Despite the widespread acceptance of guidelines related to desirable physicochemical properties of potential small-molecule drugs, key properties - such as lipophilicity - of recently developed clinical candidates and advanced lead compounds have been shown to differ significantly from those of historical leads and drugs. By analysing the physicochemical properties of a large database of hits and corresponding leads identified in the past decade, we show that this undesirable phenomenon can be traced back to the nature of high-throughput screening hits and hit-to-lead optimization practices. Conceptual and organizational adjustments may be required to enable a smooth lead-evolution process that reduces the chance of high compound-related attrition in clinical trials.

No abstract

A small molecule nonpeptide inhibitor of beta-secretase has been developed, and its binding has been defined through crystallographic determination of the enzyme-inhibitor complex. The molecule is shown to bind to the catalytic aspartate residues in an unprecedented manner in the field of aspartyl protease inhibition. Additionally, the complex reveals a heretofore unknown S(3) subpocket that is created by the inhibitor. This structure has served an important role in the design of newer beta-secretase inhibitors.

To probe the interaction between transducin (G(t)) and photoactivated rhodopsin (R*), 14 analog peptides were designed and synthesized restricting the backbone of the R*-bound structure of the C-terminal 11 residues of G(t)alpha derived by transferred nuclear Overhauser effect (TrNOE) NMR. Most of the analogs were able to bind R*, supporting the TrNOE structure. Improved affinities of constrained peptides indicated that preorganization of the bound conformation is beneficial. Cys347 was found to be a recognition site; particularly, the free sulfhydryl of the side chain seems to be critical for R* binding. Leu349 was another invariable residue. Both Ile and tert-leucine (Tle) mutations for Leu349 significantly reduced the activity, indicating that the Leu side chain is in intimate contact with R*. The structure of R* was computer generated by moving helix 6 from its position in the crystal structure of ground-state rhodopsin (R) based on various experimental data. Seven feasible complexes were found when docking the TrNOE structure with R* and none with R. The analog peptides were modeled into the complexes, and their binding affinities were calculated. The predicted affinities were compared with the measured affinities to evaluate the modeled structures. Three models of the R*/G(t)alpha complex showed strong correlation to the experimental data.