Biotechnology Center (BIOTEC), Technische Universität Dresden, 01307 Dresden, Germany.
J Chem Inf Model. 2021 May 24;61(5):2248-2262. doi: 10.1021/acs.jcim.0c01227. Epub 2021 Apr 26.
Pan-assay interference compounds (PAINS) are promiscuous compound classes that produce false positive hits in high-throughput screenings. Yet, the mechanisms of PAINS activity are poorly understood. Although PAINS are often associated with protein reactivity, several recent studies have shown that they also mediate noncovalent interactions. Aiming at a deep understanding of PAINS promiscuity, we performed an analysis of the Protein Data Bank to characterize the binding modes of PAINS. We explored the binding mode conservation of 34 PAINS classes present in 871 ligands and among 517 protein targets. The two major findings of this work are the following: First, different PAINS classes exhibit different levels of binding mode conservation. Our novel classification of PAINS based on binding mode similarity enables a rational assessment of PAINS from a structural perspective. Second, PAINS classes with variable binding modes can bind with high affinity. The evaluation of noncovalent binding modes of PAINS-like compounds sheds light on the mechanisms of promiscuous binding. Our findings could facilitate the decisions on how to deal with PAINS and help scientists to understand why PAINS produce hits in their screenings.
泛分析干扰化合物(PAINS)是一类混杂的化合物,它们在高通量筛选中会产生假阳性结果。然而,PAINS 的作用机制还了解甚少。尽管 PAINS 通常与蛋白质反应性有关,但最近的几项研究表明,它们也介导非共价相互作用。为了深入了解 PAINS 的混杂性,我们对蛋白质数据库进行了分析,以表征 PAINS 的结合模式。我们探索了 871 种配体和 517 种蛋白质靶标中存在的 34 种 PAINS 类的结合模式保守性。这项工作的两个主要发现如下:首先,不同的 PAINS 类表现出不同程度的结合模式保守性。我们基于结合模式相似性的新型 PAINS 分类可以从结构角度对 PAINS 进行合理评估。其次,具有可变结合模式的 PAINS 类可以与高亲和力结合。对 PAINS 样化合物的非共价结合模式的评估揭示了混杂结合的机制。我们的研究结果可以帮助科学家了解为什么 PAINS 在他们的筛选中产生阳性结果,并有助于他们做出如何处理 PAINS 的决策。