American Chemical Society
jm020859m_si_001.pdf (786.92 kB)

3D QSAR Analyses-Guided Rational Design of Novel Ligands for the (α4)2(β2)3 Nicotinic Acetylcholine Receptor

Download (786.92 kB)
journal contribution
posted on 2003-04-22, 00:00 authored by Holger Gohlke, Simone Schwarz, Daniela Gündisch, Maria Cristina Tilotta, Alexander Weber, Thomas Wegge, Gunther Seitz
Three-dimensional quantitative structure−activity relationship methods, the comparative molecular field analysis (CoMFA) and the comparative molecular similarity indices analysis (CoMSIA), were applied using a training set of 45 ligands of the (α4)2(β2)3 nicotinic acetylcholine receptor (nAChR). All compounds are related to (−)-epibatidine, (−)-cytisine, (+)-anatoxin-a, and (−)-ferruginine, and additionally, novel diazabicyclo[4.2.1]nonane- and quinuclidin-2-ene-based structures were included. Their biological data have been determined by utilizing the same experimental protocol. Statistically reliable models of good predictive power (CoMFA r2 = 0.928, q2 = 0.692, no. of components = 3; CoMSIA r2 = 0.899, q2 = 0.701, no. of components = 3) were achieved. The results obtained were graphically interpreted in terms of field contribution maps. Hence, physicochemical determinants of binding, such as steric and electrostatic and, for the first time, hydrophobic, hydrogen bond donor, and hydrogen bond acceptor properties, were mapped back onto the molecular structures of a set of nAChR modulators. In particular, changes in the binding affinity of the modulators as a result of modifications in the aromatic ring systems could be rationalized by the steric, electrostatic, hydrophobic, and hydrogen bond acceptor properties. These results were used to guide the rational design of new nAChR ligands such as 4852 and 54, which were subsequently synthesized for the first time and tested. Key steps of our synthetic approaches were successfully applied Stille and Suzuki cross-coupling reactions. Predictive r2 values of 0.614 and 0.660 for CoMFA and CoMSIA, respectively, obtained for 22 in part previously unknown ligands for the (α4)2(β2)3 subtype, demonstrate the high quality of the 3D QSAR models.