Identification of inhibitor binding sites of the cAMP-specific phosphodiesterase 4

Using the technique of site-directed mutagenesis, point mutants of human PDE4A have been developed in order to identify amino acids involved in inhibitor binding. Relevant amino acids were selected according to a peptidic binding site model for PDE4 inhibitors, which suggests interaction with two tryptophan residues, one histidine and one tyrosine residue, as well as one Zn2+ ion. Mutations were directed at those tryptophan, histidine, and tyrosine residues, which are conserved among the PDE4 subtypes (PDE4A-D) and lie within the high-affinity 4-[3-(cyclopentoxyl)-4-methoxyphenyl]-2-pyrrolidone (rolipram) binding domain of human PDE4A (amino acids 276-681 according to the PDE4A sequence L20965). Truncations to this region do not alter enzyme activity or inhibitor sensitivity. The mutants were expressed in COS1 cells, and the recombinant cyclic nucleotide phosphodiesterase (PDE) forms have been characterized in terms of their catalytic activity and inhibitor sensitivities. Tyrosine residues 432 and 602, as well as histidine 588, were found to be involved in inhibitor binding, but no interaction was detected between tryptophan and PDE inhibitors tested. To test the possibility that other amino acids are of importance for hydrophobic interactions, selected phenylalanine residues were also mutated. We found phenylalanine 613 and 645 to influence inhibitor binding to PDE4. The significant differences in the inhibitor sensitivities of the mutants show that the various inhibitors have different enzyme binding sites. Based on the assumption that the known side effects of PDE4 inhibitors (like emesis and nausea) are caused directly by selective inhibition of different conformation states of PDE4, our results may be a hint to differ between PDE4 inhibitors, which have emetic side effects (like rolipram), and those that do not have side effects (like N-(3,5-dichlorpyrid-3-yl)-[1-(4-fluorbenzyl)-5-hydroxy-indol-3-yl]-glyoxylateamide [AWD12-281]) by the differences of their binding sites and in that context contribute to the development of novel drugs. Furthermore, the identification of amino acid interactions proposed by the peptidic binding site model, which was used for the mutant selection, verifies the PrGen modeling as a useful method for the prediction of inhibitor binding sites in cases where detailed knowledge of the protein structure is not available. (C) 2001 Elsevier Science Inc. All rights reserved.