| Hydrolysis is one of the important degradation pathways of organic compounds, and the hydrolysis rate constant (KH) is the basic parameter for ecological risk assessment, which can be determined by experiment. However, It is time and labor consuming, as well as requiring equipments and standard substances for experimental determination of KH, so it is necessary to develop computational prediction methods based on experimental data. Quantitative structure-activity relationships (QSAR), which correlates the physicochemical properties, environmental behavior parameters and toxicological parameters of compounds with their molecular structures, is regarded as one of promising alternative techniques. Antibiotics are frequently detected in aquatic environment, and they have cause worldwide concern for their environmental behaviors and ecological risks. However, there is a lack of hydrolysis information for many antibiotics currently.According to the OECD guideline, QSAR models were developed for predicting the hydrolysis rate of esters, amides and antibiotics by using multiple linear regression (MLR) algorithm. According to the guideline of hydrolysis test suggested by United States Environmental Protection Agency (OPPTS 835.2120 Hydrolysis), the hydrolysis rates of 7 antibiotics were determined, and the hydrolysis transformation products of 3 antibiotics were characterized. The main findings were as follows:(1) The prediction model of esters had good fitness, robustness and predictive ability (R2adj =0.898,Q2LOO=10.851, Q2ext= 0.651), based on 70 base-catalyzed hydrolysis rate constants from databases and literatures. The prediction model of amides had fairly good quality of fitness, robustness and predictive ability (R2adj=0.681, Q2LOO=0.647), based on 94 hydrolysis rate constants from databases and literatures.(2) The prediction model of antibiotics had good fitness, robustness (R22adj=0.885, RMSE =0.139,Q2LOO=0.848), based on 20 hydrolysis rate constants from literatures. It was indicated from the molecular descriptors that molecular size, shape, symmetry and the distribution of atoms had important contributions on the hydrolysis of antibiotics. The prediction model of antibiotics had good predictive ability with the validation using the experimental hydrolysis data of amoxicillin and doxycycline hyclate.(3) The hydrolysis rate constants of seven antibiotics were determined. Among them, amoxicillin, doxycycline hyclate, chloramphenicol and roxithromycin were easy to be hydrolyzed, and lincomycin, ciprofloxacin hydrochloride, roxarsone were not easy to be hydrolyzed. The hydrolysis transformation products of amoxicillin, doxycycline and penicillin G were characterized using high resolution mass spectrometry, and possible hydrolysis pathways were analyzed. For amoxicillin, the hydrolysis reaction started with the breaking of C-N bond of beta-lactam, ring cleavage, and then decarboxylation and dehydration. |