SAR and QSAR in Environmental Research最新文献

Enhancer of Zeste Homolog 2 (EZH2) inhibitors have demonstrated selective efficacy, but their broader therapeutic potential remains limited, highlighting the need to clarify the structural basis of their activity. The central aim of our study is to systematically analyse the structural diversity and activity patterns of known EZH2 inhibitors to provide insights that may guide incremental scaffold optimization. We examined 531 potential EZH2 inhibitors retrieved from ChEMBL through a cheminformatics workflow encompassing clustering, scaffold identification, activity cliff detection, and chemical space visualization. Using RDKit and NetworkX, 94 clusters were generated, of which 13 contained ten or more compounds. Notably, clusters 6, 16, 20, 21, and 31 exhibited favourable balances of structural homogeneity and enrichment scores, suggesting chemical cohesiveness and biological relevance for structure - activity relationship (SAR) prioritization. Statistical analyses revealed significant differences in mean pIC₅₀ values across clusters, underscoring distinct activity distributions linked to structural groups. Scaffold analysis highlighted pyrrole - benzamide derivatives, particularly those incorporating morpholine and piperidine motifs, as enriched among potent inhibitors. Substructure evaluation further indicated that aromatic rings and aromatic amine groups were positively correlated with bioactivity. These findings delineate key SAR features of EZH2 inhibitors and provide guidance for scaffold refinement, hit identification, and lead optimization.

Peptide quantitative structure-activity relationship (pQSAR) has been widely used in the computational peptidology community to model, predict and explain the activity and function of bioactive peptides. Various amino acid descriptors (AADs) have been developed to characterize the residue building blocks of peptides at sequence level. However, a significant issue is that the total number of AAD-characterized descriptors is proportional to peptide length, thus causing inconsistency in the resulting descriptor vector matrix for a panel of length-varying peptide sequences (LVPSs), which cannot be engaged in pQSAR modelling. Currently, only one AAD-based scaling approach, termed auto-cross covariance (ACC) that was proposed thirty years ago, is available for treating such issue. In this study, we described the second AAD-based multivariate method to do so, namely Residue Descriptor-Distance Vector (RDDV). The strategy characterizes a peptide sequence by using an inter-residue pseudo-interaction potential between different pre-assigned amino acid types involved in the sequence, which results in a given (invariable) number of descriptor parameters for different LVPSs. Here, the RDDV was tested, examined and validated in an in-house pQSAR-oriented bioactive peptide data cluster, which was explored systematically with combinations of different AADs and regression tools. We also compared RDDV with the traditional ACC in multiple aspects.

Phosphorylation plays an important role in the activity of CDK2 and inhibitor binding, but the corresponding molecular mechanism is still insufficiently known. To address this gap, the current study innovatively integrates molecular dynamics (MD) simulations, deep learning (DL) techniques, and free energy landscape (FEL) analysis to systematically explore the action mechanisms of two inhibitors (SCH and CYC) when CDK2 is in a phosphorylated state and bound state of CyclinE. With the help of MD trajectory-based DL, key functional domains such as the loops L3 loop and L7 are successfully identified. The results of FEL analysis show that the binding of CyclinE significantly enhances conformational stability of key functional regions of CDK2 (such as the L3 loop, L7 loop, and αC helix), while phosphorylation modification increases conformational diversity of the CDK2-related system. Further verification by quantum mechanics/molecular mechanics-generalized Born surface area (QM/MM-GBSA) calculations shows that binding of CyclinE can enhance the binding ability of inhibitors, while phosphorylation weakens this binding effect. Residue-based free energy estimation reveals the hot spot regions of inhibitor-CDK2 binding, providing crucial target information for structure-based drug design. This study provides theoretical foundations for the development of highly selective CDK2 inhibitors and might be of great significance for cancer targeted therapy.

Brugia malayi, a causative agent of lymphatic filariasis, relies on its endosymbiont Wolbachia for survival. MurE ligase, a key enzyme in Wolbachia peptidoglycan biosynthesis, serves as a promising drug target for anti-filarial therapy. In this study, we employed a hierarchical virtual screening pipeline to identify phytochemical inhibitors targeting the MurE enzyme of the Wolbachia endosymbiont of B. malayi (wBmMurE). A validated high-quality model of wBmMurE was used to screen 17,967 phytochemicals, and the identified hits were subjected to toxicity profiling, and ADME filters to select potent drug-like candidates. Five phytochemicals such as biotin, quisqualic acid, succinic acid, 9,14-dihydroxyoctadecanoic acid, and N-isovaleroylglycine with permissible ADME profiles showed favourable binding affinities (GlideScore range: -12.86 to -10.57 kcal/mol), and stable interactions with catalytically important residues were selected from screened hits. Comparative analysis with reported MurE inhibitors validated the superior affinity and drug-like behaviour of our identified leads. Molecular dynamics simulations of 300 ns confirmed the conformational stability of ligand-bound complexes, while MM-GBSA analysis supported their favourable binding free energies. The results revealed that the identified compounds have the tendency of binding within substrate binding cavity of wBmMurE. These findings suggest that selected phytochemicals could serve as starting points for the development of novel anti-filarial agents.

Evaluating the permeability of different molecular structures across the Caco-2 cell line is crucial for drug discovery and development. The present study primarily focuses on developing machine learning-based multiclass classification models for predicting the permeability of molecules across the Caco-2 cell line. However, the class imbalance in permeability datasets poses a significant challenge for developing predictive models in the case of multiclass analysis. To address the class imbalance issue, we employed different balancing strategies, including oversampling, undersampling, and hybrid approaches, to balance the training set. A five-fold cross-validation approach was employed for optimizing the hyperparameters. After completion of the evaluation process, we concluded that the XGBoost multiclass classifier trained with ADASYN oversampling achieved the best performance (accuracy, 0.717; MCC, 0.512 on the test set). Additionally, extreme permeability classes were also classified separately, and the best model exhibited strong predictive performance (accuracy, 0.853; MCC, 0.703 on the test set). To enhance the interpretability of the best-performing models, we performed SHAP analysis to elucidate descriptor importance and provide explainability. Our findings demonstrate that appropriate data balancing strategies can significantly improve predictive performance in multiclass permeability classification, offering a valuable framework for drug permeability assessment.

The existing QSAR approaches for mammalian acute toxicity have been limited in scope, often relying on small or narrowly focused datasets and on classification endpoints. In contrast, our work leverages a sufficiently large curated dataset (9843 rat oral and 2323 intravenous LD₅₀ values) to build regression models of acute toxicity. The best-performing QSAR models developed using 2D RDKit descriptors and the Cat Boost method achieve Q² _test = 0.66 at a data coverage of at least 77% within the applicability domain (AD) during validation of test sets. All models were rigorously validated according to OECD QSAR principles with clearly defined endpoints, explicit algorithms, and a well-characterized AD. The best QSAR models are integrated into the ToxAI_assistant web platform (https://tox-ai-assistant.streamlit.app/), which includes toxicity-level prediction with allowance for AD in terms of World Health Organization (WHO). We also provide mechanistic insight by identifying key toxicophores - substructural features statistically associated with high toxicity - thereby offering a structural interpretation. In sum, these elements (large and diverse data, regression modelling, WHO-based categorization, detailed fragment analysis, and AD assessment) together address the gaps of earlier studies and constitute the core novelty of our approach.

Monoamine oxidase B (MAO-B) is a key target in Parkinson's disease treatment due to its role in dopamine metabolism. This study applied a multi-stage in silico workflow - combining 3D-pharmacophore modelling, 2D-QSAR, ADMET filtering, docking, molecular dynamics (MD), and MM/PBSA analysis - to identify selective MAO-B inhibitors. From four datasets including ZINC, DrugBank, TCM, and UNPD, 22 top candidates were selected based on docking scores and predicted selectivity over MAO-A. MD simulations (200 ns) and binding free energy calculations identified four promising compounds - ZINC21285023, ZINC79651118, ZINC58283019, and UNPD89644 (crotafuran E)- that exhibited stable binding and favourable interactions with key residues such as Cys172 and Tyr435. These compounds demonstrated performance comparable to or better than safinamide and are strong candidates for further experimental validation as selective MAO-B inhibitors.

In this study, we employed a quantitative Read-Across Structure-Activity Relationship (q-RASAR) approach to develop a statistically robust machine learning (ML)-based q-RASAR model aimed at predicting the agonistic activity of compounds targeting the α7-nicotinic acetylcholine (α7nACh) receptor, a key therapeutic target in Alzheimer's disease (AD) due to its involvement in cognitive processes and neuroprotection. We developed a rigorously validated univariate q-RASAR linear regression (LR) model using an extensive dataset of 1,727 structurally diverse heterocyclic and aromatic hydrocarbon compounds sourced from the publicly available Binding Database (www.bindingdb.org). Additionally, we explored various other ML-based q-RASAR models to further enhance predictive performance. The established LR q-RASAR model was subsequently applied to predict the Mcule database (https://mcule.com/database/), containing 1,91,94,405 chemical compounds, to identify structurally relevant candidates with potential α7nACh receptor agonistic activity. Additionally, molecular docking analysis and molecular dynamics (MD) simulations for 100 ns were conducted to investigate interactions between the target protein and ligands. Overall, this investigation highlights the critical influence of hydrophobicity, electronic effects, ionization degree, and steric factors as key determinants in the design of potential anti-Alzheimer's agents.

Dual-specificity tyrosine phosphorylation-regulated kinases (DYRKs) play crucial roles in regulating cell growth and brain development. Dysregulation of these kinases is linked to disorders like Down syndrome and cancers. The selective inhibition of DYRK1A over other isoforms remains a significant challenge due to their high structural similarity. This study investigates the selectivity of Abemaciclib, an FDA-approved CDK4/6 inhibitor known to target DYRK1A, against other DYRK family isoforms. We employed molecular docking and molecular dynamics simulations, coupled with the Molecular Mechanics Poisson-Boltzmann Surface Area method, to evaluate the selectivity profile of Abemaciclib. Results showed that it binds strongest to DYRK1B, followed by DYRK1A, DYRK4, DYRK3 and DYRK2, which is validated with the statistical analysis. Enhanced selectivity for DYRK1B arises from stronger van der Waals and electrostatic interactions. Hydrophobic contacts and hydrogen bonds, especially within the kinase's hinge region, help stabilize the complex. Notably, Leu241 in DYRK1A and its identical residues in other isoforms play a pivotal role in these stabilizing interactions. Key residue differences, like Phe170, Glu239 and His285 in DYRK1A, contribute to specific interactions that underpin the molecular binding pattern. By identifying conserved and isoform-specific interactions, our study provides valuable insights for the rational design of potent and selective DYRK inhibitors.

Given the widespread presence of emerging contaminants in the environment, assessing and ensuring their biosafety is urgent. Under the Globally Harmonized System (GHS), the LD₅₀ parameter of acute oral toxicity (AOT) is crucial for chemical safety classification. Animal testing limitations have highlighted the need for alternative methods, and machine learning offers a new approach to predict LD₅₀ through quantitative structure-activity relationship (QSAR) models. This study developed and optimized a machine learning model for LD₅₀ classification of emerging contaminants based on data from more than 6000 known AOT. Using molecular descriptors and fingerprints, the model achieves an accuracy above 0.86 and a recall score over 0.84, outperforming previous models. The model's robustness was confirmed across various types of emerging contaminants. Shapley additive explanations (SHAP) identified key descriptors like BCUTp_1h, ATSC1pe, and SLogP_VSA4, while the information gain (IG) method highlighted alert substructures [P-O, P-S]. These findings suggest that compounds with high polarizability, mean electronegativity and significant surface area may adversely affect rats. This model enhances understanding of acute toxicity mechanisms and serves as a tool for early screening of safer compounds, promoting the design of greener chemicals.