Activity Prediction Modeling Of Carbamates Pesticides And Molecular Simulation Of Their Interaction Mechanism With Acetylcholinesterase

Carbamates(CMs)pesticides are the most commonly used pesticides in agricultural production,and their continuous abuse has posed a serious threat to the survival and safety of organisms in the entire ecosystem.Therefore,it is of great significance to realize the rapid assessment of the biological activity of CMs pesticides and perform in-depth study of their toxicity mechanism,which plays a crucial role in their potential risk assessment,scientific regulatory decision-making,rational molecular design,and effective disposal of poisoning events.With the continuous expansion of the existing compound warehouse,the accumulation of experimental data and the rapid improvement of software and hardware performance,computer simulation has become the mainstream technology,providing a scientific,efficient,stable and reliable technical approach for compound properties prediction and chemical reaction process simulation.However,there are still many problems in the activity prediction and mechanism of action of carbamate pesticides.Mainly as follows:(1)Most of the existing activity prediction models were developed on the basis of small amount of data,linear fitting algorithms,and insufficient verification,which limited their application scope and prediction abilities;(2)Most of their modeling endpoints were biological activities on insects or animals,lacking research on the potential risks that CMs pesticides might cause to the humans;(3)There are few theoretical studies on the mechanism of action between CMs pesticides and their main target,acetylcholinesterase(ACh E).The general laws of the mechanism of action of pesticides still need to be further explored.Focusing on the above-mentioned problems,this paper carried out the activity prediction of CMs pesticides as well as the mechanism simulation study of their reaction process with ACh E,via collection of compound information data and application of quantitative structure-activity relationship(QSAR)and hybrid quantum mechanics/molecular mechanics molecular dynamics simulation technology(QM/MMMD).The main results are as follows:(1)Based on the CMs information data collected from the open-source databases Pubchem and ChEMBL and related scientific literature,and applying the Java Web framework and My SQL database management technology,a carbamate pesticide information database system(CMPBank V1.0)with data retrieval,export and maintenance management was independently developed.It includes the basic information,physicochemical properties,cholinesterase inhibitory activity,and multispecies and multi-pathway acute toxicity of existing CMs pesticides and other CMs compounds.At present,the CMPBank V1.0 database contains 906 biological activity data and more than 1000 physical and chemical properties data of 503 CMs molecules.The subsequent research on its biological activity prediction model,and also provided sufficient data reference for researchers engaged in CMs research and government agencies dedicated to the supervision of such dangerous compounds.(2)The modeling process of QSAR regression model is constructed by applying the KNIME platform and three types of algorithms: random forest,extreme gradient boosting and support vector machine.Based on the information data originated from CMPBank V1.0 and different descriptors,a series of regression prediction models for the inhibitory activity of CMs pesticides on insects’ ACh E were developed,and then three high-quality models were selected to construct a consistent model.The obtained consistency model shows satisfactory both internal and external prediction performance,which can provide reliable activity prediction results for evaluating CMs pesticides with unknown insecticidal activity.In addition,by comparing the performance of QSAR models obtained by combining different machine learning algorithms with different kinds of descriptors,it is found that the descriptors from MOE can achieve similar internal and external prediction performance to the descriptors from Pa DEL with a smaller number of descriptors,while the models constructed by quantum chemical descriptors is poor.In addition,there is no significant difference in the performance of the models trained by the three nonlinear learning algorithms.This suggests to some extent that the use of quantum chemical descriptors alone is not sufficient to reflect the relationship between the molecular structure of CMs pesticides and their ACh E inhibitory activity,but the model performance is not sensitive to the three non-machine learning algorithms chosen.(3)Using the KNIME platform and the extreme gradient boosting algorithm,the modeling process of the QSAR classification model of the inhibitory activity of CMs pesticides on human ACh E was constructed.Based on this process,combined with three different descriptors,and using 1000 n M and 100 n M as activity thresholds,a series of global QSAR binary classification models were developed for a total of 3645 compounds including CMs pesticide molecules and other ACh E inhibitors.Among them,the internal and external prediction AUC of the classification model with a threshold of 1000 n M can reach more than 0.9,and the internal and external prediction AUC of the classification model with a threshold of 100 n M can also reach more than 0.8,indicating that the developed classification model has relatively good prediction performance.It can be used to assess the potential risks posed by CMs pesticides to humans.In addition,the three developed classification models were used to screen CMs molecules of unknown activity in the Chem Div datasets,yielding 625,1378 and 1215 potentially highly active molecules with predicted activities within 100 n M,respectively.These compounds can be used as early warning information to identify high-risk molecules with high human toxicity in novel candidate pesticides in advance.(4)Based on the QM/MM MD molecular simulation technology,the reaction process between three typical CMs pesticide molecules and human ACh E was simulated respectively.From the stability of the enzyme-substrate complex structure,it is found that the positive charge on the side chain of the substrate molecule can stabilize the complex through long-range action,which increases the affinity of the enzyme to the molecule and induces a conformation favorable for the reaction.In addition,by simulating the mechanism of enzyme intoxication and hydrolytic reactivation,all three substrate molecules were found to be more reactive than the natural substrate ACh.The mechanism study of the hydrolytic reactivation process showed that the rate-determining step of the reaction between carbamate molecules and ACh E was the second step of deamination,and the activation energy was 8.7 kcal/mol.This study explains the effects of the chemical structure of the substrate molecule and related charges on the affinity and activity of the enzyme,and provides more theoretical basis for the rational transformation and design of more precise targets of CMs pesticides.This paper focuses on the related issues of carbamate pesticides,and studies the rapid assessment of the biological activity and molecular mechanism of carbamate pesticides from four aspects: data set construction,QSAR model construction,early warning compound screening,and toxicity mechanism simulation.The research of this paper has laid a theoretical foundation and provided practical application techniques for the potential risk assessment,scientific regulatory decision,structural modification and poisoning treatment of CMs pesticides.