Preferred antagonist binding state of the NMDA receptor: synthesis, pharmacology, and computer modeling of (phosphonomethyl)phenylalanine derivatives.

A series of substituted [phosphono-, sulfo-, carboxy-, and (N-hydroxycarbamoyl)methyl]phenylalanines were synthesized as probes for the investigation of the preferred antagonist state of the NMDA receptor antagonists. The potency of these compounds was evaluated by measuring electrophysiological responses induced by NMDA in cultured mouse cortical neurons. 3-(Phosphonomethyl)phenylalanine [1(m)] a formal AP7 analogue, has been shown to be the most potent antagonist in this study with an IC50 of around 5 microM. The isomeric 2-(phosphonomethyl)phenylalanine [1(o)] was about half as active as 1(m) and as active as compound 5(3), a derivative which is cis-hydrogenated on the phenyl ring of 1(m). Replacement of a phosphono by a sulfo group led to a large reduction in the ability of these compounds to antagonize NMDA responses, although the ortho and meta isomers retained some activity in their reduced forms. In both series the para isomers were almost completely inactive at 100 microM. Introduction of a carboxyl or a bidentate HONHCO group in place of the phosphono moiety in the 3-position results in compounds devoid of activity. The active and inactive compounds of this study were used in conjunction with the most potent linear and cyclic phosphono-containing NMDA antagonists reported to date to determine, via computer modeling techniques, a three-dimensional model corresponding to a antagonist preferring state of the NMDA binding site. This structure defines a pharmacophore which is characterized by (i) well-defined distances between the central atoms of the polar groups PO3H-, NHn+, (n = 2, 3), and COO- (P-N = 5.89 +/- 0.12 A, P-C = 6.66 +/- 0.08 A, and N-C = 2.28 +/- 0.01 A), (ii) a sterically allowed region between the C5 methylene and the PO3H- group, and (iii) a molecular electrostatic field in which the positive, neutral, and negative potential zones are self-contained--with the negative potential zone connecting the PO3H- and COO- groups as the largest. We have compared our results to a preliminary model of the NMDA antagonist site by Hutchison et al. and to a topological model of the NMDA-glycine receptor site by Cordi et al. Our proposed steric-electrostatic pharmacophore which refines, simplifies, and improves these models has now to be validated by the design of new NMDA antagonists.