Ballante Flavio, Caroli Antonia, Wickersham Richard B, Ragno Rino
Rome Center for Molecular Design, Dipartimento di Chimica e Tecnologie del Farmaco, Sapienza Università di Roma , P. le A. Moro 5, 00185 Roma, Italy.
J Chem Inf Model. 2014 Mar 24;54(3):956-69. doi: 10.1021/ci400759t. Epub 2014 Mar 4.
The multichaperone heat shock protein (Hsp) 90 complex mediates the maturation and stability of a variety of oncogenic signaling proteins. For this reason, Hsp90 has emerged as a promising target for anticancer drug development. Herein, we describe a complete computational procedure for building several 3-D QSAR models used as a ligand-based (LB) component of a comprehensive ligand-based (LB) and structure-based (SB) virtual screening (VS) protocol to identify novel molecular scaffolds of Hsp90 inhibitors. By the application of the 3-D QSAutogrid/R method, eight SB PLS 3-D QSAR models were generated, leading to a final multiprobe (MP) 3-D QSAR pharmacophoric model capable of recognizing the most significant chemical features for Hsp90 inhibition. Both the monoprobe and multiprobe models were optimized, cross-validated, and tested against an external test set. The obtained statistical results confirmed the models as robust and predictive to be used in a subsequent VS.
多分子伴侣热休克蛋白（Hsp）90复合物介导多种致癌信号蛋白的成熟和稳定性。因此，Hsp90已成为抗癌药物开发的一个有前景的靶点。在此，我们描述了一个完整的计算程序，用于构建几个三维定量构效关系（3-D QSAR）模型，作为基于配体（LB）和基于结构（SB）的综合虚拟筛选（VS）方案的基于配体（LB）部分，以识别Hsp90抑制剂的新型分子骨架。通过应用3-D QSAutogrid/R方法，生成了八个SB PLS 3-D QSAR模型，得到了一个最终的多探针（MP）3-D QSAR药效团模型，该模型能够识别Hsp90抑制作用中最重要的化学特征。单探针和多探针模型均经过优化、交叉验证，并针对外部测试集进行了测试。获得的统计结果证实这些模型稳健且具有预测性，可用于后续的虚拟筛选。
