Zhang Na, Zhao Hongtao
College of Life Science and Bioengineering, Beijing University of Technology, Beijing 100124, China.
Lephar Research, Rindögatan 21, 11558 Stockholm, Sweden.
Bioorg Med Chem Lett. 2016 Aug 1;26(15):3594-7. doi: 10.1016/j.bmcl.2016.06.013. Epub 2016 Jun 7.
By deconvoluting 238,073 bioactive molecules in the ChEMBL library into extended Murcko ring systems, we identified a set of 2245 ring systems present in at least 10 molecules. These ring systems belong to 2221 clusters by ECFP4 fingerprints with a minimum intracluster similarity of 0.8. Their overlap with ring systems in commercial libraries was further quantified. Our findings suggest that success of a small fragment library is driven by the convergence of effective coverage of bioactive ring systems (e.g., 10% coverage by 1000 fragments vs. 40% by 2million HTS compounds), high enrichment of bioactive ring systems, and low molecular complexity enhancing the probability of a match with the protein targets. Reconciling with the previous studies, bioactive ring systems are underrepresented in screening libraries. As such, we propose a library of virtual fragments with key functionalities via fragmentation of bioactive molecules. Its utility is exemplified by a prospective application on protein kinase CK2, resulting in the discovery of a series of novel inhibitors with the most potent compound having an IC50 of 0.5μM and a ligand efficiency of 0.41kcal/mol per heavy atom.
通过将ChEMBL库中的238,073种生物活性分子解卷积为扩展的Murcko环系统,我们鉴定出一组至少存在于10个分子中的2245个环系统。通过ECFP4指纹图谱,这些环系统属于2221个簇,簇内最小相似度为0.8。我们进一步量化了它们与商业库中环系统的重叠情况。我们的研究结果表明,一个小片段库的成功取决于生物活性环系统有效覆盖范围的趋同(例如,1000个片段覆盖10%,而200万个高通量筛选化合物覆盖40%)、生物活性环系统的高富集度以及低分子复杂性,从而提高与蛋白质靶点匹配的概率。与之前的研究一致,生物活性环系统在筛选库中的代表性不足。因此,我们通过生物活性分子的片段化提出了一个具有关键功能的虚拟片段库。其效用通过在蛋白激酶CK2上的前瞻性应用得到例证,从而发现了一系列新型抑制剂,其中最有效的化合物的IC50为0.5μM,每个重原子的配体效率为0.41kcal/mol。