eEF2K蛋白同源模建及其抑制剂小分子的虚拟筛选研究

Study on Homology Modeling of eEF2K Protein and Virtual Screening of Its Inhibitors Molecules

目的:筛选潜在的真核生物延伸因子2激酶(eEF2K)抑制剂小分子,为eEF2K抑制剂的设计和研发提供参考。方法:采用同源模建技术构建eEF2K蛋白晶体结构模型,并进行Loop优化和分子动力学优化,借助SAVES在线服务器从Verify3D、EERAT和拉氏图等3个方面对上述模型进行评估。收集55个eEF2K抑制剂小分子,使用InsightⅡ软件以其中的28个(编号为奇数,设为训练集)为基础构建具有活性预测能力的Hypogen药效团模型,以另外27个(编号为偶数,设为测试集)进行验证,通过拟合活性[即半数抑制浓度的负对数(p IC50)]预测值与真实值并借助Ligand profiler热图筛选最优药效团模型。结合上述药效团模型和Lipinski五规则、分子对接方法进行eEF2K抑制剂小分子的虚拟筛选。结果与结论:所建eEF2K蛋白晶体结构模型的整体质量因素得分为93.697,其中83.33%的氨基酸Verify3D得分≥0.2,且位于不允许区的氨基酸占氨基酸总数的1.7%,其氨基酸构象及骨架结构合理,模型可靠性高。共构建了9个具有活性预测功能的Hypogen药效团模型(02~10号),其中03号药效团模型包含2个氢键受体和2个共轭芳香环,可更好地区分活性及非活性分子,其pIC50预测值与真实值拟合最好(相关系数为0.665 3),具有较好的预测能力和较高的可靠性。通过虚拟筛选最终获得9个潜在的eEF2K抑制剂小分子(pIC50预测值为1.074~1.185,分子与蛋白相互作用的Dcoking-score得分为-9.730^-7.467),其中Pro268、Asp267、Gln171、Phe121、Glu212可能是e EF2K抑制剂与靶点蛋白相互作用的关键氨基酸,作用方式包括氢键、盐桥、疏水等。上述分子有望成为e EF2K抑制剂研发的先导化合物。

OBJECTIVE:To screen potential eEF2 K inhibitor molecules,and to provide reference for the design and R&D of eEF2 K inhibitor. METHODS:The eEF2 K crystal structure model was constructed by homology modeling technique. The model was optimized by Loop optimization and molecular dynamics. With the help of SAVES online server,the above models were evaluated from three aspects such as Verify3 D, EERAT and Laplace diagram. Totally 55 eEF2 K inhibitor molecules were collected. Hypogen pharmacophore model with activity prediction ability was constructed based on 28 of them(odd number,as training set)by Insight Ⅱ software and validated by other 27(even number,as test set). The optimal pharmacophore model was screened by fitting the predicted and experimental values of activity [i.e. negative logarithm of half inhibitory concentration(pIC50)]and using Ligand profiler thermogram. The virtual screening of small molecules of eEF2 K inhibitors was carried out by combining the above pharmacophore model,Lipinski’s five rules and molecular docking method. RESULTS & CONCLUSIONS:The overall quality factor score of the crystal structure model of eEF2 K protein was 93.697. Among them,83.33% of the amino acid Verify3 D score was more than or equal to 0.2,and 1.7% of the total amino acids were located in the non-permissible region. The amino acid conformation and skeleton structure of the model were reasonable and the reliability of the model was high. Totally 9 Hypogen pharmacophore models(No. 02-10)with active predictive function were constructed,among which No. 03 pharmacophore model included 2 hydrogen bond receptors and 2 conjugated aromatic rings,which could better distinguish active and inactive molecules.The predicted value of pIC50 fitted the experimental value best(the correlation coefficient was 0.665 3),and it had good predictive ability and high reliability. Finally,9 potential eEF2 K inhibitor molecules were obtained through virtual screening(pIC50 ranged from 1.074 to 1.185,and Dcoking-score of protein-molecule interaction ranged from-9.730 to-7.467). Pro268,Asp267,Gln171,Phe121 and Glu212 may be the key amino acids for the interaction between eEF2 K inhibitors and target proteins,including hydrogen bonds,salt bridges and hydrophobicity. These 9 molecules are expected to be the lead compounds for the development of e EF2 K inhibitors.