A new computational model for the prediction of toxicity of phosphonate derivatives using QSPR

Abstract

Structural and electronic properties of a series of 25 phosphonate derivatives were analyzed applying density functional theory, with the exchange-correlation functional PBEPBE in combination with the 6-311++G** basis set for all atoms. The chemical reactivity of these derivatives has been interpreted using quantum descriptors such as frontier molecular orbitals (HOMO, LUMO), Hirshfeld charges, molecular electrostatic potential, and the dual descriptor [$\Delta f(r)$]. These descriptors are directly related to experimental median lethal dose ($\text {LD}_{50})$, expressed as its decimal logarithm [${A}_{\mathrm{obs}}= \text {log}_{10}$($\text {LD}_{50})$] through a multiple linear regression equation. The proposed model predicts the toxicity of phosphonates in function of the volume (V), the load of the most electronegative atom of the molecule (q), and the eigenvalue of the molecular orbital HOMO (${E}_{\mathrm{HOMO}})$. The obtained values in the internal validation of the model are: ${R}^{2}= 82.71$%, ${R}^{2}_{\mathrm{ADJ}} = 80.24$%, $F= 33.5$, $\delta {K}=0.169$, $\delta {Q}=0.011$, ${R}^{\mathrm{P}}=0.423$, ${R}^{\mathrm{N}} = -\,0.025\,(-\,0.311)$, and ${Q}^{2}_{\mathrm{boot}} = 75.45$%. The toxicity of nine phosphonate derivatives used as test molecules was adequately predicted by the model. The theoretical results indicate that the oxygen atom of the O=P group plays an important role in the interaction mechanism between the phosphonate and the acetylcholinesterase enzyme, inhibiting the removal of the proton of the ser-200 residue by the his-440 residue.