Modeling photoinduced toxicity of PAHs based on DFT-calculated descriptors

Quantitative structure-activity relationships (QSARs) were established for photoinduced toxicity of polycyclic aromatic hydrocarbons (PAHs) to two aquatic species. Partial least squares (PLS) regression and molecular structural parameters calculated by density functional theory (DFT) were employed for model development. Two QSAR models were established and their high R-2 and Q(CUM)(2) values indicated their good goodness-of-fit, robustness and internal predictive power. The descriptors that describe the partition behavior, light absorbance, and generation of reactive free radicals were found to be successful in modeling the photoinduced toxicity. The average molecular polarizability (alpha), energy gap (E-GAP) between the energy of the lowest unoccupied molecular orbital and the highest occupied molecular orbital, lowest triplet excitation energy (E-T1) and vertical electron affinity at the lowest excited triplet (VEA(T1)) were the main molecular structural factors. Polarizability which determines the partition of PAHs between water and lipid governs the photoinduced toxicity of selected PAHs. Moreover, the photoinduced toxicity increased with the decreasing of E-GAP probably due to better spectral overlap. The parameter, VEA(T1) that characterizes the ability of PAH anion radical (PAH(center dot-)) generation from excited triplet state PAH ((3)PAH(center dot)), is also related with the photoinduced toxicity. This investigation will make us gain more insight into the photoinduced toxicity mechanism and assess the applicability of various DFT-based descriptors to toxicological QSARs. (C) 2009 Elsevier Ltd. All rights reserved.