药物吸收与代谢的理论化学研究

【作者】 胡桂香；

【导师】 俞庆森； 商志才；

【作者基本信息】 浙江大学 ， 物理化学， 2003， 博士

【摘要】 本论文主要针对化合物药动学特征的吸收和代谢过程进行了理论方面的研究,建立了药物透过人体小肠、Caco-2单细胞层、人体皮肤、血脑屏障等生物膜的吸收数学模型,并对大量化合物进行预测;对药物由于代谢而产生的氧化羟化影响因素进行了初步探讨。口服药物都要经过小肠吸收。我们首先用表述分子三维结构的VolSurf参数建立了药物小肠吸收模型,取得较好的结果(模型的相关性为r2 = 0.82,留一交叉验证方法证明模型的预测能力为q2 =0.67)。对其它数据不太可靠的化合物按照数据评价的质量进行分类预测,存在整体误差,能根据数据来源进行较好的解释,证明所建立的模型具有较好的预测能力。对参数进行分析表明有利于人体小肠吸收的药物分子特征包括:分子量中心与亲水、疏水中心高的不平衡性,分子内较大的疏水区域,分子内少的氢键给体以及低的亲水-亲脂平衡,即疏水性在分子中占主导地位。药物透过Caco-2单细胞层的渗透系数的数学模型给出了较好的结果(r2 = 0.95,q2 = 0.75)。对其它类药物进行预测,发现用结构不同的化合物所建立的模型对肽类药物以及同系物分子均有一定的预测能力,但是当分子的柔性键多,可能形成分子内氢键时,模型的预测能力大大下降。药物透过Caco-2单细胞层的影响因素较人体小肠复杂,分子内较多的氢键给体和受体均对Caco-2单细胞层的渗透性不利,对Caco-2单细胞层渗透性高的药物分子必须具有合适的氢键给体和受体。分子内局部能量极小值越大,对渗透越有利。同时,体积大且球形性高的分子对渗透有利。药物经皮吸收的主要屏障部位在角质层,由于该层为死细胞层,使得药物经皮渗透影响因素与其它生物膜有较大的不同。药物经皮吸收的数学模型较好(r2 = 0.86,q2 = 0.80)。参数分析表明具有高疏水性的分子有利于经皮吸收,分子内亲水区域较大时不利于经皮吸收。当分子内三个能量极小值较小,且它们之间的相互距离较大时有利于经皮吸收。同时,当分子体积的增大引起分子亲水-疏水平衡减小时会导致药物经皮渗透性的增大。一定的亲水区域对经皮吸收有利。以所建立的模型作为训练集,对其他各类药物进行预测。结果表明模型对小分子的结构不同的化合物预测能力较好,但对甾类化合物的预测出现了系统偏差。对有争议的数据进行了预测,取得与文献报道相一致的结果。在VolSurf参数与药物血脑屏障渗透性之间建立模型,相关性及预测能力均较好,表现为r2=0.92,q2=0.78。以所建立的模型对其他各类数据来源不同的药物进行预测,结果表明模型对所考察的两个预测集均有较好的预测能力。参数分析结果表明药物分子量中心与疏水区域之间的不平衡性有利于BBB渗透性;只有合适的疏水区域才对BBB渗透性有利。较多的氢键给体和受体对BBB渗透性不利,即分子内较多的亲水区是不利因素。分子体积V,表面积S,褶皱程度R与BBB渗透性正相关,分子量MW与分子球形性G则为负相关。药物的水溶解度与吸收密切相关。用VolSurf参数与药物水溶解度之间建立相关模型,得到较好的结果(r2=0.90,q2=0.77)。参数分析表明三个局部能量最小值越小,且它们之间的距离越大,对其水溶解度越有利;分子内较大的亲水区域对水溶解度有利,疏水区域情况相反。同时,大分子的溶解度较小分子低。对强亲水性分子难以穿透生物膜问题利用人们提出的分子伞模型进行了理论计算,结果表明真空下屏蔽构象和暴露构象的能量相差不大,且低能构象交替出现,验证了谷胱甘肽分子伞在不同的环境中能够呈现不同的构象这一膜输送先决条件。当考虑溶剂效应,水溶液中暴露构象的能量较屏蔽构象显著低,前者可以在水溶液中稳定存在而后者不能,从理论上解释了谷胱甘肽分子伞在水溶液中呈现暴露构象的原因。利用VolSurf参数分析了分子伞以屏蔽构象穿透磷脂双分子层的影响因素,结果表明屏蔽结构较小的两亲矩及较大的分子褶皱程度是其能够穿透细胞膜的主要影响因素,与构象的绝对疏水区域无关。从量子化学角度出发,分别对3,5-二芳基-s-三氮唑类以及DL204-IT及其化学修饰结构两类非甾类终止妊娠化合物进行了研究,并建立了相关模型,相关性较好(r分别为0.911,0.850)。两个模型具有高度一致性,结果互相验证,具有较强的说服力。根据所建立的模型,化合物生成热(HF)以及发生氧更多还原

【Abstract】 The thesis is about theoretical study on absorption and metabolism process of pharmacokinetics characters. The mathematical models are built for prediction of drug absorption through biological membranes including human small intestine, Caco-2 cell monolayers, human skin and blood-brain barrier. The preliminary study is performed on influencing factors of hydroxylation for drug metabolism.All oral drugs are absorbed through small intestine. The mathematical model of intestinal absorption is built with VolSurf descriptors, which describe molecular three-dimensional structure. The good result is gained (the correlation coefficient (r2) is 0.82 and the predictivity by cross validation (q2) is 0.67). We predict other unreliable data according to their quality with the model. The entire difference can be explained well on the basis of data source, which shows that the model owns the good predictivity. The molecular characters favorable to human intestinal absorption include the high imbalance between the center of mass of a molecule and the barycentre of its hydrophilic and hydrophobic regions, high hydrophobic regions, less hydrogen bond donors, as well as low hydrophilic-lipophilic balance, that is hydrophobic region dominates in the molecule.The mathematical model of the permeability coefficients of drugs through Caco-2 cell monolayers shows the good result (r2 = 0.95 and q2 = 0.75). The prediction results demonstrate that the model has some predictivity for peptide drugs and homologues. But the predictivity of the model decreases when more flexible bonds in the molecule result in intramolecular hydrogen bonds. The influencing factors are more complicated than human intestine. Not the more but the suitable hydrogen bond donors and acceptors are favorable to the permeability. The large local energy minima are beneficial to the permeability of drugs. At the same time, large volume and high globularity are favorable factors.Because the principal barrier, the stratum corneum, is in essence a dead layer in the living organism, the influencing factors of human percutaneous absorption of drugs are different from other biological membranes. The good mathematical model of percutaneous absorption shows r2 is 0.86 and q2 is 0.80. High hydrophobicity is favorable and large hydrophilic region is unfavourable to skin penetration. Low three local energy minima and far distances between them are beneficial to penetration. Meanwhile, when molecular volume increasing brings about hydrophilic-lipophilic balance decreasing, percutaneous permeability increases. Some hydrophilic regions are favorable to penetration. Other kinds of drugs are predicted with the training set. The good results are gained from small diverse molecules. But the systematic difference exists for steroids. The predicted results for argumentative data are in agreement with the reference.The mathematical model is built between VolSurf descriptors and drug permeability of blood-brain barrier. The result shows r2 = 0.92 and q2 = 0.78. The model possesses good predictivity to two test sets from different sources. Analysis to descriptors shows the imbalance between the center of mass of a molecule and the barycentre of its hydrophobic regions is favorable to BBB permeability. Only suitable hydrophobic regions are beneficial更多还原