Juhl P Benjamin, Trodler Peter, Tyagi Sadhna, Pleiss Jürgen
Institute of Technical Biochemistry, University of Stuttgart, Allmandring 31, 70569 Stuttgart, Germany.
BMC Struct Biol. 2009 Jun 3;9:39. doi: 10.1186/1472-6807-9-39.
Previously, ways to adapt docking programs that were developed for modelling inhibitor-receptor interaction have been explored. Two main issues were discussed. First, when trying to model catalysis a reaction intermediate of the substrate is expected to provide more valid information than the ground state of the substrate. Second, the incorporation of protein flexibility is essential for reliable predictions.
Here we present a predictive and robust method to model substrate specificity and enantioselectivity of lipases and esterases that uses reaction intermediates and incorporates protein flexibility. Substrate-imprinted docking starts with covalent docking of reaction intermediates, followed by geometry optimisation of the resulting enzyme-substrate complex. After a second round of docking the same substrate into the geometry-optimised structures, productive poses are identified by geometric filter criteria and ranked by their docking scores. Substrate-imprinted docking was applied in order to model (i) enantioselectivity of Candida antarctica lipase B and a W104A mutant, (ii) enantioselectivity and substrate specificity of Candida rugosa lipase and Burkholderia cepacia lipase, and (iii) substrate specificity of an acetyl- and a butyrylcholine esterase toward the substrates acetyl- and butyrylcholine.
The experimentally observed differences in selectivity and specificity of the enzymes were reproduced with an accuracy of 81%. The method was robust toward small differences in initial structures (different crystallisation conditions or a co-crystallised ligand), although large displacements of catalytic residues often resulted in substrate poses that did not pass the geometric filter criteria.
以前，人们已经探索了调整用于模拟抑制剂 - 受体相互作用的对接程序的方法。讨论了两个主要问题。第一，在尝试模拟催化作用时，底物的反应中间体预计比底物的基态能提供更有效的信息。第二，纳入蛋白质柔性对于可靠的预测至关重要。
在此，我们提出了一种预测性且稳健的方法，用于模拟脂肪酶和酯酶的底物特异性和对映选择性，该方法使用反应中间体并纳入蛋白质柔性。底物印记对接从反应中间体的共价对接开始，随后对所得的酶 - 底物复合物进行几何优化。在将相同底物第二轮对接至几何优化结构后，通过几何过滤标准识别出有效构象，并根据其对接分数进行排序。应用底物印记对接来模拟：（i）南极假丝酵母脂肪酶B及其W104A突变体的对映选择性；（ii）皱落假丝酵母脂肪酶和洋葱伯克霍尔德菌脂肪酶的对映选择性和底物特异性；以及（iii）乙酰胆碱酯酶和丁酰胆碱酯酶对底物乙酰胆碱和丁酰胆碱的底物特异性。
实验观察到的酶的选择性和特异性差异的重现准确率为81%。该方法对初始结构的微小差异（不同的结晶条件或共结晶配体）具有稳健性，尽管催化残基的大幅位移常常导致底物构象无法通过几何过滤标准。
