Androgen signaling is a confounding factor for β-catenin-mediated prostate tumorigenesis

Abstract

Emerging evidence has demonstrated the critical roles for both androgen and Wnt pathways in prostate tumorigenesis. A recent integrative genomic analysis of human prostate cancers (PCas) has revealed a unique enrichment of androgen and Wnt signaling in early-onset PCas, implying their clinical significance in the disease. Additionally, interaction between the androgen receptor (AR) and β-catenin has long been detected in PCa cells. However, the consequence of this interaction in prostate tumorigenesis is still unknown. Because mutations in adenomatous polyposis coli, β-catenin and other components of the destruction complex are generally rare in PCas, other mechanisms of aberrant Wnt signaling activation have been speculated. To address these critical questions, we developed Ctnnb1^L(ex3)/+/R26hAR^L/+:PB-Cre4 mice, in which transgenic AR and stabilized β-catenin are co-expressed in prostatic epithelial cells. We observed accelerated tumor development, aggressive tumor invasion and a decreased survival rate in Ctnnb1^L(ex3)/+/R26hAR^L/+:PB-Cre4 compound mice compared with age-matched Ctnnb1^L(ex3)/+:PB-Cre4 littermate controls, which only have stabilized β-catenin expression in the prostate. Castration of the above transgenic mice resulted in significant tumor regression, implying an essential role of androgen signaling in tumor growth and maintenance. Implantation of the prostatic epithelial cells isolated from the transgenic mice regenerated prostate intraepithelial neoplasias and prostatic adenocarcinoma lesions. Microarray analyses of transcriptional profiles showed more robust enrichment of known tumor- and metastasis-promoting genes: Spp1, Egr1, c-Myc, Sp5, and Sp6 genes, in samples isolated from Ctnnb1^L(ex3)/+/R26hAR^L/+:PB-Cre4 compound mice than those from Ctnnb1^L(ex3)/+:PB-Cre4 and R26hAR^L/+:PB-Cre4 littermate controls. Together, these data demonstrate a confounding role of androgen signaling in β-catenin-initiated oncogenic transformation in prostate tumorigenesis.

Introduction

Androgen signaling, mediated through the androgen receptor (AR) and its ligands, testosterone and 5α-dihydrotestosterone, is essential for prostate tumorigenesis.^{1, 2, 3} Androgen-deprivation therapy inhibits androgen signaling-mediated cell growth and survival and thus has been widely used to treat prostate cancer (PCa).⁴ The AR is expressed in most PCa samples before and after the therapy.⁵ In fact, AR gene amplification appears in one-third of PCa samples after androgen-deprivation therapy.⁶ Additionally, global gene-expression profiling shows that the AR is one of the several genes to be consistently upregulated in castration-resistant prostate cancer,^{7, 8} underscoring the significance of androgen signaling in prostate tumorigenesis.

Likewise, Wnt signaling pathways have a significant role in prostate tumorigenesis.⁹ Abnormal expression of Wnt ligands, receptors and effectors has been identified in PCa cells and cells of the surrounding microenvironments, suggesting paracrine-regulatory mechanisms in prostate tumorigenesis.^{10, 11} In the canonical signaling pathway, secreted Wnt ligands bind to Frizzled proteins and regulate the stability of β-catenin, a key component of Wnt signaling.¹² In the absence of a Wnt signal, β-catenin is constitutively downregulated by a multicomponent destruction complex containing glycogen synthase kinase 3β, axin and adenomatous polyposis coli (APC), promoting phosphorylation on serine and threonine residues in the N-terminal region of β-catenin following ‘priming’ phosphorylation of Ser45 by Casein Kinase I and thereby targeting it for degradation via the ubiquitin proteasome pathway.^{13, 14, 15, 16} Increase in nuclear β-catenin has been shown to promote PCa cell proliferation.¹⁷ In mice, expression of stabilized β-catenin in the prostate induces the development of squamous metaplasia, prostate intraepithelial neoplasias (PIN)^{18, 19} and prostate adenocarcinoma.²⁰ Conditional deletion of APC in the mouse prostate elevates nuclear β-catenin and results in prostatic adenocarcinoma formation.²¹ Intriguingly, mutations in APC, β-catenin and other components of the β-catenin destruction complex are rare in human PCa,^{22, 23} suggesting alternative mechanisms of β-catenin dysregulation in prostate tumorigenesis. One of such alternative mechanisms is the hypermethylation of the APC promoter region, which ultimately results in aberrant activation of β-catenin/Wnt signaling pathway. In fact, APC promoter hypermethelation in PCa has been correlated with poor clinical prognosis²⁴ and recently has been proposed as one of the prognostic PCa markers.²⁵

Despite the elusiveness of aberrant Wnt activation, an interaction between the AR and β-catenin proteins has been identified in PCa cells.^{26, 27, 28} A recent integrative genomic study revealed a significant enrichment of the androgen and Wnt signaling pathways in the early-onset PCa samples but not in the elderly-onset PCa samples,²⁹ which further implicates a clinical importance of both androgen and Wnt signaling in prostate tumorigenesis. However, the biological significance of this interaction in the pathogenesis of PCa remains unknown. To address this critical question and recapitulate the AR and β-catenin interaction in vivo, we developed Ctnnb1^L(ex3)/+/R26hAR^L/+:PB-Cre4 mice, in which transgenic AR and stabilized β-catenin are co-expressed in prostatic epithelial cells. We observed accelerated tumor development, aggressive tumor invasion and a decreased survival rate in these compound mice versus control littermates. Castration of these mice resulted in significant tumor regression. Moreover, implantation of prostatic epithelial cells isolated from the compound mice regenerated much severe PIN and prostatic adenocarcinoma lesions in tissue recombination assays. Furthermore, we detected increased expression of known pro-tumorigenic and pro-metastatic genes Spp1, Egr1, c-Myc and Sp6 in the prostates of the compound mice. These data demonstrate a confounding role of androgen signaling in β-catenin-mediated oncogenic transformation in prostate tumorigenesis.

Results

Activation of transgenic AR expression accelerates oncogenic transformation in the mouse prostate

Although an interaction between the androgen and Wnt signaling pathways has been identified in PCa cells, the significance of this interaction in prostate tumorigenesis remains unclear. Specifically, we still do not know whether these two pathways can synergistically promote PCa initiation and progression. To address this question, we developed Ctnnb1^L(ex3)/+/R26hAR^L/+:PB-Cre4 mice, in which expression of human AR transgene and stabilized β-catenin were activated through the probasin promoter-driven Cre expression in the mouse prostate (Figure 1a). Using genomic PCR and western blotting analyses, we confirmed LoxP/Cre-mediated recombination events (Figure 1b), resulting in human AR and stabilized β-catenin protein expression in mouse prostate tissues (Figure 1c). Ctnnb1^L(ex3)/+/R26hAR^L/+:PB-Cre4, Ctnnb1^L(ex3)/+:PB-Cre4 and R26hAR^L/+:PB-Cre4 mice were born at the expected Mendelian ratios and appeared normal with no obvious differences from their wild-type littermates at birth. We systematically examined male mice at 3, 4, 5, 6 and 8 weeks of age (Supplementary Figure S1). Consistent with previous reports,^{30, 31} we did not observe any obvious abnormalities in 2- to 12-month-old R26hAR^L/+:PB-Cre4 mice (data not shown). Adhering to recommendations of the Mouse Models of Human Cancers Consortium Prostate Pathology Committee,³² we first identified prostatic hyperplasia and mPIN1 in 4-week-old Ctnnb1^L(ex3)/+:PB-Cre4 mice (Figure 1d1; Supplementary Figures S1B1 and B2). These lesions originated predominantly in the dorsal/lateral prostate and ventral prostate lobes. Over time, it appears that these low-grade mPIN lesions progressed toward high-grade mPIN lesions (mPIN3 and mPIN4) (Supplementary Figures S1D1, D2, E1 and E2). Notably, Ctnnb1^L(ex3)/+/R26hAR^L/+:PB-Cre4 compound mice, with transgenic AR and stabilized β-catenin co-expression in the prostate, developed PIN lesions earlier than Ctnnb1^L(ex3)/+:PB-Cre4 mice. We observed mPIN1 and mPIN2 lesions at 3 weeks of age (Supplementary Figure S1A4), and mPIN3 and mPIN4 lesions at 5 weeks in Ctnnb1^L(ex3)/+/R26hAR^L/+:PB-Cre4 mice (Supplementary Figures S1C3 and C4). Additionally, squamous metaplasia appeared in both Ctnnb1^L(ex3)/+:PB-Cre4 and Ctnnb1^L(ex3)/+/R26hAR^L/+:PB-Cre4 mice (Supplementary Figure S1C3). Expression of stabilized β-catenin was confirmed in prostatic epithelial cells in both Ctnnb1^L(ex3)/+:PB-Cre4 and Ctnnb1^L(ex3)/+/R26hAR^L/+:PB-Cre4 mice (Figures 1d5 and e5; Supplementary Figures S2A1–3 and C1–3). Interestingly, we noted atypical cell foci with stabilized β-catenin expression within PIN lesions of Ctnnb1^L(ex3)/+:PB-Cre4 mice (arrows, Figure 1d5; Supplementary Figures S2A1 and A3). Transgenic AR expression was only detected in prostatic epithelial cells in Ctnnb1^L(ex3)/+/R26hAR^L/+:PB-Cre4 compound mice (Figure 1e6; Supplementary Figures S2D1–3). To establish the nature of interaction between stabilized β-catenin with endogenous and transgenic AR, we performed co-immunoprecipitation experiments using whole cell lysates isolated from prostate tissues of 3-month-old Ctnnb1^L(ex3)/+:PB-Cre4 and Ctnnb1^L(ex3)/+/R26hAR^L/+:PB-Cre4 mice. As shown in Figure 1f, both endogenous and transgenic AR interact with stabilized and wild-type β-catenin. The results further demonstrate the protein–protein interaction between stabilized β-catenin with both transgenic and endogenous AR and implicate that a collaborative relationship between androgen and Wnt signaling pathways may synergistically promote the initiation of oncogenic transformation in the mouse prostate.

Figure 1

The conditional expression of human AR transgene and stabilized β-catenin allele in mouse prostate. (a) A scheme of the conditional human AR transgene targeting construct is shown on the left figure. A PGK-neomycin cassette with flanked loxP sites (LSL cassettes) was inserted between the CAG promoter and a FLAG-tagged human AR coding sequence containing a nine-polyglutamine repeat tract. Red triangles sandwiching PGK-neomycin cassette (a dotted box) indicate loxP sequences. A targeting strategy for the exon 3 deletion of β-catenin gene is shown on the right figure. Exons are shown as filled boxes (E2–E9), whereas intronic sequences are shown as solid lines. Cre-recombined alleles for both constructs are shown below the targeting strategies. (b) Genomic PCR was performed to confirm the targeting and PB-Cre4 alleles using the tail tips of the mice. (c) Western blotting was performed to confirm the protein expression levels of stabilized β-catenin and transgenic human AR using prostate tissues isolated from the 8-week-old mice of the different genotypes. (d and e). Low- and high-magnification photomicrographs of anterior prostate (AP), dorsal and lateral prostate (D/LP) and ventral prostate (VP) lobes from 5-week-old Ctnnb1^L(ex3)/+:PB-Cre and Ctnnb1^L(ex3)/+/R26hAR^L/+:PB-Cre mice stained with hematoxylin and eosin or with the antibodies against β-catenin and human AR in which tissues were also counterstained with hematoxylin. Note the presence of low-grade mPIN lesions in the Ctnnb1^L(ex3)/+:PB-Cre mice (d5 and d6) and high-grade mPIN lesions in Ctnnb1^L(ex3)/+/R26hAR^L/+:PB-Cre mice (e5 and e6). (f) Cell lysates from the prostate of 3-month-old Ctnnb1^L(ex3)/+:PB-Cre or Ctnnb1^L(ex3)/+/R26hAR^L/+:PB-Cre mice were immunoprecipitated with the AR (upper panel) or β-catenin antibody (lower panel) and analyzed by western blotting (IB) with the indicated antibodies.

Co-expression of transgenic AR and stabilized β-catenin induces invasive prostate adenocarcinoma development

Following the consensus that high-grade mPIN lesions can progress to prostate adenocarcinomas, we continued examining both Ctnnb1^L(ex3)/+:PB-Cre4 and Ctnnb1^L(ex3)/+/R26hAR^L/+:PB-Cre4 mice for extended periods of time. We observed prostatic tumor formation in these mice beginning at 3 months of age. Gross examination revealed extensive prostatic tumor masses with multiple lobules in Ctnnb1^L(ex3)/+/R26hAR^L/+:PB-Cre4 mice than those in age-matched Ctnnb1^L(ex3)/+:PB-Cre4 littermates (Figure 2b1 versus 2a1 and 2b2 versus 2a2). Histologically, larger and more extensive prostatic tumor masses appeared in Ctnnb1^L(ex3)/+/R26hAR^L/+:PB-Cre4 than in Ctnnb1^L(ex3)/+/R26hAR^L/+:PB-Cre4 mice (Figure 2d1 versus 2c1 and 2d2 versus 2c2). By 3 months of age, Ctnnb1^L(ex3)/+/R26hAR^L/+:PB-Cre4 developed multi-focal intracystic adenocarcinomas with extensive microinvasion in all prostatic lobes (Figure 2d1). These lesions further progressed into invasive prostatic adenocarcinomas in four of the four Ctnnb1^L(ex3)/+/R26hAR^L/+:PB-Cre4 mice by 4 months of age (Figure 2d2). In contrast, Ctnnb1^L(ex3)/+:PB-Cre4 mice showed milder changes, featuring high-grade PINs and scattered intracystic adenocarcinomas at 3 months (Figure 2c1) and more frequent intracystic andenocarcinomas with rare microinvasion at 4 months (Figures 2c2). These data demonstrate that Ctnnb1^L(ex3)/+/R26hAR^L/+:PB-Cre4 mice have accelerated tumor growth compared with Ctnnb1^L(ex3)/+:PB-Cre4 mice, suggesting a synergistic effect of stabilized β-catenin and transgenic AR co-expression in enhancing oncogenesis and accelerating invasive prostatic adenocarcinoma formation.

Figure 2

The synergistic effect of AR and Wnt signaling in mouse prostate tumor progression. Corresponding gross photographs (a1–b2), subgross photomicrographs (c1–d2), and photomicrographs (e–h) of prostrate gland with seminal vesicles and urinary bladder from 3- to 4-month-old Ctnnb1^L(ex3)/+:PB-Cre and Ctnnb1^L(ex3)/+/R26hAR^L/+:PB-Cre mice. On the gross and subgross views, there is prostate adenocarcinoma development in both Ctnnb1^L(ex3)/+:PB-Cre and Ctnnb1^L(ex3)/+/R26hAR^L/+:PB-Cre mice, and the tumors are larger in the Ctnnb1^L(ex3)/+/R26hAR^L/+:PB-Cre mice, based on volume and cross-sectional area). Microscopically, the tumors in both the transgenic mice are represented by the development of intracystic prostatic adenocarcinomas and high-grade mPIN lesions. In the Ctnnb1^L(ex3)/+:PB-Cre mice, there are few scattered intracystic prostatic adenocarcinomas with limited microinvasion into the surrounding stroma. In contrast, the larger tumors in the Ctnnb1^L(ex3)/+/R26hAR^L/+:PB-Cre mice correspond with the presence of larger numbers of intracystic prostatic adenocarcinomas with surrounding microstromal invasion. (i) Pathological abnormalities in R26hAR^L/+:PB-Cre, Ctnnb1^L(ex3)/+:PB-Cre and Ctnnb1^L(ex3)/+/R26hAR^L/+:PB-Cre mice.

Ctnnb1^L(ex3)/+/R26hAR^L/+:PB-Cre4 compound mice showed a shortened survival rate

Systematic analysis showed that Ctnnb1^L(ex3)/+/R26hAR^L/+:PB-Cre4 mice have a significantly earlier onset and faster progressing tumor phenotype than R26hAR^L/+:PB-Cre4 and Ctnnb1^L(ex3)/+:PB-Cre4 mice (Figure 2i). Three of the three Ctnnb1^L(ex3)/+/R26hAR^L/+:PB-Cre4 mice developed multifocal intracystic adenocarcinomas with microinvasive lesions as early as at 2 months of age. At 4 months of age, four of the four Ctnnb1^L(ex3)/+/R26hAR^L/+:PB-Cre4 mice showed massive intracystic adenocarcinomas with extensive invasion in multiple prostatic lobes. Ctnnb1^L(ex3)/+/R26hAR^L/+:PB-Cre4 mice die spontaneously between 6 and 9 months, which is significant shorter than Ctnnb1^L(ex3)/+:PB-Cre4 and R26hAR^L/+:PB-Cre4 littermates (Figure 3a). The average weight of prostate tumor tissues at 7 months in Ctnnb1^L(ex3)/+/R26hAR^L/+:PB-Cre4 mice was significantly heavier than those of age-matched R26hAR^L/+:PB-Cre4 and Ctnnb1^L(ex3)/+:PB-Cre4 littermates (Figure 3b). Based on systematic gross and histological examination by a board-certified veterinary pathologist, prostatic tumor burden was the most likely cause of death for all of these mice. Shorter survival rate in Ctnnb1^L(ex3)/+/R26hAR^L/+:PB-Cre4 mice, as compared with controls, further demonstrate that Ctnnb1^L(ex3)/+/R26hAR^L/+:PB-Cre4 mice have accelerated prostate tumor development and more aggressive tumor phenotype.

Figure 3

The mouse survival rate and prostatic cell proliferation. (a) Kaplan–Meier survival curve of Ctnnb1^L(ex3)/+/R26hAR^L/+:PB-Cre, Ctnnb1^L(ex3)/+:PB-Cre and R26hAR^L/+:PB-Cre mice. (b) Wet weights of prostates of 7-month-old R26hAR^L/+:PB-Cre, Ctnnb1^L(ex3)/+:PB-Cre and Ctnnb1^L(ex3)/+/R26hAR^L/+:PB-Cre mice. **P<0.01. (c–e) Prostate sections isolated from 3-month-old R26hAR^L/+:PB-Cre, Ctnnb1^L(ex3)/+:PB-Cre and Ctnnb1^L(ex3)/+/R26hAR^L/+:PB-Cre mice were stained for Ki67. (f) A total of 1000 epithelial cells in each lesion from five different lesions from three mice of each genotype were evaluated for Ki67 immunoreactivity. *P<0.05 and **P<0.01.

Transgenic AR and stabilized β-catenin co-expression synergistically enhances cell proliferation

In this study, we examined the effect of co-expressing stabilized β-catenin and transgenic AR in promoting cell proliferation in the transgenic mice using Ki67 immunohistochemistry. A significant increase of Ki67-positive cells was observed in the samples of Ctnnb1^L(ex3)/+/R26hAR^L/+:PB-Cre4 versus Ctnnb1^L(ex3)/+:PB-Cre4 mice (Figures 3c1–e2). We quantified Ki67-positive cells per 1000 epithelial cells from five fields on three different slides that were prepared independently from three mice of each genotype. As shown in Figure 3f, the epithelial proliferative index was the highest in the tumor lesions of Ctnnb1^L(ex3)/+/R26hAR^L/+:PB-Cre4 mice, suggesting a synergistic role of transgenic AR and stabilized β-catenin co-expression in promoting cell proliferation in the mouse prostate.

Transgenic AR and stabilized β-catenin co-expression induces oncogenic transformation predominantly in prostatic luminal cells

To further define the cellular properties of prostatic pathologies developed in Ctnnb1^L(ex3)/+:PB-Cre4, and Ctnnb1^L(ex3)/+:R26hAR^L/+:PB-Cre4 :mice, we comprehensively analyzed different cellular markers in both higher-grade PIN and adenocarcinoma lesions (Figure 4). Most atypical and tumor cells showed positive immunoreactivity for E-cadherin (Figure 4d1) and CK8 (Figure 4e1), secretory epithelial markers, but showed no immunoreactivity for the neuroendocrine cell marker synaptophysin (Figure 4h1). Immunoreactivity for p63, a cellular marker for prostatic basal epithelial cells, appeared mainly in the basal compartment of prostatic glands with PIN lesions but was lacking in tumor lesions of both Ctnnb1^L(ex3)/+:PB-Cre4 and Ctnnb1^L(ex3)/+/R26hAR^L/+:PB-Cre4 mice (Figure 4g1). CK5 antibodies demonstrated similar staining patterns as those of p63 (Figure 4f1). Positive cytoplasmic and nuclear reactivity with β-catenin antibodies appeared in both PIN and tumor sections isolated from Ctnnb1^L(ex3)/+:PB-Cre4 and Ctnnb1^L(ex3)/+/R26hAR^L/+:PB-Cre4 mice (Figure 4c1). In contrast, positive reactivity with the human AR antibody was only observed in PIN and tumor lesions of Ctnnb1^L(ex3)/+/R26hAR^L/+:PB-Cre4 mice (Figures 4b2 and b4). Taken together, these data imply that most atypical and tumor cells in both Ctnnb1^L(ex3)/+:PB-Cre4 and Ctnnb1^L(ex3)/+/R26hAR^L/+:PB-Cre4 mice contain prostatic luminal cellular markers.

Figure 4

Immunohistochemistry analyses of the prostate tumors in Ctnnb1^L(ex3)/+:PB-Cre and Ctnnb1^L(ex3)/+/R26hAR^L/+:PB-Cre mice. Adjacent prostate tissue slides were prepared from prostatic tumor samples of 1.5- and 6-month-old Ctnnb1^L(ex3)/+:PB-Cre (1 and 3) and Ctnnb1^L(ex3)/+/R26hAR^L/+:PB-Cre mice (2 and 4). (a1–a4) Hematoxylin and eosin staining; (b1–h4) immunohistochemistry with different antibodies as labeled in the figure.

Effect of abnormal activation of AR and β-catenin in inducing prostatic oncogenesis in tissue recombinants

In this study, we assessed the oncogenic potential of transgenic AR and stabilized β-catenin-expressing cells to regenerate cancerous cells using a tissue recombination system. In the double-fluorescent mT/mG Cre reporter mouse strain, the expression of membrane targeted tandem dimer Tomato (mT) or membrane-targeted green fluorescent protein (mG) is controlled by Cre-mediated recombination, where Cre-expressing cells contain mG and can be sorted from mT cells.³³ We first generated and analyzed specific mG protein expression in prostatic luminal epithelial cells controlled by PB-Cre-mediated recombination, allowing for the selection of recombined cells, in mT/mG:PB-Cre mice (Supplementary Figure S3). We then produced both mT/mG/Ctnnb1^L(ex3)/+:PB-Cre4 and mT/mG/ Ctnnb1^L(ex3)/+/R26hAR^L/+:PB-Cre4 mice to specifically label stabilized β-catenin only or stabilized β-catenin/transgenic AR-positive cells (Figures 5a and b). Single-cell suspensions prepared from prostate tumors of 3-month-old mT/mG/Ctnnb1^L(ex3)/+:PB-Cre4 or mT/mG/Ctnnb1^L(ex3)/+/R26hAR^L/+:PB-Cre4 mice mixed with embryonic urogenital sinus mesenchymal (UGSM) cells were transplanted under the renal capsule of NOD/SCID mice and analyzed after 12 weeks. Grafts composed solely of UGSM cells appeared small and fibrous and mainly contained stromal-like cells. However, large translucent grafts were observed in grafts containing both UGSM and prostatic epithelial cells. Histological analysis of these grafts showed tubular epithelial structures with prostatic-like ducts filled with fluid (Figures 5c and d). We observed typical high-grade PIN or multi-focal intracystic adenocarcinoma lesions in tissue recombinant grafts prepared with prostatic epithelial cells isolated from mT/mG/Ctnnb1^L(ex3)/+:PB-Cre4 or mT/mG/Ctnnb1^L(ex3)/+/R26hAR^L/+:PB-Cre4 mice (arrows, Figures 5c or d and c’ and d’). These were very similar to the lesions presenting in the prostates of Ctnnb1^L(ex3)/+:PB-Cre4 and Ctnnb1^L(ex3)/+/R26hAR^L/+:PB-Cre4 mice. Of note were the more severe pathological changes in grafts using prostatic epithelial cells isolated from mT/mG/Ctnnb1^L(ex3)/+/R26hAR^L/+:PB-Cre4 versus mT/mG/Ctnnb1^L(ex3)/+:PB-Cre4 mice. Co-expression of transgenic AR and/or stabilized β-catenin, as well as mG protein, was detected in atypical and tumor cells of all graft samples (Figures 5e and p). These data further demonstrate the oncogenic capacity of transgenic AR and stabilized β-catenin in regenerating cancerous glandular structures in the mouse prostate.

Figure 5

Regenerating activity of tumor cells from mT/mG/Ctnnb1^L(ex3)/+:PB-Cre and mT/mG/Ctnnb1^L(ex3)/+/R26hAR^L/+:PB-Cre prostates. (a and b) Schematic illustration of labeling PB-Cre-positive cells with the mT/mG reporter. (c–p) Dissociated prostatic tumor cells from mT/mG/Ctnnb1^L(ex3)/+/R26hAR^L/+:PB-Cre and mT/mG/Ctnnb1^L(ex3)/+:PB-Cre mice were transplanted with urogenital sinus mesenchymal cells under the renal capsule in NOD/SCID mice. After 12 weeks, grafts were harvested and examined. (c–d) Hematoxylin and eosin staining of the outgrown tissues. (e–p) Immunofluorescent staining with different antibodies as labeled in the figure.

Androgen signaling is a confounding factor for β-catenin-initiated oncogenic transformation

Early-onset and rapidly progressing tumor phenotype in Ctnnb1^L(ex3)/+/R26hAR^L/+:PB-Cre4 compound mice suggests a promotional role of androgen signaling in β-catenin-mediated oncogenic transformation. Thus we evaluated the effect of castration on tumor development, maintenance and progression in the transgenic mice. Given the fact that PINs developed as early as 3 weeks in Ctnnb1^L(ex3)/+/R26hAR^L/+:PB-Cre4 mice, we castrated R26hAR^L/+:PB-Cre4, Ctnnb1^L(ex3)/+:PB-Cre4 and Ctnnb1^L(ex3)/+/R26hAR^L/+:PB-Cre4 mice at 2 weeks of age and analyzed them at 16 weeks of age to determine the effect of androgens in initiating oncogenic transformation in the prostate (Figure 6a). We observed underdeveloped prostate lobes in all three different genotypes of mice, featuring markedly atrophic glandular/ductal profiles in combination with a relatively prominent concentric peri-glandular/-ductular fibromuscular stroma (Figures 6b and d’). Importantly, there were no proliferative, dysplasic and mPIN pathological changes in Ctnnb1^L(ex3)/+:PB-Cre4 nor Ctnnb1^L(ex3)/+/R26hAR^L/+:PB-Cre4 mice, contrasting to those in age- and genotype-matched intact mice (Figure 1), which implicates an essential role of androgens in initiating oncogenic transformation of the prostate. To assess androgens in promoting tumor growth and progression, we castrated Ctnnb1^L(ex3)/+:PB-Cre4 and Ctnnb1^L(ex3)/+/R26hAR^L/+:PB-Cre4 mice at 6 weeks, after high-grade PINs have already developed, and then analyzed them at 16 weeks (Figure 6a). Prostate lobes were grossly smaller than the age-matched intact mice, and histological analysis showed some residual mPIN-like lesions as well as remnants of sloughed, intraluminal keratin. However, the epithelial cells lining these mPIN-like lesions consisted of 1–2 cell layers of uniform cuboidal cells with minimal nuclear atypia (Figures 6e and h’). Extensive peri- and inter- glandular/-ductular inflammatory responses appeared in both castrated Ctnnb1^L(ex3)/+:PB-Cre4 and Ctnnb1^L(ex3)/+/R26hAR^L/+:PB-Cre4 mice. We continued the study by castrating Ctnnb1^L(ex3)/+:PB-Cre4 and Ctnnb1^L(ex3)/+/R26hAR^L/+:PB-Cre4 mice at 16 weeks when they developed prostatic adenocarcinomas. At 26 weeks of age, we observed regression of tumor masses in all prostate lobes (Figure 6a). Although residual lower-grade mPIN lesions still remained, no tumor masses were observed in all the examined samples isolated from Ctnnb1^L(ex3)/+/R26hAR^L/+:PB-Cre4 mice castrated at 16 weeks (Figures 6i and l’). We also detected massive cell death in all prostatic lobes, accompanied with extensive immune responses. To further explore the effect of androgens on tumor growth in Ctnnb1^L(ex3)/+:PB-Cre4 and Ctnnb1^L(ex3)/+/R26hAR^L/+:PB-Cre4 mice, we supplemented androgen pellets into the mice that were castrated at 16 weeks of age (Supplementary Figure S5A). Importantly, we observed that more severe PIN lesions re-developed in castrated Ctnnb1^L(ex3)/+/R26hAR^L/+:PB-Cre4 than in Ctnnb1^L(ex3)/+:PB-Cre4 mice at 30 weeks of age, 4 weeks after androgen pellet supplement (Supplementary Figure S5E versus B). Transgenic hAR expression was only detected in the prostates of Ctnnb1^L(ex3)/+/R26hAR^L/+:PB-Cre4 mice (Supplementary Figures S5F and F’) but not in the ones of Ctnnb1^L(ex3)/+:PB-Cre4 mice (Supplementary Figures S5C and C’). Stabilized β-catenin was observed in the prostates of both Ctnnb1^L(ex3)/+:PB-Cre4 and Ctnnb1^L(ex3)/+/R26hAR^L/+:PB-Cre4 mice (Supplementary Figures S5D–G’). These experiments further demonstrate a reliance on androgens in β-catenin-mediated oncogenic transformation in the mouse prostate.

Figure 6

Androgen action is necessary for tumor formation and maintenance. (a) Schematic illustration of castration. (b–d’) Photomicrographs of prostate tissues from Ctnnb1^L(ex3)/+/R26hAR^L/+:PB-Cre, Ctnnb1^L(ex3)/+:PB-Cre and R26hAR^L/+:PB-Cre mice castrated at 2 weeks and killed at 16 weeks and stained with hematoxylin and eosin. Note the presence of only small numbers of prostate glandular profiles, which are widely scattered and are of small diameter (interpreted as atrophy after hormone-withdrawal). (e–h’) Photomicrographs of prostate tissues from Ctnnb1^L(ex3)/+/R26hAR^L/+:PB-Cre and Ctnnb1^L(ex3)/+:PB-Cre mice castrated at 6 weeks and killed at 16 weeks and stained with hematoxylin and eosin. The sections are highly cellular due mainly to the presence of abundant inflammation and fibroplasia, among which are small numbers of remnant prostate gland profiles. These glands are of small diameter and predominantly lined by a single or double layer of uniform, quiescent epithelial cells (interpreted as regression of mPIN lesions and/or prostatic adenocarcinoma lesions). Some remnant keratin and cellular/nuclear debris is present in the lumina of these glands. (i–l’) Photomicrographs of prostate tissues from Ctnnb1^L(ex3)/+/R26hAR^L/+:PB-Cre and Ctnnb1^L(ex3)/+:PB-Cre mice castrated at 16 weeks and killed at 26 weeks and stained with hematoxylin and eosin. These sections appear similar to those noted in the Ctnnb1^L(ex3)/+/R26hAR^L/+:PB-Cre and Ctnnb1^L(ex3)/+:PB-Cre mice castrated at 6 weeks, only differing by the additional presence of occasional prostatic glands that contain regressing mPIN lesions.

Identification of downstream targets that promote prostate tumor initiation and progression

Next we performed microarray analyses to investigate the molecular basis by which transgenic AR and stabilized β-catenin expression may promote prostate tumorigenesis in these transgenic mice. We carefully isolated prostatic tumor tissues from 4-month-old Ctnnb1^L(ex3)/+:PB-Cre4 and Ctnnb1^L(ex3)/+/R26hAR^L/+:PB-Cre4 mice and microscopically confirmed that the tumor tissues used to prepare RNA samples for microarray studies were composed of >80% tumor cells. Analyses of gene-expression profiles yielded differential expression of more than eightfold in 625 genes in samples from Ctnnb1^L(ex3)/+/R26hAR^L/+:PB-Cre4 and Ctnnb1^L(ex3)/+:PB-Cre4 mice in comparison to those from age- and sex-matched R26hAR^L/+:PB-Cre4 mice, reflecting that the change in the expression of these genes resulted from stabilized β-catenin expression. The top 100 upregulated and downregulated genes are shown in Supplementary Figure S6. With real-time quantitative reverse transcriptase–PCR (qRT–PCR), we confirmed an upregulation of Spp1, Egr1, Bex1, Sp6, Lef1, Axin2, Cd44 and c-Myc among those genes (Figures 7a–c). Of note, expression of Spp1, Egr1, Bex1, c-Myc and Sp6 transcripts in prostate tissues isolated from Ctnnb1^L(ex3)/+/R26hAR^L/+:PB-Cre4 mice was significantly higher than those from Ctnnb1^L(ex3)/+:PB-Cre4 or R26hAR^L/+:PB-Cre4 mice (Figures 7b and c), implying the need for both transgenic AR and stabilized β-catenin expression in regulating transcription of these genes. Intriguingly, previous studies have implicated the promotional roles of these gene products in PCa initiation and progression.^{34, 35, 36} Using immunohistochemistry, we demonstrated an increased expression of Spp1 and Egr-1 proteins in the prostates of Ctnnb1^L(ex3)/+/R26hAR^L/+:PB-Cre4 mice (Figure 7d). Expression levels of Lef1, Axin2 and Cd44, which have been documented as β-catenin downstream targets, were higher in the samples isolated from Ctnnb1^L(ex3)/+:PB-Cre4 mice than those from Ctnnb1^L(ex3)/+/R26hAR^L/+:PB-Cre4 mice (Figure 7c). Additionally, to further explore the possible molecular basis for stabilized β-catenin and transgenic AR in regulating gene expression, we performed chromatin immunoprecipitation (ChIP) assay with prostatic cells isolated from 3-month-old Ctnnb1^L(ex3)/+:PB-Cre4 and Ctnnb1^L(ex3)/+/R26hAR^L/+:PB-Cre4 mice. We analyzed the mouse c-Myc locus, one of the upregulated genes in prostate tissues isolated from Ctnnb1^L(ex3)/+/R26hAR^L/+:PB-Cre4 mice with the immunoprecipitated genomic DNA by qPCR using specific primers spanning the mouse c-Myc locus (Figure 7e).³⁷ We observed the recruitment of both AR and β-catenin at β-catenin/TCF4-binding regions on the mouse c-Myc locus (Figure 7f). Interestingly, the enrichments of β-catenin and AR at β-catenin/TCF4-binding regions were higher in the samples isolated from Ctnnb1^L(ex3)/+/R26hAR^L/+:PB-Cre4 mice than ones from Ctnnb1^L(ex3)/+:PB-Cre4 mice. Overall, the data indicate that co-activation of transgenic AR and β-catenin expression regulates the transcription of a unique subset of genes, which may contribute to prostate tumor initiation and progression.

Figure 7

Co-expression of transgenic AR and stabilized β-catenin induces a subset of gene expression, which may contribute to prostate tumor initiation and progression. (a) Gene microarray analysis showing the upregulated genes in Ctnnb1^L(ex3)/+:PB-Cre and Ctnnb1^L(ex3)/+/R26hAR^L/+:PB-Cre prostates. RNAs extracted from 4-month-old Ctnnb1^L(ex3)/+/R26hAR^L/+:PB-Cre, Ctnnb1^L(ex3)/+:PB-Cre and R26hAR^L/+:PB-Cre prostates were used for microarray analysis. (b, c) qRT–PCR confirmation of the gene alterations in Ctnnb1^L(ex3)/+:PB-Cre and Ctnnb1^L(ex3)/+/R26hAR^L/+:PB-Cre prostates. (d) Adjacent prostate tissue slides of 4-month-old Ctnnb1^L(ex3)/+/R26hAR^L/+:PB-Cre, Ctnnb1^L(ex3)/+:PB-Cre and R26hAR^L/+:PB-Cre mice were stained with Spp1 and Egr-1 antibodies. (e) Schematic representation of the mouse c-Myc locus containing two β-catenin/TCF4-binding regions. (f) Cells isolated from prostate tissues of 3-month-old Ctnnb1^L(ex3)/+/R26hAR^L/+:PB-Cre and Ctnnb1^L(ex3)/+:PB-Cre mice were subjected to ChIP using β-catenin or AR antibodies or normal IgG. Immunoprecipitated chromatin fragments were then analyzed by qPCR using specific primers for amplicon “A” (f, left panel) and amplicon “B” (f, right panel) spanning the mouse c-Myc locus as indicated in panel e. Results are presented as the percentage immunoprecipitated over input. **P<0.01.

Discussion

It has been over a decade since the interaction between AR and β-catenin has been demonstrated in PCa cells.^{26, 27, 28, 38} However, the biological consequences of this interaction in prostate tumorigenesis remain unclear. In this study, we directly addressed this long-term unclear question using a series of newly developed mouse models. We observed accelerated invasive tumor development and a shortened survival in Ctnnb1^L(ex3)/+/R26hAR^L/+:PB-Cre4 mice, in which the expression of both transgenic AR and stabilized β-catenin proteins is concurrent in prostate epithelial cells. In contrast, Ctnnb1^L(ex3)/+:PB-Cre4 mice, which only have stabilized β-catenin expression in the prostate, displayed a milder tumor phenotype. Because endogenous AR expression exists in prostatic luminal cells of Ctnnb1^L(ex3)/+:PB-Cre4 mice and castration prior to puberty did not result in tumor development, our observation implies an essential role for androgen signaling in enhancing β-catenin-mediated prostate tumor initiation. Importantly, a recent study has shown an enrichment of AR and Wnt signaling pathways in early-onset PCa, but not in elderly-onset PCa,²⁹ highlighting a clinical significance for this study. As mutations in APC, β-catenin and other components of the β-catenin destruction complex have rarely been detected in PCa cells, results from this current study suggest a possible mechanism underlying abnormal activation of Wnt/β-catenin signaling in prostate tumorigenesis. As >100 AR co-regulators have been identified,³⁹ with their roles in prostate tumorigenesis remaining largely uncharacterized, this study provides a novel model that recapitulates the biological roles of AR and β-catenin, its co-regulator, during the course of PCa initiation and progression in vivo.

To determine the extent of the role that androgens have in prostate tumorigenesis, we pursued a series of castration experiments in this study. When mice were castrated at 2 weeks, prostate gland development was impaired in all mice with different genotypes and no PIN or other malignant lesions was observed in both Ctnnb1^L(ex3)/+/R26hAR^L/+:PB-Cre4 and Ctnnb1^L(ex3)/+:PB-Cre4 mice at 16 weeks of age, indicating an essential role of androgens in β-catenin-mediated oncogenic transformation. To further examine the effect of androgens on maintenance and progression of prostate tumors, we analyzed both Ctnnb1^L(ex3)/+/R26hAR^L/+:PB-Cre4 and Ctnnb1^L(ex3)/+:PB-Cre4 mice that were castrated at 6 or 16 weeks of age, when PINs or prostate adenocarcinomas had already developed (Figure 6a). We observed significant regression of PIN and prostatic adenocarcinomas in both Ctnnb1^L(ex3)/+/R26hAR^L/+:PB-Cre4 and Ctnnb1^L(ex3)/+:PB-Cre4 mice. Importantly, when androgen pellets were implanted into castrated mice, we observed a re-growth of transgenic AR and stabilized β-catenin-positive prostate tumor cells (Supplementary Figure 5S). Histological analyses of those prostate tissues showed more severe pathological changes in the prostate tissues isolated from Ctnnb1^L(ex3)/+/R26hAR^L/+:PB-Cre4 than from Ctnnb1^L(ex3)/+:PB-Cre4 mice. These data highlight the importance of androgen signaling in maintaining and promoting β-catenin-mediated prostatic tumor formation and progression and also correlate these mouse models with human early-onset PCa, which are thought to have a unique AR signature.²⁹ These features that mimic some unique aspects of early-onset PCa may also enable further study of these pathways potentially resulting in more target therapy development for PCa.

We observed either high-grade PIN or prostatic adenocarcinoma lesions in regenerated samples using epithelial cell suspensions isolated from Ctnnb1^L(ex3)/+:PB-Cre4 or Ctnnb1^L(ex3)/+/R26hAR^L/+:PB-Cre4 mice, respectively. Because prostatic cells with transgenic AR and stabilized β-catenin expression were also labeled with GFP expression, we confirmed that the regenerated PIN and adenocarcinoma lesions were derived solely from transgenic AR and stabilized β-catenin-expressing epithelial cells. In this set of experiments, regenerated PIN and cancerous lesions were very similar to the malignant changes that appeared in the prostates of Ctnnb1^L(ex3)/+:PB-Cre4 or Ctnnb1^L(ex3)/+/R26hAR^L/+:PB-Cre4 mice. The cell suspensions isolated from Ctnnb1^L(ex3)/+/R26hAR^L/+:PB-Cre4 developed prostatic adenocarcinomas, whereas suspensions from Ctnnb1^L(ex3)/+:PB-Cre4 mice only produced high-grade PIN lesions. This result further demonstrates a synergistic effect of co-expression of transgenic AR and stabilized β-catenin in promoting oncogenic transformation in prostate cells. In this study, we also directly implanted mouse PCa tissues in the grafting assays⁴⁰ and observed the similar results (Supplementary Figure S4).

Using microarray approaches, we searched for the molecular targets that are responsible for PIN and prostatic adenocarcinoma development. We observed higher expression of several β-catenin downstream targets, including Lef1, Axin2 and Cd44, in the prostatic tissues of Ctnnb1^L(ex3)/+:PB-Cre4 compared with those from Ctnnb1^L(ex3)/+/R26hAR^L/+:PB-Cre4 and R26hAR^L/+:PB-Cre4 mice. We also observed a subset of pro-tumorigenic and pro-metastatic genes that show higher expression levels in prostate tissues of Ctnnb1^L(ex3)/+/R26hAR^L/+:PB-Cre4 compound mice, including Spp1, Egr-1, Bex1, c-Myc and Sp6. SPP1 is a pro-metastasis invasion gene identified in human PCa that has been suggested as a prognostic marker for disease recurrence and lethal metastasis.³⁴ Increased Spp1 expression has been shown to contribute to the metastatic phenotype in both in vitro and in vivo studies. EGR-1 (early growth response-1) was upregulated in human prostate tumors and related to tumor progression.³⁵ Interestingly, Bex1 and Sp6 have also been implicated in breast cancer therapy resistance and linked to the regulation of Wnt-BMP signaling pathway.^{41, 42} Because both AR and β-catenin are transcriptional regulators, we assessed their collaborative role in regulating transcription using our mouse models. We detected the recruitment of both AR and β-catenin on the endogenous mouse c-Myc promoter region. Intriguingly, the intensity of PCR fragments containing the β-catenin-binding regions was higher in the samples of Ctnnb1^L(ex3)/+/R26hAR^L/+:PB-Cre4 mice than in those of Ctnnb1^L(ex3)/+:PB-Cre4 mice, suggesting that co-recruitment of the AR and β-catenin on the c-Myc promoter increases transcription activity in PCa cells of Ctnnb1^L(ex3)/+/R26hAR^L/+:PB-Cre4 mice. Currently, we are pursuing more mechanistic in-depth experiments to investigate the collaborative regulation of AR and β-catenin in distinct gene expression.

Finally, we observed that the most atypical and tumor cells in PIN and adenocarcinoma lesions of both Ctnnb1^L(ex3)/+:PB-Cre4 and Ctnnb1^L(ex3)/+/R26hAR^L/+:PB-Cre4 mice were E-cadherin and CK8 positive, but synaptophysin negative, suggesting that oncogenic transformation may be initiated in luminal epithelial cells. It has been shown that prostatic luminal epithelial cells are AR positive and can function as tumor-initiating cells,^{43, 44} a feature of the tumor cells that we confirmed in our regeneration assays. Given the evidence that the probasin promoter is primarily active in prostatic luminal cells, our current data suggest that the dysregulation of androgen and β-catenin signaling pathways results in prostatic oncogenic transformation predominately in luminal epithelial cells. Furthermore, as demonstrated in this study, identification of a synergistic effect of AR and β-catenin on PCa development and progression suggests a novel mechanism for dysregulation of AR and β-catenin expression that is different from that of single gene, AR or β-catenin, alterations that potentially recapitulates features observed in early-onset PCa tumorigenesis. Therefore, more in-depth investigations using these newly developed mouse models should further enhance our knowledge regarding androgen and Wnt signaling pathways in prostate tumorigenesis.

Materials and methods

Mouse breeding, genotyping and castration

All animal experiments performed in this study were approved by the Administrative Panel on Laboratory Animal Care at Stanford University. All mice were used in this study were on a C57BL/6 background. Mice containing the conditional Ctnnb1 allele (Ctnnb1^lox(ex3)) were obtained from Dr Makoto M Taketo.⁴⁵ R26hAR^loxP/wt:PB-Cre4 mice were generated by crossing R26hAR^loxP/wt (Zhu et al.³¹) and PB-Cre4 mice.⁴⁶ Ctnnb1^{lox(ex3)/lox(ex3)} female mice were intercrossed with R26hAR^loxP/wt:PB-Cre4 males to generate Ctnnb1^L(ex3)/+/R26hAR^L/+:PB-Cre4 and Ctnnb1^L(ex3)/+:PB-Cre4 littermates. Mice were genotyped by PCR as described previously.^{31, 47, 48} The forward primer, 5′-TCCTCAGAGAGCCTCGGCTAGGTAG-3′, was used with the reverse primer, 5′-CCGTAAGTTATGTAACGCGGAACTC-3′ or 5′-TCTGTCTAGGGGTTGGATAAGCCAG-3′, to detect the AR target or wild-type allele,³¹ respectively. The Ctnnb1 conditional and wild-type alleles were detected with the forward 5′-AACTGGCTTTTGGTGTCGGG-3′ and reverse primers 5′-TCGGTGGCTTGCTGATTATTTC-3′.⁴⁵ The forward primer 5′-TTGCCTGCATTACCGGTCGATGCA-3′ and the reverse primer 5′-GATCCTGGCAATTTCGGCTAT-3′ were used to detect the Cre transgene. Genomic DNA fragments were amplified at 95 °C for 3 min, 95 °C for 45 s, 58 °C for 40 s and 72 °C for 60 s for 35 cycles and then at 72 °C for 5 min.

For castration, the mice were anesthetized by intraperitoneal injection of ketamine and xylazine. Both testicles and epididymis were removed through a scrotal approach. The distal end of the spermatic cord was ligated with silk thread as described previously.⁴⁹ For androgen treatment, testosterone pellets (12.5 mg, Innovative Research of America, Sarasota, FL, USA) were placed in castrated mice subcutaneously to restore serum testosterone level.

In vivo prostate regeneration assays

We obtained the double-fluorescent mT/mG:Cre reporter strain from Dr Liqun Luo³³ and used it to generate mT/mG/Ctnnb1^L(ex3)/+/R26hAR^L/+:PB-Cre4 and mT/mG/Ctnnb1^L(ex3)/+:PB-Cre4 mice. Prostate tissues from the above mice were minced into small pieces of 1 mm³, digested in Dulbecco’s modified Eagle’s medium/Collagenase/fetal bovine serum for 3 h at 37 °C and 0.25% Trypsin-EDTA (Invitrogen, Carlsbad, CA, USA) on ice for 1 h. Digested cells were pipetted vigorously to dissociate cell clumps and then passed through 70-μm cell strainers (BD Biosciences, San Jose, CA, USA) to obtain single-cell suspensions. Mouse UGSM cells were prepared as described previously.⁵⁰ UGSM cells were cultured in Dulbecco’s modified Eagle’s medium/5% fetal bovine serum/5% Nu-serum/5 μg/ml insulin/10⁻⁸M 5α-dihydrotestosterone. Approximately, 1 × 10⁵ of prostatic epithelial and UGSM cells were mixed and resuspended in transplantation solution, 1:1 (v/v) phosphate-buffered saline:Matrigel (BD Biosciences) and transplanted under the renal capsule of 6–8-week-old male NOD/SCID mice.

Histological analyses and immunohistochemistry

In this study, we used the new guidelines recommended by The Mouse Models of Human Cancers Consortium Prostate Pathology Committee in 2013 for our pathological analyses.³² Mouse tissues were fixed and processed as described in our previous study.³¹ Tissue slides were exposed to different first antibodies in phosphate-buffered saline with 1% goat serum at 4 °C overnight, including 1:500 dilution of anti-human AR (sc-7305, Santa Cruz, Dallas, TX, USA), 1:500 dilution of anti-mouse/human AR (sc-816, Santa Cruz), 1:250 dilution of anti-p63 (sc-8431, Santa Cruz), 1:3000 dilution of anti Ki67 (NCL-ki67, Novacastra, Buffalo Grove, IL, USA), 1:500 dilution of anti-β-catenin (sc-7199, Santa Cruz), 1:300 dilution of anti-E-cadherin (c20820, Transduction Laboratories, San Jose, CA, USA), 1:800 dilution of anti-CK5 (PRB-160P, Covance, San Diego, CA, USA), 1:800 dilution of anti-CK8 (MMS-162P, Covance), 1:200 dilution of anti-synaptophysin (18-0130, Invitrogen), 1:100 dilution of anti-SPP1 (91655, Abcam, Cambridge, MA, USA) and 1:50 dilution of anti-Egr1 (4153, Cell Signaling, Danvers, MA, USA) antibodies. Slides were then incubated with biotinylated anti-rabbit or anti-mouse secondary antibody (BA-1000 or BA-9200, Vector Laboratories, Burlingame, CA, USA) for 1 h, horseradish peroxidase streptavidin (SA-5004, Vector Laboratories) for 30 min and then visualized by the DAB Kit (SK-4100, Vector Laboratories). All samples were subsequently counterstained with 5% (w/v) Harris Hematoxylin (Sigma-Aldrich Corp., St Louis, MO, USA).

RNA isolation and qRT–PCR assays

Mouse prostate tissues were homogenized in RNA-Bee (TEL-TEST, Inc., Friendswood, TX, USA), and total RNA was isolated as recommended by the manufacturer. RT was carried out following our previous report.⁵¹ For quantitative PCR, cDNA samples were mixed with SYBR GreenER qPCR Super Mix Universal (11762, Invitrogen) and specific primers in the MX 3005P thermocycler (Stratagene, Santa Clara, CA, USA) and performed quantitative PCR according to the manufacturer’s protocol. Relative mRNA levels were calculated by Delta Delta C(T) method.⁵² Reactions were done in triplicate, and the values were normalized by GAPDH (glyceraldehyde 3-phosphate dehydrogenase) expression levels. Primers for Spp1 (5′-ATCTCACCATTCGGATGAGTCT-3′; 5′-TGTAGGGACGATTGGAGTGAAA-3′), Egr1 (5′-TCGGCTCCTTTCCTCACTCA-3′; 5′-CTCATAGGGTTGTTCGCTCGG-3′), Sp5 (5′-TGGGTTCACCCTCCAGACTTT-3′; 5′-CCGGCGAGAACTCGTAAGG-3′), Sp6 (5′-CCTGCAACCTCTCCAGACATA-3′; 5′-GCCCTGTGAAAAGTCTACCTCC-3′), Bex1 (5′-ATGGAGTCCAAAGATCAAGGCG-3′; 5′-CTGGCTCCCTTCTGATGGTA-3′), Ect2 (5′-TTGGGCCACCAGTTATACTCA-3′; 5′-CCAGGTTCAGCATACTCGTACAA-3′), Lef1 (5′-TGTTTATCCCATCACGGGTGG-3′; 5′-CATGGAAGTGTCGCCTGACAG-3′), c-Myc (5′-CCCTATTTCATCTGCGACGAG-3′; 5′-GAGAAGGACGTAGCGACCG-3′), Axin2 (5′-ATGAGTAGCGCCGTGTTAGTG-3′; 5′-GGGCATAGGTTTGGTGGACT-3′), Cd44 (5′-TCGATTTGAATGTAACCTGCCG-3′; 5′-CAGTCCGGGAGATACTGTAGC-3′), and Gapdh (5′-AGGTCGGTGTGAACGGATTTG-3′; 5′-TGTAGACCATGTAGTTGAGGTCA-3′) were synthesized and used in the qPCR reactions, respectively.

Gene expression microarray and analysis

RNA samples were isolated from age-matched mice of different genotypes. Fluorescently labeled cDNA probes were prepared from extracted RNA samples and hybridized to Affymetrix Mouse Gene ST 2.0 arrays (Affymetrix, Santa Clara, CA, USA). Array hybridization, washing and scanning were carried out per the manufacturer's instructions. Data processing and analysis were performed with the Affymetrix Expression and Transcriptome Analysis Console software (Affymetrix).

Co-immunoprecipitaion and western blotting

Mouse prostate tissues were isolated and processed as described previously.^{31, 47, 48} Whole cell lysates were precleared for 20 min with 10 μl of protein-A/G-agarose beads (Pierce, Waltham, MA, USA) and then incubated with preequilibrated protein-A/G-agarose beads with AR (sc-816, Santa Cruz) or β-catenin antibody (610154, BD Biosciences) at 4 °C for 3 h. The beads were washed three times in lysis buffer and eluted by boiling in SDS-PAGE (sodium dodecyl sulfate-polyacrylamide gel electrophoresis) sample buffer. After SDS-PAGE, proteins were transferred to nitrocellulose (Schleicher & Schüll, Keene, NH, USA) and blocked in TBS-T (50 mm Tris-HCl, 150 mM NaCl and 0.08% Tween 20) with 5% dry nonfat milk. Membranes were probed with β-catenin (610154, BD Biosciences), human AR (sc-7305, Santa Cruz) or tubulin (MS-581, Lab Vision, Fremont, CA, USA) antibody. Anti-rabbit or mouse immunoglobulin G conjugated to horseradish peroxidase were used as secondary antibodies (Promega, Madison, WI, USA). Detection was performed with enhanced chemiluminescence reagents (Amersham Biosciences, Pittsburgh, PA, USA).

ChIP assays

ChIP assays were performed as described previously ⁵³ In brief, single-cell suspensions of prostatic cells isolated from 3-month-old Ctnnb1^L(ex3)/+:PB-Cre and Ctnnb1^L(ex3)/+/R26hAR^L/+:PB-Cre mice were incubated with 1% formaldehyde at room temperature. The crosslinking was quenched with 0.1 M glycine for 5 min. The cells washed sequentially with cold phosphate-buffered saline, wash buffer I (0.25% Triton X-100, 10 mM EDTA, 0.5 mM EGTA and 10 mM HEPES at pH 6.5) and wash buffer II (200 mM NaCl, 1 mM EDTA, 0.5 mM EGTA and 10 mM HEPES at pH 6.5). Cells were then lysed in lysis buffer (1% SDS, 10 mM EDTA, 50 mM Tris (pH 8.1). The chromatin was sheared to an average size of 500 bp by sonication, diluted 10-fold in ChIP dilution buffer (2 mM EDTA, 150 mM NaCl, 20 mM Tris-HCl (pH 8.1) and 1% Triton X-100), subjected to immunoprecipitation with either AR (sc-816, Santa Cruz) or β-catenin antibody (610254, BD Biosciences) overnight at 4 °C and recovered with protein-A/G-agarose beads (Pierce). The immunoprecipitates were serially washed with different TSE (Triton/SDS/EDTA)-based buffers and eluted for PCR analysis. The immuno-complexes were eluted from the beads through incubation with 10 × bead volume of elution buffer (1% SDS and 0.1 M NaHCO₃). Crosslinks were reversed by incubating the elution samples at 65 °C for 6 h, and chromatin DNA fragments were purified with a PCR Purification Kit (Qiagen, Valencia, CA, USA). The above samples and inputs were analyzed by real-time qPCR using specific primers for the β-catenin/TCF4-binding regions on mouse c-Myc locus;³⁷ primers for amplicon 'A', 5′-ACTCATTCGTTCGTCCTTC-3′ and 5′- CCTCGCTCCACACAATAC-3′, and for amplicon 'B', 5′-CTCACTGGAACTTACAATCTG-3′ and 5′-CAACGCCCAAAGGAAATC-3′, were synthesized and used in the qPCR reactions.

Statistical analyses

Data are shown as the mean±s.d. Differences between groups were examined by two-tailed Student’s t test or two-way analysis of variance for comparisons among multiple groups. For all analyses, P<0.05 was considered statistically significant.