Bactericide Screening And Validation Potential Target Of Erwinia

Erwinia carotovora subsp. carotovora is a wide-range host plant pathogenic bacterium that has caused enormous economic lossesorldwide. At present, the main measures to treat and prevent E. carotovora are breed improvement and pesticides. Considering the long period of breed improvement and single target of pesticides, developing new pesticides and discovering new drug targets for E. carotovora are urgent. This study attempts use computer-aided drug design to screen antibacterial lead compounds and verify whether alkaline phosphatase is a potential drug target of E. carotovora. First, support vector machine and artificial neural network were used to construct classifiers to identify potential antimicrobial compounds from compound libraries for rapid classification. The potential antibacterial compounds obtained through model classification were then virtually screened. Finally, the screened compounds underwent antibacterial experiment in vitro to determine the inhibition activity against their targets.First, a fungicides identification model was bulit using the support vector machine and artificial neural network. The train and internal test sets were constructed from4,423antibacterial compounds from MDDR database and4,423non-antibacterial compounds from the ACD database. Approximately314compounds from the DrugBank database were selected as positive and negative samples of the external test set to evaluate the generalization ability of the models. The support vector machine and artificial neural network were used to build potential antibacterial compound classifier using18molecular descriptors as feature vectors. The accuracies of the train test, internal test and external test sets built by the support vector machine were92.28%,89.09%, and90.12%, respectively. The accuracies of the train test, internal test set and external test sets built the artificial neural network were91.43%,88.42%, and91.40%, respectively. The two models were used to classify the Specs company of approximately200,000compounds. Approximately75,000potential bactericides were obtained.Alkaline phosphatase, cell glycosyltransferase, riboflavin synthase and pantothenate kinase were then used to virtually screen the75,000potential bactericides. The enzymatic3D structures were obtained through homology modeling, and semi-flexible molecular docking was used to virtual screen the classified Specs compounds. The top100compounds were selected as the virtual screen result. The last three enzymes acting as potential targets in bacteria were not reported, so we selected the compounds screened by alkaline phosphatase as target for experimental validation. Finally, the screened compounds by alkaline phosphatase underwent antibacterial experiment, to determine the inhibition activity against alkaline phosphatase in vitro. Three antibacterial compounds were obtained in the antimicrobial experiment (i.e., AG-664/15584557, AA-768/34978042, and AE-641/30177051). Their half inhibitory concentrations were65.88±1.65μg/mL,63.04±1.75μg/mL and42.87±1.88μg/mL. The enzymology experiment indicated that the three bactericides had a certain effect on the alkaline phosphatase. These results show that alkaline phosphatase is a potential target for Erwinia carotovora subsp. carotovora.