【作者】 王红青；

【导师】 俞庆森；

【作者基本信息】 浙江大学 ， 化学， 2006， 硕士

【摘要】 本论文对蛋白激酶C抑制剂、有机磷酸酯和吡咯抗真菌类化合物在衍生化纤维素类手性固定相CDMPC上的拆分进行了理论计算研究，探讨了其拆分机理。根据体系不同，全文共分三个部分，概述如下：第一部分，主要用定量结构-对映体保留关系的方法研究蛋白激酶C抑制剂在拆分过程中的影响因素。首先用TSAR软件计算蛋白激酶C抑制剂的参数（立体、静电、疏水和其他的参数），再用多元线性回归方法建立回归模型，回归方程的相关系数和交叉验证的相关系数都取得了较好的结果（logk₁的交叉验证回归系数q²=0.823，模型的线性回归系数为r²=0.946；logk₂的q²=0.727，r²=0.943）。然后用HQSAR、CoMFA和CoMSIA方法对蛋白激酶C抑制剂进行定量结构-对映体保留关系的研究，得到了与多元线性回归相类似的结果，都说明了增大“亚甲基氨”位置上的疏水基团的体积对于提高该类化合物的分离是有利的。在HQSAR模型中，logk₁模型的q²达到0.868，r²达到0.984，而logk₂的q²和r²分别为0.908、0.986；CoMFA模型中，q²达到了0.6以上，而CoMSIA方法中，q²大于0.7，都说明建立的模型具有良好的线性相关性和预测未知化合物的能力。第二部分，主要用定量结构-对映体保留关系的方法研究有机磷酸酯化合物在拆分过程中的影响因素。首先用TSAR软件计算有机磷酸酯化合物的参数（立体、静电、疏水和其他的参数），再用多元线性回归方法建立回归模型，回归方程的相关系数和交叉验证的相关系数都取得了较好的结果（logk₁的交叉验证回归系数q²=0.760，模型的线性回归系数为r²=0.945；logk₂的q²=0.643，r²=0.898）。然后用HQSAR、CoMFA和CoMSIA方法对有机磷酸酯化合物进行定量结构-对映体保留关系的研究，得到了与多元线性回归相类似的结果，并且都说明增大“磷酸酯基团中氧原子”位置上的疏水基团的体积对于提高该类化合物的分离是不利的，碳链越长越不利。在HQSAR模型中，logk₁的q²达到0.950，r²达到0.984，而logk₂的q²和r²分别为0.840、0.978，说明建立的模型具有良好的线性相关性和预测未知化合物的能力。但是CoMFA模型中，q²在0.4左右，而CoMSIA方法中，q²大于0.4，结果不理想。和HQSAR结果对比，可能是选取的分子构象导致的。第三部分，主要用定量结构-对映体保留关系的方法研究吡咯抗真菌类化合物在拆分过程中的影响因素。首先是用TSAR软件计算吡咯抗真菌类化合物的参数（立体、静电、疏水和其他的参数），再用多元线性回归方法建立回归模型，logk₁的交叉验证回归系数q²=0.377，模型的线性回归系数为r²=0.761；logk₂的q²=0.161，r²=0.717。这个结果很不理想，可能是此类化合物的结构差异很小，分离因子的数值范围也较小的缘故。然后用HQSAR、CoMFA和CoMSIA方法对吡咯抗真菌类化合物进行定量结构-对映体保留关系的研究，得到了比多元线性回归好的结果。在HQSAR模型中，logk₁的q²达到0.758，r²达到0.922，而logk₂的q²和r²分别为0.557、0.864；CoMFA和CoMSIA模型中，q²都大于0.6，说明建立的模型具有较好的线性相关性和一定的预测未知化合物的能力。在本论文中，我们首次将定量结构-对映体保留关系方法用于纤维素-三（3，5-二甲基苯基氨基甲酸酯）手性固定相（CDMPC）拆分手性化合物的机理研究中，各种方法相互印证，能很好地解释实验结果，说明了研究方法的可行性和建立的模型的可靠性。并且，识别模型与定量结构-对映体保留关系研究结果的结合，对修饰和设计新的手性固定相有一定的指导意义。更多还原

【Abstract】 The thesis is about theoretical calculation on chiral separations of enantiomer compounds of protein kinase C inhibitors、organic phosphonate esters and azole antifungal agents on cellulose tris（3,5’dimethylphenyl carbamate） chiral stationary phase （CDMPC）. There are three main parts, as follows:In the first part, the focus is the Quantitative Structure Enantioselective Rentention Relationship （QSERR） study on protein kinase C inhibitors. Firstly, we calculated the descriptors of protein kinase C inhibitors including those obtained by quantum chemical calculation, and multiple linear regression （MLR） analyses was used to study the effect factor in chiral separation and gained good results （cross-validation coefficient and linear correlation coefficient of logk₁ equal to 0.823 and 0.946; and those of logk₂ equal to 0.727 and 0.943）. Furthermore, HQSAR、CoMFA and CoMSIA was applied to this study, and satisfying results were obtained, indicating that our models are all successful, and results show increasing the volume of the substituents R on -CH₂NH-R will facilitate separation. In the models of HQSAR, cross-validation coefficient and linear correlation coefficient of logk₁ equal to 0.868 and 0.984; those of logk₂ equal to 0.908 and 0.986. The results indicate that the HQSAR models are successful too.In the second part, the same methods were used to perform the QSERR study of organic phosphonate esters. The analyses show the 2D-QSAR result is accordant with that of the 3D-QSAR. Firstly, we calculated the descriptors of organic phosphonate esters including those obtained by quantum chemical calculation, and multiple linear regression （MLR） analyses was used to study the effect factor in chiral separation and gained good results （cross-validation coefficient and linear correlation coefficient of logk₁ equal to 0.760 and 0.945; and those of logk₂ equal to 0.643 and 0.898）. Furthermore, HQSAR、CoMFA and CoMSIA was applied to this study, and satisfying results were obtained, indicating that our models are all successful, and results show increasing the volume of the substituents on oxygen atom of organic phosphonate esters will go against separation. In the models of HQSAR, cross-validation coefficient and linear correlation coefficient of logk₁ equal to 0.950 and 0.984; those of logk₂ equal to 0.840 and 0.978. The results indicate that the HQSAR models are successful too.In the third part, the QSERR of azole antifungal agents was performed using the TSAR（MLR）、HQSAR、CoMFA and CoMSIA methods, and obtained satisfying results. Firstly, we calculated the descriptors of protein kinase C inhibitors including those obtained by quantum chemical calculation, and multiple linear regression （MLR） analyses was used to study the effect factor in chiral separation, cross-validation coefficient and linear correlation coefficient of logk₁ equal to 0.377 and 0.761; and those of logk₂ equal to 0.161 and 0.717. The results are not satisfactory being likely to nuance of structures and of separation factors. Furthermore, HQSAR、CoMFA and CoMSIA was applied to this study, and satisfying results were obtained, indicating that our models are all successful, and better than that of MLR. In the models of HQSAR, cross-validation coefficient and linear correlation coefficient of logk₁ equal to 0.758 and 0.922; those of logk₂ equal to 0.557 and 0.864. The results indicate that the HQSAR models are successful.For the first time, we used the QSERR methods to research the chiral separation mechanism of chiral compounds on CDMPC, and fortunately, the results indicate that our recognition models do work on the explaining and validating the experiments. These models will be useful to speculate the combining mode between the CDMPC and these compounds as well as provide theoretical foundation for designing and synthesizing new type CSPs.更多还原