药物代谢性质（ADME）早期快速预测技术的研究及系列化合物的评价

【作者】 庄笑梅；

【导师】 阮金秀；

【作者基本信息】 中国人民解放军军事医学科学院 ， 药理学， 2006， 博士

【摘要】 随着组合化学和高通量筛选技术的发展，药物研发速度发生了质的飞跃，但其中40％化合物会因代谢性质较差而被淘汰。本课题的目的就是为了紧跟国际前沿和加快国内创制新药的步伐，在两项基金的资助下，率先在国内建立起快速的早期药代预测体系，解决关键技术，推动药物研发进程。 整个课题研究内容分为两大部分：遵循快速、简单、通用、有效的原则，首先建立了计算机虚拟、体外及体内三个层次快速药代性质预测模型；进而应用该体系对几个创新系列化合物进行筛选，得到了一些结构-代谢效应关系的规律，评选出几个有开发前景的候选药物，并验证了不同模型之间的相关性。 模型构建： 计算机虚拟预测采用VolSurf及Pallas软件。先用已知的阳性结果验证软件的预测能力，其中Caco-2细胞透膜性及血脑屏障通透性都与文献报道的结果有较好的相关性，相关系数分别为r~2=0.8998，r~2=0.7714。我们还应用VolSurf软件首次构建了预测P-gp的底物模型。通过模型提供的信息，得出一些有意义的结论：1．化合物大小、形状及容量性质与P-gp ATP酶活性成正相关；2．化合物的POL(极性)和氢键容量(HB)也明显影响P-gp ATP酶活性；3．P-gp底物可能具有两亲性。 在体外快速预测技术中，我们用阳性药物作为质控标准构建了国际公认的用于大量化合物透膜性质评价的Caco-2细胞模型、人工膜模型和用于评价代谢稳定性的肝微粒体模型，为了提高体外评价的通量，还在人工膜模型和肝微粒体模型上建立了盒式多药(n in one)筛选技术。 在体内快速评价技术中，建立了整体动物同时给多个药物(盒式给药)，并与单独给药的结果进行比较，发现两种给药方式的药代动力学参数没有明显的差别。在生物利用度的快速评价中，采用多次采血，一次测定的实验方法“一点法”，比较同系列化合物的生物利用度，结果与传统方法的生物利用度排序基本一致。 药物评价： 对AGEs裂解剂系列化合物从计算机、体外、体内三个层次进行了评价。通过对计算机预测结果与结构进行比较发现：化合物的透膜性主要与临界堆积参数和D(疏水探针与靶标分子相互作用产生的疏水性区域)成正比，而C_w(容量因子，亲水区／分子表面积)与其成反比。在体外实验中，分别应用Caco-2细胞、人工膜模型、肝微粒体模型进行评价，并考察了计算机、Caco-2细胞模型评价结果的相关性。更多还原

【Abstract】 With the development of Combinatorial Chemistry and high throughput screening technology, the speed of drug discovery has improved dramatically. However, it has been estimated that approximately 40% of compounds have failed in the past due to problems in pharmacokinetics and drug delivery. The objective of this paper was to in pace with the latest development of international pharmaceutical industry and to accelerate the national drug research. Surported by two foundations, we have taken the lead in constructing a systematic screening system and established the key technology to speed up the development of drug discovery.This dissertation consists of two parts. Firstly, in accordance with the principle of rapidness, simplicity, generality and effectiveness, an integrated screening ADME properties system including three phases of in silico, in vitro and in vivo was constructed. Secondly, we evaluated several series of novel compounds by using the platform and validated the relationships among different models.Model Construction:VolSurf and Pallas soft wares were applied in computational prediction. As the first step, we validated the predictive ability of these soft wares by using the known compounds published on literature or from our past experiments. Liner regression analysis showed a reasonable correlation (r~2=0.8998, r~2=0.7714). A new two components partial least squares discriminant analysis (PLS) model for the prediction of P-glycoprotein-associated ATPase activity of drugs was built by using VolSurf compute theoretical molecule descriptors derived from 3D molecule interaction fields. The results effectively investigated that properties associated with the volume, polarizability, and hydrogen bond could have important impact on the P-glycoprotein-associated ATPase activity.On in vitro ADME screening system, Caco-2 cell model and PAMPA model were built to evaluate the permeability, and liver microsome model was developed to appraise the metabolic stability. We optimized the experimental conditions to standardize the models. In order to increase the in vitro throughput, 5 in 1 approach in PAMPA and liver更多还原