A role for androgen receptor variant 7 in sensitivity to therapy: Involvement of IGFBP-2 and FOXA1

Highlights

• •The presence of ARV7 reduces the sensitivity to treatments including cabazitaxel and enzalutamide.
• •The presence of ARV7 promotes cell proliferation and motility.
• •The interplay between IGFBP-2 and FOXA1 may contribute to the effects of ARV7.
• •IGFBP-2 may be a target gene, that is differentially regulated by AR and ARV7.

Abstract

Prostate cancer (PCa) is one of the leading causes of cancer-related deaths in men. Localised PCa can be treated effectively, but most patients relapse/progress to more aggressive disease. One possible mechanism underlying this progression is alternative splicing of the androgen receptor, with AR variant 7(ARV7) considered to play a major role. Using viability assays, we confirmed that ARV7-positive PCa cells were less sensitive to treatment with cabazitaxel and an anti-androgen-enzalutamide. Also, using live-holographic imaging, we showed that PCa cells with ARV7 exhibited an increased rate of cell division, proliferation, and motility, which could potentially contribute to a more aggressive phenotype. Furthermore, protein analysis demonstrated that ARV7 knock-down was associated with a decrease in insulin-like growth factor-2 (IGFBP-2) and forkhead box protein A1(FOXA1). This correlation was confirmed in-vivo using PCa tissue samples. Spearman rank correlation analysis showed significant positive associations between ARV7 and IGFBP-2 or FOXA1 in tissue from patients with PCa. This association was not present with the AR. These data suggest an interplay of FOXA1 and IGFBP-2 with ARV7-mediated acquisition of an aggressive prostate cancer phenotype.

Introduction

Worldwide, prostate cancer (PCa) is the most commonly diagnosed cancer in males, contributing to 6.6% of all deaths from cancer [1,2] . Localised PCa can be treated effectively by radical prostatectomy together with radio- and hormone therapy. Advanced and metastatic PCa is managed using androgen deprivation therapy (ADT), androgen receptor (AR) agonists/antagonists or inhibitors like enzalutamide (MDV-3100) that block translocation of AR to the nucleus therefore preventing ligand binding [3,4] or with chemotherapy using taxanes, like cabazitaxel [5]. A recent systematic review showed that the treatment sequence for metastatic castration resistant PCa (mCRPC) patients of abiraterone followed by enzalutamide improved progression free survival (PFS) that was quicker and less costly [6,7]. After an initial response, most patients develop resistance and relapse mainly due to alternative mechanisms of AR signalling (AR amplification, alternative splicing, intra-tumoral androgen production and others) and progress to CRPC that accounts for the majority of PCa-related deaths [8]. Alternative splicing of the AR results in generation of AR variants, including AR variant7 (ARV7)(Fig 1Ai). These variants lose their AR binding domains, are constitutively activated, and maintain their nuclear localisation sequence and DNA-binding domain [9], [10], [11], [12]. ARV7 is generated by splicing inclusion of CE3, a cryptic exon found in intron 3 [13] that results in active nuclear ARV7 protein regardless of androgen presence [14]. The expression of ARV7 correlates with androgen-independent cell proliferation and progression to CRPC [15], [16], [17] that has been further reviewed recently [18,19].

In addition to the established role of AR in PCa, other key molecules are recognised to play a part. For example, circulating insulin-like growth factor binding protein-2(IGFBP-2) is increased in patients with PCa and positively correlates with stage/grade of the tumour [20,21]. We also previously reported that IGFBP-2 induces PCa cell growth and contributes to chemoresistance [22,23]. More recently, we reported a novel correlation between IGFBP-2 and a pioneer transcription factor, FOXA1 in prostate cancer cells [24], that opens the chromatin for binding of additional transcription factors, such as the AR and its variants [25], [26], [27].

FOXA1 is considered an oncogene in PCa, involved for example in promoting proliferation and migration of PCa cells [28,29]. We reported that FOXA1 interacted with the IGFBP-2 gene in normal prostate epithelial cells that negatively regulated IGFBP-2. However, in cancer cells FOXA1 associating with the IGFBP-2 gene was minimal, which suggested a loss of the negative regulation [24]. This study aimed to further understand the role of ARV7 in prostate cancer and to interrogate potential links with and regulation of FOXA1 and IGFBP-2 with a view to identifying novel ways of optimising sensitivity to current treatments.

Materials and methods

Materials

All chemicals, unless otherwise stated, were purchased from Sigma, UK. Enzalutamide was bought from ApexBio, US (Cat No. A3003), Cabazitaxel from Cambridge Bioscience, UK(CAY22262) and foetal bovine serum (FBS) from Invitrogen, USA.

Cell culture

LNCaP, 22Rv1 and VCaP cells were obtained from the American Type Culture Collection (ATCC) and have been authenticated by short tandem repeat (STR) analysis and are routinely tested for mycoplasma (Cat No PK-CA20-700-20). LNCaP and VCaP cells were cultured as described previously [22]. 22Rv1 cells were maintained in RPMI1640 media with 10%v/v FBS, 1%v/v L-glutamine solution(2mM). Doxycycline-inducible 22Rv1 cells targeting GFP (22Rv1GFP) or ARV7 (c22Rv1shARV7) were a kind gift from Cato's laboratory and were maintained as described previously [30]. Dead cells were assessed using trypan blue cell counting as described previously [31].

Cell viability assay (MTT)

To evaluate cell viability MTT assay was conducted according to our previously published protocol [32]. In brief, all cell lines were seeded at 3 × 10⁴ cells/ well in a 96-well plate and incubated overnight in growth media (GM) for 24h and transferred to serum-free media(SFM) supplemented with sodium bicarbonate(1mg/mL), bovine serum albumin(0.2mg/mL) and transferrin(0.01mg/mL) for a further 24h. Cells in each well were then treated with enzalutamide (0–100µM) or cabazitaxel (0-640nM) for 24 or 48 hours, after which 10 μL of the MTT solution (7.5 mg/mL) were added. Cells were then incubated for an additional 2 hours at 37°C in the dark and the reaction was stopped by addition of 50µl/well of acidified triton buffer (0.1M HCL, 10% v/v Triton X-100 in water). Tetrazolium crystals were dissolved by mixing on a plate shaker for 20min. The absorbance was determined using a microplate reader at 595 nm.

Western immunoblotting

Equal amounts of proteins from whole cell lysates, determined using a BCA protein assay (Thermo Fisher Scientific, 23225) were separated on SDS-PAGE gels as described previously [22]. The membranes were probed with antiserum against: AR (1:1000 Cell Signaling, D5F11), ARV7 (1:500: Precision Antibody AG10008), FOXA1 (1:1000 Thermo Fisher, PA5-27157), IGFBP2 (1:1000 Abcam, ab109284), GAPDH (1:5000 Millipore, MAB 374). A secondary anti-rabbit or anti-mouse antibody was used (1:2000 or 1:5000 dilution respectively). Images were quantified using ImageJ and relative optical densities were calculated by adjusting to a loading control of GAPDH. Peroxidase binding was visualised by enhanced chemiluminescence and detected using ChemiDoc XRS+ System and analysed using Image Lab Software (BioRad, 170-8265). Western immunoblots were quantified using BioRad Quantity One 4.6.5 1-D Analysis Software. Images were quantified using ImageJ and relative optical densities were calculated by adjusting to a loading control of GAPDH.

Quantitative RT PCR

Total RNA from cells seeded at 1  ×  10⁶ cells/T25 flasks was extracted using TRIzol reagent (Invitrogen) as previously described [22]. ARV7 primers for PCR were used with the following sequences: forward 5′-TGTCCATCTTGTCGTCTTCG -3′ and reverse 5′-CAGCCTTTCTTCAGGGTCTG -3′ (primer size 162 bp). GAPDH primers with the following sequences were used for normalization: forward 5′- CATCTTCTTTTGCGTCGCCA -3′ and reverse 5′- TTAAAAGCAGCCCTGGTGACC-3′ (primer size 140 bp) (both purchased from Sigma).

Live cell imaging using phase holographic imaging

Cells were plated on 24-well Lumox plates (Sarstedt, 94.6000.014) and imaged using a Holomonitor M4 live cell imaging system (Phase Holographic Imaging, PHI) inside an incubator (set to 37°C and 5% CO2). Analysis of data was performed using HoloMonitor App Suite 3.5.1 (PHI AB, Lund, Sweden). In depth analysis based on selected position per treatment was performed using HStudio (PHI AB, Lund, Sweden).

The prostate cancer: evidence of exercise and nutrition trial (PrEvENT)

Participants: Details of the 96 men participating in PrEvENT trial (14/SW/0056) were published previously [33,34].

Tissue: Prostate tissue was formalin-fixed, and paraffin embedded (FFPE), cut and sections collected on adhesive slides. Patient's clinical/pathological data were extracted and analysed anonymously from medical records for the purpose of this study.

Immunohistochemistry

Ventana BenchMark Ultra Immunostainer system (Ventana Medical Systems) was used to stain tissue sections according to the manufacturer's protocol. In brief, 4µm paraffin sections were dried at 60°C/1h before immunostaining. After deparaffinisation and pre-treatment with cell conditioner, slides were incubated with primary antibody against AR (1:150, Santa Cruz), ARV7 (1:1000, Precision Antibody), FOXA1 (1:450, Abcam) and IGFBP-2 (1:2000, Abcam) for 2h. Slides were counterstained with Haematoxylin(8min) and bluing reagent(4min) (Ventana Medical Systems), then dehydrated and mounted in distyrene plasticizer and xylene mountant (DPX, SIGMA). An Allred scoring system was used to assess AR and ARV7 staining as described previously [35].

Statistical analysis

Data were analysed with SPSS 24.0 presented as the mean ± S.E.M of a minimum of three independent experiments (or SD for holomonitor experiments). Independent sample t-test was used to compare means of two groups and for the means of more than two groups, one-way analysis of variance (ANOVA) followed by least significant difference (LSD) post hoc test was used. For both tests, a p-value of equal or less than 0.05 was considered statistically significant. Comparison between discrete variables like scoring for tissue markers was performed using Spearman rank correlation. GraphPad Prism 9.0 (GraphPad Software, La Jolla, CA, USA) was used to calculate IC50 for viability assays.

Results and discussion

The presence of ARV7 alters the sensitivity to cabazitaxel and enzalutamide

AR mRNA and protein are still detected during the switch to androgen-independence [36,37] with ARV7 reported to contribute to a more aggressive PCa phenotype. Antonarakis et al. showed that circulating tumour cells (CTCs) expressing ARV7 mRNA are linked with resistance to enzalutamide and abiraterone [38,39]. A recent systematic review and meta-analysis showed that 70% of ARV7 positive PCa cases, had a Gleason score >=8 compared to ARV7 negative ones (50%, OR 1.68, 95% CI 1.25-2.25, p < 0.001). The analysis also confirmed higher rates of bone metastasis for ARV7-positive versus ARV7-negative cancers (81.7% versus 69.0%; OR 1.97, 95% CI 1.44-2.69, p < 0.001) [40].

In our study, we assessed the level of endogenous ARV7 protein in LNCaP (castration-sensitive [41]), VCaP (hormone refractory PCa [42]), and 22Rv1 (derived from a human prostatic carcinoma xenograft [43]) cells. ARV7 protein was detected in VCaP, but not in LNCaP cells (Fig 1Aii). ARV7 was present in the 22Rv1GFP- control cells but not in the 22Rv1shARV7 cells, with ARV7 stably silenced (Fig 1Aiii), that was also reflected at the mRNA level (Fig 1Aiv). Notably, as was reported by Cato et al [30], we confirmed that stable knockdown of ARV7 in the 22RV1 cells had no significant effects on levels of the AR in either the cytoplasm or the nucleus (Supp. Fig 1).

We then investigated the impact of ARV7 in relation to chemosensitivity. Cabazitaxel dose-dependently decreased metabolic activity with half maximal inhibitory concentration (IC50) reaching 50μM versus 61μM for LNCaP (ARV7 negative) and VCaP (ARV7 positive) cells, respectively (Fig 1B i and ii). Similarly, with 22Rv1 cells, the IC50 for 22Rv1 shARV7 was 124μM compared with 228μM for the control 22Rv1 GFP cells (Fig 1B iii and iv). Treatment with an anti-androgen enzalutamide, also resulted in a dose-dependent inhibition of metabolic activity, with an IC50 of 56μM for LNCaP and 77μM for VCaP cells (Fig 1B v and vi). Enzalutamide also inhibited the MTT activity of 22Rv1 cells, with an IC50 of 78.4μM for 22Rv1 shARV7 and 118μM for the control cells (Fig 1Bvii and viii). These data suggest that the presence of ARV7 contributes to resistance to both chemo- and anti-hormone therapies.

Fig. 1.

A (i) Schematic illustrating the structure of ARV7 generated by splicing cryptic exon 3 after exons 1-3. Western blot showing abundance of ARV7 proteins in relation to housekeeping GAPDH in (ii) LNCaP, VCaP (iii) 22Rv1 GFP and 22Rv1 shARV7 and (iv) PCR assessment showing fold change of mRNA levels relative to GAPDH (n=3). The blots are representative of experiments repeated three times. The PCR is the mean of three independent experiments. Melt curves were performed for each RT-PCR analysis to ensure that no non-specific amplification was occurring (data not shown). B. Indicates graphs showing a dose-dependent response to cabazitaxel with 50% inhibitory concentrations (IC50) in LNCaP (i), VCaP(ii), 22Rv1 GFP (iii) and 22Rv1 shARV7 (iv) cell lines (n=3 or more) and to enzalutamide with IC50 in LNCaP (v), VCaP (vi), 22Rv1 GFP (vii) and 22Rv1 shARV7 (viii) cell lines (n=3 or more).

One possible explanation for such resistance to taxanes, could be supported by a mechanism proposed by Yu et al. whereby they proposed that ARV7 was responsible for inactivation of mitotic spindle assembly checkpoints (SAC) and therefore ARV7-expressing cells were more likely to withstand treatment with taxanes [44]. Another paper supporting a role for ARV7 in therapy resistance suggested that a combination of cisplatin or carboplatin with enzalutamide might degrade ARV7 via ubiquitination revealing a potentially targetable Malat1/SF2 splicing complex in CRPC, with carboplatin having less adverse effects [45]. Wilson et al. also proposed resveratrol (RSV), a polyphenol transhydroxystilbene found in grapes and red wine, to be important in downregulating ARV7 also by enhancing ubiquitination [46]. These data support that the presence of ARV7 contributes to therapy resistance and therefore may serve as a potential targetable candidate in restoring sensitivity to treatment.

The presence of ARV7 impacts on the proliferation and motility of prostate cancer cells

To assess the effects of altering ARV7 levels on proliferation and motility we focussed on the 22Rv1 cells with ARV7 (22Rv1 GFP) or without (22Rv1 shARV7) and assessed them using live digital holographic microscopy. The average motility of 22Rv1 GFP cells (in μm) was higher, than those without ARV7 (Fig 2 Ai). Tracking each individual cell allowed their movement trajectories to be mapped and are presented as rose plots. Each coloured line represents the path that an individual cell travelled. A smaller “rose” plot was noted with ARV7 negative cells (Fig 2Cii) versus positive (Fig 2Aiii) suggesting an increase of movement in ARV7 positive cells. Next, proliferation rate of these cells was assessed and 22Rv1 GFP cells had a higher average number of cells in comparison with the ARV7 knock-down cells (Fig 2 Bi). The rate of cell division was also recorded and similarly, ARV7 positive cells divided at a higher rate than cells without ARV7 (Fig 2Bi and ii, respectively). These data suggest that ARV7 promotes the rate of cell division, consequent proliferation, and motility, potentially contributing to aggressiveness of the cancer. Previous publications showed that ARV7 promoted PCa migration and upregulation of genes involved in epithelial to mesenchymal transition (EMT) and to increase colony formation [47,48]. ARV7 knock-down resulted in cell cycle arrest in the G2/M phase [44]. A recent study found a positive correlation with ARV7 and B7-H3, an immune checkpoint molecule, overexpressed in PCa that associates with poor cancer prognosis [49]. Together with our study it leads to the overall indication that ARV7 promotes oncogenic behaviour and clearly suggests the importance of AR signaling in the transition to a more aggressive phenotype.

Fig. 2.

Time-lapse was set to capture images every 10min over 48h, selecting at least 5 different regions per well. Average cell number or cell motility (um) per each region/per timepoint was normalised to the control at 0h. A. Graphs showing (i) changes in average cell motility in um over time for 22Rv1 GFP and shARV7 cells. Data points with SD from 3 or more repeats. (ii & iii) Rose plots representing cell movement trajectories for 22Rv1 shARV7 cells and for 22Rv1 GFP cells respectively. Movement trajectory of each cell overlayed in a xyz- position plot, was tracked over time and rose plots containing bigger “rose” indicated increased migration. Each colour of the lines represented the migration trajectory of individual cells. B. Graphs showing (i) changes in average cell proliferation over time normalized to 0h control for 22Rv1 GFP and shARV7 cells. Data points with SD from 3 or more repeats. Cell family tree plots representing cell division rate for 22Rv1 shARV7 cells (ii) and 22Rv1 GFP cells (iii) over time. Cell family plots were derived based on tracking each individual cell, together with its daughter cells after division. This gives information about the time of each cell division per position/ per treatment. Each coloured cell line represents individual cell and each fork on the line represents cell division.

Associations of ARV7 with IGFBP-2 and FOXA1

Knowing that the presence of ARV7 impacts on the behaviour of prostate cancer cells, our next aim was to identify potential targetable pathways or key players that could underly the mechanism of action of ARV7. FOXA1 is a well-established PCa oncogene and modulator of steroid hormones [50,51] and our previous publications have reported the importance of IGFBP-2 in PCa cell proliferation and chemosensitivity [22,23]. We also proposed FOXA1 to be a negative regulator of IGFBP-2 in normal prostate cells which was lost in cancer cells as we could no longer observe the association between FOXA1 and the IGFBP-2 gene [24].

In this study, we observed that IGFBP-2 and FOXA1 abundance was significantly reduced (by 35% and 34%) in ARV7 knocked-down cells in comparison to the 22Rv1 GFP control (Fig 3A), which suggests that ARV7 may regulate the expression of IGFBP-2 and FOXA1. We assessed the levels of ARV7, AR, IGFBP-2 and FOXA1 proteins in our PCa cohort and scoring results are presented in Table 1. Most (79.4%) of the patients were categorised as medium grade (Gleason grade=7) representing an intermediate cancer stage. Low (Gleason grade =6) and high (Gleason grade 8 or 9) grades represented only 13.7% and 6.9% of patients, respectively [52]. Spearman rank correlation analysis for all patients revealed as anticipated a significant positive correlation between AR and ARV7 protein levels within our PCa cohort (Fig 3B). In addition, we also assessed the association between ARV7 levels with Gleason grade, tumour stage, PSA, age, and body mass index in benign or tumour tissue and found no significant correlations (data not shown). We did suspect this would be the case as our cohort represents predominantly localised medium grade PCa and not aggressive CRPC.

Fig. 3.

A. Shows representative western blots indicating the abundance of FOXA1 and IGFBP-2 proteins in relation to housekeeping GAPDH in 22Rv1 GFP versus 22Rv1 shARV7. Graphs on the right show relative fold changes of optical density (OD) (n=3). B. Represents the correlations between tissue-based ARV7 and AR peptide abundance from samples of prostate collected during the PrEvENT trial. Spearman correlation is shown with r, p values.

Table 1.

Table summarising total scoring results of immunohistochemical staining of AR, ARV7, IGFBP-2 and FOXA1 proteins from prostate tissue (PrEvENT cohort). Data for all patients or stratified by Gleason grade with “Low grade” equal to Gleason grade 6, “medium grade” to Gleason grade 7 and “high grade” to Gleason grade 8 or 9.

				AR Total score		ARV7 Total score		IGFBP-2 Total score		FOXA1 Total score
Groups	Gleason Grade (N, %)	PSA (SD) N	Mean age (SD) N	Benign (SD) N	Cancer (SD) N	Benign (SD) N	Cancer (SD) N	Benign (SD) N	Cancer (SD) N	Benign (SD) N	Cancer (SD) N
All patients	7 (102, 94%)	10.3 (31.3) 102	64.4 (6.3) 104	6.4 (0.87) 93	6.6 (0.91) 93	5.6 (1.4) 97	5.9 (1.3) 97	2.4 (2.6) 99	4.2 (2.3) 97	5.9 (2.2) 101	6.5 (1.2) 101
Low grade	6 (14, 13.7%)	12 (17.7) 13	65.1 (4.6) 14	6.8 (0.83) 12	6.4 (1.2) 12	6 (1.3) 12	6.5 (1) 12	1.7 (2.3) 14	3.75 (2.8) 12	6.1 (2.3) 13	6.1 (1.6) 13
Medium grade	7 (81, 79.4%)	10.3 (9.4) 76	64.2 (6.3) 78	6.4 (0.83) 72	6.55 (0.86) 72	5.81 (1.4) 76	5.87 (1.4) 76	2.52 (2.7) 77	4.2 (2.3) 76	5.9 (2.2) 79	6.6 (1.2) 79
High grade	8 or 9 (7, 6.9%)	13.4 (6.1) 7	68.4 (5.3) 7	6.2 (1.2) 6	6.83 (0.98) 6	6 (1.5) 6	6.2 (1.2) 6	2.83 (2.6) 6	4.5 (1.8) 6	5.9 (2.8) 7	6.7 (0.76) 7

Notably though, the correlations of ARV7 and AR individually with FOXA1 and IGFBP-2 were markedly different. With all 96 patients, significant correlations were observed between ARV7 and IGFBP-2 in benign (r=0.3, P<0.01) and tended towards significance in cancer tissue r=0.2, P=0.07, respectively) (Fig 4i and ii). FOXA1 was also significantly positively associated with ARV7 in benign (r=0.24, P=0.02) and cancer tissue (r=0.29, P<0.01) (Fig 4iii and iv). Such correlations were not present when comparing associations between AR and FOXA1 or IGFBP-2, although AR and FOXA1 tended towards significance in both benign and cancer tissue (Fig 4 Bi-iv).

Fig. 4.

A. Shows correlations between tissue-based ARV7 and IGFBP-2 (i & ii) or FOXA1 (iii & iv) peptide abundance and B Shows correlations between tissue-based AR and IGFBP-2 (i & ii) or FOXA1 (iii & iv) peptide abundance from samples of prostate collected during the PrEvENT trial. Spearman correlation represented with r, p values.

Although the numbers are too small in the low and high Gleason grade groups, we performed correlation analysis within each Gleason grade subgroup for comparison (Table 2) and interestingly found that a significant association between AR and FOXA1 was seen in the low Gleason grade cancer tissue. Whilst, notable, this would need to be validated in a bigger cohort.

Table 2.

Table showing correlations between total score of immunohistochemical staining for ARV7 or AR and IGFBP-2 or FOXA1 proteins from prostate cancer tissue samples: analysis in all patients or stratified by Gleason grade, with “Low” equal to Gleason grade 6, “medium” to Gleason grade 7 and “high” to Gleason grade 8 or 9. r – Spearman's rank correlation coefficient, * p value < 0.05, ** p value < 0.01, *** p value < 0.001.

		ARV7/IGFBP-2		ARV7/FOXA1		AR/IGFBP-2		AR/FOXA1
		r (95% CI)	p value (N)	r (95% CI)	p value (N)	r (95% CI)	p value (N)	r (95% CI)	p value (N)
All patients	Benign	0.3	0.003**(93)	0.24	0.02* (96)	0.01	0.89 (89)	0.17	0.09 (92)
	Cancer	0.19	0.07 (93)	0.29	0.003** (96)	(-)0.04	0.72 (89)	0.18	0.08 (92)
Low (Gleason grade=6)	Benign	0.11	0.73 (12)	0.12	0.7 (12)	0.17	0.59 (12)	(-)0.14	0.66 (12)
	Cancer	0.1	0.76 (11)	0.48	0.11 (12)	0.28	0.39 (11)	0.61	0.03* (12)
Medium (Gleason grade=7)	Benign	0.36	0.001**(75)	0.09	0.39(76)	0.04	0.71 (70)	0.19	0.11 (72)
	Cancer	0.22	0.06 (74)	0.27	0.02* (76)	(-) 0.07	0.52 (69)	0.07	0.53 (72)
High (Gleason grade=8 or 9)	Benign	(-)0.11	0.86 (5)	0.77	0.07 (6)	0.02	0.95 (5)	0.66	0.14 (6)
	Cancer	(-)0.17	0.77 (5)	0.78	0.06 (6)	0	1(5)	0.56	0.24 (6)

We selected example tissue sections from two patients for each subgroup (3+4 or 4+3) of Gleason group 7 to illustrate the significant positive correlations and to demonstrate within this grade, that we noted large variation in the levels of ARV7 (Figure 5).

Fig. 5.

Representative images of immunohistochemical staining (IHC) of ARV7, FOXA1 and IGFBP-2 from samples of prostate collected during the PrEvENT trial. Two patients from 3+4 and two patients from 4+3 grades were selected.

These data suggest an interplay of IGFBP-2 and FOXA1 with ARV7-mediated acquisition of aggressive prostate cancer phenotype. Given the fact that our tissue samples predominantly represent localised medium grade PCa, we speculate that a cohort of more aggressive cancer would strengthen the associations.

The regulation of a unique and or overlapping set of genes by ARV7 may enable tumours that have switched from an AR to an ARV7 transcriptional programme to promote cancer progression and this has been suggested previously [50,53,54]. Literature suggests that for AR to exert its action it needs to interact with coactivators and corepressors, whereas ARV7 interacts predominantly with nuclear receptor corepressors (NCoR) [30]. We speculate that IGFBP-2 may be one target gene, that is differentially regulated by AR and ARV7 that may be dependent upon FOXA1, but this remains to be determined.

These data may contribute to the complex mechanism underlying ARV7-mediated increases in proliferation and motility in PCa cells, suggesting roles for FOXA1 and IGFBP-2 in the presence of ARV7.

Author contributions

KB performed all the work and wrote the manuscript. RB performed and analysed the IHC. AB provided clinical expertise. CP supervised and directed the project and contributed to writing the paper. All authors read and approved the final manuscript.

Concept attributable to KB, AB & CP. Data acquisition and analysis KB, RB, AS & CP. KB and CP wrote the paper. AB advised on clinicals aspects of the work. All authors commented on drafts and approved the final version of the paper.

Funding

KMB is supported by a West Wales Prostate Cancer Support Group and by Cancer Research UK (C18281/ A29019) program grant (The Integrative Cancer Epidemiology Programme) and the NIHR Bristol Biomedical Research Centre (BRC-1215-20011). The MRC Integrative Epidemiology Unit is supported by the Medical Research Council and the University of Bristol (MC_UU_12013/6, MC_UU_12013/9).

The views expressed are those of the authors and not necessarily those of the NHS, the NIHR or the Department of Health.

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.