Hormone-induced enhancer assembly requires an optimal level of hormone receptor multivalent interactions

Abstract

Transcription factors (TFs) activate enhancers to drive cell-specific gene programs in response to signals, but our understanding of enhancer assembly during signaling events is incomplete. Here, we show that androgen receptor (AR) forms condensates through multivalent interactions mediated by its N-terminal intrinsically disordered region (IDR) to orchestrate enhancer assembly in response to androgen signaling. AR IDR can be substituted by IDRs from selective proteins for AR condensation capacity and its function on enhancers. Expansion of the poly(Q) track within AR IDR results in a higher AR condensation propensity as measured by multiple methods, including live-cell single-molecule microscopy. Either weakening or strengthening AR condensation propensity impairs its heterotypic multivalent interactions with other enhancer components and diminishes its transcriptional activity. Our work reveals the requirement of an optimal level of AR condensation in mediating enhancer assembly and suggests that alteration of the fine-tuned multivalent IDR-IDR interactions might underlie AR-related human pathologies.

Keywords: androgen receptor; condensate formation; condensation; enhancer; hormone-induced enhancer assembly; intrinsically disordered region; multivalent interaction; phase separation.

Figure 1.. AR requires its NTD to form condensates and to activate transcription in response to hormone stimulation.

(A-B) Representative droplet formation images and phase diagram of purified GFP-ARNTD at indicated conditions. Scale bar: 10 μm. (C) Representative images of LNCaP cells transiently transfected with GFP-ARwt or GFP-ARΔIDR treated with vehicle or 100 nM DHT. White boxes indicate the zoomed regions shown on the right. NTD deletion (ΔIDR) abolished DHT-induced AR foci formation. Scale bar: 10 μm. (D) Quantification of LNCaP cells transiently transfected with GFP-ARWT or GFP-ARΔIDR with indicated treatments. Percentages of cells showing AR foci in GFP-positive cells were plotted. Statistics: one-way ANOVA, ***P < 0.001. (E-H) Heatmaps and aggregate plots of ATAC-seq data of LNCaP cells stably expressing Dox-inducible exogenous ARwt or ARΔIDR and transduced with control or AR shRNA to knock down endogenous AR. Exogenous ARwt, but not ARΔIDR, promoted chromatin opening and rescue the reduced chromatin accessibility caused by shAR. (I-J) Genome browser view of ATAC-seq signals on the enhancers of AR target genes KLK2 and KLK3 (highlighted in light yellow). IDR deletion abolished the enhancer activation function of AR. See also Figures S1 and S2.

Figure 2.. Disrupting the multivalent interactions of AR compromises AR-mediated transcription.

(A-B) Representative images and quantification of LNCaP cells transiently transfected with GFP-ARwt and treated with indicated treatments. 1,6-HD treatment significantly disrupted DHT-induced AR foci formation. White boxes indicate the zoomed regions shown on the right. Scale bar: 10 μm. Percentages of cells showing AR foci in GFP-positive cells were plotted. Statistics: one-way ANOVA, ns: non-significant, ****P < 0.0001. (C-D) Heatmaps and aggregate plots of ATAC-seq signal on active AR enhancers derived from LNCaP cells treated with vehicle, DHT, DHT+1,6-HD, or DHT+2,5-HD. DHT treatment increased chromatin accessibility of AR enhancers and 1,6-HD inhibited chromatin opening in response to DHT. (E) Representative genome browser view of ATAC-seq signals on the enhancers of AR target genes KLK2 and KLK3 (highlighted in light yellow). (F-G) Heatmaps and aggregate plots of ATAC-seq signal on non-AR enhancers. DHT or 1,6-HD treatment did not affect the chromatin accessibility of non-AR enhancers. (H) Representative genome browser view of ATAC-seq signals on GAPDH gene locus as a negative control showing 1,6-HD did not affect the chromatin accessibility of a non-AR target gene. (I-L) Heatmaps and genome browser views of GRO-seq signals around the centers of active AR enhancer regions (I and J) or non-AR enhancer regions (K) or non-AR target gene GAPDH (L). DHT treatment promoted eRNA transcription on AR enhancers and this DHT effect was diminished by 1,6-HD treatment. AR enhancers were highlighted by light yellow color in (J). See also Figure S3.

Figure 3.. Aromatic residue mutation in NTD weakens multivalent AR-AR interactions and disrupts AR transcriptional activity.

(A) Representative droplet formation images of purified GFP-NTDwt and GFP-NTD7FS at indicated protein concentrations. Scale bar: 10 μm. (B) Quantification of the size of droplets formed by purified GFP-NTDwt and GFP-NTD7FS at indicated protein concentrations. Statistics: one-way ANOVA, ***P < 0.001. (C-D) Representative images and quantification of LNCaP cells transiently transfected with GFP-ARwt or GFP-AR7FS and treated with vehicle or DHT. Mutating the aromatic residues inhibited DHT-induced AR condensate formation. White boxes indicate the zoomed regions shown on the right. Scale bar: 10 μm. Percentages of cells showing AR foci in GFP-positive cells were plotted. Statistics: one-way ANOVA, ns: non-significant, ****P < 0.0001. (E) Schematic illustration of the LacO array system to test AR self-interactions. eYFP-AR-LacI is recruited to LacO array through protein-DNA binding and mCherry-AR can be recruited to the LacO array through AR-AR interactions. (F) Representative images of LacO array-containing U2OS cells co-expressing indicated proteins. Scale bar: 10 μm. (G) Quantification of mCherry-AR recruitment to the LacO hub through AR-AR self-association. Enrichment of mCherry above relative level of 1 suggests AR-AR self-association. Statistics: one-way ANOVA, ***P < 0.001. (H-I) Heatmaps and aggregate plots of ATAC-seq signal on active AR enhancers derived from LNCaP cells with indicated treatments. Exogenous expression of AR7FS failed to rescue the reduced chromatin accessibility on AR enhancers caused by shAR. (J) Genome browser view of ATAC-seq signals on the enhancers of AR target genes (highlighted in light yellow) showing that 7FS mutation abolished the enhancer activation function of AR. See also Figure S4.

Figure 4.. AR NTD can be functionally substituted by FUS and TAF15 IDRs, but not ERα IDR.

(A) Schematic illustration of ARwt, ARΔIDR, and swapped AR mutant proteins that were expressed in LNCaP cells for rescue experiments in Figures 4, S5, and S6. (B-F) Representative images of LNCaP cells transiently transfected with the indicated AR constructs and treated with vehicle or DHT. White boxes indicate the zoomed regions shown on the right. Scale bar: 10 μm. (G) Quantifications of LNCaP cells with AR condensates (top) and the fringe visibility of AR condensates (bottom). In the top panel, percentages of cells showing AR foci in GFP-positive cells were plotted. In the bottom panel, fringe visibility values of randomly selected AR foci from 6 cells (5 foci/cell) were plotted for each condition. Statistics: one-way ANOVA, ns: non-significant, *P < 0.05, ***P < 0.001, ****P < 0.0001. (H) Luciferase reporter assay to examine transcriptional activity of the indicated AR proteins and conditions. 293T cells were co-transfected with the luciferase reporter vector containing 3xARE and one of the AR constructs followed by luciferase activity measurement. Statistics: one-way ANOVA, ns: non-significant, ****P < 0.0001. See also Figures S5 and S6.

Figure 5.. PolyQ expansion leads to more stable AR condensates.

(A) Representative droplet formation images of purified GFP-NTDwt (pQ23) and GFP-NTDpQ69 at indicated protein concentrations. Scale bar: 10 μm. (B) Quantification of the size of droplets formed by purified GFP-NTDwt and GFP-NTDpQ69 at indicated protein concentrations. n.d.: non-detectable. Statistics: student’s t-test, *P < 0.05, ***P < 0.001. (C) Schematic diagram of droplets sedimentation and western-blot assays. (D) Representative image of the droplet sedimentation and western blot analyses. Supernatant (S) indicates the free proteins. Pellet (P) indicates proteins inside droplets. The top band represents GFP-ARNTD and the lower band represents a truncated GFP-ARNTD fragment. (E) Quantification of the ratio of protein in pellet. The intensity of the top band in panel (D) was measured using ImageJ and the pellet fraction ratio P/(S + P) was plotted. Statistics: one-way ANOVA, *P < 0.05. (F-G) Representative images and quantification of LNCaP cells transiently transfected with GFP-ARpQ69 and treated with vehicle, DHT, or DHT+1,6-HD. Scale bar: 10 μm. GFP-ARpQ69 foci were more resistant to 1,6-HD treatment. Statistics: one-way ANOVA, ns: non-significant, ***P < 0.001. (H-I) Representative images and quantification of Fluorescence Recovery After Photobleaching (FRAP) analyses on GFP-ARwt and GFP-ARpQ69 condensates formed in LNCaP cells in response to DHT treatment. Scale bar: 10 μm. Error bar: standard deviation. See also Figure S7 and Supplemental Videos 1 and 2.

Figure 6.. PolyQ length correlates with AR condensation propensity and PolyQ track expansion reduces AR transcriptional activity.

(A) Representative images of U2OS cells transfected with GFP-AR and Halo-ARNTD-NLS. Shown are individual frames from two-color movies of GFP-AR (green) and Halo-ARNTD-NLS labeled with 5 nM Halo ligand PA-JF646 (magenta). A white dashed line outlines the nucleus. Scale bar: 10 μm. (B) Schematic illustration of the AR proteins used in the single-particle tracking (SPT) analysis. The residence time of ARNTD in the puncta formed by their corresponding full-length variants (ARwt and ARpQ69, respectively) was measured. (C) The residence times of ARNTDwt and ARNTDpQ69 in the puncta formed by their corresponding full-length variants were plotted. The value for WT protein is averaged from 29 cells and its mutant from 26 cells measured in more than three independent transfection and imaging sessions. Asterisk indicates a significant difference between the two proteins (Wilcoxon rank-sum test). (D-E) Heatmaps and aggregate plots of ATAC-seq signals around the centers of active AR enhancers under the indicated conditions. ARpQ69 was not able to substitute for ARwt for AR function in activating AR enhancers. (F) Genome browser view of ATAC-seq signals on the enhancers of AR target genes KLK2 and KLK3 (highlighted by light yellow color) showing that polyQ track expansion abolished the enhancer activation function of AR. (G) Representative images of LNCaP cells transiently transfected with the indicated AR constructs and treated with vehicle or DHT. White boxes indicate the zoomed regions shown on the right. Scale bar: 10 μm. (H) Quantifications of LNCaP cells with AR condensates. PolyQ length positively correlates with the percentage of cells with AR condensates at 30 minutes post DHT treatment. Statistics: one-way ANOVA, ns: non-significant, *P < 0.05, **P < 0.01, ***P < 0.001. (I) Luciferase reporter assay to examine transcriptional activity of the indicated AR variants and conditions in 293T cells. PolyQ length negatively correlates with AR transcriptional activity in 293T cells. LNCaP cells were not used for the luciferase reporter assay due to the high endogenous AR expression level. Statistics: one-way ANOVA, **P < 0.01, ***P < 0.001. See also Figure S7 and Supplemental Videos 1 and 2.

Figure 7.. An optimal level of AR multivalent interactions is critical for the assembly of AR enhancer complex.

(A-B) Heatmaps and aggregate plots of AR ChIP-seq data showing the binding of ARwt, ARΔIDR, and ARpQ69 on active AR enhancers. ChIP-seqs were performed using BirA-BLRP LNCaP stable cell lines with endogenous AR knocked down and doxycycline-induced exogenous expression of BLRP-tagged AR proteins. (C) Genome browser view of ChIP-seq signals on the enhancers of AR target genes (highlighted by light yellow color) showing that IDR deletion abolished AR binding but polyQ track expansion did not affect AR binding on AR enhancers. (D) A list of enhancer complex components identified from our BioID analyses using ARwt, ARΔIDR or ARpQ69 as a bait protein. Normalized peptide numbers of each identified protein were listed for each experiment. (E) Representative images of LacO array-containing U2OS cells co-expressing indicated proteins. Scale bar: 10 μm. (F) Quantification of mCherry-MED1 recruitment to the LacO hub through AR-MED1 association. Statistics: one-way ANOVA, *P < 0.05, ***P < 0.001. (G) Schematic working model for AR enhancer assembly. An optimal level of AR-AR or AR-cofactor multivalent interactions mediated by wildtype AR IDR promotes the assembly of enhancer complex to activate transcription. Disturbing (e.g., ΔIDR or 7FS mutation) or overly-strengthening AR condensation (e.g., pQ69) leads to defective enhancer assembly and disrupted transcriptional activation. See also Figure S8.