New treatment options are needed to prolong the life of patients with end-stage prostate cancer (PCa) that is resistant to current AR antagonists. These AR inhibitors all bind in the ligand binding pocket (LBP) of AR resulting in an inactive receptor. Mutations within this LBP can circumvent the action of these AR antagonists rendering them inactive or even active as an agonist. The discovery of AR ligand binding domain (LBD) dimerization is an opportunity for alternative targeting of the AR where an inhibitor binds to the surface of the LBD. The physiological relevance of LBD dimerization has been demonstrated in vivo by the infertility and absence of accessory sex organs in male mice as a consequence of a mutated AR LBD dimerization interface.

We therefore explored LBD dimerization as a target for AR inhibition. A druggable surface was identified within the dimerization interface and named the DIM pocket. Using a pharmacophore query for the DIM pocket, we screened for inhibitors in silico. Twenty-nine compounds were tested for AR inhibition in a luciferase reporter cell line. Three structurally similar compounds demonstrated anti-androgenic properties with DIM20 as the least toxic molecule. The mechanism of action of DIM20 was further explored.

DIM20 inhibits dimerization of the full-length AR as well as the isolated LBDs in nanoBiT complementation assays. Using biolayer interferometry, we confirmed reversible dose-dependent binding to the AR LBD. Binding in the LBP was excluded since DIM20 did not displace ³H-DHT in a whole cell competition assay. The non-competitive nature of DIM20 was further confirmed by its synergistic effects on transcription and proliferation of AR-dependent PCa cell lines when DIM20 was combined with Enzalutamide. Surprisingly, DIM20 also inhibits the activity and dimerization of heterodimers of the AR and a truncated NTD-DBD variant.

In conclusion, these findings demonstrate that the AR LBD dimerization interface is a relevant therapeutic target for PCa therapy.