【作者】 孔韧；

【导师】 王存新；

【作者基本信息】 北京工业大学 ， 生物医学工程， 2008， 博士

【摘要】 1983年,科学家Barre-Sinoussi等首次确认1型人类免疫缺陷病毒(Human Immunodeficiency virus 1, HIV-1)是引起获得性免疫缺陷综合症(Acquired immunodeficiency syndrome, AIDS)的病原体。自从1981年6月发现首例艾滋病患者到2007年底,这种疾病已导致超过两千万人死亡。尽管现有的药物以及高效抗逆转录病毒联合疗法(Highly Active Antiretroviral Therapy, HAART)的应用改善了病人临床治疗效果,但无法根除病毒、耐药性病毒株的出现以及毒副作用是其难以克服的困难。迄今为止,疫苗的研制仍未获得成功,因此针对新靶点研发安全高效的抗HIV药物非常紧迫。2003年上市的Fuzeon(T-20)证明了阻断病毒进入细胞可抑制艾滋病,病毒进入细胞过程成为当前药物设计研究的热点。众所周知,新药研发是一项耗资巨大的工程,为了提高研发效率、缩短研发周期、降低成本,计算机辅助药物设计(Computer-aided drug design, CADD)方法近年来被广泛地应用于现代新药开发中。该方法主要分为两类:基于结构的药物设计以及基于配体的药物设计。本论文结合CADD方法,以设计新型抗艾滋病药物为目的,以病毒进入细胞过程为研究对象,从配体和受体出发进行了一系列应用基础研究工作。论文内容主要包括以下几个方面:(1) HIV-1包膜蛋白gp120与抑制剂BMS-378806结合模式的研究BMS-378806是新发现的可有效抑制gp120蛋白与CD4受体结合的小分子,它与gp120的结合模式尚未阐明。本论文联合使用分子对接(AutoDock 3.0)和分子动力学(Molecular dynamics, MD)模拟方法,探索了BMS-378806与gp120蛋白的结合模式。对接能较低的结合模式Mode I和Mode II被选为代表构象。进一步的复合物MD模拟表明,Mode II中小分子与gp120结合更为稳定。分析Mode II的动力学平均结构并和实验结果进行对照,发现小分子氮杂吲哚与Trp427残基的接近可能导致了环上C-4位置取代的敏感性。在Phe43口袋内部的残基向大侧链残基的突变,如S375W和T257R可能会对结合造成空间位阻。口袋中的其他残基Ser256、Thr257、Asn425、Met426和Val430与小分子有较强的相互作用。复合物的平均结构与残基突变实验较为吻合。BMS-378806与结合态gp120的作用模式类似于CD4上Phe43残基与gp120的结合。模拟结果从原子水平上给出了这类抑制剂的结合模式。(2)以HIV-1跨膜蛋白gp41为靶点的抑制剂设计(a)基于配体的药效团模型构建根据目前已知的抑制跨膜蛋白gp41六螺旋结构形成的小分子结构,用GASP软件构建了基于配体的药效团模型。挑选的模型Model 01由两个疏水特征、两对氢键受体特征和一对氢键给体特征组成,将活性分子与药效团模型叠合发现,分子中的苯环、吡咯环等对应了模型中的疏水特征,羧基、羟基等对应了模型的氢键受体和给体特征。(b)结合模式研究和基于受体的药效团模型构建用分子对接的手段研究已有的小分子抑制剂与gp41五螺旋结构的结合模式,并根据分子对接的结果,进行了基于受体的药效团模型构建,模型包括四个疏水特征,三个氢键受体特征和一个氢键给体特征。(c)药效团模型的验证通过对比基于配体的药效团模型Model 01、小分子的结合模式及基于受体的药效团模型,发现三者之间比较一致,基于配体结构得到的模型Model 01可认为是基于受体的药效团模型的一部分。从ACDSC库中随机挑选部分分子,并加入已知活性抑制剂构成测试数据库,使用该库对Model 01进行筛选能力测试,发现该模型可有效的将活性化合物挑选出来(富集度为5.04)。(d)针对gp41的虚拟筛选设计了综合考虑已知配体信息和受体结构信息的针对gp41的虚拟筛选策略。对包括CMC、ACDSC、NCI、MDDR以及中国中草药数据库(CHDD)在内的总共250万个分子进行了虚拟筛选、毒性预测和象药性分析,挑选出36个小分子化合物,已合成两个化合物,在体外细胞水平实验上表现出一定的抑制HIV-1 IIIB型病毒感染的能力。(e) NB-2分子的结构优化以gp41的五螺旋晶体结构为靶点,使用从头设计方法(LeapFrog)在已知象药性小分子抑制剂NB-2的基础上对其结构进行了优化,并用分子对接程序对新分子进行了结合模式和结合自由能预测。新分子的计算结合自由能普遍低于NB-2分子,化合物的合成正在进行中。(3) CCR5受体拮抗剂的设计(a)可预测活性的药效团模型构建用已知的吡咯烷与正丁烷类CCR5拮抗剂的结构和活性信息构建了一个三维药效团模型。最好的药效团模型(Hypo 1)由两个正离子化特征和三个疏水特征组成,训练集预测的相关系数为0.924,null cost值与total cost值的差异为63.67 bits。用该模型预测了由74个分子组成的测试集活性,结果表明该模型可以提供较好的活性预测结果(R = 0.703)。使用该模型预测了虚拟组合化学库中小分子的活性值,挑选出部分预测活性较好的分子进入到合成阶段。(b) CoMFA和CoMSIA模型的构建选择72个吡咯烷分子作为训练集,39个吡咯烷分子作为测试集,进行了CoMFA(Comparative Molecular Field Analysis)和CoMSIA(Comparative Molecular Similarity Indices Analysis)模型构建。最佳CoMFA模型(CoMFA2)的非交叉验证回归系数r2为0.952,交叉验证回归系数q2为0.637,并可对测试集分子作出较好的预测(相关系数R = 0.785)。最佳CoMSIA模型(CoMSIA2)的非交叉验证回归系数r2为0.958,交叉验证回归系数q2为0.677,对测试集分子作出了较好的预测(相关系数R = 0.806)。同时等势面图提供了立体场、静电场、氢键场和疏水场的可视化图像。本研究可为这类化合物的结构优化提供线索。本文以HIV-1病毒进入细胞过程的关键蛋白gp120、gp41和CCR5为靶点,使用分子模拟的手段对抑制剂与蛋白的结合模式、药效团模型以及定量构效关系进行了一系列的研究。获得了可解释实验结果的抑制剂与蛋白的合理结合模式,可有效挑选活性分子的药效团模型以及具有一定预测能力的定量构效关系模型,设计了一批针对不同靶点的新化合物,部分分子被合成并测定具有抗病毒活性。本文的研究工作可为新型抗艾滋病药物的开发提供有用信息。更多还原

【Abstract】 In 1983, Barre-Sinousi et al. identified the human immunodeficiency virus (HIV-1) as the source of infection of acquired immunodeficiency syndrome (AIDS) for the first time. Since the first patient of AIDS was discovered in June 1981, this fatal disease has taken over 20 million people away. Although the application of current drugs and highly active antiretroviral therapy has achieved great successes, there still exist a number of difficulties, such as the unavailable eradication of the virus, the emergence of drug resistant virus and the strong side effects. Up till now, the medical research on vaccine also failed to break through. Therefore, it is of great urgency to design the novel drugs with safety and effectiveness against the new targets in the life cycle of HIV-1. The Fuzeon (T-20) which came into the market as an anti-HIV-1 drug in 2003 proved the entry process as potential and effective targets. Recently, it also becomes the focus of the medical research for the purpose to discover safe and effective inhibitors dealing with the entry stage of HIV-1 infection. It is well known that the discovery of new drugs is quite a costly and time consuming task. The application of computer-aided drug design (CADD) in the process of drug discovery is expected to save both money and time. In recent years, it is widely used in current drug discovery research. CADD is commonly separated into two general categories: Structure-Based Drug Design and Ligand-Based Drug Design. In this thesis, a series of research work has been done based on the ligand and receptor structures by using the CADD methods to study the entry process of the virus infection, with the aim to design the lead compounds against AIDS. The content of the thesis contain the following major aspects:(1) Study on binding mode between BMS-378806 and HIV-1 envelope protein - gp120BMS-378806 is a newly discovered small molecule that effectively blocks the binding of CD4 with gp120. The binding mode of this type of inhibitors remains unknown. AutoDock 3.0 in conjunction with molecular dynamics (MD) simulation was used to explore the binding mode between BMS-378806 and gp120. Two structures, Mode I and Mode II, with the lowest docking energy were selected as different representative binding modes. The analysis for the MD simulation data of the complex indicates that the binding of BMS-348806 with gp120 in Mode II is more stable. The average structure of Mode II was analyzed and compared with the experimental data. The sensitive of C-4 substitution on the azaindole ring may result from its occurence in the vicinity of Trp427. The Large side chain mutations inside the Phe43 cavity, such as S375W and T257R mutation, make steric hindrance for the binding. Other reisdues in the cavity such as Ser256, Thr257, Asn425, Met426 and VAL430 have considerable interactions with the small molecule. The average structure of the complex is consistent with the experimental data of mutation. The binding mode between BMS-378806 and the bounded state gp120 is similar to the binding of Phe43 in CD4 with gp120. The simulation results give an atomic insight to the possible binding mode of this kind of inhibitors.(2) Inhibitor design targeting to HIV-1 transmembrane protein - gp41(a) Pharmacophore model construction based on ligands’structuresThe software of GASP was used to generate ligand-based pharmacophore models on the basis of the identified structure information of active inhibitors interfering with the six helix bundle formation of gp41. The selected model - Model 01 was comprised of two hydrophobic features, two pairs of hydrogen bond acceptor features and one pair of hydrogen bond donor feature. The mapping of active molecules to the model showed that the hydrophobic features represented the phenyl group or pyrrole group, and the hydrogen bond features represented the hydroxyl group or carboxyl group.(b) Binding modes research and pharmacophore model construction based on receptor structuresThe binding modes of the known inhibitors with the 5-helix gp41 were obtained by using the molecule docking methods. The acceptor-based pharmacophore model was constructed according to the docking results. It is composed of four hydrophobic features, three hydrogen bond acceptor features and one hydrogen bond donor feature.(c) Pharmacophore model validation By comparison of the ligand-based pharmacophore model (Model 01), the binding modes of the inhibitors and the acceptor-based pharmacophore model, they all turned out a sort of consistency. The ligand-based pharmacophore model seemed to be a subset of the acceptor-based pharmacophore model, which is necessary for the activity. The filter property of the model was validated by a test database which was composed by molecules randomly selected from the ACDSC database and the known inhibitors. The results show that the model could filter out the active molecules efficiently with an enrichment factor of 5.04.(d) Virtual screening target to gp41A virtual screening strategy was designed to combine both the ligand and the receptor structure information. About 2,500,000 molecules from CMC, ACDSC, NCI, MDDR and CHDD were screened. The toxicity and drug-like properties were also estimated. Thirty six molecules were selected finally and two of them were synthesized. Both of the compounds showed the ability to inhibit the infection of the HIV-1 IIIB virus in vitro.(e) Structural optimization based on NB-2The de novo drug design method ( LeapFrog ) was used to optimize structure of NB-2 targeting the 5-helix structure of gp41. The docking method was adopted to recalculate the binding modes and the binding free energy of the new compounds. It was found that the binding free energy of new molecules were low than that of NB-2.The synthesis experiments are undertaken on the way.(3) Inhibitor design targeting to CCR5(a) Construction of predicable pharmacophore modelA three dimensional pharmacophore model was developed for a considerable number of pyrrolidine based and butane-based chemokine (C-C motif) receptor 5 (CCR5) antagonists. The most predictive pharmacophore model (Hypo 1), consisting of two positive ionizable points and three hydrophobic groups, had a correlation of 0.924 for the training set, and a cost difference of 63.67 bits between the null cost and the total cost. The model was applied to predict the activity of 74 compounds as a test set with a correlation factor of 0.703. The results indicated that the model was able to provide accurate activity prediction for novel antagonist design. It has been employed to filter a combinatorial database, and the molecules with good predicted activities were selected into the synthesis experiments.(b) Construction of CoMFA and CoMSIA modelA series of 1,3,4-trisubstituted pyrrolidine-based CCR5 receptor antagonists was taken as our target to construct CoMFA (Comparative Molecular Field Analysis) and CoMSIA (Comparative Molecular Similarity Indices Analysis) models. The training set and test set consisting of 72 and 39 selected molecules, respectively. For the best CoMFA model (CoMFA2), the conventional correlation coefficient r2 = 0.952, the cross-validated coeffient q2 = 0.637, and the correlation factor on the test set prediction R = 0.785, while for the best CoMSIA model (CoMSIA2), r2 = 0.958, q2 = 0.677, and R = 0.806. Furthermore the contour map also provides a visual representation for the contributions of steric, electrostatic, hydrogen-bond and hydrophobic fields. The study could provide useful information for structure optimization of this kind of compounds.In this paper, a series of research work, including binding mode prediction, pharmacophore model construction and QSAR (Quantitative structure-activity relationship) analysis has been done against the key proteins (gp120, gp41, CCR5) involved in the entry of HIV-1. A binding mode of BMS-378806 with gp120 which is consistent with the experimental data as well as effective pharmacophore models based on gp41 inhibitors and predictable QSAR models based on CCR5 antagonists were achieved. The new compounds targeting to different proteins were designed and a part of them were synthesized and proved to be active. These works may benefit the development of anti-AIDS drug candidates.更多还原