Androgen receptor (AR)–interacting proteins have previously been catalogued as nearly 1,000 factors, but how these interactions are organized across subcellular compartments and over time remains uncharacterized. Proximity labeling captures both direct binders and neighboring proteins that occupy a receptor's local environment, broadening the concept of AR-interacting proteins into a larger set of AR-proximal interacting proteins (AR-PIPs). The authors applied proximity labeling quantitative mass spectrometry (PL-qMS) to build a spatiotemporal atlas of the extranuclear AR-proximal interactome in prostate tumor cells, with the aim of defining compartment-specific and time-dependent AR proximal networks.
The study used PL-qMS to identify proteins in the immediate proximity of extranuclear AR in LNCaP prostate tumor cells. Proximity labeling approaches identify direct interactors and neighboring proteins within the receptor's local milieu, enabling capture of a broader AR-proximal population than traditional binding assays. Analyses compared cytosolic and membrane compartments and assessed changes across an androgen time course. Functional enrichment and network analysis methods were applied to the proteomic datasets to highlight enriched complexes and candidate AR‑proximal interaction networks.
PL-qMS recovered 82.2% of previously reported AR-interacting proteins, demonstrating substantial overlap with known AR binders. In addition to this recovery, the approach identified a total of 3,947 AR-PIPs across cytosolic and membrane compartments. This expanded dataset positions proximity labeling as a sensitive method to map the receptor's proximal proteome beyond classical interaction databases.
The AR-proximal proteome exhibited dynamic remodeling across the androgen exposure time course. Identified AR-PIPs were distributed across extranuclear compartments, and their relative proximities to AR changed in an androgen-sensitive manner. These findings support the concept that AR's extranuclear interaction environment is both spatially and temporally regulated in response to ligand.
Functional enrichment and network analyses of the PL-qMS datasets identified the retromer complex as an androgen-sensitive component of the AR proximal interactome. The retromer complex emerged from the enrichment and network-level interrogation of AR-PIPs, implicating endosomal trafficking machinery in extranuclear AR biology.
To confirm the proteomic finding, proximity ligation assays were performed and validated the association between AR and retromer components. These in situ assays provided orthogonal evidence supporting the proximity-based proteomic identification of the retromer as an AR‑proximal interaction.
The authors performed partial genetic disruption of VPS26A, a core retromer component, to test functional consequences of perturbing this AR-proximal network. Disruption of VPS26A attenuated transcription of canonical androgen-regulated genes. Mechanistically, VPS26A perturbation caused mislocalization of the AR coactivator TMF1, linking retromer function to nuclear transcriptional outputs via coactivator trafficking. Together these results establish a functionally validated retromer–AR–TMF1 axis within the extranuclear AR-proximal interaction network (AR-PIN).
This work establishes that subcellular proximal proteomes can serve as a spatiotemporal framework for probing AR function. By resolving AR-proximal proteins by compartment and time, the approach can reveal dynamic networks that influence canonical nuclear transcriptional programs through extranuclear processes such as trafficking and coactivator localization. The identification and validation of the retromer–AR–TMF1 axis exemplify how such atlases can pinpoint functionally relevant pathways that modulate androgen-regulated gene expression.
All reported experiments were performed in LNCaP prostate tumor cells. The manuscript is posted as a bioRxiv preprint and has not been certified by peer review, as noted by the authors. Detailed experimental parameters, dataset accessions, and additional methodological specifics were not included in the summary provided here and should be obtained from the full preprint for replication or deeper evaluation.