【作者】 徐亮；

【导师】 王希诚； 赵伟杰；

【作者基本信息】 大连理工大学 ， 药物工程， 2009， 博士

【摘要】 苯酚衍生物是在自然界分布广泛的一类环境污染物,通过不同的作用模式对生物体产生毒害作用。属于某种作用模式的苯酚衍生物虽然含有不同的取代基,但在结构或物理化学性质上具有一定的相似性。定性的分类方法和定量的结构-活性相关性分析都试图寻找这些相似性,然后构建分类器或函数关系对未知作用模式的苯酚衍生物作出预测。上述两种方法都要求用于描述结构特征的分子描述符能够反映深层次的作用机理,数目尽可能的少,并且易于解释。近几年发展的基于统计学习理论的支持向量机分类算法受到广泛关注并已成功用于解决各种分类问题,但能否对少数已知作用模式（即样本数目少）的苯酚衍生物进行分类值得研究,如何验证和解释分类的结果也是值得探索的一个问题。苯酚衍生物的生物降解是降低其对环境污染和生物体毒性的一个有效途径。细菌对芳烃的好氧降解途径中,双加氧酶催化两个关键反应:芳环双羟基化和开环。所有的芳香族化合物都是先降解为邻苯二酚,然后邻苯二酚通过邻位或间位双加氧酶的作用裂解为粘康酸半醛或粘康酸,从而使苯环开环断裂。大量的实验和理论研究都将注意力集中在活性位点附近O₂分子如何活化,邻苯二酚如何开环裂解,以及如何捕获反应的中间产物等方面,却很少关注O₂是如何到达反应中心这一问题。研究表明酶催化的反应速率不仅取决于活性位点的反应速率还取决于气体进入活性中心的速率,因此寻找酶结构中的气体分子通道逐渐成为研究酶催化活性的一个热点问题,有希望成为调控酶反应速率进而发现新的反应中间体的一种实验方法。通常酶的活性中心都深埋于蛋白质内部,通过分子动力学模拟寻找酶结构中的气体分子通道不仅能够在原子水平上观察气体进入蛋白质的动态过程,而且能够发现控制气体进入活性位点的关键氨基酸残基,为相关的突变实验提供理论指导。本文结合支持向量分类机和分子动力学对第一个问题从宏观和微观两个尺度进行的研究;运用不同的分子动力学模拟方法寻找外二元醇双加氧酶2,3-HPCD中的O₂通道,揭示了O₂通道与酶催化反应速率之间的关系,论文工作包含以下三个部分:1.利用支持向量机对含155个极性麻醉剂和19个非偶联剂的数据集用7个分子描述符进行分类测试,结果表明用于描述疏水和氢键作用的分子描述符对上述两类苯酚衍生物的分类正确率达99%以上。以典型的极性麻醉剂水杨酸（SAL）和非偶联剂五氯酚（PCP）为例,利用分子动力学模拟研究了它们在磷脂双层POPC中的平衡特性,结果显示由于疏水作用,平衡时SAL和PCP分子不能分布在空间相对宽松的POPC内部,而是分布在磷脂双层头部,靠近水相的区域。SAL能够和油酰基碳链上的羰基氧形成稳定的氢键,不仅如此,模拟还发现SAL能够形成分子内氢键;PCP则主要和水形成稳定的氢键。分子动力学模拟一方面揭示了分子描述符隐含的作用机理,另一方面也纠正了以前定量结构.活性研究中认为SAL是和带负电荷的磷酸根结合的观点。2.外二元醇双加氧酶2,3-HPCD催化降解4.硝基邻苯二酚（或4-羧基邻苯二酚）,但其四个同源亚基却具有不同的催化活性。这部分工作利用局部加强取样（LES）分子动力学模拟寻找每个亚基中的O₂通道,发现每个亚基中只含有一条供O₂在蛋白质内部扩散的通道。虽然构成O₂通道周围的氨基酸基本相同,但O₂和这些氨基酸的碰撞频率却不一样,揭示每个通道具有不同的O₂亲和性。通过计算O₂和这些残基的相互作用能进一步说明亚基A最适合O₂的扩散,与实验中在亚基A中发现催化反应的最终产物而在其它亚基中捕获到反应中间体的结论一致,表明O₂通道和亚基的催化速率有一定的相关性。分析残基的运动特征还揭示了空隙的呼吸运动是调控O₂进出蛋白质的主要因素。3.不同于LES模拟,本部分的分子动力学模拟不包含O₂,称为隐式配体取样（ILS）。先对包含2,3-HPCD的体系进行10纳秒常规分子动力学模拟,然后对每一帧轨迹文件计算与O₂的相互作用能,得到反映整个体系与O₂相互作用的自由能曲面。连接势能曲面上从水相至活性位点能量最低的区域即为可能的O₂通道。通过比较每个O₂通道的几何和能量特征,发现亚基A中的O₂通道能量最低,与LES结论一致。延长模拟至20纳秒,继续观察空隙的呼吸运动并与LES结果进行比较,发现O₂对蛋白质结构影响很小,O₂在每个亚基中的运动按照特定的通道进行,空隙呼吸运动的频率差异可能是导致O₂在不同亚基中扩散快慢的主要原因。本论文的研究工作为研究苯酚衍生物与生物大分子,如生物膜和蛋白质的相互作用提供了启发。运用分子动力学模拟解释分类、定量构效关系的结果为揭示苯酚衍生物的作用模式提供了新的思路。运用LES和ILS两种计算方法首次发现O₂通道对亚基催化性能的影响,为相关实验的研究提供了新的方向。更多还原

【Abstract】 Phenols and their derivatives are ubiquitous environmental contaminants and toxic to many organisms by interfering energy transduction in cells. With different substitutions, phenols may exert various biological activities. Besides, one kind of toxicity may be caused by the superposition of effects from different mechanisms. Both classification （qualitative） and QSAR （quantitative） approaches have currently been applied to predict the toxicity of chemicals. One of fundamental issues is how to select and interpret the molecular descriptors used in both methods. Extensive applications of support vector machines （SVM） in classification and QSAR analysis have made it one of popular machine-learning methods in recent years. However, the number of phenols with known mode of action is small, whether SVM is applicable to such problem is worth investigating, as well as how to validate and interpret the classification results.The biodegradation of phenols and their derivatives is one of the important ways to reduce their contamination and toxicity. Dioxygenases play an important role in the biodegradation of catechol and its derivatives by catalyzing the cleavage of aromatic rings. Generally, the intradiol dioxygenases require Fe³⁺ to cleave C-C bond between the phenolic hydroxy groups to produce cis, cis-muconic acid, while the extradiol dioxygenases use Fe²⁺ (or Mn²⁺) as a cofactor to cleave the C-C bond adjacent to the phenolic hydroxy groups to yield 2-hydroxymuconaldehyde. Both experimental and theoretical investigations have been focused on the detection of intermediates in the reaction cycle in order to develop a general chemical mechanism of O₂ activation and insertion. However, little is known about the mechanism of how O₂ reaches the reaction sites of the related enzymes, which raises the question whether the rate of catalysis is limited by O₂ access to the active site. The precise, atomic-resolution pathways for O₂ migration in the protein, along with predicted relative significant parts of the pathways, should help to rationalize the selectivity of specific intermediate at the active site of each subunit and greatly facilitate the selection of specific site mutations for such studies.In this paper, the combination of SVM and molecular dynamics simulations enables us to investigate the classification of phenols and their derivatives from different scales. Two approximation methods, locally enhanced sampling and implicit ligand sampling, were applied to find the O₂ pathways in the extrodioxygenase of 2,3-HPCD. These works are outlined as follows. 1. In this work, we first employed SVM on a dataset containing 155 polar narcotics and 19 uncouplers to filter the predictive hydrophobic and hydrogen bonding descriptors. The overall classifaction rate was above 99%. Molecular dynamics （MD） simulations were then conducted to investigate the behavior of salicylate （SAL） and pentachlorophenol （PCP） molecules in the context of a palmitoyl-oleoyl-phosphatidylcholine （POPC） lipid bilayer. The results demonstrated that their equilibrium properties in the lipid bilayer were closely associated with hydrophobic and hydrogen bonding descriptors. The preferred position of SAL in the POPC bilayer lies in the lipid headgroup, while PCP resides in the region between carbonyl groups and water phase （lipid-water interface）. SAL could form stable hydrogen bonds with carbonyl oxygen atom of oleoyl chain in POPC, as well as intermolecular hydrogen bonds. PCP acts as hydrogen acceptor and establishes hydrogen bonds mainly with water. The observations from molecular dynamics simulations facilitated to elucidate the mechanism of polar narcotics and uncouplers.2. The experimental determination of the structure of Fe²⁺-containing homoprotocatechuate 2,3-dioxygenase （2,3-HPCD） with X-ray crystallography showed that three different intermediates reside in different subunits of a single homotetrameric enzyme molecule. In this paper, two locally enhanced sampling molecular dynamics simulations were performed to determine the potential O₂ pathways inside a recently solved X-ray structure of homoprotocatechuate 2,3-dioxygenase. It is found that nominally identical subunits of the single homotetrameric structure contain distinct O₂ affinity diffusion pathways, which partly correlates with the observation of the simultaneous presence of three different reaction intermediates in four independent active sites. Residues that are critical for O₂ diffusion are also examined and discussed. In particular, we find that the breathing motion of internal cavity defined by these residues results in O₂ migration process.3. Based on the trajectory of a 10-ns molecular dynamics simulation, implicit ligand sampling was applied to calculate the 3D free energy map for O₂ inside the protein. The energetically optimal routes for O₂ diffusing were identified for each subunit of the homotetrameric protein structure. The O₂ tunnels formed due to thermal fluctuations were also characterized by connecting elongated cavities inside the protein. Superimposing the favorable O₂ tunnels onto the free energy map, both energetically and geometrically preferred O₂ pathways were determined, as well as the amino acids that may be critical for O₂ passage along these paths. Our results demonstrate that identical subunits possess quite distinct O₂ tunnels. The order of O₂ affinity of these tunnels is generally consistent with the order of catalytic rate of each subunit. As a consequence, the subunit containing the highest O₂ affinity pathway has the highest probability for finding product of reaction. Compared with results of LES, we find that it is the frequency of the breath motion of the cavity that limits O₂ acess to the actie site of each monomer. In contrast to randomized behavior, O₂ diffusion in the four subunits is clearly limited to specific regions located within the conserved active site domain.In summary, the combination of SVM and MD simulations provides a new strategy for elucidating the mechanism of toxic action underlying the corresponding molecular descriptors. The indentifation of O₂ pathways in 2,3-HPCD will be valuable to engineer non-heme iron dioxygenases in order to find intermediates by charactering O₂ diffusion routes.更多还原