Most protein kinases share a DFG (Asp-Phe-Gly) motif in the ATP site which can assume two distinct conformations, the active DFG-in and the inactive DFG-out states. Small molecule inhibitors able to induce the DFG-out state have received considerable attention in kinase drug discovery. Using a typical DFG-in inhibitor scaffold of Aurora A, a kinase involved in the regulation of cell division, we found that halogen and nitrile substituents directed at the N-terminally flanking residue Ala273 induced global conformational changes in the enzyme, leading to DFG-out inhibitors that are among the most potent Aurora A inhibitors reported to date. The data suggest an unprecedented mechanism of action, in which induced-dipole forces along the Ala273 side chain alter the charge distribution of the DFG backbone, allowing the DFG to unwind. As the ADFG sequence and three-dimensional structure is highly conserved, DFG-out inhibitors of other kinases may be designed by specifically targeting the flanking alanine residue with electric dipoles.

The ortho-carboxylic acid substituted bisanilinopyrimidine 1 was identified as a potent hit (Aurora A IC50 = 6.1 ± 1.0 nM) from in-house screening. Detailed structure activity relationship (SAR) studies indicated that polar substituents at the para position of the B-ring are critical for potent activity. X-ray crystallography studies revealed that compound 1 is a type-I inhibitor that binds the Aurora kinase active site in a DFG-in conformation. Structure activity guided replacement of the A-ring carboxylic acid with halogens and incorporation of fluorine at the pyrimidine 5-position led to highly potent inhibitors of Aurora A that bind in a DFG-out conformation. B-ring modifications were undertaken to improve the solubility and cell permeability. Compounds such as 9m with water-solubilizing moieties at the para-position of the B-ring inhibited the autophosphorylation of Aurora A in MDA-MB-468 breast cancer cells.

Screening of the 50,000 ChemBridge compound library led to the identification of the oxadiazole-isopropylamide 1 (PI-1833) which inhibited CT-L activity (IC50 0.60 μM) with little effects on the other 2 major proteasome proteolytic activities, T-L and PGPH-L. LC/MS-MS and dialysis show that 1 is a non-covalent and rapidly reversible CT-L inhibitor. Focused library synthesis provided 11ad (PI-1840) with CT-L activity (IC50 27 nM). Detailed SAR studies indicate that the amide moiety and the 2 phenyl rings are sensitive toward modifications. Hydrophobic residues, such as propyl or butyl, in the para-position (not ortho or meta) of the A-ring and a meta-pyridyl group as B-ring significantly improve activity. Compound 11ad (IC50 0.37 μM) is more potent than 1 (IC50 3.5 μM) at inhibiting CT-L activity in intact MDA-MB-468 human breast cancer cells and inhibiting their survival. The activity of 11ad warrants further pre-clinical investigation of this class as non-covalent proteasome inhibitors.