Aurora A is a mitotic serine-threonine protein kinase that controls chromosomal segregation and plays a key role in proper cell cycle progression and cytokinesis. The activity is frequently elevated in a variety of human cancers and Aurora A is generally considered an attractive target for therapeutic intervention.

Based on screening and subsequent hit-to-lead expansion, we have identified Tetrahydro[b]benzothiophenes as a novel and potent compound class inhibiting Aurora A. Several computational strategies have been applied to assess the binding geometry and predict the binding affinity of tetrahydro[b]benzothiophenes to the target enzyme. Molecular interaction modes with the ATP-binding site were predicted using several docking programs such as FlexX, QXP and Yeti. These direct molecular modelling techniques allowed us to predict the geometry in good agreement with the experimental results from X-ray crystallography. For a more accurate and systematic classification of structure-activity relationships, we applied multidimensional QSAR modelling using the software packages Raptor and Quasar. These techniques generate a pseudo-receptor using an ensemble of overlaid inhibitors as a template, thus imprinting the compounds’ properties on an envelope mimicking the protein interaction surface. For purposes of training of a pseudoreceptor model, observed IC50 values, multiple conformations, alternative protonation states, and tautomers of the inhibitors are taken into account, which increases the dimensionality of this method from classical 3D QSAR to 4D-, 5D- or even 6D-QSAR. Some molecular features of the compound series could be rationalized that were not evident by classical molecular modelling in combination with X-ray crystallography.

The collective information derived from combining direct and indirect molecular modelling techniques allows us to pursue the optimization of this compound series for Aurora A inhibition.

98th AACR Annual Meeting-- Apr 14-18, 2007; Los Angeles, CA