Horwitz J P, Massova I, Wiese T E, Wozniak A J, Corbett T H, Sebolt-Leopold J S, Capps D B, Leopold W R
Department of Internal Medicine, Wayne State University School of Medicine, Detroit, Michigan 48202.
J Med Chem. 1993 Nov 12;36(23):3511-6. doi: 10.1021/jm00075a004.
In vitro screening of a number of 2-(aminoalkyl)-5-nitropyrazolo[3,4,5- kl]acridines has previously indicated (Sebolt, J.S.; et al. Cancer Res. 1987, 47, 4299-4304) that these compounds, in general, exhibit selective cytotoxicity against the human colon adenocarcinoma, HCT-8, cell line, relative to mouse leukemia L1210 cells. Comparative molecular field analysis (CoMFA) was applied to HCT-8 and L1210 growth inhibition assays (IC50s) of a series (44) of the pyrazoloacridine derivatives with the objective of predicting improved solid tumor selectivity. In the absence of crystallographic data, the 9-methoxy derivative (15), which is currently in clinical study, was selected as the template molecular model. Two different structural alignments were tested: an alignment of structures based on root mean square (RMS)-fitting of each structure to 15 was compared with an alternative strategy, steric and electrostatic alignment (SEAL). Somewhat better predictive cross-validation correlations (r2) were obtained with models based on RMS vis-à-vis SEAL alignment for both sets of assays. A large change in lattice spacing, e.g., 2 to 1 A, causes significant variations in the CoMFA results. A shift in the lattice of half of its spacing had a much smaller effect on the CoMFA data for a lattice of 1 A than one of 2 A. The relative contribution of steric and electrostatic fields to both models were about equal, underscoring the importance of both terms. Neither calculated log P nor HOMO and/or LUMO energies contribute to the model. Steric and electrostatic fields of the pyrazoloacridines are the sole relevant descriptors to the structure-activity (cross-validated and conventional) correlations obtained with the cytotoxic data for both the L1210 and HCT-8 cell lines. The cross-validated r2, derived from partial least-squares calculations, indicated considerable predictive capacity for growth inhibition of both the leukemia and solid-tumor data. Evidence for the predictive performance of the CoMFA-derived models is provided in the form of plots of actual vs predicted growth inhibition of L1210 and HCT-8 cells, respectively, by the pyrazoloacridines. The steric and electrostatic features of the QSAR are presented in the form of standard deviation coefficient contour maps of steric and electrostatic fields. The maps indicate that increases or decreases in steric bulk that would enhance growth inhibition of HCT-8 cells would likewise promote growth inhibition of L1210 cells. Contour maps generated to analyze the electrostatic field contributions of the pyrazoloacridines to growth inhibition provide an essentially similar set of results.(ABSTRACT TRUNCATED AT 400 WORDS)
先前对多种2-(氨基烷基)-5-硝基吡唑并[3,4,5-kl]吖啶进行的体外筛选表明（Sebolt, J.S.等人，《癌症研究》，1987年，47卷，4299 - 4304页），相对于小鼠白血病L1210细胞，这些化合物总体上对人结肠腺癌HCT - 8细胞系表现出选择性细胞毒性。应用比较分子场分析（CoMFA）对一系列（44种）吡唑并吖啶衍生物进行HCT - 8和L1210生长抑制试验（IC50值）分析，目的是预测改善实体瘤选择性。由于缺乏晶体学数据，选择目前正在进行临床研究的9 - 甲氧基衍生物（15）作为模板分子模型。测试了两种不同的结构比对：一种是基于将每个结构与15进行均方根（RMS）拟合的结构比对，另一种是空间和静电比对（SEAL）策略。对于两组试验，基于RMS比对的模型相对于SEAL比对得到的预测交叉验证相关性（r2）略好。晶格间距的大幅变化，例如从2 Å到1 Å，会导致CoMFA结果有显著差异。对于1 Å的晶格，晶格间距偏移其一半对CoMFA数据的影响远小于2 Å晶格的情况。空间和静电场对两个模型的相对贡献大致相等，这突出了这两个因素的重要性。计算得到的log P以及HOMO和/或LUMO能量均对模型没有贡献。吡唑并吖啶的空间和静电场是与L1210和HCT - 8细胞系细胞毒性数据所获得的结构 - 活性（交叉验证和常规）相关性的唯一相关描述符。通过偏最小二乘法计算得到的交叉验证r²表明，该模型对白血病和实体瘤数据的生长抑制具有相当的预测能力。CoMFA衍生模型预测性能的证据以吡唑并吖啶分别对L1210和HCT - 8细胞实际生长抑制与预测生长抑制的绘图形式给出。QSAR的空间和静电特征以空间和静电场的标准偏差系数等高线图形式呈现。这些图表明，增加或减少空间体积以增强HCT - 8细胞生长抑制的同时，也会促进L1210细胞的生长抑制。为分析吡唑并吖啶对生长抑制的静电场贡献而生成的等高线图给出了基本相似的一组结果。（摘要截断于400字）
