A general technique to rank protein-ligand binding affinities and determine allosteric versus direct binding site...

To realize the full potential of combinatorial chemistry-based drug discovery, generic and efficient tools must be developed that apply the strengths of diversity-oriented chemical synthesis to the identification and optimization of lead compounds for disease-associated protein targets. We report an affinity selection-mass spectrometry (AS-MS) method for protein-ligand affinity ranking and the classification of ligands by binding site. The method incorporates the following steps: (1) an affinity selection stage, where protein-binding compounds are selected from pools of ligands in the presence of varying concentrations of a competitor ligand, (2) a first chromatography stage to separate unbound ligands from protein-ligand complexes, and (3) a second chromatography stage to dissociate the ligands from the complexes for identification and quantification by MS. The ability of the competitor ligand to displace a target-bound library member, as measured by MS, reveals the binding site classification and affinity ranking of the mixture components. The technique requires no radiolabel incorporation or direct biochemical assay, no modification or immobilization of the compounds or target protein, and all reaction components, including any buffers or cofactors required for protein stability, are free in solution. We demonstrate the method for several compounds of wide structural variety against representatives of the most important protein classes in contemporary drug discovery, including novel ATP-competitive and allosteric inhibitors of the Akt-1 (PKB) and Zap-70 kinases, and previously undisclosed antagonists of the M(2) muscarinic acetylcholine receptor, a G-protein coupled receptor (GPCR). The theoretical basis of the technique is analyzed mathematically, allowing quantitative estimation of binding affinities and, in the case of allosteric interaction, absolute determination of binding cooperativity. The method is readily applicable to high-throughput screening hit triage, combinatorial library-based affinity optimization, and developing structure-activity relationships among multiple ligands to a given receptor.