efficacy of PS1 inhibitors. Currently, most PS1 inhibitors are only used as a tool to study the structure and function of PS1, and a few inhibitors with a high selectivity have been tested in clinics. Less-selective PS1 inhibitors were found to not only inhibit Aβ production but also inhibit Notch cleavage, which led to serious adverse events. The archaeal presenilin homologue (PSH) is a surrogate protease of presenilin that is useful for agent screening. In this study, we performed 200 ns molecular dynamics simulations (MD) of four systems to explore the conformational changes of different ligands binding to PSH. Our results indicated that the PSH-L679 system formed 3–10 helices in TM4, loosening up TM4 and allowing substrates to enter the catalytic pocket, thereby making it less inhibitory. Additionally, we found that III-31-C can bring TM4 and TM6 closer, resulting in the contraction of the PSH active pocket. Altogether, these results provide the basis for the potential design of newer PS1 inhibitors.

Thermal stability is a limiting factor for effective application of D-psicose 3-epimerase (DPEase) enzyme. Recently, it was reported that the thermal stability of DPEase was improved by immobilizing enzymes on graphene oxide (GO) nanoparticles. However, the detailed mechanism is not known. In this study, we investigated interaction details between GO and DPEase by performing molecular dynamics (MD) simulations. The results indicated that the domain (K248 to D268) of DPEase was an important anchor for immobilizing DPEase on GO surface. Moreover, the strong interactions between DPEase and GO can prevent loop α1′-α1 and β4-α4 of DPEase from the drastic fluctuation. Since these two loops contained active site residues, the geometry of the active pocket of the enzyme remained stable at high temperature after the DPEase was immobilized by GO, which facilitated efficient catalytic activity of the enzyme. Our research provided a detailed mechanism for the interaction between GO and DPEase at the nano–biology interface.

The accumulation of cross-β-sheet amyloid fibrils is a hallmark of the neurodegenerative process of Alzheimer’s disease (AD). Although it has been reported that green tea substances such as epicatechin (EC), epicatechin-3-gallate (ECG), epigallocatechin (EGC) and epigallocatechin-3-gallate (EGCG) could alleviate the symptoms of AD and other neurodegenerative diseases, the pharmacological mechanism remains largely unexplored. This study aimed to reveal the underlying mechanism of EC, ECG, EGC and EGCG in AD using a computer-aided screening strategy. Our results showed that the four tea polyphenols interfered with the signaling pathways of AD via calcium signaling channels, neurodegeneration-multiple disease signal pathways and others. We also identified the key residues of the interaction between VEGFA and the four active components, which included Glu64 and Phe36. Overall, we have provided valuable insights into the molecular mechanism of tea polyphenols, which could be used as a reference to improve therapeutic strategies against AD.

JOURNAL/atin/04.03/02274269-202509000-00002/figure1/v/2026-04-23T113529Z/r/image-tiff Neurotoxicity refers to harmful changes in the structure and function of the central and/or peripheral nervous system caused by exposure to chemical, physical, or biological factors. These changes can manifest as organic damage, functional disorders, and behavioral alterations. Traditional biological testing methods for neurotoxicity are often expensive, labor-intensive, and time-consuming. Additionally, existing predictive models rely on small datasets and lack comprehensive, large-scale, and professional algorithms and platforms for accurately predicting brain toxicity. In this study, we integrate molecular fingerprints and descriptors—namely, MorganFP, MACCS, RDKFP, and TopologicalTorsionFP—with machine learning models such as Random Forest and Support Vector Machine, as well as deep learning models such as Graph Neural Networks, to develop a neurotoxicity prediction model. The optimal model, MorganFP-SVM, demonstrates excellent performance in predicting neurotoxicity, achieving an accuracy of 86.56%. This significantly outperforms other neurotoxicity prediction models, including ADMETlab 3.0. When compared to existing neurotoxicity prediction models DINeuroT and ADMETlab 3.0, the MorganFP-SVM model exhibits superior performance across multiple key evaluation metrics, offering greater balance and comprehensiveness as a reliable tool for neurotoxicity prediction. The DINeuroT model shows weaknesses in specificity and Matthews correlation coefficient, while ADMETlab 3.0, though strong in sensitivity, has limited capacity to accurately identify negative samples. Overall, the comprehensive advantages of the MorganFP-SVM model in neurotoxicity prediction establish it as an essential tool in the field. It not only provides higher predictive accuracy but also demonstrates strong stability and reliability across various evaluation metrics. This model offers a promising approach for assessing neurotoxicity risks related to drug development and environmental pollutants, thereby providing a scientific basis for public health decision-making.

Small molecules from edible fruits are promising resources for natural antioxidants with functional-food and nutrition-oriented potential. Here, we compiled the chemical constituents of Clausena lansium (Lour.) Skeels, an edible fruit with medicinal and nutritional value, and established a 474-compound database. We developed a multiscale workflow integrating ensemble machine learning, molecular docking, structural clustering, molecular dynamics simulations, and quantum chemical calculations. The ensemble models achieved ROC-AUC values of 0.86-0.99 across eight antioxidant assays. Through multi-criteria screening, molecular dynamics simulations, and quantum chemical calculations, Quercetin 3-arabinoside was prioritized as a candidate molecule, exhibiting stable non-covalent binding within the Keap1 Kelch domain and a HOMO-LUMO gap approximately 25% lower than that of vitamin C. Considering reference standard availability, its structural isomer Avicularin was used for experimental assessment and showed dose-dependent radical-scavenging activity and ferric reducing antioxidant power. This study provides a food-oriented strategy for rapid discovery of food-derived antioxidant candidates from edible fruits.