Published: October 27, 2011 r2011 American Chemical Society 443 dx.doi.org/10.1021/jp203491r | J. Phys. Chem. A 2012, 116, 443451 ARTICLE pubs.acs.org/JPCA Combination of Neural Networks and DFT Calculations for the Comprehensive Analysis of FDMPO Radical Adducts from Fast Isotropic Electron Spin Resonance Spectra Katerina Makarova,* ,, Ekaterina V. Rokhina, Elena A. Golovina, Henk Van As, and Jurate Virkutyte § Faculty of Pharmacy, Department of Physical Chemistry, Medical University of Warsaw, Zwirki i Wigury 61, 02-091 Warsaw, Poland Laboratory of Biophysics and Wageningen NMR Centre, Wageningen University, Dreijenlaan 3, 6703 HA Wageningen, The Netherlands § Pegasus Technical Services Inc., 46 East Hollister Street, Cincinnati, Ohio 45219, United States INTRODUCTION The production of free radicals is essential in normal metab- olism. However, in unregulated concentrations, they may cause cell injury or even death. 1 A complete understanding of biological mechanisms that involve free radicals requires ecient radical detection and therefore accurate characterization. Electron Spin Resonance (ESR) spectroscopy has been extensively used for the detection and identication of short-lived free radicals. However, the short lifetime, the high reactivity, and, as a consequence, low concentration of free radicals limit their direct detection. To overcome these drawbacks, the spin trapping method was introduced. 2 It is based on the trapping of radicals by a spin trap, leading to the formation of a more stable radical, a so-called spin adduct, that can be easily detected by ESR spectroscopy. More- over, the shape of the ESR spectra of a spin adduct can be used to identify the trapped radical. Currently, the spin trapping tech- nique is widely used to study in vitro and in vivo formation of free radicals. 3 Unfortunately, the use of the spin trapping technique to investigate radical formation in complex systems (e.g., biological systems) faces two important limitations: (1) a dierent trapping eciency of particular types of radicals and (2) a short lifetime of some spin adducts (e.g., superoxide). 4 To overcome these limitations, a number of novel spin traps have been introduced and evaluated for qualitative analysis of radical-generating systems. For instance, 4-hydroxy-5,5-dimethyl-2-triuoromethylpyrro- line-1-oxide (FDMPO), a uorinated analogue of 5,5-dimethyl- pyrroline-N-oxide (DMPO), has already demonstrated eective- ness in assessing radical production due to the high stability of the adducts (up to several days) and the high trapping rate for a wide range of free radicals (including C-centered, 3 OH, O 2 3 À , and other free radicals). 5,6 Also, high FDMPO spin-trapping eciency and its application to the trapping of oxygen and C-centered free radicals in chemical and biological systems have been studied previously. 5 Nonetheless, the identication of FDMPO radical adducts is a challenging task since the relation between the structure and the ESR spectral parameters (splitting pattern) for FDMPO spin adducts is not unique. In most cases, the dierent radical adducts exhibit very comparable splitting patterns: a triplet of which each line is split in a 1:4:4:1 quartet due to the interaction of the electron spin with the nuclear spins of the nearby N- and F-nuclei. Structural assignment of spectral components can be based on comparison of the ESR parameters of spin adducts produced in alternate ways, e.g., in dierent solvents. 7 However, the dierences between the spin adducts may be insignicant, and the changes in the ESR parameters due Received: April 14, 2011 Revised: August 23, 2011 ABSTRACT: The 4-hydroxy-5,5-dimethyl-2-triuoromethyl- pyrroline-1-oxide (FDMPO) spin trap is very attractive for spin trapping studies due to its high stability and high reaction rates with various free radicals. However, the identication of FDMPO radical adducts is a challenging task since they have very comparable Electron Spin Resonance (ESR) spectra. Here we propose a new method for the analysis and interpretation of the ESR spectra of FDMPO radical adducts. Thus, overlapping ESR spectra were analyzed using computer simulations. As a result, the N- and F-hyperne splitting constants were obtained. Furthermore, an articial neural network (ANN) was adopted to identify radical adducts formed during various processes (e.g., Fenton reaction, cleavage of peracetic acid over MnO 2 , etc.). The ANN was eective on both knownFDMPO radical adducts measured in slightly dierent solvents and not a priori knownFDMPO radical adducts. Finally, the N- and F-hyperne splitting constants of 3 OH, 3 CH 3 , 3 CH 2 OH, and CH 3 (CdO)O 3 radical adducts of FDMPO were calculated using density functional theory (DFT) at the B3LYP/6-31G(d,p)//B3LYP/6-31G++//B3LYP/EPR-II level of theory to conrm the experimental data.