The androgen receptor depends on ligand‐binding domain dimerization for transcriptional activation

Abstract

Whereas dimerization of the DNA‐binding domain of the androgen receptor (AR) plays an evident role in recognizing bipartite response elements, the contribution of the dimerization of the ligand‐binding domain (LBD) to the correct functioning of the AR remains unclear. Here, we describe a mouse model with disrupted dimerization of the AR LBD (AR^Lmon/Y). The disruptive effect of the mutation is demonstrated by the feminized phenotype, absence of male accessory sex glands, and strongly affected spermatogenesis, despite high circulating levels of testosterone. Testosterone replacement studies in orchidectomized mice demonstrate that androgen‐regulated transcriptomes in AR^Lmon/Y mice are completely lost. The mutated AR still translocates to the nucleus and binds chromatin, but does not bind to specific AR binding sites. In vitro studies reveal that the mutation in the LBD dimer interface also affects other AR functions such as DNA binding, ligand binding, and co‐regulator binding. In conclusion, LBD dimerization is crucial for the development of AR‐dependent tissues through its role in transcriptional regulation in vivo. Our findings identify AR LBD dimerization as a possible target for AR inhibition.

This study reveals the contribution of ligand‐binding domain (LBD) dimerization to androgen receptor (AR) activity. Disrupting LBD dimerization affects multiple receptor functions, proposing this interface as new therapeutic target.

Introduction

The androgen receptor (AR) is a nuclear receptor that belongs to the subfamily of steroid receptors. After binding of its cognate ligand, the AR translocates to the nucleus, binds DNA, and recruits co‐regulators and RNA polymerase II to initiate transcription (Aranda & Pascual, 2001). All steroid receptors have a conserved DNA‐binding domain (DBD), a ligand‐binding domain (LBD), and a variable N‐terminal domain (NTD) (Escriva et al, 2004). Steroid receptors assemble as homodimers to perform their classical role as transcription factors. For the AR and for other steroid receptors, dimerization via the DBD has been well characterized and is crucial for the recognition of the bipartite, inverted DNA repeats that act as androgen response elements (AREs) (Shaffer et al, 2004; Sahu et al, 2014). A second receptor dimerization mechanism is AR‐specific and is named the N/C interaction. It occurs through binding of the ²³FQNLF²⁷ motif in the NTD of the AR to the coactivator‐binding groove on the surface of the AR LBD (He et al, 2000, 2004; He & Wilson, 2002; Li et al, 2006; van Royen et al, 2007, 2012). The third dimerization mechanism, which happens via the LBDs, has been more controversial for the AR. The only evidence for AR LBD dimerization comes back from two‐hybrid protein–protein interaction assays (Doesburg et al, 1997). No further evidence became available, until Nadal et al (2017) published a unique crystal structure of AR LBD dimers. Although the dimerization surface differs from the one described for other nuclear receptors (Brzozowski et al, 1997; Tanenbaum et al, 1998; Williams & Sigler, 1998; Bledsoe et al, 2002; Nadal et al, 2017), biophysical data confirmed this dimerization interface (Nadal et al, 2017). However, the exact in vivo role of LBD dimerization in receptor functioning remains unknown and has not been studied for any steroid receptor.

Mutations in the AR LBD have been described in patients with androgen insensitivity syndrome (AIS). These patients have an intersex phenotype ranging from mild (e.g., normal male phenotype with infertility) to complete (female phenotype with undescended testes and absence of Wolffian and Müllerian ducts) AIS (Batista et al, 2018). Most AR LBD mutations can be correlated with reduced ligand and/or co‐regulator binding. However, some of these mutations are located within the LBD dimerization interface. Hence, the associated AIS phenotype might be explained by the fact that these mutations disrupt LBD dimerization. The W752R mutation, discovered in two siblings suffering from AIS (Boehmer et al, 2001), was predicted to disrupt the LBD dimer interface without strongly affecting other AR functions (Nadal et al, 2017). To study the physiological relevance of LBD dimerization for AR activity, we generated a mouse model (AR^Lmon for monomeric LBD of the AR) bearing the corresponding murine W731R point mutation.

Results

The W752R mutation disrupts AR LBD dimerization

In the center of the human AR (hAR) LBD dimer interface, Trp 752 is part of an important stabilizing π‐stack (Nadal et al, 2017). In the wild‐type (WT) LBD dimer, the two opposing tryptophans stabilize the interface via hydrogen bonds (Fig 1A). Molecular dynamics (MD) simulations were used to investigate the effect of W752 replacement by R or A on AR LBD dimer stability. Based on the model, tryptophan replacement by arginine (W752R) retains the possibility of hydrogen bond formation but induces an electrostatic repulsion of the positive charges and a steric clash due to the larger arginine side group (Fig 1A). Replacing the tryptophan by alanine (W752A) would disrupt the stabilizing π‐stack without adding electrostatic interactions. Since arginine could still form a hydrogen bond, it investigated whether it disrupts dimerization. For both W752R and W752A, MD simulations indicated that the interface was disrupted, with one protein bending away from the other (Fig EV1A–C). This is reflected in the calculated free energy of binding for the dimer interface shifting from −34.24 kJ/mol for the WT receptor to −19.38 kJ/mol for the W752R mutant and to −14.01 kJ/mol for the W752A mutant (Table EV1A and Fig EV1D and E). To study the influence of the mutations on the stability of the monomer–ligand interactions, we calculated the binding free energy of the AR LBD bound to dihydrotestosterone (DHT) in the presence of the AF2 crystal peptide. No differences in free energy were observed in both mutants, showing no major effect of the replacement on ligand‐induced receptor stabilization (Table EV1B). The W752R mutation was selected for further study. To confirm the effect of the W752R mutation on AR LBD dimerization in a cellular context, we performed acceptor‐bleaching fluorescence resonance energy transfer microscopy (abFRET) (van Royen et al, 2009). In the presence of 10 nM DHT, a clear abFRET signal was detected between two WT LBDs, whereas no abFRET signal occurred between two W752R LBDs (Fig 1B). A yeast two‐hybrid experiment confirmed the interactions between two WT LBDs, but not between two W752R LBDs (Fig EV1F).

Figure 1. Disrupting AR LBD dimerization.

• A
  Crystal structure of the human AR LBD core dimer based on PDB 5JJM (1 monomer in gray and 1 monomer in pink). W752, located at the interface of the AR LBD dimer, is involved in hydrogen bond formation with T756 of the neighboring LBD and thereby stabilizes the interface. R752 disrupt this stabilization by steric hindrance and charge repulsion. The side chains of both W and R are shown at position 752. A close‐up of the LBD‐LBD interface is given. The ligand (DHT) is depicted as spheres.
• B
  Acceptor photobleaching FRET after transfection of Hep3B cells with labeled WT LBD or labeled W752R‐mutated LBD. Representative confocal images of Hep3B cells transiently expressing WT or W752R AR in the presence of 10 nM DHT are shown below the bars. Scale bar = 10 µm. The bar graphs show means ± SEM (biological replicates, n = 54 (WT LBD) and n = 65 (W752R LBD), unpaired two‐tailed Student’s t‐test, ***P < 0.001).
• C
  Representative pictures of the AGD of 13‐week‐old WT male (upper left), WT female (lower left), AR^Lmon/Y (upper right), and AR^−/Y (lower right) mice.
• D
  Upper panel: a representative picture of the urogenital tract of a WT male and an AR^Lmon/Y mouse. Lower panel: a representative picture of the testis of a WT male and an AR^Lmon/Y mouse. Scale bar = 1 cm.
• E, F
  Serum levels of T (E) and LH (F) in WT males, and AR^Lmon/Y and AR^−/Y mice at the age of 13 weeks. The bar graphs show means ± SEM (biological replicates, n = 8, one‐way ANOVA with Tukey’s multiple comparisons test, **P < 0.01, ***P < 0.001, ns = not significant).

Figure EV1. Molecular dynamics of the AR LBD dimerization interface and yeast two‐hybrid assay.

• A
  Molecular dynamics simulations based on the WT AR LBD dimer crystal structure.
• B
  Molecular dynamics simulations based on the W752R AR LBD dimer model.
• C
  Superposed model of WT LBD dimer (pink) and W752R LBD dimer (white).
• D
  Molecular dynamics simulations based on the W752A AR LBD dimer model.
• E
  Superposed model of WT LBD dimer (pink) and W752A LBD dimer (purple).
• F
  Yeast two‐hybrid assay on human WT LBD and human W752R LBD. RFP signal normalized to 0 nM DHT is shown. The bar graphs show means ± SEM (biological replicates, n = 3, unpaired two‐tailed Student’s t‐test, ***P < 0.001, ns = not significant).

The AR^Lmon/Y mice have a feminized phenotype

The corresponding murine LBD dimer disrupting mutation, W731R, was introduced in the AR gene of C57BL/6J mice by CRISPR/Cas9 to generate AR^Lmon/Y males (Appendix Fig S1A and B). To estimate the importance of LBD dimerization in normal AR functioning, we compared the AR^Lmon/Y males with WT littermates and global AR knockout mice (AR^−/Y), in which the AR is no longer expressed (De Gendt et al, 2004). Longitudinal follow‐up of the anogenital distance (AGD), an external marker for sexual differentiation in mice (Schwartz et al, 2019), showed a notable difference between WT males and AR^Lmon/Y mice (Figs 1C and EV2A). The AGD of AR^Lmon/Y mice was up to three times reduced compared with WT males and comparable to that of WT females and AR^−/Y. Nipple development was clearly visible in AR^Lmon/Y and AR^−/Y mice, while it was absent in WT males. Body weight and composition of AR^Lmon/Y were comparable to those of WT females and AR^−/Y (Fig EV2B and C). The testes of AR^Lmon/Y mice were cryptorchid and had an intermediate weight between testes of WT males and AR^−/Y mice (Fig EV2D). Dissection of the urogenital tract revealed the absence of male reproductive organs including seminal vesicles, vas deferens, prostate, and epididymis in AR^Lmon/Y (Fig 1D). Kidneys of both AR^Lmon/Y and AR^−/Y mice weighed 16% less when compared to WT (Fig EV2E). Serum testosterone (T) and luteinizing hormone (LH) were increased by sixfold and 45‐fold in AR^Lmon/Y mice compared with WT males, respectively (Fig 1E and F). No significant increase was observed for follicle‐stimulating hormone (FSH) in AR^Lmon/Y mice (Fig EV2F).

Figure EV2. Detailed evaluation of the AR^Lmon/Y phenotype.

• A
  Evolution of the anogenital distance (AGD) over time. Average is shown, and shaded areas represent SEM (biological replicates, n ≥ 10).
• B
  Body weight followed over time. Average is shown, and shaded areas represent SEM (biological replicates, n ≥ 10).
• C
  Total amount of fat at 12 weeks of age determined by EchoMRI normalized to body weight. The bar graphs show means ± SEM (biological replicates, n ≥ 10, one‐way ANOVA with Tukey’s multiple comparisons test, *P < 0.05, ns = not significant).
• D
  Testes weight normalized to body weight of 13‐week‐old WT males, and AR^Lmon/Y and AR^−/Y mice. The bar graphs show means ± SEM (biological replicates, n = 8, one‐way ANOVA with Tukey’s multiple comparisons test, **P < 0.01, ***P < 0.001).
• E
  Kidney weight normalized to body weight of 13‐week‐old WT males, and AR^Lmon/Y and AR^−/Y mice. The bar graphs show means ± SEM (biological replicates, n = 8, one‐way ANOVA with Tukey’s multiple comparisons test, *P < 0.05, ns = not significant).
• F
  Serum levels of FSH in WT males, and AR^Lmon/Y and AR^−/Y mice at the age of 13 weeks. The bar graphs show means ± SEM (biological replicates, n = 8, one‐way ANOVA with Tukey’s multiple comparisons test, **P < 0.01, ns = not significant).
• G
  Upper panel: H&E staining on testis of a WT male, and AR^Lmon/Y or AR^−/Y mouse. Lower panel: immunofluorescence staining of the AR (green). Nuclei are shown in blue. Orange, blue, and white arrows indicate LC, SC, and peritubular myoid cells, respectively. Scale bar = 50 µm.
• H–I
  Relative contribution of seminiferous epithelium (H) and interstitium (I) in testis from 13‐week‐old mice. Proportions are expressed relative to total testis volume. The bar graphs show means ± SEM (biological replicates, n = 5, one‐way ANOVA with Tukey’s multiple comparisons test, ***P < 0.001, ns = not significant).

The testicular function is strongly affected in AR^Lmon/Y mice

The AR regulates cellular composition and histological appearance of the testis (Wang et al, 2009). To get further insight into the remaining activity of the AR^Lmon, we performed testicular analysis. Histological analysis of the testes of AR^Lmon/Y mice showed that the diameter of the seminiferous tubules was smaller compared with that of WT males, with fewer spermatogenic cells (Fig EV2G, upper panel). Nevertheless, in comparison with AR^−/Y mice, seminiferous tubules of AR^Lmon/Y mice had a larger cross‐sectional area with further developed spermatogenic cells (Fig EV2G, upper panel). Seminiferous epithelium was decreased in AR^Lmon/Y and AR^−/Y mice (Fig EV2H). In contrast, the interstitium in the testes of AR^Lmon/Y and AR^−/Y mice was significantly increased compared with WT testes (Fig EV2I). Immunofluorescence staining showed nuclear AR expression in both WT and AR^Lmon/Y testes, whereas AR was not detected in testis from AR^−/Y mice (Fig EV2G, lower panel). AR‐positive Leydig cells (LC), Sertoli cells (SC), and peritubular myoid cells were present in both WT and AR^Lmon/Y testes, with striking hyperplasia of the androgen‐producing LC in the AR^Lmon/Y (Table 1; Fig EV2G, lower panel). A higher percentage of spermatogenesis‐supporting SC was observed in the AR^Lmon/Y and AR^−/Y testes, although the absolute nuclear volume of SC per testis was reduced in both genotypes compared with WT (Table 1). Further examination of the histology of AR^Lmon/Y testes uncovered the patchy presence of differentiated elongated spermatids (ESd), albeit at very low numbers as illustrated in Table 1. Quantification of the spermatogenic cells confirmed that the relative numbers of spermatogonia (Sg) were increased in AR^Lmon/Y testes, while the percentage of the specific spermatogenic cell types decreased the further they develop toward the elongated stage (Table 1). Measurement of intratesticular T concentrations showed a significant increase in androstenedione (A‐dione), T, and DHT in AR^Lmon/Y mice compared with WT (Appendix Fig S2A–C). To gain further insight into their testicular phenotype and the expression of AR‐regulated genes, we performed RNA‐sequencing (RNA‐seq) analysis on whole testis of five mice per genotype: WT, AR^Lmon/Y, and AR^−/Y. Comparing gene expression between AR^Lmon/Y or AR^−/Y mice and WT males confirmed the more severe phenotype of AR^−/Y testes. Indeed, compared with WT, more genes were up‐ and downregulated in AR^−/Y mice and with a higher magnitude than in AR^Lmon/Y testes (Fig 2A and B). A large part of the differentially expressed genes is specific to the spermatogenic lineage. The comparison of SC‐ and LC‐specific transcripts extracted from the bulk RNA‐seq data showed clear relative differences between WT, AR^Lmon/Y, and AR^−/Y (Fig 2C). In general, relative expression levels of SC‐ and LC‐specific genes tended to be higher in AR^Lmon/Y mice compared with WT, corresponding to the higher percentage of SC and LC and fewer germ cells in AR^Lmon/Y testes. However, several well‐known AR‐regulated genes, such as the SC‐specific Rhox5 and the LC‐specific Insl3, were almost completely silenced in AR^Lmon/Y. This was confirmed by qPCR (Fig 2D). Furthermore, steroidogenic enzyme expression was dysregulated in AR^Lmon/Y testes. Indeed, the expression of genes encoding enzymes involved in the uptake of cholesterol (Star), the conversion of cholesterol into pregnenolone (Cyp11a1), and the hydroxylation of pregnenolone (Cyp17a1) were higher compared with WT (Fig 3A, confirmed by qPCR in Appendix Fig S2D). Hsd3b1, which is responsible for the formation of A‐dione, was also significantly higher expressed in testes of AR^Lmon/Y (Fig 3A), which correlated with the higher LH and A‐dione serum levels (Figs 1F and 3B). Malfunctioning of the AR in the hypothalamic–pituitary–gonadal (HPG) axis also led to increased expression levels of the LH receptor (LHR) in testes of both AR^Lmon/Y and AR^−/Y mice (Appendix Fig S2E). Surprisingly, the expression level of Hsd17b3, which is responsible for the final conversion of A‐dione into T, was found to be fivefold lower in AR^Lmon/Y than in WT (Fig 3C, confirmed by qPCR in Appendix Fig S2F) and even 13‐fold lower in AR^−/Y mice. Because the AR is a transcription factor, we performed co‐transfections of AR with a luciferase reporter under the control of the Hsd17b3 promoter. We selected the region ranging from −230 bp to +1 bp because it is most conserved between mammalians. The reporter construct was androgen‐responsive when co‐transfected with a WT mouse AR (mAR), but not in the presence of the W731R mAR (Fig 3D). Similarly, the Insl3 promoter (–132 to +1 bp) also conferred androgen responsiveness to a luciferase reporter gene, which was disrupted by the introduction of the AR^Lmon mutation (Appendix Fig S2G). Based on these observations, we hypothesize a positive feedback loop between AR activity and A‐dione to T conversion by HSD17B3 in WT testes, and postulate that the AR^Lmon mutation might interrupt this feedback mechanism (Fig 3E).

Table 1.

Relative (%) and absolute (mm3) testicular cell‐type quantification.

Genotype	LC (%)	SC (%)	Total GC (%)	Sg (%)	Sc (%)	RSd (%)	ESd (%)
WT	4.74 ± 0.86#	2.00 ± 0.11#	26.96 ± 2.58#	0.76 ± 0.05#	12.9 ± 1.33#	8.03 ± 0.94	5.26 ± 0.7
ARLmon/Y	19.37 ± 2.03*	5.54 ± 0.68*,#	13.59 ± 1.85*,#	2.14 ± 0.17*,#	10.34 ± 1.31*,#	0.95 ± 0.75*	0.16 ± 0.16*
AR−/Y	18.35 ± 0.96*	9.02 ± 0.52*	5.14 ± 0.91*	2.92 ± 0.28*	2.22 ± 0.73*	Absent	Absent

Genotype	LC (mm3)	SC (mm3)	Total GC (mm3)	Sg (mm3)	Sc (mm3)	RSd (mm3)	ESd (mm3)
WT	4.30 ± 0.75#	1.81 ± 0.08#	24.41 ± 2.21#	0.69 ± 0.04#	11.68 ± 1.14#	7.26 ± 0.79	4.78 ± 0.65
ARLmon/Y	4.07 ± 0.22#	1.16 ± 0.11*,#	3.28 ± 0.92*,#	0.47 ± 0.05*,#	2.46 ± 0.63*,#	0.31 ± 0.26*	0.05 ± 0.05*
AR−/Y	0.91 ± 0.14P*	0.46 ± 0.10*	0.27 ± 0.09*	0.15 ± 0.04*	0.12 ± 0.05*	Absent	Absent

Upper panel: percentage of the different cell types in testis from 13‐week‐old mice (biological replicates, n = 5). All percentages were acquired by normalizing the cell counts of each specific cell type to the total cell counts in each testis. Lower panel: Absolute nuclear volume of cell types in testis from 13‐week‐old mice (biological replicates, n = 5). Values are average ± SEM. LC: Leydig cells; SC: Sertoli cells; GC: germ cells; Sg: spermatogonia; Sc: spermatocytes; RSd: round spermatids; ESd: elongated spermatids. One‐way ANOVA with Tukey’s multiple comparisons test was performed. *Statistical significance compared with WT: P < 0.05. ^#Statistical significance compared with AR^−/Y: P < 0.05.

Figure 2. Transcriptome analysis of testes.

• A, B
  Volcano plots visualizing the differentially expressed genes in AR^Lmon/Y (A) and AR^−/Y (B) testes compared with testes from WT males. The dotted horizontal lines represent a q‐value of 0.05, while the dotted vertical lines represent a FC of 1.5. Genes that are downregulated compared with WT are shown using blue dots, while red dots represent the upregulated genes. Hsd17b3 and Star are indicated as yellow and green stars, respectively.
• C
  Box plots showing testicular expression from the bulk RNA‐seq data (biological replicates, n = 5) of SC‐ and LC‐specific genes extracted from the single‐cell RNA‐seq data described by Green et al (2018). Box plots are composed of a box from the 25th to 75th percentile with the median as a line and whiskers calculated via the Tukey method. Expression levels are normalized to average expression in WT for each gene. Two‐way ANOVA with Tukey’s multiple comparisons test, ***P < 0.001.
• D
  RT–qPCR analyses of Rhox5 and Insl3 of testes from 13‐week‐old WT males, and AR^Lmon/Y and AR^−/Y mice. Expression levels are normalized to WT. The bar graphs show means ± SEM (biological replicates, n = 6, one‐way ANOVA with Tukey’s multiple comparisons test, ***P < 0.001, ns = not significant).

Figure 3. Steroidogenesis in AR^Lmon/Y and AR^−/Y mice.

• A
  Expression levels of genes involved in steroidogenesis normalized to WT. The bar graphs show means ± SEM (data derived from RNA‐seq analysis of testes, biological replicates, n = 5, one‐way ANOVA with Tukey’s multiple comparisons test, *P < 0.05, **P < 0.01, ***P < 0.001, ns = not significant).
• B
  Serum levels of A‐dione in WT males, and AR^Lmon/Y and AR^−/Y mice at the age of 13 weeks. The bar graphs show means ± SEM (biological replicates, n = 8, one‐way ANOVA with Tukey’s multiple comparisons test, ***P < 0.001, ns = not significant, n.d. = not detected; below detection limit of 0.175 nmol/L).
• C
  Expression levels of Hsd17b3 normalized to WT. The bar graphs show means ± SEM (data derived from RNA‐seq analysis of testes, biological replicates, n = 5, one‐way ANOVA with Tukey’s multiple comparisons test, **P < 0.01, ***P < 0.001).
• D
  Reporter gene assays in HeLa cells co‐transfected with the mouse Insl3 promoter together with the WT or mutant mouse AR. The bar graphs show means ± SEM (biological replicates, n = 4, unpaired two‐tailed Student’s t‐test, **P < 0.01, ns = not significant).
• E
  Schematic overview of the feedback mechanism between testes and pituitary in the transcriptional regulation of steroidogenic genes.

The androgen‐regulated transcriptome is lost in kidneys of AR^Lmon/Y mice

In search of remaining androgen responses in the AR^Lmon/Y model, we investigated the AR^Lmon/Y kidneys. Kidney is indeed a well‐documented androgen‐sensitive organ (Pihlajamaa et al, 2014; Khalil et al, 2020). At 13 weeks of age, AR^Lmon/Y and WT littermates were orchidectomized and implanted with either T‐releasing (ORX + T) or empty (ORX) silastic sticks (Fig 4A). The resulting serum T values and the androgen effects on seminal vesicles were measured to validate the experimental approach of orchidectomy and T treatments (Appendix Fig S3A and B). RNA‐seq analysis was performed on kidneys collected 4 days after surgery (Fig 4A). Principal component analysis showed that the WT ORX + T group forms a separate cluster from all other groups (Fig 4B). Differential analysis revealed 1,221 genes that were differentially expressed between the ORX + T group and the ORX group in WT males (q < 0.05; FC > 1.5) (Fig 4C), whereas no genes were differentially regulated by T in orchidectomized AR^Lmon/Y mice (q < 0.05; FC > 1.5). Detailed examination of three well‐known androgen‐regulated genes, Fkbp5, Odc1, and Kap (Fig 4D–F, confirmation by qPCR in Appendix Fig S3C), confirmed that there were no remaining androgen responses in the AR^Lmon kidney. Expression levels of the AR were comparable between the four groups (Appendix Fig S3D and E).

Figure 4. Evaluation of the renal androgen response after disrupting AR LBD dimerization.

• A
  The experimental set‐up to study the androgen response in kidney.
• B
  PCA on the renal transcriptomes of WT ORX, WT ORX + T, AR^Lmon/Y ORX and AR^Lmon/Y ORX + T.
• C
  Heatmap for the genes that are differentially expressed in kidney between WT ORX and WT ORX + T (q < 0.05; FC > 1.5) and corresponding levels in AR^Lmon/Y ORX and AR^Lmon/Y ORX + T.
• D–F
  mRNA expression levels extracted from RNA‐seq data for Fkbp5 (D), Odc1 (E), and Kap (F) in kidneys used for RNA‐seq. Expression levels are normalized to WT ORX. The bar graphs show means ± SEM (biological replicates, n = 5, one‐way ANOVA with Tukey’s multiple comparisons test, **P < 0.01, ***P < 0.001, ns = not significant).

LBD dimerization is important for normal ligand and DNA binding of AR in vitro

In search of a mechanism behind the fact that LBD dimerization is crucial for AR functioning, we investigated which steps in the transcription activation process are affected. Reporter gene assays showed that both WT and mutated hAR (W752R) are active upon DHT stimulation (Fig 5A). However, a shift of the EC50 of DHT from 0.23 nM for AR WT to 5.50 nM for AR W752R was observed. The analogous mutation in mAR resulted in a similar transactivation profile. Indeed, at 10 nM DHT the mAR W731R and the hAR W752R mutant reached transcription activation levels comparable to those of the WT mAR and hAR, respectively (Appendix Fig S4A). We confirmed similar expression levels of WT and W752R via Western blot, indicating no clear effect of the mutation on AR protein steady‐state level (Appendix Fig S4B). Ligand‐binding assay showed comparable B_max values for WT and mutated AR. However, the K_d of AR^Lmon for DHT was higher than WT hAR (Fig 5B). Together with the MD predictions on the influence of the W752R mutation to the binding of DHT (Table EV1B), this binding assay suggests that ligand binding is influenced by the disruption of LBD dimerization. Interestingly, however, Western blot on both cytoplasmic and nuclear extracts confirmed nuclear translocation of both hAR W752R and mAR W731R (Fig 5C and Appendix Fig S4C, respectively). In vitro DNA binding by electrophoretic mobility assay (EMSA) showed slightly reduced binding of the AR^Lmon compared with WT AR (Fig 5D).

Figure 5. In vitro analysis of the W752R mutation.

• A
  Transactivation assay in HEK293 cells with an integrated 4xSLP‐HRE2 E1B 12 TATA Luc reporter and transiently transfected with human WT or W752R AR followed by stimulation with increasing DHT concentrations. The bar graphs show means ± SEM (biological replicates, n = 4, two‐way ANOVA with Sidak’s multiple comparisons test, ***P < 0.001, ns = not significant).
• B
  Specific ligand‐binding assay in COS cells with overload of cold ligand and increasing concentrations of radioactive labeled DHT. Values are normalized to protein expression. Individual data points are shown with nonlinear regression curve fit (n = 2).
• C
  Western blot on cytoplasmic, nuclear, and whole‐cell extracts derived from HeLa cells transfected with human WT or W752R AR followed by stimulation with increasing concentrations of DHT. Both panels represent the same blot. The blot was cut into two parts, and antibodies against HSP90 and Lamin A/C were used to confirm cellular fractionation of cytoplasmic and nuclear proteins. For AR visualization (upper panel), longer exposure time was used. # = residual AR expression (lower panel).
• D
  EMSA using nuclear extracts of COS‐7 cells transfected with AR WT or AR W752R and incubated with radiolabeled ARE, more specifically TAT‐GRE (Denayer et al, 2010). Shift (S) occurred through binding of the AR dimer on the ARE. Supershift (SS) occurred after addition of an AR antibody. S = shift and U = unbound ARE. Western blot on AR, depicted on the right, shows similar expression levels for WT and mutated receptors.

Source data are available online for this figure.

Chromatin binding of AR is reduced in AR^Lmon/Y mice

The changes in in vitro DNA binding prompted us to investigate chromatin binding in vivo. Western blot on chromatin fractions extracted from WT and AR^Lmon/Y kidneys showed the presence of AR in both genotypes, confirming nuclear translocation in the AR^Lmon/Y mice. Chromatin binding was, however, reduced in the AR^Lmon/Y kidneys (Fig 6A). We complemented this approach by performing chromatin immunoprecipitation (ChIP) on kidneys from AR^Lmon/Y and WT littermates using validated antibodies (Appendix Fig S5A–K). ChIP‐qPCR on specific ARBS (Pihlajamaa et al, 2014) of known AR‐regulated genes in the kidney (Fkb5 and Kap) showed that disrupting LBD dimerization led to loss of chromatin binding of the mutated AR (Fig 6B and Appendix Fig S6A–C). This correlated with a diminished expression of the Fkbp5 and Kap genes in AR^Lmon/Y kidneys together with a loss of the active histone mark H3K27ac at their regulatory regions (Fig 6B and Appendix Fig S6D). Conversely, the repressive histone mark H3K27me3 showed a tendency to increase in Fkbp5 and Kap regulatory regions of AR^Lmon/Y kidneys (Fig 6B and Appendix Fig S6E). AR staining on kidneys of castrated WT and AR^Lmon/Y mice, supplemented with either vehicle or supraphysiological T (Fig 4A), showed complete nuclear translocation of the AR in WT males, while the mutated AR is present in both nucleus and cytoplasm upon stimulation with T (Fig 6C). This finding is in line with the in vitro nuclear translocation assay, where the DHT‐induced nuclear translocation of WT AR is completed at a lower concentration of DHT (10 nM) than that of AR W752R (100 nM) (Appendix Fig S4D).

Figure 6. Chromatin binding of the AR^Lmon .

• A
  Western blot on whole‐cell extracts and chromatin fractions extracted from kidneys of WT males and AR^Lmon/Y mice. Antibodies against GAPDH and histone 3 were used to confirm cellular fractionation. Relative intensities are indicated, whereby the upper bands (AR) are normalized to the lower bands (GAPDH or histone 3). Both panels represent the same blot. The blot was cut into two parts, and a longer exposure time was used for AR visualization (upper panel).
• B
  AR occupancy (blue) and levels of the histone marks H3K27ac (green) and H3K27me3 (red) at ARBS within the regulatory regions of Fkbp5, Kap (both androgen‐regulated in kidney), and Tox3 (androgen‐regulated in prostate but not in kidney and hence used as a negative control) were assessed by ChIP‐qPCR in kidneys from 13‐week‐old AR^Lmon/Y mice and WT males. The bar graphs show means ± SEM (biological replicates, n = 6; one‐sample t‐test with the Benjamini–Hochberg correction for multiple testing was used to determine whether a mean Log₂ FC was statistically different from 0, *P < 0.05, **P < 0.01, ***P < 0.001). Fold enrichments are shown in Appendix Fig S6C–E.
• C
  Representative pictures of immunofluorescence AR staining (green) in kidneys of castrated WT males and AR^Lmon/Y mice supplemented with vehicle or T. Nuclei are shown in blue. Scale bar = 20 µm.

Source data are available online for this figure.

LBD dimerization influences the hormone‐dependent interactome of AR

In a mammalian double‐hybrid assay, we subsequently tested for the interaction of the W752R LBD with an LxxLL peptide, derived from steroid receptor coactivator‐1 (SRC1). Even at high DHT concentrations (100 nM), the binding is clearly reduced when compared to WT LBD (Fig 7A). To determine whether LBD dimerization affects the agonist‐induced protein interactions of AR in intact cells, we performed a proximity‐dependent biotin identification (BioID) assay with receptors fused to the mutated E. coli biotin ligase (BirA*) that covalently attaches biotin to primary amines within 10‐nm range (Uetz et al, 2000). To that end, we generated tetracycline‐inducible cell lines expressing the WT AR or AR W752R as C‐terminal fusions to BirA* and first checked the cellular localization of the fusion proteins by confocal microscopy. In the absence of DHT, both the WT and the mutated AR‐BirA* fusions reside in the cytoplasm. Upon DHT stimulation, transfer of AR W752R‐BirA* to the nucleus appears less complete than that of WT AR‐BirA* (Fig 7B), which is in line with the previous findings (Fig 6C). Furthermore, confocal imaging showed that the cellular localizations of the biotinylated proteins match that of the BirA*‐fused proteins. Biotinylated proteins from three biological replicates were affinity‐purified with streptavidin and detected by mass spectrometry (MS). BirA*‐tagged eGFP was used to control unspecific interactions, and Significance Analysis of INTeractome (SAINT) (Choi et al, 2011) was utilized to determine the statistical significance of the detected interactions by setting BirA*‐EGFP as the control. Comparison of the thereby obtained interactome of the WT AR with that of the AR W752R showed that the W752R mutation significantly weakens the interaction with twelve high‐confidence interactor proteins of WT AR, including four subunits of SWI/SNF (BAF) chromatin remodeling complex (ARID1A, ARID1B, SMARCE1, and SMARCA4) and lysine demethylases KDM1A (LSD1) and JMJ1DC (Fig 7C). Since only two cytoplasmic proteins (DBT and GREPL1) were considered high‐confidence interactors of AR W752R‐BirA* (Fig 7C), we compared all proteins in WT AR‐ and AR W752R‐BirA* samples showing ≥twofold stronger signal than in EGFP controls and ≥ twofold induction of their signals by DHT to reveal more subtle differences between the two AR forms. The latter comparison revealed weaker interaction of AR W752R with several other transcriptional co‐regulators, such as NCOA6, NCOR1, NCOR2, lysine methyltransferase KMT2D (MML4), and ATP‐dependent chromatin remodeler CHD7 (Dataset EV1). The markedly altered interactome of the AR W752R is in line with the reduced interactions with the LXXLL motif in the double‐hybrid assay (Fig 7A) on the one hand and with the severe phenotype observed in the AR^Lmon/Y mice on the other.

Figure 7. Comparison of WT AR and AR W752R interactomes.

• A
  Double‐hybrid assay in COS cells transfected with WT LBD, W752R LBD, or empty vector (negative control) in the presence of the LxxLL fragment of SRC1. Average is shown, and shaded areas represent SEM (biological replicates, n = 4, Tukey’s multiple comparisons test, **P < 0.01, ***P < 0.001).
• B
  Confocal fluorescence microscopy images of AR‐WT‐BirA* and AR W752R‐BirA*expressing HEK293 cells treated in the presence of 50 µM biotin with 100 nM DHT or vehicle (ethanol) and with or without 0.03 µg/ml TET as indicated. AR‐BirA*s were detected with anti‐AR (red) and biotinylated proteins with fluorescently labeled streptavidin (green). Nuclei were visualized using DAPI. Scale bar = 20 µm.
• C
  Heatmap showing the MS spectral counts of high‐confidence interactors (FDR < 0.05 after SAINT analysis) identified with WT AR‐BirA* and AR W752R‐BirA* upon DHT or vehicle exposure. Values for three biological replicates from DHT‐ and vehicle‐exposed samples are shown. GRPEL1 and DBT were the only high‐confidence interactors of AR W752R‐BirA*. On the right, FDRs and spectral count averages are shown for the DHT‐treated samples. Spectral counts have been normalized to those of AR in each sample.

Discussion

Like for most nuclear receptors, dimerization of the AR is considered crucial for its canonical functioning as transcription factor via AREs (Claessens et al, 2008). The best studied nuclear receptor dimer interface is the second zinc finger of the DBD (Shaffer et al, 2004), but the AR contains an additional N/C interaction, which could serve as dimer interface (He et al, 2004). A crystal structure of the dimeric AR LBD revealed the details of a third dimer interface. However, the exact role of LBD dimerization in AR functioning remained unknown (Hay & McEwan, 2012; Nadal et al, 2017). Here, we introduced a mutation (hW752R/mW731R) that disrupts the LBD dimerization without major effect on the LBD 3D structure (Nadal et al, 2017). Although LBD dimerization is well accepted for most nuclear receptors (Billas & Moras, 2013; Louw, 2019), its in vivo contributions to receptor functioning has never been proven. To determine the physiological relevance of AR LBD dimerization, we introduced this mutation by CRISPR/Cas9 in mice.

The AR^Lmon/Y mice present an external female phenotype, based on body composition, nipple development, and AGD. This phenotype is reminiscent of human AIS, of rodent testicular feminization models, and of global AR knockout mouse models (Brown, 1995; Galani et al, 2008; Kerkhofs et al, 2009). Moreover, the AR^Lmon/Y mice have absence of the Wolffian duct derivatives (epididymis, seminal vesicles, and prostate), but the presence of intra‐abdominal testes. This is expected since the initial development of testes is AR‐independent, while other secondary sex organs are AR‐dependent (Parker et al, 1999). The severity of the AR^Lmon/Y phenotype demonstrates the dependence of the AR on its LBD dimerization for its proper in vivo functioning. Moreover, the AR^Lmon/Y model seems to phenocopy the corresponding human AIS mutation, although a detailed analysis of the phenotype of the affected siblings is lacking (Brinkmann et al, 1995; Boehmer et al, 2001). Surprisingly, the AR^Lmon/Y animals show sixfold higher serum T levels compared with WT males, which is also much higher than in AR^−/Y mice. As expected, these high T levels are unable to act on the HPG axis due to malfunctioning of the AR in pituitary and hypothalamus, explaining the high LH levels. These increased LH levels stimulate steroidogenesis in the LC. RNA‐seq from the testes shows that all enzymes that act upstream of HSD17B3 are strongly upregulated, which is a known response to high LH levels. HSD17B3 is the last enzyme in the synthesis of T and responsible for the conversion of A‐dione into T. In AR^Lmon/Y testis, this final step is affected, resulting in an accumulation of A‐dione. Similar to AR^Lmon/Y, it was recently shown that a HSD17B3 knockout mouse model also has very high circulating A‐dione levels, in combination with elevated serum T values, indicating that other enzymes can compensate for HSD17B3 in T synthesis (Rebourcet et al, 2020; Sipilä et al, 2020). In contrast, the disruption of the HPG axis in the AR^−/Y mice leads to a more severe testicular phenotype: while LH levels are increased compared with WT males, T synthesis remains low although the defective LC cells still have some response as evidenced by the circulating A‐dione (De Gendt et al, 2004) and (Fig 3B). Reporter assays revealed that the Hsd17b3 promoter is androgen‐responsive and that AR^Lmon is unable to induce its activity. In conclusion, our results show that androgens regulate their own levels by controlling the final step in the androgen synthesis (Fig 3D). Our data clearly demonstrate that disrupting AR LBD dimerization prevents sexual development, disrupts the HPG axis, and results in a near‐complete arrest of spermatogenesis. The fact that sperm cells of later stages are found in the AR^Lmon/Y mice while not in the AR^−/Y mice might indicate some remaining activity of AR^Lmon in the testis, possibly due to the high testicular T levels. This is also indicated by the residual expression of Rhox5 and Insl3.

To study how the mutation affects AR function, we studied the androgen response in the kidneys of the AR^Lmon/Y mice. Androgens have no effect on the transcriptome of AR^Lmon/Y kidney, which strongly indicates that the mutation completely inactivates the canonical actions of the receptor. However, this is in stark contrast with the AR activity in reporter assays. Indeed, mAR W731R and hAR W752R are still able to activate reporter genes in vitro. Ligand‐binding assay showed an increase in Kd of the mutated AR; however, nuclear translocation still occurred at low DHT concentration (10 nM) in vitro and was complete at high concentrations (100 nM, Fig 5C). Surprisingly, DNA binding was attenuated by the AR W752R mutation. Although DBD and LBD are known to be able to function independently, functional communication between these two domains has been postulated before (Chandra et al, 2008; Helsen et al, 2012). The EMSA data suggest that AR^Lmon should still be able to interact with DNA in vivo. Western blot indeed showed the presence of the AR^Lmon in the chromatin fraction of kidneys derived from AR^Lmon/Y mice. However, in ChIP assays on kidneys from ORX+T‐treated AR^Lmon/Y mice, the AR is unable to bind ARBS, suggesting a defect in the occupation of the chromatinized androgen‐responsive enhancers when LBD dimerization is lost. Furthermore, mapping of the protein interaction landscape of AR by BioID revealed that disruption of LBD dimerization severely compromises interaction with transcriptional co‐regulators, including components of the chromatin remodeling SWI/SNF complex. Our data thus revealed that a mutation that disrupts AR LBD dimerization contributes to multiple processes during the transcription activation process, including ligand binding, nuclear translocation, DNA binding, co‐regulator recruitment, and ultimately enhancer binding.

In conclusion, the AR^Lmon/Y mouse phenotyping shows that the disruption of the AR LBD dimerization largely inactivates the AR. This is in contrast to the in vitro data on this mutant receptor, which binds ligand, translocates to the nucleus, and even activates androgen reporter genes at the androgen levels in the circulation of AR^Lmon/Y mice. The inability of the AR^Lmon to transactivate genes in the kidneys is most likely explained by its inability to bind ARBS in the chromatin and to a reduced interaction with some of the co‐regulator complexes. As a consequence, our study points at the LBD dimer interface as a target for the development of a new type of AR inhibitors. Such inhibitors could be of use in advanced prostate cancer, resistant against the current AR antagonists that target the ligand‐binding pocket.

Materials and Methods

Generation of transgenic mice

The AR^Lmon/Y mice were generated in the Transgenic Mouse Core Facility (VIB, Ghent) (Appendix Fig S1A). Cas9 mRNA and protein (Sigma) were microinjected together with a cr/tracrRNA duplex (5’‐AGTGAAGGACCGCCAACCCA‐3’, IDT) into the pronucleus of fertilized C57BL/6J oocytes. The single‐stranded DNA oligo with sequence 5’‐GATGACCAGATGGCGGTCATTCAGTATTCCTGGATGGGACTGATGGTATTTGCCATGGGTCGACGGTCCTTCACTAATGTCAACTCCAGGATGCTCTACTTTGCACCTGACTTGGTTTTCAAT‐3’ containing the Trp (TGG) to Arg mutation (CGA) was co‐injected. Introduction of the mutation generated a diagnostic SalI restriction site for genotyping purposes. The knock‐in founder mouse (only one out of 24 try‐outs) generated two heterozygous offspring (AR^Lmon/+) (Appendix Fig S1B), of which one was crossed with WT C57BL/6J males until the third generation. The AR gene was resequenced to confirm the presence of the W731R mutation and absence of other mutations. AR^−/Y mice were generated by De Gendt et al (2004).

Animal care and PCR genotyping

Mice were group‐housed in the Animal Housing Facility of the KU Leuven under constant temperature, humidity, and day/night cycle. They had ad libitum access to tap water and a standard diet (1% calcium, 0.76% phosphate). All procedures were approved by the Animal Ethical Committee of the KU Leuven (P197/2017). Mice were identified via PCR with an appropriate primer pair flanking the region of interest (FW: 5’‐CAGCCCCACCATTCAGACTT‐3’; REV: 5'‐GACTGACAGAAGTCCCCAGC‐3') on genomic DNA extracted from ears, resulting in a band of 384 bp. Heterozygote females were distinguished from WT females after restriction with SalI (Thermo Fisher Scientific). To identify AR^Lmon/Y mice, the presence of the Y chromosome was confirmed by PCR amplification of the Zfy gene (Kunieda et al, 1992).

Surgical procedures

To determine androgen response in kidney, a cohort of AR^Lmon/Y mice and WT littermates underwent orchidectomy (ORX; abdominal approach) under isoflurane anesthesia at 13 weeks of age. Mice were subcutaneously implanted with medical‐grade silicone tubing (Silclear) sealed with medical adhesive silicone (Silastic) in the nuchal region. The silicone tubes were either filled with T (Sigma‐Aldrich) or remained empty. As previously described, a silastic implant of 1 cm delivers a daily dose of 23 µg T in vivo, which is supraphysiological (Vanderschueren et al, 2000). Mice were sacrificed after 4 days by cardiac puncture (serum collection), and kidneys were harvested for RNA isolation. The effectiveness of hormone replacement was verified by measuring serum T values in both groups (Appendix Fig S3A), and by weighing of the androgen‐responsive seminal vesicles of WT littermates (not present in AR^Lmon/Y; Appendix Fig S3B).

Whole‐body analysis

Body weight, AGD, and body composition of AR^−/Y, AR^Lmon/Y, WT male, and WT female littermates were evaluated every two weeks until the age of 16 weeks. AGD was measured using an automatic caliper. Body composition was determined by quantitative magnetic resonance (EchoMRI‐100H Analyzer; Echo Medical Systems LLC, Houston, TX, USA) according to the manufacturer’s instructions. Based on the body weight, the absolute amount of body fat was converted into percentage of fat.

Organ and serum collection

Mice were sacrificed at 13 weeks of age via cardiac puncture followed by dissection of the urogenital tract. Organs were weighed immediately after removal. Organs were either snap‐frozen in liquid nitrogen for extraction of DNA, RNA, and proteins via NucleoSpin TriPrep column purification (Macherey‐Nagel) or fixated for histochemical analysis. Testes were fixated in Bouin’s fluid at 4°C overnight and subsequently transferred to 70% ethanol for storage at 4°C. Kidneys were fixed in 2% PFA at 4°C overnight and subsequently transferred to PBS for storage at 4°C. Coagulated blood was centrifuged at 130,000 g for 10 min, and serum was stored at −20°C until steroid measurements.

Hormone measurements

Total serum levels of A‐dione and T were measured at the Leuven University Hospital by LC‐MS/MS without derivatization using a two‐dimensional liquid chromatography system and a 5500 tandem mass spectrometer in atmospheric pressure chemical ionization‐positive mode as previously described (Pauwels et al, 2013). LH and FSH were determined by ELISA and RIA, respectively, at the University of Virginia by the Center of Research in Reproduction, Ligand Assay, and Analysis Core (Charlottesville, VA, USA). Intratesticular steroid concentrations were determined by GC‐MS/MS (Nilsson et al, 2015).

Histochemical techniques and testicular cell counts

Fixed testes and kidneys were embedded in paraffin wax using standard procedures, and 5‐µm sections were stained with hematoxylin/eosin. Immunofluorescence staining was performed with antibodies directed against the AR (SP107, Abcam). Antigen was retrieved by incubation in 0.01 M citrate buffer using a pressure cooker. Bound antibodies were visualized using a Tyramide Signal Amplification Kit (PerkinElmer; fluorescein for AR), followed by nuclear counterstaining with DAPI (4’,6‐diamidino‐2‐phenyl‐indole dihydrochloride). Samples incubated without primary antibody were used as negative controls. Images were captured using a slide scanner microscope Axio Observer (Zeiss). Cell counts were performed on testis cross‐sections stained by hematoxylin/eosin as previously described (Tan et al, 2005). In brief, cross‐sections of testes were examined using 63x objective fitted to Zeiss AxioScope A1 microscope and a 121‐point eyepiece graticule. Using a systematic clock‐face sampling pattern from a random starting point, 32 microscopic fields (3872 points) were counted. Points falling over LC, SC, or germ cell nuclei, seminiferous epithelium, interstitium, and seminiferous tubule lumen were scored and they were expressed as relative (%) volume per testis. Values for percent nuclear volume were converted to absolute nuclear volumes per testis by reference to testis volume (=weight) because shrinkage was minimal.

Quantitative RT–PCR

Total RNA was extracted from tissue samples via NucleoSpin TriPrep Column Purification (Macherey‐Nagel). RNA was reversed‐transcribed to cDNA with the RevertAid Reverse Transcriptase Kit (Thermo Fisher Scientific). Primer sequences are described in Appendix Table S1. All primers were checked for efficiency (standard curve) and the generation of single amplicons (melting curve). PCR mixtures (10 µL) contained 5 µl Fast SYBR Green Master Mix (Applied Biosystems). The StepOnePlus sequence detector PCR detection system (Applied Biosystems) was used to quantify gene expression. Three housekeeping genes, namely Hprt, Actb, and Gapdh, served as endogenous controls.

Protein extraction

Whole‐cell extracts: Cells were lysed in Passive Lysis Buffer (Promega) and incubated at room temperature for 5 min. After centrifugation, protein concentration was measured in supernatant using the Pierce™ Coomassie Protein Assay Kit (Thermo Fisher Scientific). Kidneys were cut into small pieces in ice‐cold PBS and pressed through a 70‐µm cell strainer followed by a few passages through 18G and 21G needles. The tissue homogenate was subsequently lysed in ice‐cold lysis buffer containing 25 mM HEPES (pH 7.5), 300 mM NaCl, 1.5 mM MgCl2, 20 mM b‐glycerol phosphate, 2 mM EDTA, 2 mM EGTA, 1 mM DTT, 1% Triton X‐100, 10% glycerol, 10 µg/ml leupeptin, 5 µg/ml aprotinin, 1 mM PMSF, 1 mM Na3VO4, and 50 mM NaF, followed by three freeze–thaw cycles. After centrifugation, protein concentration was measured in supernatant using the Pierce™ BCA Protein Assay Kit (Thermo Fisher Scientific). Cytoplasmic and nuclear extracts: Extracts were obtained using the NE‐PER^TM Nuclear and Cytoplasmic Extraction Kit (Thermo Fisher Scientific) according to the manufacturer’s protocol. Chromatin fraction: Kidneys were cut into small pieces in ice‐cold PBS and pressed through a 70‐µm cell strainer followed by a few passages through 18G and 21G needles. Part of the homogenate was used for preparation of whole‐cell extracts as described above. The rest of the homogenate was pelleted by centrifugation at 400 g for 5 min, lysed in Buffer A (50 mM HEPES, pH 7.5, 10 mM KCl, 1.5 mM MgCl2, 340 mM sucrose, 10% glycerol, 1 mM DTT, and 1x PIC from Roche), and incubated for 10 min at 4°C. Samples were centrifuged at 1,300 g for 5 min at 4°C, and supernatants were discarded. Nuclear pellets were washed with Buffer A and subsequently lysed in solution B (3 mM EDTA, 0.2 mM EGTA, 1 mM DTT, and 1x PIC from Roche). After incubation for 30 min at 4°C, samples were centrifuged at 1,700 g for 5 min at 4°C and supernatants were discarded. Chromatin pellets were washed with solution B, resuspended in Buffer C (50 mM Tris–HCl, pH 8.0, 1 mM MgCl2, and 83 U/µL benzonase), and incubated for 20 min at 4°C.

Western blot

Cells were lysed in Passive Lysis Buffer (Promega) and incubated at room temperature for 5 min. After centrifugation, protein concentration was measured in supernatant using the Pierce™ Coomassie Protein Assay Kit (Thermo Fisher Scientific). Proteins were denaturized by adding LDS (Invitrogen) and reducing agent (Invitrogen) at 72°C for 10 min, separated on a NuPAGE Novex 4–12% Bis‐Tris Gel (Invitrogen), and blotted onto a PVDF membrane (GE Healthcare). Membranes were blocked with 5% non‐fatty dry milk in 0.1% TBST or PBS for 15 min. Incubation with primary antibody occurred at 4°C overnight. The following antibodies were used: AR (in‐house (Dubois et al, 2014)), AR (SP107‐ab236225, Abcam), GAPDH (sc‐32233, Santa Cruz), β‐Actin (A5441, Sigma‐Aldrich), HSP90 (sc‐13119, Santa Cruz), H3 (3638S, Cell Signaling), and Lamin A/C (sc‐376248, Santa Cruz). Immunodetection was performed with an ImageQuant Las4000 using the Western Lightning Plus ECL Reagent (PerkinElmer), after incubation with HRP‐conjugated secondary antibodies (Dako) at room temperature for 1 h.

RNA‐sequencing analysis

Total RNA of five mice for each genotype was extracted from tissue samples via NucleoSpin TriPrep Column Purification (Macherey‐Nagel). The RNA concentration was measured using a NanoDrop™ Spectrophotometer (Thermo Fisher Scientific), and subsequently, RNA integrity was evaluated with an Agilent Bioanalyzer. The RNA samples were processed by the Genomics Core Leuven (Belgium). Libraries were generated with the Illumina TruSeq‐Stranded mRNA Sample Preparation Kit and subsequently sequenced on the Illumina HiSeq 4000. Reads of 50 bp with single end were generated, and an average of 20 million reads was obtained. The reads were mapped against the mouse genome mm10. Differential gene expression and GSEA using default settings were performed with the Qlucore Omics Explorer v3.6.

Cell culture

Hela, HEK293, and Hep3B cells were obtained from the American Type Culture Collection (ATCC). HEK293 cells were modified with stable integration of 4x SLP‐ARE (Denayer et al, 2010). HeLa and HEK293 cells were cultured in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10% fetal calf serum (FCS). Hep3B cells were seeded in minimum essential medium (MEM)‐α medium supplemented with L‐glutamine and 5% FCS.

Transactivation assays

One day before transfection, HeLa or HEK293 cells were seeded in 96‐well plates (10,000 cells per well) in DMEM with 5% charcoal‐striped serum. GeneJuice Transfection Reagent (VWR International) was used to transfect 100 ng of reporter construct (pGL3basic containing the Insl3 or Hsd17b3 promoter), 10 ng of AR expression vector (with pCMV‐IE promoter), and 5 ng of pCMV‐β‐gal expression vector (Stratagene) to correct for transfection efficiency. The next day, medium was replaced by DMEM with 5% charcoal‐striped serum or DHT (Sigma‐Aldrich). The following day, cells were harvested in Passive Lysis Buffer (Promega) and luciferase activity was measured with a Luminoskan luminometer (De Bruyn et al, 2011).

Nuclear translocation assay

One day before transfection, HeLa cells were seeded in DMEM with 5% charcoal‐striped serum in 96‐well plates. GeneJuice Transfection Reagent (VWR International) was used to transfect 10 ng of GFP‐fused receptor plasmid and 80 ng empty vector. The next day, medium was replaced by DMEM with 5% charcoal‐striped serum or DHT (Sigma‐Aldrich) for 2‐h incubation. Cells were fixated, and nuclei were stained with Hoechst 33342 (Sigma‐Aldrich). The average intensity of the GFP signal was calculated for nucleus and cytoplasm of each transfected cell. The ratio of average intensity in nucleus/cytoplasm is given. To investigate the AR presence in cytoplasmic and nuclear extracts, HeLa cells were seeded in T75 flasks in DMEM with 5% charcoal‐striped serum (4*10^6 cells per flask). The next day, cells were transfected with 2 µg receptor (WT hAR, W752R hAR, WT mAR, or W731R mAR) and 18 µg empty vector using X‐treme gene (Sigma‐Aldrich). The next day, cells were stimulated with DHT for 2 h and extracts were provided using the NE‐PER™ Nuclear and Cytoplasmic Extraction Kit (Thermo Fisher Scientific) according to the manufacturer’s protocol.

Acceptor photobleaching fluorescence resonance energy transfer

The coding sequence for hAR LBD (residues 612–919) was linked to the coding sequence for enhanced yellow fluorescent protein (EYFP) or enhanced cyan fluorescent protein (ECFP) on the N‐ or C‐terminal end. Two days before microscopic analysis, Hep3B cells were seeded on a coverslip in a 6‐well plate at a density of approximately 30,000 cells per well. Cells were co‐transfected with constructs encoding EYFP‐ and ECFP‐labeled AR LBD. Cells transfected with both free EYFP and ECFP or with the ECFP‐EYFP fusion construct served as negative and positive controls, respectively. All transfected cells were stimulated with 10 nM DHT (Sigma‐Aldrich) in αMEM supplemented with charcoal‐stripped FCS for at least 16 h. Image collection and analysis were performed as described before (Nadal et al, 2017).

Electrophoretic mobility shift assay (EMSA)

EMSAs were performed with nuclear extracts of COS‐7 cells expressing WT AR or mutated AR in the presence of 10 nM DHT. The previously described protocol was used (De Bruyn et al, 2011). The binding of the receptors was performed on TAT‐GRE (5’‐AGAACAtccTGTACA‐3’) sequence containing double‐stranded oligonucleotides, which were radioactively labeled with [α‐32P]dCTP. The confirmation of the specific interaction between the AR and the probe was performed with an in‐house AR antibody.

Ligand‐binding assay

One day before transfection, COS‐7 cells were seeded in a 48‐well plate (30 000 cells per well) in DMEM with 5% charcoal‐stripped serum. GeneJuice Transfection Reagent (VWR International) was used to transfect 375ng of AR expression vector (with pCMV‐IE promoter). Two days later, a series of ³H‐R1881 concentrations (0.1 nM, 0.3 nM, 1 nM, 3 nM, 10 nM, 30 nM, and 100 nM) was added, one time in the presence of a > 100‐fold molar excess of DHT (non‐specific binding) and one time in the absence of cold ligand (total binding). Cells were incubated for 90 min at 37°C followed by three washing steps with ice‐cold PBS. Cells were harvested in Passive Lysis Buffer (Promega), and ³H‐signal in the lysates was measured after adding Ultima Gold XR (PerkinElmer). Data were normalized for the average dpm of all data points. Specific binding was calculated by extracting the non‐specific binding from the total binding. K^d and B^max were calculated using GraphPad Prism v8.

Mammalian double‐hybrid assay

COS‐7 cells were seeded at a density of 10,000 cells per 96 well in DMEM, 5% stripped serum, 1% Glutamax, and 1% penicillin–streptomycin. The next day, 100 ng (Gal4)5 TATA Luc reporter plasmid, 50 ng of plasmid encoding Gal4 DBD‐AR (1–36) or Gal4 DBD‐SRC1a (1241–1441), 10 ng of plasmid expressing VP16 AD‐AR LBD (640–919), and 5 ng of pCMV‐B‐Gal are co‐transfected per 96 well. The plasmid containing the Gal4 DBD (1–147) fused to the LxxLL fragment of SRC1a (1241–1441), which has been described before (Kalkhoven et al, 1998). The (GAL4)5TATA‐luc luciferase reporter plasmid was a kind gift of M.G. Parker (Imperial Cancer Research Fund, London, UK). To obtain the VP16 AD fused to AR LBD (640–919), a PCR fragment containing the hAR LBD (640–919) was cloned into the pSNATCHII vector described in Alen et al (1999). For stimulation and harvesting of the cells, the protocol for the transactivation study was followed. Data were normalized for the average luciferase activity of all data points per independent experiment.

Construction of tetracycline‐inducible cell lines expressing C‐terminal BirA* fusions of WT AR and AR W752R

Via In‐Fusion technology, PCR‐generated fragments of 3xFlag‐AR and (GA)6‐BirA* were recombined with the pcDNA5/FRT/TO backbone in order to create the AR WT with C‐terminal BirA* linked to each other by six repeats of Gly‐Ala. To insert the W752R mutation, the HindIII and AsuII restriction sites within the AR were used to exchange the WT DBD‐LBD fragment for a DBD‐LBD W752R fragment isolated from the AR^Lmon expression vector mentioned above. Co‐transfection of each plasmid with the pOG44 Flp‐Recombinase Expression Vector resulted in the stable integration of the pcDNA5/FRT/TO‐3xFlag‐AR‐(GA)6‐BirA* wt or W752R in the FRT site of HEK293‐FLP‐IN‐TREX cells. Positive clones were selected via hygromycin B (100 µg/ml).

Proximity‐dependent biotin identification (BioID) and data analysis

BioID experiments were performed as described previously for N‐terminal BirA*‐AR fusion (Lempiainen et al, 2017). After growing in serum‐depleted medium for 24 h, HEK293‐FLP‐IN‐TREX‐AR‐BirA* and HEK293‐FLP‐IN‐TREX‐AR W752R‐BirA* cells were induced with tetracycline (0.03 µg/ml, Sigma‐Aldrich) for the next 18 h, after which biotin (50 µM, Sigma‐Aldrich) with either vehicle (ethanol) or DHT (100 nM) was added for 6 h. HEK293‐FLP‐IN‐TREX‐ BirA*‐eGFP cells were used as the background control and treated identically. AR‐specific interactors from three biological replicates were discriminated from background contaminants by using six individual BirA*‐EGFP control purifications as the control. Significance Analysis of INTeractome (SAINT) (Choi et al, 2011) with default settings was used to determine the statistical significance of the detected interactions. SAINT input and output files are in Dataset EV1. Interactions with FDR < 0.05 were considered significant with the following exceptions: Acetyl‐CoA carboxylase 2 (ACACB, endogenously biotinylated), keratins (KRT2, KRT5, and KRT14) and trypsin (unspecific interactors), and tubulin beta‐4A chain (TUBB4A, non‐specific mapping of peptides to different tubulin isoforms). Additionally, data were analyzed manually by accepting all proteins whose signal (raw spectral count) was ≥ 1 in at least 2/3 of biological WT AR or AR W752R replicates. Raw spectral counts (SPCs) from these proteins were normalized to corresponding bait (WT AR or AR W752R), and values were filtered into WT AR‐ or AR W752R‐specific interactors by criterion that SPC was ≥ twofold higher than in EGFP control, and then filtered into DHT‐specific interactors by criterion that SPC in DHT was ≥ twofold higher than in vehicle.

Confocal microscopy

Confocal microscopy was used to analyze the cellular localization of AR‐BirA* or AR W752R‐BirA* and the biotinylated proteins as described in Lempiäinen et al (2017). AR was detected with anti‐AR antibody (1:100 dilution, sc7305, Santa Cruz) and secondary antibody rhodamine Red‐X (715‐295‐150; Jackson ImmunoResearch Laboratories Inc., West Grove, PA). DAPI was added to coverslips for 5 min, before they were mounted with ProLong^® Diamond antifade reagent (Thermo Fisher Scientific) and imaged with Zeiss LSM 800 confocal microscope.

Chromatin Immunoprecipitation (ChIP)

Kidneys from 12 AR^Lmon/Y mice and 12 WT littermates (one kidney per mouse) were used for ChIP and processed according to a protocol adapted from Dubois et al (2020) and Paakinaho et al (2014). Kidneys were pooled two by two to obtain six biological replicates per genotype. First, kidneys were cut into small pieces in ice‐cold PBS and pressed through a 70‐µm cell strainer followed by a few passages through 18G and 21G needles. Part of the homogenate was used for RNA isolation and quantitative RT–PCR as described above. The rest of the homogenate was fixed for 10 min at room temperature with 1% formaldehyde followed by a 10‐min incubation with 125 mM glycine. After a wash with ice‐cold PBS, cells were resuspended in lysis buffer (50 mM Tris–HCl, pH 8.0, 10 mM EDTA, 1% SDS, and 1x PIC from Roche) and sonicated for 15 min (15 cycles 30 s ON / 30 s OFF using Bioruptor NGS from Diagenode). Chromatin (40 µg for H3K27ac and H3K27me3 ChIP, 250 µg for AR ChIP) was diluted 10‐fold in RIPA Buffer (1× PBS containing 1% NP‐40, 0.5% sodium deoxycholate, 0.1% SDS, and 1X PIC) and incubated overnight at 4°C with 2 µg of H3K27ac antibody (Active Motif, #39685), 2 µg of H3K27me3 antibody (Millipore, #07‐449), or 4 µl of K183 (a polyclonal rabbit antiserum raised against the full‐length rat AR and described in Karvonen et al (1997)). The next day, Magna ChIP Protein A Magnetic Beads (Sigma‐Aldrich), pre‐incubated overnight at 4°C with 5 mg/ml BSA and 40 µg/ml yeast tRNA, were added during 4 h at 4°C in the presence of 70 µg/ml yeast tRNA. Beads were washed four times with LiCl IP Wash Buffer (100 mM Tris pH 7.5, 500 mM LiCl, 1% NP‐40, and 1% sodium deoxycholate) containing 10 µg/ml yeast tRNA and twice with TE Buffer (10 mM Tris–HCl pH 8.0, 1 mM EDTA). DNA was then eluted in 100 mM NaHCO₃ containing 1% SDS and incubated overnight at 65°C in the presence of 20 µg/ml proteinase K for reverse crosslinking. DNA purification was performed using the MinElute PCR Purification Kit (Qiagen, #2800), and samples were subjected to qPCR analyses. The primer sequences are listed in Appendix Table S1. Before performing ChIP on AR^Lmon/Y kidneys, all antibodies were first validated using kidneys from chemically castrated male WT mice with or without T supplementation described in Kim et al (2020, 2021) (Appendix Fig S5A–K).

Molecular dynamics simulations

The crystal structure of the AR LBD homodimer (PDB: 5JJM) (Nadal et al, 2017) was used as a template in which the W752R mutation was introduced. Both WT and mutated structures were prepared using Molecular Operating Environment (MOE) software (Chemical Computing Group, Quebec, Canada). Afterward, these structures were simulated by Gromacs (Pronk et al, 2013) with the amber99sb force field (Hornak et al, 2006), 100 ps V‐rescale thermostat NVT ensemble, 100 ps V‐rescale thermostat and Parrinello‐Rahman barostat NPT ensemble, and 20 ns Berendsen thermostat and barostat MD production. Following the simulation, 100 snapshots per system from the last 10‐ns trajectory were used to calculate free energy of binding of the interaction between the two monomers in which the MM‐GBSA method implemented in Amber16 was applied (Case et al, 2016).

Yeast two‐hybrid assay

The modified BY4741 Saccharomyces cerevisiae strain [MATa his3Δ1 leu2Δ0 met15Δ0 ura3Δ0 can1Δ:: [pPGI1 _yeCitrine_tCYC1] ho1Δ:: [p(8x LexA‐min pCYC1)_mCherry_tDIT1] was transformed with a single plasmid encoding both bait [LexA(DBD)‐AR (612–919)] and prey [AR (612–919)‐VPR] protein of the WT AR or W752R mutant under control of the TEF1 promoter and CYC1 terminator via standard protocols (Amberg et al, 2005). Three independent transformants of both WT and W752R AR were incubated for 24 h at 30°C in SC medium in the presence of vehicle or 10 nM DHT and diluted 1/200 prior to the flow cytometry (Attune NxT Flow Cytometer; Thermo Fisher). The resultant mCherry fluorescence intensities are normalized by the corresponding yeCitrine signal.

Statistical analysis

Statistical analyses were performed using GraphPad Prism v8 and R (R Core Team, 2015). Unpaired two‐tailed Student’s t‐test was used to compare two groups. To compare three groups or more, one‐way ANOVA with Tukey’s multiple comparisons test was performed. Two‐way ANOVA with Bonferroni’s post hoc test was used in experiments with more than one independent variable. One‐sample t‐test with the Benjamini–Hochberg correction for multiple testing was used to determine whether a mean Log₂ FC was statistically different from 0. P values < 0.05 were considered statistically significant and represented as follows: *P < 0.05, **P < 0.01, and ***P < 0.001. Data are represented as mean ± standard error of the mean (SEM).

Author contributions

SEK, CH, DV, and FC designed the study. SEK, FH, CH, VD, MEvR, AH, EP, TN, K‐ML, LH, SP, JJP, MP, RE, and GC performed data acquisition and analysis. SEK, VD, NA, AV, CO, DV, CH, and FC interpreted the data. All authors drafted, revised, and approved the manuscript content.

Conflict of interest

The authors declare that they have no conflict of interest.

Supporting information

Appendix

Expanded View Figures PDF

Table EV1

Dataset EV1

Source Data for Expanded View and Appendix

Source Data for Figure 5C

Source Data for Figure 6A

EMBO reports (2021) 22: e52764.

Data availability

RNA‐Seq data are available on Gene Expression Omnibus GSE172020‐RNA sequence on kidneys of wild‐type C57BL/6J and AR^Lmon/Y mice that underwent orchidectomy and testosterone (or vehicle) replacement for 4 days and GSE172026–RNA sequence on testes of wild‐type C57BL/6J, AR^Lmon/Y, and AR^−/Y mice.