2.1 Cell lines

The human prostate cancer cell lines, the castration resistant cell line 22Rv1 and androgen-sensitive LNCaP cells were obtained from American Type Culture Collection (ATCC, Manassas, VA). Cells were maintained in RPMI 1640 (Invitrogen, Grand Island, NY) and 10% fetal bovine serum (FBS), 100units/ml penicillin and 100 μg/ml streptomycin in a 5% CO2 incubator at 37 °C. The androgen-independent LNCaP95 cell line (derivative cells from LNCaP) was a generous gift from Dr. Stephen Plymate (University of Washington, Seattle, WA). For experiments examining responses to androgen, cells were seeded in 10% charcoal-stripped serum and were stimulated for 24 h by growth medium containing 1 nM dihydrotestosterone (DHT) (Sigma–Aldrich, St. Louis, MO) or R1881.

2.2 Antibodies

The following antibodies were used for the various experiments. Antibodies against E-cadherin, β-tubulin, Androgen Receptor (N-20), Dynein IC1/2 proteins were purchased from Santa Cruz Biotechnology (Santa Cruz, CA). The antibodies against N-cadherin and CD31 were obtained from AbCam Cell Signaling (Cambridge, UK); antibodies against PARP-1, Vimentin, GAPDH, Snail, cofilin and β-catenin proteins were obtained from Cell Signaling Technology (Danvers, MA).

2.3 Cell viability assay

The effect of the various treatments on prostate cancer cell viability was evaluated using the (Thiazolyl Blue Tetrazolium bromide (MTT) assay). Cells were seeded into 24-well plates and after grown to 60–75% confluency, were treated with vehicle control (DMSO, Sigma–Aldrich, St. Louis, MO), Docetaxel (DOC, 1 μM), EPI-002 (25 μm), or combination (1 μM DOC + 25 μM EPI-002) in RPMI 1640 with 10% CSS (Charcoal Stripped Serum) for 24 h. At termination of exposure, cells were aspirated and rinsed with PBS then treated with 250 μl/well MTT (1 mg/ml) for 30 min at 37 °C. After incubation, MTT was aspirated and formazan crystal was solubilized with DMSO. Absorbance was measured at 570 nm using μQuant Spectrophotometer (Biotech Instruments Inc., Winooski, VT).

2.4 Cell proliferation

Cells seeded into 96-well plates were pre-treated for 1hr with vehicle or EPI-002 (25 μM). After 16 h, cells were treated with Docetaxel (0.5 μM) (alone or combination with EPI-002) before addition of R1881 (0.1 nM) or vehicle under serum-free and phenol red-free conditions, and subsequently incubated for 23 h. After pulse-labeling with 10 μM BrdU (for 2 h), BrdU-labeled cells were identified with the anti-BrdU-POD (Roche). BrdU incorporation was measured at 570 nm using a VersaMax ELISA Microplate Reader (Molecular Devices). The results from three independent experiments performed triplicate were analyzed.

2.5 Western blot analysis

Total cellular protein was extracted from cell pellets by homogenization with RIPA buffer (Cell Signaling Technology, Danvers, MA). Protein samples were loaded into 4%–12% SDS-polyacrylamide gels (Bio-rad, Hercules, CA) and subjected to electrophoretic analysis and blotting. The following antibodies were used against these specific proteins: E-cadherin, N-cadherin, β-tubulin, the AR (N-20), Snail, Vimentin, PARP-1, β-catenin and GAPDH proteins. Membranes were incubated with the specific primary antibody (overnight at 4 °C) and were subsequently exposed to the relevant secondary antibody (90 min, room temperature). For signal detection, membranes were incubated with the Amersham ECL Plus Western Blotting Detection System (Amersham, GE Healthcare, Buckinghamshire, UK) for 5 min and auto-radiographed using X-ray film (Denville Scientific, South Plainfield, NJ). All protein expression bands were normalized to GAPDH expression (used as loading control).

2.6 Quantitative RT-PCR analysis

In vitro samples: RNA was extracted with the Trizol© reagent (Life Technologies, Grand Island, NY) and RNA samples (1 μg) were subjected to reverse transcription using the Reverse Transcription System (Promega, Madison, WI). TaqMan real time reverse transcriptase-PCR (Life Technologies, Grand Island, NY) analysis of the cDNA samples was conducted in an ABI7700 Sequence Detection System (Applied Biosystems, Inc, Branchburg, NJ), using the following specific primers: for Prostate Specific Antigen (KLK3; Hs02576345_m1), E-cadherin (CDH1; Hs01023894_m1), N-cadherin (CDH2; Hs00983056_m1), Vimentin (VIM; Hs00185584_m1), Snail (SNAI1; Hs00195591_m1), Twist (TWIST1; Hs01675818_s1), UGT2B17 (Hs00854486_sH), and 18S rRNA (4319413E) (Applied Biosystems, Life Technologies, Grand Island, NY). For the qRT-PCR experiments, each sample was analyzed in duplicate and data represent average values from three independent experiments. Numerical data for transcript levels were normalized to 18s rRNA in controls and expressed relative to untreated controls.

2.7 Luciferase reporter gene assays

LNCaP95 cells (1.5 × 105) and 22Rv1 cells (9.5 × 105 in 10% FBS phenol red free RPMI) were transfected with the AR responsive PSA (6.1 kb)-luciferase, probasin (PB)-luciferase and ARR3-luciferase reporters in serum-free media. At 5 h post-transfection, cells were pre-treated with vehicle or EPI-002 (EPI) (25 μM) for 16 h, and were subsequently exposed to Docetaxel (DOC) (0.5 μM), or combination of EPI and Docetaxel (EPI + DOC). Cells were incubated with R1881 (1 nM) for 23 h. After 24 h of treatment, cells were lysed and subjected to luciferase reporter activity analyses that were normalized to protein concentrations using the Glomax luminometer (Promega Corporation; Madison, WI). All transfection assays were performed in at least three independent experiments in triplicate wells.

2.8 Immunofluorescent confocal microscopy

Cells were plated (1 × 105) in chamber slides coated with fibronectin (Invitrogen). After 24–48 h, cells were exposed to medium (RPMI 1640 with 10% CSS) in the presence of DHT (1 nM), Docetaxel (DOC: 1 μM), EPI-002 (EPI: 25 μM) or in combination of the two agents. Following treatment, cells were fixed in 4% (v/v) paraformaldehyde and permeabilized with 0.1% Triton X-100 in sterile phosphate buffer saline (PBS). Fixed cells were incubated overnight with primary antibody specific for AR (N-20), Dynein IC1/2 and Tubulin, (at 4 °C) with gentle rocking and the appropriate Alexa-Fluor (Life Technologies, Grand Island, NY) fluorescent secondary (1.5 h, room temperature). Slides were mounted using Vectashield mounting medium with DAPI and were visualized using an FV1000 Confocal Microscope (Markey Cancer Center Core, University of Kentucky).

2.9 In vivo tumor targeting studies

All animal experiments were performed in accordance with the guidelines approved by the Animal Care and Use Committee of the University of British Columbia. NOD-SCID mice (6–8 weeks old) were castrated before any drug treatments. At 7 days post-castration, mice were subcutaneously injected with 22Rv1 cells (3 × 106 cells plus Matrigel™ media). Mice were subsequently divided into two groups of 10 mice each: the control group receiving 1% medium/0.1% Tween-20 daily via oral gavage; the other 10 mice in the treatment group received EPI-001 (100 mg/kg twice daily) via oral gavage. EPI-001 had no effect as a monotherapy so when tumors reached approximately 140 mm³, 5 mice in each group was intraperitoneally injected (I.P.) with Docetaxel (15 mg/kg) on day 1 and day 5; and the combination treatment of EPI-001 (100 mg/kg twice daily) and Docetaxel (15 mg/kg) for 11 days; Tumors were measured twice a week and the volume calculated by using the formula, length × height × width × 0.5236. Prostate tumor xenografts were harvested two days after the last treatment and tissue specimens were subjected to histopathological analysis. Immunostaining was conducted for the expression and cellular localization of AR, cytoskeleton organization, EMT, vascularity (CD31), and apoptosis (TUNEL, EMD Millipore, Billerica, MA). TUNEL analysis for detection of apoptotic cells was performed as previously (Pu et al., 2009).

2.10 Immunohistochemical analysis

Tissue specimens from human prostate tumor 22Rv1 xenografts were formalin fixed and paraffin-embedded; serial sections (5 μ) were subjected to immunohistochemical analysis using antibodies against E-cadherin, N-cadherin, β-tubulin, Androgen Receptor (N-20), cofilin, and CD31. After blocking nonspecific binding (1.5%NGS/TBS-Triton), sections were incubated with primary antibody (overnight, 4 °C). and subsequently incubated with biotinylated goat anti-rabbit IgG (2 h, room temperature) and horseradish peroxidase-streptavidin (EMD Millipore, Billerica, MA). Color detection was achieved with SigmaFast 3,3′-Diaminobenzidine tablets (Sigma–Aldrich, St. Louis, MO) and counterstained with haematoxylin. Images were captured via light microscopy (40× and 100×) using an Olympus BX51 microscope (Olympus America, Center Valley, PA). Content and intensity of immuno-reactivity were recorded by two independent observers (S.K.M. and N.K.).

2.11 Statistical analysis

Student's t-test or one-way ANOVA were performed using Graph Pad Prism 6 software to determine the statistical significance of difference between means. All numerical data are presented as Mean ± Standard error of the mean (SEM). Statistical significance was set at P value <0.05. ImageJ software was used for calculation of densitometry.

3.1 Combination of Docetaxel and EPI maximizes blockade of AR activity and impairs human castration resistant prostate cancer growth in vitro and in vivo

Considering that clinically significant, constitutively active AR splice variants V7 and V567es (Hornberg et al., 2011) regulate the expression of both canonical and a unique subset of gene targets that are enriched for M-phase cell cycle genes (Cao et al., 2014; Sun et al., 2010), we use two different human prostate cancer cell lines as models, harboring full length functional AR and AR variants to examine the anti-tumor effect of combined targeting of microtubules and AR NTD. The 22Rv1 human prostate cancer cells express both functional full-length AR with duplication of exon 3 and substantial levels of constitutively active AR-V7, although they do not exhibit a proliferative response to androgens (Dehm et al., 2008; Guo et al., 2009; Marcia et al., 2010). Treatment of 22Rv1 cells with Docetaxel alone (1 μM) for 24 h resulted in significant loss of cell viability (Figure 1A). Combination of EPI with Docetaxel also increased cell death (P < 0.05), while EPI as a single agent exerted no significant effect on 22Rv1 cell viability at 24 h.

To examine the anti-tumor effect of EPI alone or in combination with the microtubule-targeting chemotherapy (Docetaxel) in vivo, we used 22Rv1 xenografts in mice (castrated for 2 -weeks prior to inoculation). As shown on Figure 1B, daily treatment with the AR N-terminal inhibitor EPI (200 mg/kg), did not result in a significant growth inhibition of the 22Rv1 xenografts compared to vehicle control. The combination treatment of Docetaxel and EPI led to a significant suppression of CRPC tumor growth compared to either single treatment modality or untreated controls (Figure 1B) (P < 0.05). This dramatic prostate cancer xenograft eradication was associated with a significant increase in the number of apoptotic cells in the combination treated tumor xenografts compared to untreated control or single treatment arms, as evaluated with the TUNEL assay (Figure 1C). Assessment of the tumor vascularity based on the CD31 immuno-reactivity in xenograft tumors demonstrated a significant decrease in tumor vascularity in response to EPI and Docetaxel combination compared to untreated controls (Figure 1D; Supplemental Figure S1).

In response to Docetaxel, there was an apparent decrease in the AR immuno-reactivity in the 22Rv1 CRPC xenografts (Figure 2A). Quantitative evaluation of AR positivity and cellular localization however, did not reveal any significant changes after any of the treatments relative to untreated control (Figure 2B). As expected, treatment of mice with Docetaxel led to reduction in tubulin expression in the prostate tumor xenografts that was further enhanced by the combination treatment of Docetaxel and EPI. Quantitative analysis of cofilin immuno-reactivity revealed a significant decrease of this critical cytoskeleton protein in response to Docetaxel and the combination treatment, compared to untreated control (Figure 2C). To further interrogate these protein changes elicited by the combination treatment of Docetaxel and the NTD AR targeting, we use confocal microscopy in 22Rv1 cells in vitro. As shown on Figure 2 (panel D), treatment with Docetaxel alone (for 4 h), led to a marked reduction in β-tubulin expression and mitotic arrest (merged image, arrow heads) in CRPC. Prostate cancer cells exhibiting microtubule stabilization appear to have some cytoplasmic AR localization in 22Rv1 cells (Figure 2D). Consistent with the in vivo xenograft data (Figure 2A), there were no significant changes in the expression of AR after Docetaxel treatment (as single agent or in combination with EPI) compared to untreated control cells (Figure 2D).

Docetaxel caused a significant loss in body weight compared to animals treated with vehicle control (86% ± 3.0% vs 102% ± 1.3%, p = 0.0011), while EPI had no significant effect (98% ± 1.3%, p = 0.6088). The body weight of mice treated with EPI in combination with Docetaxel was also significantly lower than that obtained for the vehicle control (86% ± 5.4%, p = 0.0156), but not significantly different than the weight measurements from mice treated with Docetaxel monotherapy (p = 0.9529). Thus the combination treatment of EPI and Docetaxel does not appear to induce toxicity beyond that obtained from Docetaxel alone.

3.2 Impact of AR variants on taxane-mediated AR trafficking

The existence of AR variants capable of androgen-independent signaling may be critical in dictating the therapeutic response to taxanes. It was recently shown that the AR variant core consisting of the AR NTD and DNA-binding domain is sufficient for nuclear localization and androgen-independent transcriptional activation of AR target genes (Chan et al., 2012). We subsequently examined the effect of EPI and Docetaxel on the expression and cellular localization of the AR, as well as tubulin, in the 22Rv1 cells in the presence of DHT. An apparent stabilization of microtubule structures and a decrease in tubulin levels were detected in response to the combination treatment of Docetaxel and EPI (Supplementary Figure 2A). Figure 3 (panel A), reveals fluorescent images of the 22Rv1 prostate cancer cells in response to Docetaxel or EPI (as single agents), or the combination of the two agents. A modest effect of Docetaxel on androgen-induced nuclear AR translocation was observed after 4 h of treatment. This was confirmed by subcellular fractionation analysis revealing that Docetaxel treatment reduced the nuclear presence of full length AR, without causing marked changes in AR variants (nuclear vs cytosolic levels) (Supplementary Figure S2B). This is in contrast to the ability of Docetaxel and also EPI (given as single agents or in combination) to promote the cytoplasmic sequestration of AR from the nuclei in the androgen-sensitive LNCaP cells, harboring a full length AR (Supplementary Figure S3A). This impact on AR nuclear translocation by taxane and EPI-docetaxel combination in LNCaP cells was translated into a significant inhibitory effect against cell viability (Supplementary Figure S3). Western blot analysis revealed that within 6 h of treatment with EPI there was an increase in protein levels of both full-length and variant AR compared to untreated 22Rv1 cells that was suppressed by the combination treatment of EPI with Docetaxel (Figure 3B).

Dynein is a microtubule-traversing motor protein capable of efficiently facilitating nuclear transport of cytoplasmic proteins, that was previously shown to navigate AR trafficking along the microtubules and its nuclear translocation, towards induced transcriptional activity (Darshan et al., 2011). We thus examined the expression and cellular localization of dynein in 22Rv1 cells in response to the various treatments and its co-localization with AR. Androgens (DHT) increase expression of dynein and this correlates with increased AR nuclear translocation (Figure 3A; merged image). In response to Docetaxel, there was a marked alteration in dynein localization as detected by confocal microscopy (appearing as punctate sequestration), while EPI alone did not have any effect on dynein distribution/localization (Figure 3A). In marked contrast in the androgen-sensitive LNCaP cells (full length AR), treatment with Docetaxel alone or in combination with EPI, in the presence of DHT, resulted in diffuse cellular localization of dynein (Supplementary Figure S3A).

3.3 Differential effect of taxane and EPI treatment on AR vs AR-V7 target genes

To determine the impact of treatment on the functional activity of AR, the temporal induction of AR regulated genes was evaluated in 22Rv1 cells in response to Docetaxel, EPI or the combination treatment. Prostate-specific antigen (PSA/KLK3) is a gene target for the full length AR and UGT2B17 is a gene target for AR V7 variant. There was a significant increase in KLK3 mRNA expression in response to treatment (within an hour) with Docetaxel monotherapy and combination therapy (Figure 3C) (P < 0.05). There was no significant effect on KLK3 expression, in response to either EPI or Docetaxel given as single agents, or in combination (after 24 and 48 h), relative to untreated levels. A transient increase in UGT2B17 mRNA levels was detected within 1–6 h of treatment with EPI (Figure 3D). After longer treatment periods (12–48 h), upregulation of UGT2B17 in response to either EPI or combination with Docetaxel, was not sustained.

We subsequently investigated whether the combination treatment of Docetaxel and EPI, exerts an enhanced inhibitory effect of AR transcriptional activity, using three AR-driven reporter gene constructs in the 22Rv1 cells. The probasin (PB, −286/+28) reporter gene construct is the natural promoter that contains two functional androgen response elements (AREs), that comprise the androgen response region (ARR). The PSA (6.1 kb) reporter gene construct is also a natural reporter with several AREs in both the enhancer and promoter regions. The ARR3 is an artificial reporter with 3 repeats of the PB ARR in front of a minimal thymidine kinase promoter. The data shown on Figure 3, demonstrate that treatment with EPI led to significant inhibition of the androgen-induced PB-luciferase and PSA-luciferase reporters (panels E and F respectively), while unexpectedly increased the activity of the synthetic ARR3 reporter (Figure 3G) (P < 0.05); this finding resonates with the increase in AR protein levels by EPI (Figure 3B). The combination of Docetaxel and EPI consistently achieved a significant inhibitory effect for all three AR-driven reporters in 22Rv1 cells (Figure 3, panels E–G).

To validate the enhanced anti-tumor potency of the combination approach of taxanes and EPI against another androgen-insensitive prostate cancer cell line, the proliferative response of LNCaP95 human prostate cancer cells that express functional full-length AR and the AR-V7 variant was investigated (Hu et al., 2012). We found that EPI treatment alone or in combination with Docetaxel significantly inhibited cell proliferation in LNCaP95 cells, while Docetaxel as a single agent resulted in a 25% decrease in cell proliferation (Figure S5A). Furthermore analysis of the AR transcriptional activity, demonstrated that treatment of LNCaP95 cells with either EPI or Docetaxel given as single agents significantly inhibited the androgen-induced activity for all three reporters (Figure S5, panels B, C and D), prior to manifestation of cell death (Figure 4E). Consistent with the findings in 22Rv1 cells, the combination of Docetaxel and EPI resulted in maximal blockade of AR transcriptional activity in LNCaP95 cells.

3.4 Role of EMT in therapeutic response of CRPC to combined targeting of AR NTD and microtubules

The recently reported association of AR variants with clinical prostate cancer progression and therapeutic resistance to taxanes and anti-androgens (Cao et al., 2014; Li et al., 2013; Mostaghel et al., 2011; Zhang et al., 2011) may be driven by induced EMT. To assess the impact of AR NTD targeting on the EMT phenotype of prostate tumor epithelial cells after Docetaxel treatment, we subsequently conducted an expression profile of key EMT effectors in 22Rv1 cells following single agent or combination treatment. Treatment with EPI led to a marked increase of N-cadherin protein levels compared to the untreated control cells (Figure 4A); this N-cadherin up regulation was attenuated by combination with Docetaxel. By 24 h-post treatment the key phenotypic change of elevated N-cadherin associated with EMT induction was restored to control levels (Supplementary Figure S4). No significant changes in levels of Twist, E-cadherin or N-cadherin temporal expression were detected. A modest decrease in the cytoskeleton protein, cofilin was evident after 24 h of DOC or EPI monotherapy.

The EMT landscape was subsequently interrogated in the in vivo setting of 22Rv1 prostate tumor xenografts (from Figure 1B). E-cadherin expression was increased in response to both Docetaxel and EPI given as single agents, while N-cadherin levels were decreased compared to the untreated controls (Figure 4B) consistent with the phenotypic reversion of EMT to MET in prostate tumor epithelial cells. Interestingly, the combination treatment led to an apparent decrease in E-cadherin and upregulation of N-cadherin protein levels (EMT). Tissue levels of vimentin were increased after treatment with EPI or Docetaxel given as single agents, an effect that was blocked by the combination treatment (Figure 4B; Supplementary S1C).

The temporal changes in gene expression for EMT regulators were interrogated by analysis of mRNA expression levels by qRT-PCR. The results summarized in Figure 4 indicate that consistent with EPI increasing levels of N-cadherin protein at 6 h, levels of N-cadherin mRNA were also elevated within the same period of exposure to EPI, compared to levels observed for CSS control (Figure 4C). Upregulation of mRNA for vimentin and E-cadherin was also detected within 6 h of treatment with EPI. After longer treatment periods (12, 24 and 24 h), the levels for both proteins were restored to levels comparable to the controls. A significant increase in N-cadherin mRNA was detected in response to Docetaxel (as single agent) at 1 and 3 h, but with no effect on vimentin mRNA (Figure 4C). The combination treatment of EPI and Docetaxel for 24 h resulted in a significant increase in the levels of mRNA for both N-cadherin and vimentin. A significant induction of Snail mRNA, a transcriptional repressor for E-cadherin, was observed in response to both monotherapies and this was further increased response to the combination of Docetaxel with EPI (Figure 4C). Within the first hour of combination treatment, there was a significant induction in Twist gene expression (Figure 4C). Elevated expression of N-cadherin occurs in response to attenuated AR transcriptional activity or reduced levels of expression of AR (Jennbacken et al., 2010; Tanaka, 2010). Thus, these findings support the “programming” of EMT by the AR N-terminal inhibitor, to potentially drive the sensitivity/resistance of CRPC 22Rv1 cells to microtubule-targeting taxane-based chemotherapy (Figure 4D).

4.1 Conclusions

In summary, the present pre-clinical studies demonstrate that the combination strategy of targeting tubulin (by taxane chemotherapy) -and the NTD of AR variants association (by EPI) and cellular localization can impair CRPC by reversing EMT and navigating EMT-MET cycling. These findings provide a new insight into a potential mechanism for overcoming cross-resistance driven by differential status, and activity of AR, depending on its nuclear/cytosolic localization. Further understanding of the mechanisms of aberrant activation of AR and the distinct transcriptome of AR variants compared to full length wild-type AR, is central to the optimization of therapeutic targeting of CRPC and of major significance in patients who failed Docetaxel therapy. Ongoing studies focus on interrogation of the functional relationships between dynein and other partners such as HSP90, facilitating interactions between AR variants and tubulin in the cytoplasm of prostate cancer cells and the impact on cancer cell migration and invasion in CRPC (Figure 4D).