Liao Si Yan, Chen Jin Can, Qian Li, Shen Yong, Zheng Kang Cheng
School of Chemistry and Chemical Engineering, Zhongshan (Sun Yat-Sen) University, Guangzhou 510275, People's Republic of China.
Eur J Med Chem. 2008 Oct;43(10):2159-70. doi: 10.1016/j.ejmech.2007.10.033. Epub 2007 Nov 7.
The quantitative structure-activity relationship (QSAR) of 32 flavone and isoflavone derivatives with cytotoxicity expressed as pGC50, which is defined as the negative value of the logarithm of necessary molar concentration of this series of compounds to cause 50% growth inhibition against the human cervical epithelioid carcinoma cell line (HeLa), has been studied by using the density functional theory (DFT), molecular mechanics (MM2) and statistical methods. In order to obtain QSAR model with high predictive ability, the original dataset was randomly divided into a training set comprising 26 compounds and a test set comprising the rest 6 compounds. An optimal model for the training set with significant statistical quality (RA2=0.852) and predictive ability (q2=0.818) was established. The same model was further applied to predict pGC50 values of the 6 compounds in the test set, and the resulting predictive correlation coefficient Rpred2 reaches 0.738, further showing that this QSAR model has high predictive ability. It is very interesting to find that the cytotoxicities of these compounds against HeLa appear to be mainly governed by two quantum-chemical factors, i.e., the energy (ELUMO) of the lowest unoccupied molecular orbital (LUMO) and the net charges of C atom at site 6 on aromatic rings (QC6). Here the possible action mechanism of these compounds was analyzed and discussed in detail, in particular, the fact why the flavone derivatives have considerably higher cytotoxicity than isoflavone derivatives was reasonably explained. Based on this QSAR equation, 5 new compounds with higher cytotoxicity have been theoretically designed. Such results can offer useful theoretical references for experimental works.
以pGC50表示细胞毒性的32种黄酮和异黄酮衍生物的定量构效关系（QSAR）已通过密度泛函理论（DFT）、分子力学（MM2）和统计方法进行了研究。pGC50定义为该系列化合物对人宫颈上皮样癌细胞系（HeLa）产生50%生长抑制所需摩尔浓度的负对数。为了获得具有高预测能力的QSAR模型，将原始数据集随机分为包含26种化合物的训练集和包含其余6种化合物的测试集。建立了一个具有显著统计质量（RA2 = 0.852）和预测能力（q2 = 0.818）的训练集最优模型。将同一模型进一步应用于预测测试集中6种化合物的pGC50值，所得预测相关系数Rpred2达到0.738，进一步表明该QSAR模型具有较高的预测能力。有趣的是，发现这些化合物对HeLa的细胞毒性似乎主要受两个量子化学因素控制，即最低未占分子轨道（LUMO）的能量（ELUMO）和芳环上6位C原子的净电荷（QC6）。在此详细分析和讨论了这些化合物可能的作用机制，特别是合理地解释了黄酮衍生物比异黄酮衍生物具有更高细胞毒性的原因。基于此QSAR方程，理论上设计了5种具有更高细胞毒性的新化合物。这些结果可为实验工作提供有用的理论参考。
