Abstract

Prostate cancer is
the second leading cause of male cancer fatalities in Western countries. The
androgen receptor (AR) plays a crucial role in the development and progression
of prostate cancer.  A number of targeted
approaches are approved for the treatment of prostate cancer that either antagonise
the AR through binding the steroid pocket of the ligand-binding domain (LBD) (bicalutamide, enzalutimide) or by blocking
androgen biosynthesis (abiraterone). Resistance mechanisms often arise and
patients relapse, resulting in a more aggressive cancer known as castration-resistant
prostate cancer (CRPC). Of the multiple resistance mechanisms that are known, many
include point mutations within the AR-LBD or splice variants of AR that lack
the LBD. Novel drugs are required to treat CRPC.

The AF-2 pocket is induced in the activated AR and is important for AR
signaling through its recruitment of coactivator proteins and its binding of the
N-terminal region of the AR. The AR activation function-2 (ARAF-2) project
seeks to target a number of AR-dependent resistance mechanisms through the
identification of small molecule inhibitors that bind in the AF-2 pocket

Hit finding
projects that target novel pockets in proteins involved in protein-protein
interactions such as AF-2 require robust triage cascades, including a suite of biochemical
and biophysical assays with x-ray crystallography to confirm the compound
binding mode. Here we describe a hit finding campaign to discover small molecules
that target the AF-2 pocket of the AR for the treatment of CRPC.

A virtual
screen selected a large library of small molecules for screening in a
biochemical assay that measures binding of AR-LBD to a peptide derived from a
coactivator protein. In order to identify compounds that directly bind AR-LBD,
an orthogonal SPR assay was developed. A mono-biotinylated AviTag AR-LBD
bound to a streptavidin surface enabled detection of the AR N‑terminal peptide with
high sensitivity. Following screening of the hits with the SPR assay, compound
binding was grouped into one of the following three categories: no binding or
binding levels consistent with either specific or super-stoichiometric non-specific
binding.

Following a computational
analysis, compounds from a number of clusters were selected for
resynthesis and rescreening in the SPR and biochemical coactivator recruitment
assays. To further understand compound binding mode (AF-2 or allosteric pocket or non‑specific), an
SPR competition assay (using the AR N‑terminal peptide) and a soakable AR-LBD crystal system were developed. Profiling of compounds in these assays allowed non-AF-2
pocket binders to be successfully deselected during the triage process.