【作者】 孙家英；

【导师】 蔡绍皙；

【作者基本信息】 重庆大学 ， 生物制药工程， 2010， 博士

【摘要】 神经氨酸酶抑制剂(neuraminidase inhibitors, NIs)是近年来重点研究的一类新型抗流感病毒药物,NIs可选择性地抑制神经氨酸酶(neuraminidase,NA),阻止子代病毒颗粒在宿主细胞中的复制和释放,从而有效地治疗流感和缓解症状。作为目前最有效的一类抗流感药物,NIs的耐药性病例已成逐年上升趋势,因此新型NIs的研发已迫在眉睫。本文采用计算机辅助药物分子设计(Computer-Aided Drug Design, CADD)中的定量构效关系(Quantitative Structure Activity Relationship, QSAR)和分子对接(docking)对6类单一骨架和2类混合骨架NIs进行系统研究,并取得了以下重要研究成果:(1) 8个NA的序列比对及其共结晶配体的作用模式分析表明:8个NA序列中完全相同(identity)的残基占19.7%。氢键、静电和疏水是影响NA-NIs相互作用的重要因素,且活性口袋中精氨酸(Arg118、152、292和371)、天冬氨酸(Asp151)、谷氨酸(Glu227)、色氨酸(Trp178)等残基在NA-NIs相互作用中起关键作用。8类NIs的对接研究,表明氢键、静电和疏水是影响NA-NIs相互作用的重要因素。对接可较准确地反映配体在受体活性口袋中的空间取向,且NA共结晶配体的计算位置与其观测位置之间的均方根偏差(root mean squared diviation,RMSD)都小于1.50?,相似性(Similarity, a measure of similarity between solution coordinates and reference coordinates)在0.72~0.84之间。87个吡咯烷衍生物与8个NA之间的对接得分(total score)与实验活性之间具有显著的相关性,其相关系数r在0.443~0.692 (p<0.001)之间。(2) 87个吡咯烷衍生物:采用HQSAR和Almond进行QSAR建模,最优模型的R2分别为0.846和0.830、Q2分别为0.670和0.560、R2test分别为0.872和0.856。分别基于共同结构、受体和药效团三种叠合方法,进行CoMFA和CoMSIA分析,得到6个最优QSAR模型的R2、Q2、R2test分别在0.588~0.861、0.437~0.625和0.614~0.875之间。QSAR分析表明氢键对活性的贡献最大,其次是疏水、静电和立体作用,与对接结果一致。基于共同结构建立的CoMFA模型,采用LeapFrog全新药物分子设计,得到8个对接得分和预测活性都较高的新化合物。(3) 36个硫脲和噻重氮[2,3-α]嘧啶衍生物:HQSAR和基于受体叠合的CoMFA和CoMSIA分析,得到QSAR模型的R2、Q2和R2test分别在0.863~0.899、0.509~0.656和0.558~0.667之间。基于36个NIs构建了8个特征的药效团模型,并基于此模型进行叠合,得到的CoMFA和CoMSIA模型的R2和Q2分别为0.989和0.737、0.470和0.401。采用3D HoVAIFA得到30个多取代硫脲的最优模型,其模型的R2、Q2和R2test分别为0.849、0.724和0.689。采用LeapFrog、UNITY数据库3D搜索及虚拟筛选,分别得到了7个和10个对接得分和预测活性都较高的新化合物。(4) 40个黄酮衍生物:HQSAR、Almond、CoMFA和CoMSIA(分别基于共同结构、受体和药效团叠合)建模分析,得到8个QSAR的R2、Q2、R2test和Q2ext分别在0.872~0.988、0.420~0.772、0.214~0.935和0.272~0.782之间。利用LeapFrog、UNITY数据库3D搜索及虚拟筛选,分别得到了对接得分和预测活性都较高的8个和6个新化合物。(5) 68个环己烯衍生物:HQSAR、CoMFA和CoMSIA建模研究,得到5个QSAR模型的R2、Q2、R2test和Q2ext分别在0.847~0.920、0.413~0.652、0.526~0.841和0.531~0.838之间。LeaFrog、UNITY数据库3D搜索及虚拟筛选,分别得到了15个和13个对接得分和预测活性都较高的新化合物。(6) 45个苯甲酸衍生物:HQSAR和基于药效团叠合所建CoMFA和CoMSIA分析,得到3个QSAR模型的R2、Q2和SEE分别在0.724~0.758、0.418~0.466和0.418~0.466之间。基于药效团模型,UNITY数据库3D搜索及虚拟筛选,得到了对接得分和得分一致性值CScore都较高的20个新化合物。(7) 38个环戊烷衍生物:采用HQSAR、CoMFA和CoMSIA建模,得到3个QSAR模型的R2、Q2和SEE分别在0.783~0.963、0.256~0.413和0.152~0.375之间。基于药效团模型, UNITY数据库3D搜索和虚拟筛选,得到了对接得分和得分一致性值CScore都较高的12个新化合物。(8) 124个混合骨架NIs:通过HQSAR、HoVAIFA、Almond、CoMFA和CoMSIA研究,得到7个QSAR模型的R2、Q2、R2test和Q2ext分别在0.775~0.952、0.456~0.818、0.518~0.748和0.516~0.799之间。采用LeapFrog、UNITY数据库3D搜索和虚拟筛选,分别得到了预测活性和对接得分都较高的5个和11个新化合物。(9) 41个混合骨架NIs:HQSAR、Alomnd、CoMFA和CoMSIA建模分析,得到了4个QSAR模型的R2、Q2、R2test和Q2ext分别在0.661~0.910、0.390~0.554、0.396~0.812和0.300~0.744之间。基于药效团模型,UNITY数据库3D搜索和虚拟筛选,得到了对接得分和CScore值都较高的12个新化合物。(10) 127个设计的新化合物的ADME性质预测结果,表明所设计的新化合物中约三分之一的化合物都具有良好的渗透能力、蛋白结合亲和力、体积分布以及代谢稳定性。更多还原

【Abstract】 Recently, neuraminidase inhibitors (NIs) are a focus of new anti-influenza virus drugs. NIs can selectively inhibit neuraminidase (NA), and prevent progeny virosome from replication and release in the host cell, thus take precautions against influenza and the alleviation symptom effectively. As an effective kind of anti-influenza drugs, there is a rising trend toward resistance of NIs due to emergence of new influenza virus. So, it is necessary to research and develop new NIs.In the dissertation, Quantitative structure-activity relationship (QSAR) and molecular docking in computer-aided drug design (CADD) were employed to systematically investigate six classes with the same skeleton and two classes with different skeletons of NIs. The main conclusions were as follows:(1) Sequence alignment of 8 NA indicated that the identical residues were 19.7%. Hydrogen bonding, hydrophobic and electrostatic interactions were important factors which had effect on NA-NIs interactions. Moreover, Arginine (118, 152, 292 and 371), Aspartate 151, Glutamate 227, Tryptophan 178 and so on were the key residues in the active site.Docking results between 8 classes of NIs and NA showed that hydrogen bonding, hydrophobic and electrostatic interactions were key factors which affected NA-NIs interactions. RMSD (root mean squared deviation) between the position calculated and that observed of crystal structure in NA were all smaller than 1.50?, and similarity (a measure of similarity between solution coordinates and reference coordinates) were from 0.72 to 0.84. Docking studies between 8 NA and 87 pyrrolidine derivatives showed that the hydrogen bonding, hydrophobic and electrostatic interactions affected on the activity of NIs. There was a significant correlation between docking score (total scores) and the experimental activities of 87 pyrrolidine derivatives, correlation coefficient r=0.443~0.692 (p<0.001).(2) 87 pyrrolidine derivatives: R2, Q2 and R2test of HQSAR and Almond models were 0.846 and 0.830, 0.670 and 0.560, 0.872 and 0.856, respectively. R2, Q2 and R2test of CoMFA and CoMSIA (common structures, receptor and pharmacophore superposition) models were from 0.588 to 0.861, from 0.437 to 0.625, and from 0.614 to 0.875, respectively. Based on CoMFA model (common structure alignment), 8 new compounds with high docking score and predicted activity were obtained by LeapFrog. (3) 36 acyl(thio)urea and thiadiazolo [2,3-a] pyrimidine derivatives: R2, Q2 and R2test of HQSAR, CoMFA and CoMSIA (based on receptor alignment) models were from 0.863 to 0.899, from 0.509 to 0.656, and from 0.558 to 0.667, respectively. At the same time, the pharmacophore model with 8 features was obtained by GALAHAD. Based on pharmacophore superposition, R2 and Q2 of CoMFA and CoMSIA models were 0.989 and 0.737, 0.470 and 0.401, respectively. In addition, R2, Q2 and R2test of HoVAIFA QSAR for 30 acyl(thio)urea derivatives were in turn 0.849, 0.724 and 0.689. 7 and 10 new compounds with high docking scores and predictive activity were obtained by LeapFrog and UNITY 3D search, respectively.(4) 40 flavonoids derivatives: R2, Q2, R2test and Q2ext of HQSAR, Almond, CoMFA and CoMSIA (common structure, receptor and pharmacophore alignment) models were from 0.872 to 0.988, from 0.420 to 0.772, from 0.214 to 0.935, and from 0.272 to 0.782, respectively. By LeapFrog and UNITY database 3D Search, 8 and 6 new compounds with high docking score and predicted activity were obtained, respectively.(5) 68 cyclohexene derivatives: R2、Q2、R2test and Q2ext of HQSAR, CoMFA and CoMSIA (pharmacophore and receptor superposition) models were from 0.847 to 0.920, from 0.413 to 0.652, from 0.526 to 0.841, and from 0.531 to 0.838, respectively. 15 and 13 new compounds with high docking score and predicted activity were obtained using LeapFrog and UNITY database 3D search, respectively.(6) 45 benzoic acid derivatives: R2、Q2 and SEE (standard error of estimate) of HQSAR, CoMFA and CoMSIA models were from 0.724 to 0.758, from 0.418 to 0.466, and from 0.418 to 0.466, respectively. 20 new compounds with high docking scores and CScore values were obtained by UNITY database 3D search and virtual screening.(7) 38 cyclopentane derivatives: R2, Q2 and SEE of HQSAR, CoMFA and CoMSIA models were from 0.783 to 0.963, from 0.256 to 0.413, and from 0.152 to 0.375, respectively. Based pharmacophore model, 12 new compounds with docking score and docking consistency values CScore were obtained through UNITY database search and virtual screening.(8) 124 NIs with different skeletons: R2, Q2, R2test and Q2ext of HQSAR, HoVAIFA, Almond, CoMFA and CoMSIA models were from 0.775 to 0.952, from 0.456 to 0.818, from 0.518 to 0.748, and from 0.516 to 0.799, repectively. 5 and 11 new compounds with high predicted activity and docking score were obtained by LeapFrog and UNITY database 3D search, respectively.(9) 41 NIs with different skeletons: R2, Q2, R2test and Q2ext of HQSAR, Alomnd, CoMFA and CoMSIA, HQAR and CoMSIA models were from 0.661 to 0.910, from 0.390 to 0.554, from 0.396 to 0.812, and from 0.300 to 0.744, respectively. Based pharmacophore model, 12 new compounds with high docking score and CScore values were obtained through UNITY database 3D search and virtual screening.(10) ADME researches indicated that one third of 127 new designed compounds had good permeation ability, high protein binding affinity, volume distribution and metabolic stability.更多还原