Toxicity Prediction Of Organophosphates And Molecular Simulation Of Their Interaction Mechanisms With AChE

Organophosphates(OPs)are hazardous chemicals that must be paid high attention to in public safety.The continued illegal abuse of pesticides and nerve agents has caused serious harm to humans and the ecological environment.Rapid and accurate evaluation of the OPs toxicity as well as exploration of their toxicity mechanisms are of great significance for their early risk assessment,scientific and efficient supervision,and rapid disposal to address corresponding safety threats.With the accumulation of toxicity data,the increase in computing power,development of intelligent algorithms,and improvement of the molecular simulation technology have facilitated the wide application of computer modeling in toxicity prediction and molecular simulation to elucidate toxicity mechanisms.This effectively compensates for the shortcomings of experimental toxicity assessments such as small flux,long cycle,and the difficulty in determining toxicity mechanisms solely by relying on experimental techniques.However,there are still some deficiencies and unaddressed problems in terms of toxicity prediction and mechanisms associated with OPs.Existing toxicity prediction models comprise insignificant data on OPs,single structures,and model validation,which limit their practical toxicity prediction ability and application scope.In the simulation of the interaction mechanisms between OPs and acetylcholinesterase(ACh E),the relationships between phosphorylation and the leaving groups,the aging reaction and the aging groups are still unclear and need to be further explored.Thus,focusing on the above problems,toxicity prediction of OPs and their interaction mechanisms with ACh E were carried out in this study based on the quantitative structure-toxicity relationship(QSTR),quantum mechanics/molecular mechanics(QM/MM),and molecular dynamics(MD)simulation techniques.The research results are as follows:(1)Based on the data of OPs collected from public and commercial databases as well as professional literatures,an OP database system(OPPTox)with data browsing,query,management,and maintenance functions has been independently developed.The database system contains 621 OPs,5,870 toxicity data,9,836 basic physicochemical property data,and 37 toxic species,which lays a solid foundation for the subsequent development of toxicity prediction models and mechanism simulation research.Furthermore,50 quantum chemical(QC)descriptors,which can reflect the thermodynamic properties,reactivity,polarity,key atomic charges,and chemical bond orders of OPs,were designed under the framework of density functional theory and combined with the characteristics of the chemical structure of OPs and their interactions with ACh E.The Quantum software with batch automatic calculation and extraction of the designed QC descriptors was also independently developed,which effectively solved the problem of using the QC method to characterize the structure of OPs.Nine QC and 206two-dimensional(2D)molecular descriptors were added using the Multiwfn software and molecular operating environment(MOE)software,respectively.The extracted descriptor set provided sufficient structural characteristic parameters for the subsequent construction of toxicity prediction models for OPs.(2)Strictly following the modeling principles of the Organization for Economic Cooperation and Development(OECD),ten QSTR models were successfully developed to predict the acute toxicity of OPs via multiple administration routes in rats and mice along with their inhibitory activity against human acetylcholinesterase(h ACh E).These robust(Q²_cv=0.524-0.854)QSTR models were designed by combining relatively comprehensive OPs acute toxicity dataset,the QC and 2D molecular descriptor set,and different machine learning algorithms.They all exhibit good external prediction abilities(Q²_{ex t}=0.604-0.877)and can provide important theoretical and technical support for early risk assessment as well as scientific and efficient supervision of OPs.Based on the performance of the QSTR models,similarity of characteristic parameters,and correlation analysis of the shared OPs experimental data in the dataset,we found that administration routes have a greater effect on the acute toxicity of OPs than that of the type of species(rats and mice).The two high-quality(Q²_cv(29)0.65,Q²_{ex t}(29)0.69)consensus QSTR models for the oral acute toxicity of OPs in rats and mice and their inhibitory activity against h ACh E led to the discussion of the problem of in vitro molecular toxicity and in vivo animal toxicity of OPs for the first time.It was found that the oral acute toxicity of OPs could not be directly assessed simply using experimental data of the in vitro enzyme inhibitory activity.These new findings,which are derived from mammalian species,administration routes,and molecular level enzyme inhibitory activities,not only contribute to further our understanding of the acute toxicity of OPs,but also provide important references for the toxicity prediction of other hazardous chemicals.In addition,based on the open-source Konstanz information miner(KNIME)platform,we built a QSTR modeling workflow containing six machine learning algorithms,which greatly facilitates the construction and improvement of the QSTR models as well as the acute toxicity prediction of unknown OPs.(3)The mechanisms of phosphorylation and the aging reaction between anatoxin-a(s),soman,cyclosarin,tabun,VR,and h ACh E were simulated based on the constructed QM/MM and MD computational simulation workflow for OP and h ACh E interaction mechanisms.For the phosphorylation reaction,it was theoretically found for the first time that anatoxin-a(s)with the S configuration has a higher phosphorylation reactivity(reaction energy barrier is 1.3 kcal·mol^-1 lower than that of the R configuration).The protonation state of the nitrogen atom in the tertiary amine has little effect on the reaction process.Furthermore,the structure of the adduct after phosphorylation is in a post-aging state,which is difficult to be reactivated;this may be the primary reason for the rapid killing and lethality of anatoxin-a(s).In addition,the influence of the leaving ability of the leaving group on the phosphorylation mechanism is also explained;i.e.,the OPs containing easy leaving groups tend to adopt a concerted reaction mechanism,while the OPs containing difficult leaving groups are more likely to adopt a stepwise reaction mechanism.For the aging reaction,the structure of the aging group has little influence on the aging mechanism.Except for anatoxin-a(s),the aging processes of the other four OPs are concerted mechanisms mediated by Glu202 with the participation of water molecules.The aging barriers calculated by QM/MM are relatively similar to that observed in the conversion obtained by the experiment(the errors are approximately 2 kcal·mol^-1),and the magnitude trend of them are also consistent.Therefore,the aging barriers calculated by QM/MM are expected to be used for the virtual evaluation of the aging rates for unknown OPs.The discovery of novel toxicity mechanisms of OPs at the molecular level not only provides important reference values for the analysis of the toxicity mechanisms of OPs with new structures,but also offers a more targeted theoretical guidance for the research and development of novel antidotes.