Quantitative structure–activity study of fluorides′ toxicity

• Different methods are performed.
• B3LYP/6-31G* method is the best method, the QSAR model has good stability and prediction ability.
• Toxic mechanism is electronic reaction.

The molecular geometries and electronic structures of 14 fluorides are optimized using density functional theory (DFT) B3LYP/6-31G*, B3P86/6-31G* and Møller–Plesset perturbation theory MP2/6-31G* methods. EHOMO, ELUMO, ΔE, μ, QF, QM and V are chosen as the structural descriptors. The acute toxicity (−lg LC50) of fluorides to rats along with the above structural parameters are used to establish the quantitative structure–activity relationships (QSARs). The results show that the models based on the DFT-B3LYP/6-31G* is better than that based on B3P86/6-31G* and MP2/6-31G* methods. The toxic mechanism is controlled mainly by electronic factor QF (the maximum net atomic charge on fluorides). The QSAR relationship between the electronic parameters QF and the toxicity −lg LC50 is: −lg LC50 = 0.032 + 7.037QF (Radj2=0.869, SE = 0.37097, F = 27.578, P = 0.013). An overall cross-validation correlation coefficient is then obtained using leave-one-out method and the calculated cross-validation coefficient (q2) value is 0.660. The predicted toxic values using the above equation are in good agreement with the experimental values, and the QSAR model has good stability and strong prediction ability.

The models established based on DFT-B3LYP method is better than those based on B3P86 and MP2 methods. The QSAR model has good stability and strong prediction ability. Using the QSAR model, the fluorides which have no experimental toxic values are predicted.Figure optionsDownload as PowerPoint slide