Main content area

QSAR modelling for predicting the toxic effects of traditional and derived biomass solvents on a Danio rerio biomodel

Zuriaga, Estefanía, Giner, Beatriz, Valero, Marta S., Gómez, Manuel, García, Cristina B., Lomba, Laura
Chemosphere 2019 v.227 pp. 480-488
Crustacea, Danio rerio, abnormal development, acute toxicity, alcohols, algae, bacteria, biomass, computer software, ecotoxicology, edema, energy, environmental impact, furfural, heart, heat, invertebrates, lactic acid, lethal concentration 50, model validation, models, prediction, quantitative structure-activity relationships, regression analysis, solvents, sublethal effects, toxic substances
The increasing interest in the development of ecofriendly solvents has led to the synthesis of benign alternative chemicals with minimized environmental impacts. These kinds of chemicals are known as Green solvents. In this work, we selected three families of solvents (furfural, lactate and levulinate families) derived from biomass that are structurally related. Most of the previous ecotoxicological studies of these solvents have focused on invertebrate models such as bacteria, algae and crustaceans. To complete this information, in this work, the acute toxicity of these solvents was studied in Danio rerio (D. rerio). Sublethal and lethal effects were also observed, and the LC50 was obtained. The LC50 values ranged from 13.21 to 12073 mg L−1, with furfural being the most toxic compound and tetrahydrofurfuryl alcohol the least toxic. Furthermore, the results indicated that a frequent sublethal effect was heart oedema or malformation, even in some cases at concentrations lower than the LC50.A QSAR analysis was also performed to model the toxicological effect towards D. rerio for the studied solvents obtained from biomass and traditional solvents. A total of 15 molecular descriptors of the solvents were obtained using Gaussian 03 software. Finally, we also used the physicochemical property Log P, calculated from ACD/LogP, for QSAR modelling. Multivariable regression analysis showed that the minimum set of independent variables that leads to the best regression is Log P, the energy of the lowest unoccupied molecular orbital (ELUMO) and the heat capacity (CV). The proposed model was validated using several internal and external methods.