QSAR Analysis of Indole Analogues as Monoamine Oxidase Inhibitors A. E. Medvedev, A. S. Ivanov,* A. V. Veselovsky, V. S. Skvortsov, and A. I. Archakov Laboratory of Biogenic Amines and Laboratory “Computers in Biochemistry,” Institute of Biomedical Chemistry, Russian Academy of Medical Sciences, Pogodinskaya Street 10, Moscow, 119832, Russia Received September 15, 1995 X The quantitative structure-activity relationship (QSAR) analysis with comparative molecular field analysis (CoMFA) of indole derivatives-monoamine oxidase (MAO) inhibitors were done. The pharmacophore model included four features: two hydrophobic rings, one donor atom, and one acceptor site. The predictive values (cross-validated r 2 ) of QSAR analysis for the inhibition of MAO-A and MAO-B were 0.743 and 0.603, respectively. The contributions of steric and electrostatic fields in the interaction between inhibitors and enzymes were equal. The three-dimensional arrangement of these fields for MAO-A and MAO-B suggests that structures of active site for both enzymes are considerably differed from each other. INTRODUCTION Monoamine oxidase (E.C. 1.4.3.4), an integral protein of the outer mitochondrial membrane, catalyses a reaction of oxidative deamination of neurotransmitter monoamines in central nervous system and peripheral tissues. 1 The enzyme exists in two forms, MAO-A and MAO-B, initially distin- guished by sensitivity to low concentrations of acetylenic inhibitors, clorgyline and deprenyl, respectively. Strong evidence now exists that these forms of MAO are different proteins, encoded by two different genes. 1,2 In spite of essential progress in elucidation of peculiarities of primary structure, biosynthesis of MAO and mechanisms of its insertion into the outer mitochondrial membrane, 3,4 precise information about steric structures of substrate biding sites of MAO-A and B is still absent. This stimulates computer-modeling studies based on experimental data of MAO interaction with substrates and inhibitors. One of the first computer analysis of series of potent MAO inhibitors revealed that common arrangement of these inhibitors includes aromatic ring and a side chain with an amino group located in the ring’s plane. 5 The distance between the center of nitrogen and the ring was 0.50-0.55 nm. 6 The discovery of MAO-mediated bioactivation of MPTP to neurotoxin MPP 7 which causes a development of major symptoms of parkinsonism in experimental animals gave new rise for quantitative structure-activity relationship studies with relevance to some structural features influencing the binding to active sites of MAO-A and MAO-B. 8,9 Indoles are a large group of nitrogen-containing hetero- cyclic compounds widely distributed in living world and in its chemical environment. Some indole amines as serotonin and tryptamine are substrates of MAO. Recently it has been shown that isatin (2,3-dioxindole) acts as a relatively selective inhibitor of MAO-B than MAO-A. 10 5-Hydroxyisatin ex- hibited selective MAO-A inhibition, whilst indole and other isatin analogues were less potent MAO-A and B inhibitors. 11 In the present report we have analyzed our experimental data on inhibition of MAO-A and MAO-B by indole and isatin analogues obtained in our and other laboratories. 10-13 The action of these inhibitors was readily reversible. 12 Previous pilot QSAR analysis of their inhibitory potency revealed that molecules of selective MAO A and B inhibitors had different sizes: 11.5 × 5.6 × 1.8 and 8.5 × 5.1 × 1.8 Å, respectively. Here we have investigated further their quantitative structure-activity relationship (QSAR) with molecular field analysis (CoMFA). MATERIALS AND METHODS Experimental data of MAO inhibitors were taken from previous reports. 11-13 Computer Modeling. The study was carried out using Sybyl 6.1 software (Tripos GmbH, Munich, Germany) running on Silicon Graphics workstation Indigo-2 (R4400, XZ). The molecular models were constructed, and their geometry were minimized using the standard Tripos force field. The atomic charges were calculated by Gasteiger- Huckel method. They were used in the subsequent analysis. PowerSearch program for Windows (Tripos GmbH, Mu- nich, Germany) was employed for the conformation search for flexible bonds of molecules. In our study we have used the conformers with the lowest energy as the most probable conformation. The correctness of atomic charges’ distribution, calculated by empirical Gasteiger-Huckel method, was also validated for some molecules by quantum mechanical AM1, MNDO, and MINDO3 methods. 14 CoMFA Analysis. The inhibitors were aligned by fitting the indole common structure atom by atom. Initial phar- macophore model was constructed using the DISCO pro- gram. 15 This model was made by matching all structures of inhibitors, with a range of tolerance from 0.5 to 5.0 by 0.5 and features requirement from 3 to 7. Then the CoMFA analysis was carried out in QSAR option of Sybyl. The region was generated automatically by the program. Both steric and electrostatic fields were taken into investigations. The steric and electrostatic potentials were generated by using an sp 3 carbon probe and a +1 charge. QSAR analysis was made using PLS technique in two steps. At first using 10 components and cross-validation groups equal to a number of compounds the optimal number of components was determined. The second step was made with the determined optimal number of components and without cross-validation. X Abstract published in AdVance ACS Abstracts, June 15, 1996. 664 J. Chem. Inf. Comput. Sci. 1996, 36, 664-671 S0095-2338(95)00126-0 CCC: $12.00 © 1996 American Chemical Society + +