Androgen receptor (AR) is expressed in ~90% of estrogen receptor-positive (ER+) breast tumours, and AR activation opposes estrogen-driven growth in breast cancer (1,2). This tumour suppressive activity has renewed interest in selective androgen receptor modulators (SARMs) for AR-positive/ER-positive (AR+/ER+) breast cancer (1,3). SARMs built on different chemical scaffolds are not interchangeable, however; they engage the same pocket yet differ in efficacy, and the structural basis for that difference is poorly defined (3,4). We asked whether chemically distinct SARMs stabilise different AR conformational states, and whether those states track with co-regulator recruitment, transcriptional output, and breast cancer cell growth.
Crystal structures were solved for the AR ligand-binding domain (AR-LBD) bound to four structurally distinct SARMs and to the endogenous agonist 5α-dihydrotestosterone (DHT). Small-angle X-ray scattering reported on receptor behaviour in solution. Structural data were interpreted alongside co-regulator interaction assays, AR-responsive reporter assays, and proliferation assays in AR+/ER+ breast cancer cell lines.
All four ligands supported an agonist-compatible AR-LBD fold; however, each SARM adopted a distinct binding mode, producing local differences in pocket geometry and side-chain packing. Deviations though modest, yet they were not confined to residues contacting the ligand: changes extended to regions connected to helix 12 and the activation function-2 (AF-2) surface. Because helix 12 is a core component of AF-2 (5), ligand-specific packing within the buried pocket has a plausible route to the external surface that recruits co-regulators.
AR agonism therefore does not correspond to a single invariant conformation. Different chemotypes appear to favour related but distinguishable conformational ensembles, with consequences for the stability and accessibility of AF-2. Solution scattering and co-regulator binding measurements agreed with this interpretation, indicating that the crystallographic differences were not artefacts of crystal packing.
In cells, each SARM increased AR-dependent reporter activity and inhibited AR+/ER+ breast cancer cell proliferation in a concentration-dependent manner, but the magnitude of response varied between ligands. Receptor occupancy alone therefore cannot account for SARM activity. Efficacy appears instead to depend on how effectively a given scaffold couples the ligand-binding pocket to the receptor surfaces that drive transcription.
These results link SARM chemistry to pocket organisation, long-range conformational communication, co-regulator recognition, and cellular response, and offer a structural explanation for why ligands directed at the same orthosteric site behave as full or partial agonists. For AR-directed drug development in ER+ breast cancer, the implication is that binding affinity is an incomplete selection criterion: the conformation a compound stabilises and the co-regulators it engages matter as well. Collectively, these findings define structural determinants of selective AR modulation and establish a mechanistic framework linking ligand chemotype, receptor conformation, and tumour suppressive AR activity, thereby informing the rational development of next-generation AR-directed therapeutics for ER+ breast cancer.