An update view on the substrate recognition mechanism of phosphodiesterases: a computational study of PDE10 and PDE4 bound with cyclic nucleotides.

Early studies strongly implied that the specificity of cyclic nucleotide phosphodiesterases (PDEs) toward its endogenous substrates can be uniquely determined by the amido orientation of the invariant glutamine locating in the binding pocket of the enzyme. However, recently solved crystal structures of PDE4 (cAMP specific) and PDE10 (dual specific) in the presence of endogenous substrates have revealed that their invariant glutamine orientations are very similar despite exhibiting different substrate specificities proven physiologically. To understand this subtle specificity issue in the PDE family, here several experimentally inaccessible PDE-substrate complex models have been studied computationally, and the results are juxtaposed and compared in detail. Modeling results show that PDE10 in fact favors cAMP energetically but still can bind to cGMP owing to the robust hydrogen-bond network in the vicinity of the invariant glutamine side chain. PDE4 fails to accommodate cGMP is correlated to the weakening of this same hydrogen-bond network but not owing to any steric strain in the binding pocket. An Asn residue in the binding pocket of PDE4 has enhanced the specificity of the binding to cAMP sideway as observed in our computer simulation. Further to the previously studied syn- versus anti-conformational specificity of cAMP in PDE10, the unexpected substrate-binding mode in PDE10 versus PDE4 as reported here strongly suggested that there are remaining uncertainties in the substrate orientation and recognition mechanism in the PDE families. The molecular details of the binding pocket observed in this study provide hints for more optimal PDE4 and PDE10 inhibitor design.