DNA-binding domain as the minimal region driving RNA-dependent liquid-liquid phase separation of androgen receptor

Abstract

Androgen receptor (AR) is a nuclear hormone receptor that regulates the transcription of genes involved in the development of testis, prostate and the nervous system. Misregulation of AR is a major driver of prostate cancer (PC). The primary agonist of full-length AR is testosterone, whereas its splice variants, for example, AR-v7 implicated in cancer may lack a ligand-binding domain and are thus devoid of proper hormonal control. Recently, it was demonstrated that full-length AR, but not AR-v7, can undergo liquid-liquid phase separation (LLPS) in a cellular model of PC. In a detailed bioinformatics and deletion analysis, we have analyzed which AR region is responsible for LLPS. We found that its DNA-binding domain (DBD) can bind RNA and can undergo RNA-dependent LLPS. RNA regulates its LLPS in a reentrant manner, that is, it has an inhibitory effect at higher concentrations. As RNA binds DBD more weakly than DNA, while both RNA and DNA localizes into AR droplets, its LLPS depends on the relative concentration of the two nucleic acids. The region immediately preceding DBD has no effect on the LLPS propensity of AR, whereas the functional part of its long N-terminal disordered transactivation domain termed activation function 1 (AF1) inhibits AR-v7 phase separation. We suggest that the resulting diminished LLPS tendency of AR-v7 may contribute to the misregulation of the transcription function of AR in prostate cancer.

Keywords: DNA binding; RNA binding; androgen receptor; biomolecular condensates; liquid-liquid phase separation; prostate cancer.

FIGURE 1

Sequence‐based bioinformatics predictions for AR‐v7. Various prediction tools were used to assess potential LLPS‐related features of the AR‐v7 sequence (UniProt: P10275‐3). For LLPS propensity, we used (a) PScore ²⁵ [and (b) catGRANULE ²⁶ direct LLPS‐propensity predictions. (c) Charge distribution at pH 7.4 was smoothed with a 20‐aa sliding window, whereas for RNA‐binding propensity (d) PPRINT ⁴² RNA‐binding scale was used. AR, androgen receptor; LLPS, liquid–liquid phase separation

FIGURE 2

Interaction between DNA‐binding domain (DBD) and RNA/ARE (DNA). (a) The interaction of AR‐v7's DBD was measured with Cy5‐labeled U30 at concentration of 300 nM in Tris buffer (50 mM Tris, 150 mM NaCl, 0.05% Tween 20, pH 7.5) by MST. Interaction of immobilized (b) biotinylated poly U; (c) tRNA; (d) ARE‐1; (e) ARE‐2 on streptavidin sensors and DBD in solution measured by BLI. ARE, androgen response element; BLI, biolayer interferometry; DBD, DNA‐binding domain; LLPS, liquid–liquid phase separation; poly U, polyuridylic acid

FIGURE 3

Phase separation of DNA‐binding domain (DBD) with polyuridylic acid (poly U). (a) OD 600 at different concentrations of DBDin presence of 0.05 μg/μl of poly U (b) Hydrodynamic radius of DBD measured by dynamic light scattering (DLS) in the presence (filled circles) and absence of 0.05 μg/μl poly U (filled squares). (c) Dylight® 488‐labeled DBD(green) was used at 50 μM and mixed with Cy5‐labeled poly U (red) for final concentration of 0.05 μg/μl of poly U. Panels C show the droplets (upper snapshots) right after phase separation, and (lower snapshots) after 1 h

FIGURE 4

Phase separation of different constructs of AR‐v7 in the presence of RNA. (a) OD 600 for different constructs of AR‐v7 (as defined in the text, AF1: activation function 1, F5: fragment 5, DBD: DNA‐binding domain, F5‐DBD: F5 + DBD fused, AF1‐DBD: AF1 + DBD fused, and AR‐v7: AR splice variant v7) at a concentration of 25 μM, in the presence of 0.05 μg/μl poly U with buffer (50 mM phosphate, 50 mM NaCl, 0.5 mM TCEP, pH 7.0). Control is measured in the absence of RNA. (b) DBD with 0.05 μg/μl of different types of RNA (tRNA: transfer RNA, rRNA: ribosomal RNA, control: in the absence of RNA); (c, d, e) DBD with different concentrations of salt, hexanediol and poly U. ARE, androgen response element; DBD, DNA‐binding domain; poly U: polyuridylic acid, poly A: polyadenylic acid, poly G: polyguanylic acid

FIGURE 5

Colocalization of DNA‐binding domain (DBD) and androgen response element (ARE) DNA. Dylight® 488‐labeled DBD (green) mixed with DBD for final concentration of (a) 25 μM, then Cy5‐labeled (red) ARE‐1 was added at 2.5 μM in buffer (50 mM phosphate, 50 mM NaCl, 0.5 mM TCEP, pH 7.0). (b) 12.5 μM of ARE‐1 was mixed with DBD (2.5 μM of Cy5‐labeled ARE‐1 with 10 μM of unlabeled ARE‐1). Phase separation was induced by 0.05 μg/μl poly U. Scale bar is 10 μm. Microscopy snapshots were taken ~ 1 min after polyuridylic acid (poly U)‐induced liquid–liquid phase separation (LLPS). (c) Competition assay for DBD between DNA and RNA measured by MST. Cy5‐labeled poly U was used as fluorophore at a stable concentration and the binding to DBD as well as to DBD‐ARE‐1 and DBD‐ARE‐2 complex (1:1) was measured in buffer (50 mM Tris, 150 mM NaCl, 0.05% Tween 20, pH 7.5. (d) Competition assay for DBD between DNA and RNA measured by biolayer interferometry (BLI). RNA mimic poly U was immobilized on sensors and dipped in 10 μM of DBD or DBD‐DNA complex (1:1) in buffer (50 mM Tris, 150 mM NaCl, 0.05% Tween 20, pH 7.5. (e) OD 600 for DBD‐poly U with different concentrations of ARE‐1 and ARE‐2 in buffer (50 mM phosphate, 50 mM NaCl, 0.5 mM TCEP, pH 7.0)