Machine learning models to select potential inhibitors of acetylcholinesterase activity from SistematX: a natural products database,Molecular Diversity

当前位置： X-MOL 学术 › Mol. Divers. › 论文详情

Our official English website, www.x-mol.net, welcomes your feedback! (Note: you will need to create a separate account there.)

Machine learning models to select potential inhibitors of acetylcholinesterase activity from SistematX: a natural products database
Molecular Diversity ( IF 3.9 ) Pub Date : 2021-06-16 , DOI: 10.1007/s11030-021-10245-z
Chonny Herrera-Acevedo _{1,

2} , Camilo Perdomo-Madrigal ₃ , Kenyi Herrera-Acevedo ₄ , Ericsson Coy-Barrera ₂ , Luciana Scotti ₁ , Marcus Tullius Scotti ₁

Affiliation

Abstract

Alzheimer’s disease is the most common form of dementia, representing 60–70% of dementia cases. The enzyme acetylcholinesterase (AChE) cleaves the ester bonds in acetylcholine and plays an important role in the termination of acetylcholine activity at cholinergic synapses in various regions of the nervous system. The inhibition of acetylcholinesterase is frequently used to treat Alzheimer’s disease. In this study, a merged BindingDB and ChEMBL dataset containing molecules with reported half-maximal inhibitory concentration (IC₅₀) values for AChE (7032 molecules) was used to build machine learning classification models for selecting potential AChE inhibitors from the SistematX dataset (8593 secondary metabolites). A total of seven fivefold models with accuracy above 80% after cross-validation were obtained using three types of molecular descriptors (VolSurf, DRAGON 5.0, and bit-based fingerprints). A total of 521 secondary metabolites (6.1%) were classified as active in this stage. Subsequently, virtual screening was performed, and 25 secondary metabolites were identified as potential inhibitors of AChE. Separately, the crystal structure of AChE in complex with (–)-galantamine was used to perform molecular docking calculations with the entire SistematX dataset. Consensus analysis of both methodologies was performed. Only eight structures achieved combined probability values above 0.5. Finally, two sesquiterpene lactones, structures 15 and 24, were predicted to be able to cross the blood–brain barrier, which was confirmed in the VolSurf+ quantitative model, revealing these two structures as the most promising secondary metabolites for AChE inhibition among the 8593 molecules tested.

Graphic abstract

A consensus analysis of classification models and molecular docking calculations identified four potential inhibitors of acetylcholinesterase from the SistematX dataset (8593 structures).

中文翻译：

从 SistematX 中选择潜在乙酰胆碱酯酶活性抑制剂的机器学习模型：天然产物数据库

摘要

阿尔茨海默病是最常见的痴呆形式，占痴呆病例的 60-70%。乙酰胆碱酯酶 (AChE) 裂解乙酰胆碱中的酯键，并在终止神经系统不同区域胆碱能突触的乙酰胆碱活性中起重要作用。乙酰胆碱酯酶的抑制常用于治疗阿尔茨海默病。在本研究中，合并的 BindingDB 和 ChEMBL 数据集包含报告的半数最大抑制浓度（IC ₅₀) AChE（7032 个分子）的值用于构建机器学习分类模型，以从 SistematX 数据集（8593 个次级代谢物）中选择潜在的 AChE 抑制剂。使用三种类型的分子描述符（VolSurf、DRAGON 5.0 和基于位的指纹）获得了交叉验证后准确率超过 80% 的总共 7 个五倍模型。在此阶段，共有 521 种次级代谢物（6.1%）被归类为有活性的。随后，进行了虚拟筛选，确定了 25 种次级代谢物为 AChE 的潜在抑制剂。另外，AChE 与 (–)-galantamine 复合的晶体结构用于对整个 SistematX 数据集进行分子对接计算。对两种方法进行了共识分析。只有八个结构的组合概率值高于 0.5。最后，两个倍半萜内酯，结构15和24，预计能够穿过血脑屏障，这在 VolSurf + 定量模型中得到证实，揭示这两种结构是测试的 8593 分子中最有希望抑制 AChE 的次级代谢物。