The CXCL12/CXCR4 Axis Promotes Ligand-Independent Activation of the Androgen Receptor

Abstract

The molecular mechanisms responsible for the transition of some prostate cancers from androgen ligand-dependent to androgen ligand -independent are incompletely established. Molecules that are ligands for G protein coupled receptors (GPCRs) have been implicated in ligand-independent androgen receptor (AR) activation. The purpose of this study was to examine whether CXCL12, the ligand for the GPCR, CXCR4, might mediate prostate cancer cell proliferation through AR-dependent mechanisms involving functional transactivation of the AR in the absence of androgen. The results of these studies showed that activation of the CXCL12/CXCR4 axis promoted: The nuclear accumulation of both wild-type and mutant AR in several prostate epithelial cell lines; AR-dependent proliferative responses; nuclear accumulation of the AR co-regulator SRC-1 protein; SRC-1:AR protein:protein association; co-localization of AR and SRC-1 on the promoters of AR-regulated genes; AR- and SRC-1 dependent transcription of AR-regulated genes; AR-dependent secretion of the AR-regulated PSA protein; P13K-dependent phosphorylation of AR; MAPK-dependent phosphorylation of SRC-1, and both MAPK- and P13K-dependent secretion of the PSA protein, in the absence of androgen. Taken together, these studies identify CXCL12 as a novel, non-steroidal growth factor that promotes the growth of prostate epithelial cells through AR-dependent mechanisms in the absence of steroid hormones. These findings support the development of novel therapeutics targeting the CXCL12/CXCR4 axis as an ancillary to those targeting the androgen/AR axis to effectively treat castration resistant/recurrent prostate tumors.

1. INTRODUCTION

Prostate cancer is the leading cause of newly diagnosed cancers, and the second leading cause of cancer-related deaths, in American men [Jemal et al., 2011]. Prostate tumors are initially dependent on androgen signaling and can be successfully controlled by a series of strategies that deplete endogenous androgen expression or interfere with androgen receptor (AR)-mediated signaling. These include androgen ablation and anti-androgen therapeutics, such as finasteride, which inhibits the conversion of testosterone to the active form, dihydrotestosterone, or bicalutimde, which directly antagonizes androgen signaling at the level of the receptor [Debes and Tindall, 2001; Pienta and Bradley, 2006]. Therapies that directly target the AR are most commonly used either as single therapy or in combination with pharmaceuticalor surgical castration or ionizing radiation. However, all of these therapeutic approaches are only effective for early stage androgen-dependent prostate cancer, as progressive prostate tumors develop alternative strategies to survive and grow despite anti-androgen therapy. Eventually, such tumors develop into lethal, metastatic androgen-refractory prostate cancers [Pienta and Bradley, 2006].

The molecular mechanisms responsible for the transition from androgen-dependent to androgen-independent prostate cancer are incompletely established. However, it is known that, despite their insensitivity to anti-androgen and androgen-deprivation therapies, the survival and growth of castration resistant/recurrent prostate cancers (hereafter referred to as CRPCs) are likely dependent upon AR signaling. Evidence that AR activity may be required for the growth and progression of castration resistant prostate tumors is elicited from studies that describe AR gene transcript and protein over-expression; AR gene mutations conferring receptor hyperactivity; AR co-regulator protein dysregulation producing increased AR activity, and chromatin remodeling conducive to more efficient transcription of AR-regulated genes, in such tumors [Chen et al., 2002; Cronauer et al., 2003; Culig et al., 1998; Debes and Tindall, 2002; Jia et al., 2006; Pienta and Bradley, 2006; Scher et al., 2004; Taplin and Balk, 2004]. Taken together, these studies suggest that AR signaling plays an important role in progressive castration resistant disease, and that these activities can, and do, occur in the presence of very low levels, or absence of, circulating androgen.

Several studies have reported hormone-independent AR signaling in prostate cancer cells. Diverse types of molecules, including peptide growth factors (EGF, KGF, and IGF1) [Craft et al., 1999; Culig et al., 1994; Shi et al., 2001; Torring et al., 2000; Wen et al., 2000; Yeh et al., 1999], neuropeptides, including neurotensin and bombesin [Dai et al., 2002; Debes and Tindall, 2002; Lee et al., 2001], and inflammatory mediators such as the interleukins IL-4 and IL-6 [Culig et al., 2005; Hobisch et al., 1998; Lee et al, 2003; Lee et al., 2005; Lee et al., 2008; Lee et al., 2009; Malinowska et al., 2009; Ueda et al., 2002], and the chemokine CXCL8 (IL-8) [Araki et al., 2007; Lee et al., 2004; MacManus et al., 2007; Seaton et al., 2008], have been implicated in hormone-independent AR activation. A common thread among some of these molecules is that they are ligands for G protein coupled receptors (GPCRs). Neuropeptides as well as interleukins and chemokines are ligands for GPCRs, and activation of GPCRs has been linked to hormone-independent AR activation and prostate tumor growth [Begley et al., 2005; Daaka, 2004]. Of particular interest to this study is evidence that interactions between the CXC-type chemokine, CXCL8, and the GPCRs that recognize it, CXCR1 and CXCR2, promote the hormone-independent proliferation of prostate cancer LNCaP cells. Moreover, CXCL8-mediated hormone-independent prostate cancer cell proliferation can be inhibited by pretreatment with bicalutamide, implying that AR activation is involved in this response [Seaton et al., 2008]. Previous work accomplished in our laboratory has shown that several CXC-type chemokines, including CXCL1, CXCL5, CXCL6, and CXCL12, promote the proliferation of both non-transformed and transformed prostate epithelial cells [Begley et al., 2005; Begley et al., 2007]. Like CXCL8, CXCL1, CXCL5 and CXCL6 recognize and effect proliferation through binding the GPCRs CXCR1 and/or CXCR2 [Rot and von Andrian, 2004]. However, CXCL12 is the ligand for a different CXCR-type GPCR, CXCR4 (and, to a lesser extent, CXCR7) [Bleul et al., 1996; Burns et al., 2006]. This raises the possibility that CXCL12/CXCR4 interactions might also mediate prostate cancer cell proliferation through AR-dependent mechanisms. To test this hypothesis, we have examined whether CXCL12/CXCR4 interactions functionally transactivate the AR to mediate cellular proliferation in the absence of androgen, and determined which intracellular signaling events mediate this process. The results of these studies provide new directions to develop therapeutics to effectively halt the progression, and consequent mortality, of castration resistant prostate tumors.

2. MATERIALS AND METHODS

2.1. Cell Culture

The androgen-sensitive LNCaP and 22Rv1 cell lines were maintained in 10% RPMI media or serum free, phenol-red free RPMI supplemented with antibiotics as described [Horoszewicz et al., 1980; Sramkoski et al., 1999], VCaP cells were maintained in 10% DMEM or 5% DMEM (without phenol red, and with charcoal stripped serum) supplemented with antibiotics as described [Korenchuk et al., 2001], and LAPC4 cells (kindly provided by Dr. Charles L. Sawyers, Memorial Sloan-Kettering Cancer Center) were maintained in 10% IMDM with 10nM R1881 or serum free IMDM (without phenol red) supplemented with antibiotics as described [Craft et al., 1999; Klein et al., 1997]. LNCaP cells were used between passages 20–40 to ensure androgen sensitivity [van Bokhoven et al., 2003]. The AR is wild type in LAPC4 and VCaP cells, but mutant in LNCaP and 22Rv1 cells [Arnold et al., 2008; Attardi et al., 2004; Denmeade et al., 2003; Furr et al., 1987; Takizawa et al., 2010]. Unless otherwise noted, all studies examining the effects of supplementation with R1881 or CXCL12 utilize serum-free, phenol red-free media.

2.2. Proliferation Assay

Proliferation assays were conducted as previously described [Begley et al., 2005, 2007, 2008]. For experiments examining growth factor-mediated proliferative responses, LNCaP cells were grown in phenol red-free, serum-free media supplemented with R1881, CXCL12, or both, at concentrations described in text for 48hrs. Average cell numbers and standard deviations were calculated and data expressed as proliferation relative to serum free media or vehicle as appropriate. For experiments examining the effect of the anti-androgen bicalutamide [Furr et al., 1987] (Casodex, Zeneca Pharmaceuticals, Germany) and/or anti-CXCR4, AMD3100 [Donzella et al, 1998] (#A5602, Sigma), LNCaP and 22Rv1 cells were grown in phenol red-free, serum-free supplemented with R1881 or CXCL12 for 24 hr at concentrations as described in the text with or without vehicle (0.1% DMSO for bicalutamide, water for AMD3100), 10uM bicalutamide, or 10uM AMD3100. The extent of proliferation was assessed at 24hrs by WST assay as described by the manufacturer (Roche, USA) and data expressed as proliferation relative to serum free media or vehicle as appropriate.

2.3. Immunoblot, Protein Analysis, and Antibodies

Cells were lysed, proteins resolved by electrophoresis, and immunoblotted as described previously [Begley et al., 2005, 2007]. Primary antibodies utilized included those that detected the N-terminal domain of AR (Santa Cruz, #SC-816), phospho-AR Ser210 (Imgenex, #IMG-561), pan-phospho Serine (Zymed, #61–8100), pan-phospho Threonine (Cell Signaling, #9386), SRC-1 (Santa Cruz #sc-8995), PSA (Dako, #ER-PR8), Akt (Cell Signaling, #9272), pSer473-Akt (Cell Signaling #9271), CXCR4 (Abcam, #ab2074), Erk 1 / 2 (Cell Signaling #9102), phosphor-Erk 1 / 2 (Cell Signaling #9101), Src family kinase (Invitrogen #KHO0161), pY418Src (Invitrogen #44660G) tubulin (Upstate, #DM1A), actin (Santa Cruz, #sc-1615), GAPDH (Cell Signaling #2118), or hnRNP (Santa Cruz, #15386) in conjunction with an ECL detection system. Secondary antibodies included goat anti-rabbit (Cell Signaling, #7074) goat anti-mouse (Santa Cruz, #SC-2005), mouse anti-goat (Santa Cruz, #A3105), and were used at a 1:5000 concentration. Immunoblots shown are representative of replicate experiments. To prepare nuclear or cytoplasmic protein extracts, cells were grown to 75% confluency in maintenance medium, washed with PBS and incubated in a phenol red-free, serum-free RPMI for LNCaP cell line, and in phenol red-free DMEM containing 5% charcoal stripped serum for VCaP cell line for overnight. The cells were then treated with R1881 or CXCL12 at the indicated concentration for 24, and 48 hrs. Nuclear and cytosolic extracts were prepared from a single plate by using the NE-PER Nuclear Extraction Kit (Pierce, Rockford, IL) according to manufacturer’s instructions and stored in aliquots at −80°C until use.

2.4. Immunoprecipitation Analysis

LNCaP cells were plated in a 150mm dish in complete media then switched to phenol-red free, serum-free media for 12 hr. The effect of exogenous R1881 or CXCL12 on AR:SRC-1 association was determined by treating the overnight serum-deprived cells with vehicle, 1nM R1881 or 50pM CXCL12 for 48hrs, and preparing the cytoplasmic and nuclear extracts as described above. In brief, 1mg of cytoplasmic and nuclear extracts was pre-cleared with 1ug of isotype control IgG (BD Pharmingen #550875) and protein G Agarose (Invitrogen #15920-010) for 1hr at 4°C. To 1mg of pre-cleared cytosolic and nuclear extracts (at 1mg/ml in RIPA with protease inhibitors, sodium orthovanadate and sodium fluoride), 1ug of AR (Santa Cruz, #SC-816) or SRC-1 (Santa Cruz, #8995) antibody were added and incubated overnight at 4°C with rotation. Protein AAgarose beads were then added, incubated for 1hr at 4°C with rotation, and then the beads were washed 5× times with RIPA (plus protease and phosphatase inhibitors). Finally, the bound proteins were eluted by adding 2X SDS loading buffer with β-mercaptoethanol and boiling at 100°C for 5 min.

2.5. Immunofluorescence

LNCaP cells (1.5×10⁵ cells) were plated in a fibronectin coated (Sigma, #F-1141; 5ug/ml) chamber slides. Cells were grown in phenol red-free, serum free media for 12 hr, then treated with vehicle, 1nM R1881, or 50pM CXCL12 for 48 hr. At 70% confluence, the cells were washed 2X times with PBS on ice and fixed with freshly prepared 2% formaldehyde (Sigma, #F-8775) on ice for 5min, followed by ice cold 100% ethanol (Arcos, #64-17-5) for 2–5 min. Cells were then washed 2X times with TBST (0.25% Tween) to remove fixatives, blocked overnight at 4°C with TBST (0.25% Tween) containing 15% Goat serum (Sigma, #G9023), and 5% milk, stained for 12 hrs at 4°C with anti-AR (1:100 in blocking solution), washed 3X times for 5 min each with TBST, stained at room temperature for 1hr with goat anti-rabbit FITC (1:100 in blocking solution), washed 3X times for 5 min each with TBST. The cells were then counterstained for 2 min with DAPI (Molecular Probes, #D-1306; 10ug/ml in TBST), washed 3X times for 5 min each with TBST, and mounted in an Aqua-mount (Lerner Laboratories, PA). Images were taken on Olympus BX51 fluorescence microscope.

2.6. ELISA Assay

To assess PSA secretion, 1×10⁶ LNCaP cells were plated in 60mm dishes in complete media. The effect of exogenous R1881 or CXCL12 on PSA secretion was determined by treating the overnight serum-deprived cells with vehicle, 1nM R1881 or 50pM CXCL12 for 24 and 48hrs. The effect of bicalutamide on R1881- and CXCL12- mediated PSA secretion was examined by pre-treating the cells overnight with 0.1% vehicle or 10uM bicalutamide in phenol red-free, serum free media, followed by treatment with 1nM R1881 or 50pM CXCL12 for 24 and 48hrs. The effects of the PI3 kinase inhibitor, wortmannin (Sigma, #W1628), the MAPK MEK/ERK inhibitor, U0126 (Cell Signaling, #9903), the PKC inhibitor, bisindolylaeimide I, BISI, (Calbiochem, #203290), the Src family kinase inhibitor, PP1 (Invitrogen) and its inactive analog, PP3 (Calbiochem, #529574) on CXCL12- stimulated PSA secretion was examined by pre-treating the overnight sera deprived cells with 1uM wortmannin, 10uM U0126, 10uM BISI, 10uM PP1, 10uM PP3 or vehicle (0.1% DMSO), for 1hr, followed by treatment with 1nM R1881 or 50pM CXCL12 for 48hrs. The effect of AMD3100 (Sigma, #A5602), a CXCR4-inhibitor, on CXCL12-stimulated PSA secretion was examined by pre-treating the overnight sera deprived cells with 10uM AMD3100 for 1hr, followed by treatment with vehicle, 1nM R1881 or 50pM CXCL12 for 48hrs. To assess PSA secretion in VCaP cells, 2×10⁶ cells were plated in a 100mm dish in complete media. The effect of exogenous R1881 or CXCL12 on PSA secretion was determined by treating the overnight cells in 5% charcoal stripped serum with vehicle, 100pM, 1000pM R1881, 100pM or 1000pM CXCL12 for 24 and 48hrs. The conditioned media was then harvested, concentrated and tested by ELISA to quantitate PSA (Human Kallikrein 3/PSA DuoSet, R&D Systems # DY1344), which was normalized to ng protein/ml/million cells

2.7. Phosphorylation of AR and SRC-1

The effect of inhibition of MAPK and/or PI3K signaling pathways on the phosphorylation of AR and SRC-1 was determined by preparing LNCaP cells as described for immunoprecipitation analysis, then pre-treating the cells for 1hr with vehicle (DMSO), 10uM of the MEK inhibitor, U0126, or 1uM of the PI3K inhibitor, wortmannin, followed by treatment with vehicle or 50pM CXCL12 for 60 min. Phosphorylation of AR was monitored by immunoblot analysis of whole cell lysate against phospho-AR Ser210 (at 10 and 60 min CXCL12 treatment), or by immunoprecipitation of AR (at 60 min CXCL12 treatment) from whole cell lysate followed by immunoblot analysis against pan phospho-Serine and pan phospho-Tyrosine antibodies, and of SRC-1 by immunoprecipitation of SRC-1 (at 60 min CXCL12 treatment) from whole cell lysate followed by immunoblot analysis against pan phospho-Serine and pan phospho-Threonine antibodies. The effect of inhibition of MAPK and/or PI3K signaling pathways on the phosphorylation of Erk and Akt was determined by preparing LNCaP cells as described above, then treating with vehicle or 50pM CXCL12 for 10 and 60 min.

2.8. Quantitative Real Time PCR

All quantitative real-time assays were conducted as previously described using an Applied Biosystems 7900HT instrument and reagents [Begley et al., 2005, 2007, 2008; Kasinaet al, 2009]. Cells were grown to 70% confluence in 6-well dishes prior to RNA purification using Trizol reagent (Invitrogen, Carlsbad, CA). For experiments examining the effect of the anti-androgen bicalutamide (Casodex, Zeneca Pharmaceuticals, Germany) on R1881- and CXCL12- stimulated transcription, cells were grown for 12 hr in phenol red-free, serum-free media, then pre-treated with vehicle (DMSO) or 10uM casodex for 4hr, followed by treatment with 1nM R1881, 50pM CXCL12 or no treatment for 4hr. For experiments examining the effect of transient knock down of AR and SRC-1 on R1881- and CXCL12- stimulated transcription, cell were grown as just described then treated with vehicle, 1nM R1881, 50pM CXCL12, or 1nM R1881 + 50pM CXCL12 for 4hrs. For all experiments, 1 ug of RNA was reverse transcribed using Superscript III reverse transcriptase (Invitrogen). Resulting cDNA was diluted to the ratio of 1:100. Real-time PCR was performed using Assays on Demand (Applied Biosystems, Foster City, CA, USA) according to the manufacturer’s instructions. Reactions within individual experiments were performed in triplicate, including no template controls and an endogenous control probe, RPLPO (ribosomal protein, large, PO), to assess template concentration. Cycle numbers to threshold were calculated by subtracting averaged control from averaged experimental values, and fold gene expression was calculated by raising these values to log 2. 6-carboxy-fluorescine-conjugated, gene-specific assays were Hs00171172_m1 for AR, Hs00377590_s1 for KLK3 (PSA), Hs00237175 for TMPRSS2, Hs00237052 for CXCR4 and Hs99999902_m1 for the control, RPLPO.

2.9. Gene Reporter Assays

Cells were plated at 2.5K cells per 96-well dish and incubated at 37°C overnight. Cells were then transfected with a construct containing the promoter of the PSA gene in-frame with a firefly luciferase reporter gene, pGL3[PSA-luc] [Burnstein K.L., 1993] (kindly provided by Kerry Burnstein, University of Miami, FL), a firefly luciferase empty vector control pGL3[empty-luc] (Promega #E1741), or a renilla luciferase transfection efficiency plasmid, pGL4[hRluc-TK] (Promega #E6921) using Fugene 6 transfection reagent at 6:1 (Fugene6:DNA) following manufacturer’s protocol. After 24hr, the media was changed to phenol red-free, SF RPMI, and cells were treated with 1nM R1881, 50pM or 100pM CXCL12 for 48hr. Luciferase expression was determined by using Dual Glo luciferase assay system following manufacturer’s protocol on a SpectraMax M5 plate reader. Renilla luciferase was used as an internal control for transfection efficiency. Samples were assayed in triplicates and firefly luciferase activity was normalized to renilla luciferase activity.

2.10. siRNA Transfection

siRNA against AR (siGENOME SMARTpool Human AR # M-003400-02) and SRC-1 (siGENOME SMARTpool Human NCOA1 # M-005196-03) and negative control siRNA (#SR30004) were purchased from Dharmacon (Lafayette, CO), and Origene (Rockville, MD) respectively. LNCaP cells were seeded at 40% density the day before transfection in 6-well plates, and 5nM siRNA or mock transfections were performed with DharmaFECT 2 transfection reagent from Dharmacon according to manufacturer’s instructions. Cells were then switched to phenol red-free, serum-free media for 12 hr, followed by treatment with vehicle, 1nM R1881, 50pM CXCL12, or 1nM R1881 + 50pM CXCL12 for 4hr. RNA isolation and quantitative real time PCR analysis were performed as described above.

2.11. Chromatin immunoprecipitation (ChIP) Assays

All ChIP assays were performed using EZ-Magna ChIP A kit (#17–408, Millipore) according to manufacturer’s instructions. For Re-ChIP assays, protein-protein interactions were stabilized by treating the cells with crosslinker 5mM DTBP (#D2388, Sigma) for 30min, inactivated with 100mM Tris, 150mM NaCl pH8.0 for 10min, elution of the first ChIP was done with 50mM NaHCO₃ and 1% SDS solution, eluate was diluted 10 times with ChIP dilution buffer, pre-cleared with isotype control IgG and precipitated with second antibody as above. In brief, LNCaP cells were hormone deprived for 2 days in phenol red-free medium with 10% charcoal-stripped serum, followed by overnight serum deprivation in phenol red-free medium, and then treated with vehicle, 1nM R1881 or 50pM CXCL12 for 48hrs. Antibodies used are PG-21 (Millipore) for androgen receptor and sc-8995 (Santa Cruz) for SRC-1. Primers used for the PSA ARE III region are fwd: TGGGACAACTTGCAAACCTG and rev: CCAGAGTAGGTCTGTTTTCAATCCA and for TMPRSS2 ARE III region are fwd: CCAGAAGAATACAATGATTAAAAGGCT and rev: TGGAACTGAAGTATTGGAAAACCA [D'Antonio et al., 2010]

3. RESULTS

3.1. CXCL12/CXCR4 Axis Activation Mediates AR Accumulation in the Nucleus

We have previously shown that activation of the CXCL12/CXCR4 axis stimulates the proliferation of both androgen-sensitive and androgen-insensitive cells [Begley et al., 2005, 2008]. However, we had not previously investigated a potential mechanistic role for AR in CXCL12/CXCR4-mediated proliferation in androgen-sensitive cells in the presence or absence of hormone. Therefore, an initial study was pursued to test whether activation of the CXCL12/CXCR4 axis may promote AR activation. Like other members of the steroid/nuclear receptor superfamily, AR activation promotes its translocation from the cytoplasm to the nucleus, where it binds to canonic androgen response elements in the promoters of AR-regulated genes [Chan and O’Malley, 1976; Ham et al., 1988]. Therefore we examined whether activation of the CXCL12/CXCR4 axis in the absence of androgen stimulated the translocation of AR from the cytoplasm to the nucleus and the accumulation of AR in the nucleus. To accomplish this, LAPC-4 or VCaP cells expressing wild-type AR [Klein et al., 1997; Korenchuk et al., 2001], or 22Rv1 or LNCaP cells expressing different forms of mutant AR [Sramkosi et al., 1999; Veldscholte et al., 1990] were grown in serum-free media and treated with vehicle, 1nM synthetic androgen R1881, or 50pM CXCL12 for 48hr. These growth factor concentrations were chosen because they have been shown by our studies and those of others to promote the proliferation of LNCaP cells [Begley et al., 2005; Horoszewicz et al., 1980]. The time point of 48hr was chosen because nuclear accumulation of the AR in response to CXCL12, though evident at earlier timepoints, was most robust at 48hr (Supplementary Figure S1). Cellular proteins were separated into nuclear and cytosolic fractions and assessed for the accumulation of nuclear AR protein by immunoblotting. As shown in Figure 1A, treatment with R1881 or CXCL12 was associated with nuclear accumulation of AR in all 4 cell lines tested. The observed nuclear accumulation of AR was most robust in LNCaP and VCaP cells, suggesting that both mutant and wild-type AR translocated into the nucleus in response to CXCL12/CXCR4 axis activation. However, for all cell lines tested, nuclear accumulation of AR in response to CXCL12/CXCR4 axis activation was consistently less robust than that observed with synthetic androgen. As another approach to test whether activation of the CXCL12/CXCR4 axis could promote AR nuclear translocation and accumulation, LNCaP cells were treated with 1nM R1881, 50pM CXCL12, or vehicle, then stained with DAPI to visualize cell nuclei and with a FITC-tagged secondary antibody that recognizes a primary antibody specific for AR. These studies clearly demonstrated increased nuclear localization of the AR in R1881- and CXCL12-treated cells compared to vehicle-treated cells (Figure 1B). Moreover, this increase in nuclear accumulation of the AR could not be ascribed solely to increased cell number over a 48 hour period, as LNCaP cells do not proliferate when grown in the stringent serum-free media conditions utilized in these studies, and exhibit an increase in cell number of only 20–30% in response to 50pM CXCL12 when grown in serum-free media (Supplementary Figure S2). Co-treatment of LNCaP cells with 1nM R1881 and 50pM CXCL12 promoted nuclear accumulation of the AR to levels higher than those achieved when the cells were treated with R1881 or CXCL12 alone (Figure 2A). Taken together, the results of these studies showed that CXCL12 mediated AR nuclear translocation in the absence of androgen and augmented AR accumulation in the presence of androgen.

Figure 1. CXCL12/CXCR4 Axis Activation Mediates AR Accumulation in the Nucleus.

A. LAPC-4 or VCaP cells expressing wild-type AR or 22Rv1 or LNCaP cells expressing mutant AR were grown in phenol red-free, serum-free (SF) RPMI for 12 hr then treated with vehicle, 1nM R1881, or 50pM CXCL12 for 48 hr. The cellular proteins were subjected to nuclear and cytoplasmic fractionation, electrophoresis, and immunoblotting. All 4 cell lines treated with R1881 or CXCL12 demonstrate nuclear accumulation of AR at levels higher than those for vehicle-treated cells. Loading and fractionation controls include hnRNP (nuclear fraction), tubulin (cytoplasmic fraction) or actin (both fractions). Autoradiographs shown are representative of triplicate experiments.

B. Photomicrographs of LNCaP cells seeded on fibronectin-coated slides and treated with 1nM R1881, 50 pM CXCL12, or vehicle for 48 hrs, then incubated for 12 hr with a FITC-labeled antibody against AR and then counterstained for nuclei with DAPI for 2 min. DAPI-stained cellular nuclei emit blue spectrum fluorescence, FITC emits green spectrum fluorescence, and cells co-staining for nuclear localization of AR emit both as a blue/green fluorescence (DAPI+FITC). Treatment with R1881 or CXCL12, but not vehicle, induces nuclear AR accumulation. Magnification is 400X. Photomicrographs shown are representative of triplicate experiments.

Figure 2. CXCL12/CXCR4 Axis-Mediated Cellular Proliferation is AR-Dependent.

A. LNCaP cells were grown in phenol red-free, serum-free (SF) RPMI for 12 hr then treated with vehicle, 1nM R1881, 50pM CXCL12, or 1nM R1881+50pM CXCL12 for 48 hr. The cellular proteins were subjected to nuclear and cytoplasmic fractionation, electrophoresis, and immunoblotting. Cells co-treated with R1881 and CXCL12 demonstrate nuclear accumulation of AR at levels higher than those for singly- or vehicle-treated cells. Loading and fractionation controls include hnRNP (nuclear fraction), GAPDH (cytoplasmic fraction) or actin (both fractions). Autoradiograph shown is representative of triplicate experiments.

B. LNCaP cells were grown for 48 hr in phenol red-free, SF RPMI media alone or in combination with 10pM R1881, 100pM R1881, or 50pM CXCL12. Compared to cells grown in media alone, cells grown in media supplemented with 10pM R1881 or 50pM CXCL12 achieved a modest but significant 10–15% increase in cell number. Cells grown in media supplemented with 10pM R1881 + 50pM CXCL12 achieved a 30–50% increase in cell number over that observed for cells grown in media alone, suggesting that these low levels of R1881 and CXCL12 acted in an additive fashion to promote cellular proliferation. This level of proliferation was similar to that observed for cells grown in 100 pM R1881. Cells grown in 100pM R1881 + 50pM R1881 grew to levels 50–90% higher than that achieved in media alone, but this increase was not significantly higher than that achieved in 100pM R1881 alone. Cellular proliferation in complete 10% RPMI is shown for comparison. Error bars reflect deviations from the mean for replicate measures. Statistically significant differences (p<.05) are indicated by *.

C. LNCaP cells were grown for 24 hr in phenol red-free, serum free media supplemented or unsupplemented with 1nM R1881, 10uM bicalutamide, the solvent vehicle for bicalutamide, DMSO (indicated as +DMSO), 10uM AMD3100, or the solvent vehicle for AMD3100, water (indicated as −DMSO). Cells grown in media supplemented with R1881 in the absence of bicalutamide proliferated to levels 40–50% above those observed in media alone (p<.05, *). These levels were significantly reduced in the presence of bicalutamide, but not AMD3100 (p<.05, #). Error bars reflect deviations from the mean for replicate measures.

D. 22Rv1 cells were grown as described in C. Cells grown in media supplemented with R1881 in the absence of bicalutamide proliferated to levels 10–15% above those observed in SF RPMI (p<.05, *). These levels were significantly reduced in the presence of 10uM bicalutamide (p<.05, #), but not 10uM AMD3100. Cellular proliferation was further inhibited in the presence of both inhibitors. Error bars reflect deviations from the mean for replicate measures.

E. LNCaP cells were grown as described in B. except that phenol red-free, SF RPMI was supplemented or unsupplemented with 50pM CXCL12, 10uM bicalutamide, the solvent vehicle for bicalutamide, DMSO (indicated as +DMSO), 10uM AMD3100, or the solvent vehicle for AMD3100, water (indicated as −DMSO). Cells grown in media supplemented with CXCL12 in the absence of bicalutamide proliferated to levels 15–20% above those observed in media alone (p<.05, *). These levels were significantly reduced in the presence of either 10uM bicalutamide or 10 uM AMD3100, and decreased below basal levels in the presence of both inhibitors (p<.05, #). Error bars reflect deviations from the mean for replicate measures.

F. 22Rv1 cells were grown as described in C. Cells grown in SF RPMI supplemented with 50pM CXCL12 in the absence of AMD3100 or bicalutamide to levels 5–10% above those observed in SF RPMI (p<.05, *). These levels were unaffected by the addition of 10uM bicalutamide or 10uM AMD3100, alone or in combination. Error bars reflect deviations from the mean for replicate measures.

3.2. CXCL12/CXCR4 Axis-Mediated Cellular Proliferation is AR-Dependent

Castration resistant/recurrent prostate cancers (CRPCs) are insensitive to hormone ablation therapy. However, it is likely that, despite surgical or chemical castration, low levels of androgens persist or are synthesized in the prostatic microenvironment of these tumors [Geller, 1995; Mohler, 2008; Mohler et al., 2011]. CRPCs may harbor amplified and/or mutant ARs with altered ligand sensitivity and altered interaction profiles with diverse co-regulators [Gregory et al., 2001; Karpf et al., 2009; Lamont and Tindall, 2010; Linja et al., 2001; Steketee et al., 2002; Tan et al., 1997]. Therefore, it was necessary to test whether activation of the CXCL12/CXCR4 axis promoted a functional transactivation of the AR, i.e., initially defined as one that was associated with cellular proliferation, and to determine whether such activation was observed in the presence or absence (or both) of androgen. To address these questions, LNCaP cells were grown for 48 hr in phenol red-free, SF RPMI media alone or in combination with 10pM R1881, 100pM R1881, or 50pM CXCL12. Compared to cells grown in SF RPMI alone, cells grown in media supplemented with 10pM R1881 or 50pM CXCL12 achieved a modest but significant 10–15% increase in cell number. Cells grown in SF RPMI supplemented with both 10pM R1881 and 50pM CXCL12 achieved a 30–50% increase in cell number over that observed for cells grown in SF RPMI alone, suggesting that these low levels of R1881 and CXCL12 acted in an additive fashion to promote cellular proliferation. This level of proliferation was similar to that observed for cells grown in 100 pM R1881. Cells grown in SF RPMI supplemented with both 100pM R1881 and 50pM R1881 grew to levels 50–90% higher than that achieved in SF RPMI alone, but this increase was not significantly higher than that achieved in 100pM R1881 alone (Figure 2B). These observations suggested that low levels of CXCL12 and androgen produced an additive effect on cellular proliferation, but this effect was muted at higher androgen levels. Thus, activation of the CXCL12/CXCR4 axis promoted cellular proliferation in the absence of androgen and augmented cellular proliferation in the presence of low levels of androgen. To investigate whether CXCL12/CXCR4 axis activation mediated cellular proliferation in an AR-independent or AR-dependent manner, LNCaP cells were grown as just described except that cells were first pre-treated for 24 hr with vehicle, the anti-androgen bicalutamide, or the CXCR4 small molecule inhibitor AMD3100. As shown in Figure 2C, LNCaP cells grown in media supplemented with 1 nM R1881 proliferated to levels 40–50% above those observed in SF RPMI. This proliferative response was significantly reduced in the presence of bicalutamide, but not AMD3100. Cells grown in media supplemented with CXCL12 proliferated to levels 15–20% above those observed in SF RPMI. These levels were significantly reduced in the presence of either bicalutamide or AMD3100, and were reduced below basal levels in the presence of both inhibitors (Figure 2E). These results showed that CXCL12/CXCR4 axis-mediated proliferative responses were both CXCR4- and AR-dependent, whereas androgen/AR axis-mediated proliferative responses were solely AR-dependent. For comparison, similar studies were carried out using 22Rv1 cells, which harbor a mutated form of the AR different from that expressed by LNCaP cells [Sramkoski et al., 1999; Veldscholte et al., 1990]. As shown in Figure 2D, 22Rv1 cells proliferated less robustly than LNCaP in response to 1nM R1881, and this response was muted in response to bicalutamide but not AMD3100. However, 22Rv1 cells do not appreciably proliferate in response to CXCL12/CXCR4 axis activation, and this low proliferative response is only minimally impacted by the administration of either bicalutamide or AMD3100 (Figure 2F). Taken together, these data demonstrate that CXCL12/CXCR4 axis-mediated cellular proliferation is AR-dependent in hormone sensitive cells in the absence of androgen and may be modulated by the type of AR mutant expressed by such cells.

3.3. Activation of the CXCL12/CXCR4 Axis Promotes the Transcription of AR-regulated Genes in an AR-Dependent Manner

The previous studies had shown that activation of the CXCL12/CXCR4 axis promoted nuclear accumulation of the AR and cellular proliferation in an AR-dependent manner. Because the AR is a transcription factor, it was reasonable to test whether, like androgen, CXCL12 could promote the transcription of genes harboring androgen response elements (AREs) in their promoter regions. Preliminary studies had shown that LNCaP cells treated with 50pM CXCL12 transcribed the AR-regulated PSA gene, and that PSA transcript levels increased steadily over a 4hr period (Supplementary Figure S3). To further investigate this, LNCaP cells were grown in phenol red-free, SF RPMI and the media supplemented first with bicalutamide for 4 hr, then with R1881. RNA purified from the cells was subjected to qRT-PCR to detect the transcripts of two AR-regulated genes, PSA [Riegman et al., 1991] and TMPRSS2 [Lin et al., 1999], as well as the relevant ligand receptors, AR and CXCR4, or for the housekeeping gene, RPLPO, as a template control. Treatment with R1881 induced ~250% higher transcript levels for the PSA and ~400% higher levels for theTMPRSS2 genes, as well as ~50–100% higher levels for the AR and CXCR4 genes compared to vehicle-treated cells after normalization to RPLPO template controls. Transcript levels for both the PSA and TMPRSS2 genes were significantly reduced subsequent to bicalutamide treatment (Figure 3A). LNCaP cells were then similarly treated but with 50pM CXCL12 rather than 1 nM R1881. Treatment with CXCL12 induced ~100% higher transcript levels for the PSA and ~50% higher levels for theTMPRSS2 genes, as well as ~100–125% higher levels for the AR and CXCR4 genes compared to vehicle-treated cells after normalization to RPLPO template controls. Moreover, transcript levels for all 4 genes were significantly reduced subsequent to bicalutamide treatment (Figure 3B). Taken together, these studies show that CXCL12/CXCR4 axis activation promotes the transcription of AR-regulated genes, and of the AR itself, in an AR-dependent manner. As another approach to examine this activity, activation of PSA gene expression was further delineated in LNCaP cells transfected with either a recombinant plasmid carrying the promoter of the PSA gene driving firefly luciferase expression, a parallel empty vector construct lacking the PSA gene sequences, or a transfection efficiency plasmid expressing renilla luciferase. 24 hours post-transfection, the cells were treated with vehicle, 1nM R1881 or 50pM CXCL12 for an additional 24 hrs. As shown in Figure 3C, both CXCL12-and R1881-treated cells demonstrated significantly higher levels of PSA promoter-driven firefly luciferase expression than vehicle-treated cells. Immunoblot analysis showed that R1881-treated cells demonstrated up-regulation of the AR, but not CXCR4, proteins, whereas CXCL12-treated cells did not demonstrate up-regulation of either the AR or CXCR4 proteins above levels observed for vehicle-treated cells (Figure 3D). Comparison of the results of these experiments showed that activation of the CXCL12/CXCR4 axis promoted an AR-dependent transcription of AR-regulated genes, and that this activity was clearly significant but also clearly weaker than that promoted by synthetic androgen. Although an up-regulation of transcripts encoding both the AR and CXCR4 transcripts was observed for both R1881- and CXCL12-treated cells, these effects were not readily apparent at the protein level.

Figure 3. Activation of the CXCL12/CXCR4 Axis Promotes the Transcription of AR-regulated Genes in an AR-Dependent Manner.

A. LNCaP cells were pre-treated with vehicle alone or vehicle + 10 uM bicalutamide for 4 hr, then with vehicle or vehicle + 1 nM R1881 for 4 hr. RNA was purified from cell lysates and subjected to qRT-PCR. Cycle numbers to threshold were calculated by subtracting averaged RPLPO template control transcript levels from averaged transcript levels for PSA, TMPRSS2, AR, or CXCR4, and percent change (increase or decrease) in transcript levels for treated cells compared to vehicle-treated cells was determined. Treatment with R1881 induced ~250% higher transcript levels for the PSA and ~400% higher levels for the TMPRSS2 genes, as well as ~50–100% higher levels for the AR and CXCR4 genes compared to vehicle-treated cells after normalization to RPLPO template controls (p<.05, *). Transcript levels for the PSA and TMPRSS2 genes were significantly reduced subsequent to 10 uM bicalutamide treatment (p<.05, #). Error bars reflect deviations from the mean for replicate measures.

B. LNCaP cells were similarly treated but with 50pM CXCL12 rather than R1881. Treatment with CXCL12 induced ~100% higher transcript levels for the PSA and ~50% higher levels for the TMPRSS2 genes, as well as ~100–125% higher levels for the AR and CXCR4 genes compared to vehicle-treated cells after normalization to RPLPO template controls (p<.05, *). Transcript levels for all 4 genes were significantly reduced subsequent to bicalutamide treatment (p<.05, #). Error bars reflect deviations from the mean for replicate measures.

C. CXCL12/CXCR4 axis activation of PSA gene expression was further examined in LNCaP cells transfected with a recombinant plasmid carrying the promoter of the PSA gene driving firefly luciferase expression (pGL3[PSA-luc]), a parallel empty vector construct lacking the PSA gene sequences (pGL3[empty-luc]) or a transfection efficiency plasmid expressing renilla luciferase, pGL4[hRluc-TK]. 24 hours post-transfection, the cells were treated with 1nM R1881 or 50pM CXCL12 for an additional 24 hrs. Firefly luciferase expression was determined and normalized to renilla luciferase, then expressed as fold PSA transcript (normalized to 100%). Error bars reflect deviations from the mean for replicate measures. Both CXCL12- and R1881-treated cells demonstrated significantly (*, p<.05) higher levels of firefly luciferase expression than vehicle-treated cells.

D. LNCaP cells were grown in phenol red-free SF RPMI for 12 hr then treated with vehicle (0.1% DMSO), 1nM R1881, or 50pM CXCL12 in the absence or presence of bicalutamide for 48 hr. The cellular proteins were subjected to electrophoresis and immunoblotting against antibodies specific for SRC-1, AR, CXCR4, or actin (as loading control). R1881-treated cells demonstrated up-regulation of the AR, but not CXCR4, proteins, whereas CXCL12-treated cells did not demonstrate up-regulation of either the AR or CXCR4 proteins above levels observed for vehicle-treated cells. Pre-treatment with bicalutamide did not modulate SRC-1, AR, or CXCR4 protein levels in vehicle-, R1881-, or CXCL12-treated cells. Error bars reflect deviations from the mean for replicate measures. Significant differences (p<.05) are indicated by *.

3.4. Activation of the CXCL12/CXCR4 Axis Promotes PSA Secretion in an AR-Dependent Manner

LNCaP cells are known to secrete PSA protein in response to androgen stimulation [Fong et al., 1992]. We therefore sought to determine whether the observed promotion of AR-dependent PSA gene transcription could be observed at the protein level, e.g., whether, like androgen, CXCL12 could stimulate PSA protein synthesis and secretion in LNCaP cells. As shown in Figure 4A, immunoblot analysis of protein lysates and conditioned media prepared from LNCaP cells grown in phenol red-free SF RPMI media supplemented with vehicle, 1nM R1881 or 50pM CXCL12 detected higher levels of PSA protein in conditioned media from both R1881- and CXCL12-treated cells compared to vehicle-treated cells. To elucidate whether the observed CXCL12/CXCR4 axis-mediated increased secretion of PSA protein was AR- and CXCR4-dependent, LNCaP cells were pre-treated with vehicle, bicalutamide, or AMD3100 in phenol red-free, SF RPMI media, followed by treatment with 1nM R1881 or 50pM CXCL12 for 24 or 48hrs. Treatment with R1881or CXCL12 (Figure 4B, 4C) promoted significantly higher levels of PSA protein secretion than treatment with vehicle. Moreover, pre-treatment with bicalutamide significantly repressed both R1881- and CXCL12-promoted PSA secretion (Figure 4B), whereas pre-treatment with AMD3100 significantly repressed CXCL12-, but not R1881-, promoted PSA protein secretion (Figure 4C). To determine whether activated CXCL12/CXCR4 axis-mediated AR-dependent PSA protein secretion was limited to cells harboring a mutant AR, VCaP cells, which harbor a wild-type AR, were examined. Like LNCaP cells, VCaP cells secrete PSA in response to androgen treatment [Korenchuk et al., 2001; Lu et al., 2004]. VCaP cells treated with 1nM R1881 secreted high levels of PSA protein whereas VCaP cells treated with either 100pM or 1nM CXCL12 secreted PSA protein to levels 25–50% above baseline levels (Figure 4D). Thus, CXCL12/CXCR4 axis activation promoted a cellular response from VCaP cells consistent with modest AR activation. Taken together, these data suggested that CXCL12/CXCR4 axis activation promoted the transactivation of both wild-type and mutant AR, and promoted AR-dependent PSA secretion in the absence of androgen.

Figure 4. Activation of the CXCL12/CXCR4 Axis Promotes PSA Secretion in an AR-Dependent Manner.

A. Immunoblot analysis of protein lysates and conditioned media prepared from LNCaP cells grown in phenol red-free SF RPMI media supplemented with vehicle, 1nM R1881 or 50pM CXCL12 for 24 and 48 hrs. Immunoblots were probed for PSA in lysates and media, and for actin in lysates. PSA is detected at higher levels in conditioned media from both R1881- and CXCL12-treated cells at 48 hr.

B. LNCaP cells were pre-treated with vehicle (0.1% DMSO) or 10uM bicalutamide in phenol red-free, SF RPMI media, followed by treatment with 1nM R1881 for 24 or 48 hrs. The media was collected and the amount of PSA secreted was determined by ELISA. PSA secreted in to the media is expressed as fold over vehicle treated cells (the latter is normalized to 0%). Error bars reflect deviations from the mean for replicate measures. Treatment with R1881 or CXCL12 induced significantly higher (*, p<.05) PSA secretion than treatment with vehicle. Pre-treatment with bicalutamide significantly repressed (#, p<.05) both R1881- and CXCL12-promoted PSA secretion. Error bars reflect deviations from the mean for replicate measures.

C. The experiments described in B. were repeated except that 10uM AMD3100 was used in place of bicalutamide. Treatment with R1881 or CXCL12 induced significantly higher (*, p<.05) PSA secretion than treatment with vehicle. Pre-treatment with AMD3100 significantly repressed (#, p<.05) CXCL12-, but not R1881-, promoted PSA secretion. Error bars reflect deviations from the mean for replicate measures.

D. VCaP cells were pre-treated with vehicle (0.1% DMSO), 100pM or 1nM R1881 or 100pM or 1nM CXCL12 for 24 or 48 hrs. The media was collected and the amount of PSA secreted was determined by ELISA. PSA secreted in to the media is expressed as fold over vehicle treated cells (the latter is normalized to 0%). Error bars reflect deviations from the mean for replicate measures. Treatment with 1nM R1881 or CXCL12 induced significantly higher (*, p<.05) PSA secretion than treatment with vehicle for 24 or 48 hrs though this effect was more robust for R1881- than CXCL12-treated cells. Error bars reflect deviations from the mean for replicate measures.

3.5. CXCL12/CXCR4 Axis Activation Mediates SRC-1 Nuclear Translocation and Promotes SRC-1:AR Association

Like other members of the steroid/nuclear receptor family, efficient transcriptional activation of AR also requires activation of one or more nuclear receptor co-activators. SRC-1 is the first cloned nuclear receptor co-activator and is known to interact with the amino-terminal domain of the AR [Bevan et al., 1999; Ma et al., 1999; Onate et al., 1995]. Therefore we examined whether CXCL12/CXCR4 axis activation was associated with the nuclear translocation and accumulation of SRC-1, and SRC-1:AR protein:protein association. To accomplish this, LAPC-4, VCaP, 22Rv1, or LNCaP cells were treated with vehicle, 1nM synthetic androgen R1881, or 50pM CXCL12 for 48hr in serum-free, phenol red-free media. Cellular proteins were separated into nuclear and cytosolic fractions and assessed for the accumulation of nuclear SRC-1 protein by immunoblotting. As shown in Figure 1A, treatment with R1881 or CXCL12 was promoted the nuclear accumulation of SRC-1 in VCaP and LNCaP cells, but not in LAPC-4 or 22Rv1 cells.

We next tested whether CXCL12/CXCR4 axis activation promoted AR:SRC-1 protein:protein association. For these studies, LNCaP cells were treated as described above, then the nuclear and cytosolic protein extracts were isolated and immunoprecipitated using antibodies against AR or SRC-1. The immunoprecipitated proteins were electrophoresed and immunoblotted against antibodies specific for AR, SRC-1, hnRNP or actin. As shown in Figure 5A, both R1881- and CXCL12-treated cells exhibited nuclear accumulation of both AR and SRC-1 compared to vehicle-treated cells. AR:SRC-1 protein:protein association was also evident, with higher levels of AR:SRC-1 associated proteins evident in the nuclear fraction of CXCL12-compared to R1881- or vehicle-treated cells (densitometry values shown in Supplementary Figure S4).

Figure 5. CXCL12/CXCR4 Axis-Mediated Transcriptional Responses are AR- and SRC-1 Dependent.

A. LNCaP cells were treated in phenol red-free SF RPMI for 48 hr with vehicle (V), 1nM R1881 (R) or 50pM CXCL12 (C). The nuclear (nuc) and cytosolic (cyto) protein extracts were isolated and immunoprecipitated using antibodies against AR or SRC-1. The immunoprecipitated proteins were electrophoresed and immunobloted against antibodies specific for AR, SRC-1, hnRNP or actin. Both R1881- and CXCL12-treated cells exhibited nuclear accumulation for protein levels for both the AR and SRC-1 compared to vehicle-treated cells. AR:SRC-1 protein:protein association was also evident. Loading controls include hnRNP for nuclear and actin for cytoplasmic extracts, respectively.

B. LNCaP cells were mock-transfected (M) or transfected with 5nM siRNA non-targeting sequence control (NT), siRNA against AR (sA) or against SRC-1 (sS), the cells lysed, and proteins electrophoresed and immunoblotted against antibodies specific for SRC-1, AR, or actin to confirm AR or SRC-1 transcript silencing.

C. LNCaP cells prepared as in B. were grown in phenol red-free, SF RPMI for 12 hrs, then treated with vehicle (V), 1nM R1881 (R), 50 pM CXCL12 (C) or both (R+C) for 4 hr. RNA was isolated and assessed by qRT-PCR for AR, CXCR4, PSA and TMPRSS2 transcript levels. Cycle numbers to threshold were calculated by subtracting averaged RPLPO template control transcript levels from averaged transcript levels for PSA, TMPRSS2, AR, or CXCR4, and Fold Transcript levels were determined as the ratio transcript levels in treated : untreated cells, with vehicle-treated cells set at 0%. Both R1881- and CXCL12-stimulated NT-transfected cells expressed significantly higher transcript levels for the PSA and TMPRSS2 genes, CXCL12-stimulated NT-transfected cells expressed significantly higher levels of AR and PSA gene transcripts, and NT-transfected cells stimulated with both R1881 and CXCL12 expressed higher transcript levels for all 4 genes tested compared to vehicle-treated cells. In contrast, siAR-transfected cells were repressed for these same transcripts for cells treated with R1881 or CXCL12 alone or in combination, whereas siSRC-1-transfected cells expressed PSA and TMPRSS2 transcripts in response to R1881 treatment at levels similar to those of NT-transfected cells but were repressed for expression of these same transcripts in CXCL12-treated cells. Graphs shown depict representative date from replicate measures.

D. LNCaP cells were grown in phenol red-free SF RPMI medium treated with vehicle (V), 1nM R1881 (R), 50 pM CXCL12 (C) for 48 hrs and subjected to ChIP using antibodies specific for AR or Re-ChIP using antibodies specific against AR followed by SRC-1. The precipitated chromatin was amplified by PCR using primers flanking AREs in the promoter region of the PSA and TMPRSS2 genes. Controls are as indicated. Input reactions were subjected to PCR but not immunoprecipitation, IgG IP reactions were subjected to PCR after immunoprecipitation with IgG isotype antibody, AR IP reactions were subjected to PCR after immunoprecipitation with AR antibody, and SRC-1 IP reactions were subjected to PCR after immunoprecipitation with AR followed by SRC-1 antibodies. Higher levels of AR- and AR:SRC-1 proteins were detected associated with the PSA and TMPRSS2 promoters after treatment with R1881 or CXCL12 compared to Vehicle-treated cells.

We then wished to determine whether the observed activated CXCL12/CXCR4 axis-mediated AR:SRC-1 protein complex was functional and perhaps required for the observed CXCL12/CXCR4 axis-mediated AR-dependent transcription of the PSA and TMPRSS2 genes (Figure 3B). To accomplish this siRNA approaches were utilized to selectively knock-down AR or SRC-1 expression, then monitor the ability of the activated CXCL12/CXCR4 axis to promote transcription of the PSA and TMPRSS2 genes. For these experiments, LNCaP cells were mock-transfected, transfected with siRNA non-targeting sequence control (NT) or siRNA against AR (siAR) or against SRC-1 (siSRC-1). The cells were lysed, and proteins electrophoresed and immunoblotted against antibodies specific for SRC-1, AR, or actin. These studies confirmed near-complete AR and SRC-1 transcript silencing and protein ablation (Figure 5B). These cells were then grown in phenol red-free, SF RPMI, and RNA was isolated and subjected to qRT-PCR as described previously. As shown in Figure 5C, R1881- -stimulated NT-transfected cells expressed elevated transcript levels for the PSA and TMPRSS2 genes, CXCL12-stimulated NT-transfected cells expressed elevated levels of AR and PSA gene transcripts, and NT-transfected cells stimulated with both R1881 and CXCL12 expressed elevated higher transcript levels for all 4 genes tested compared to vehicle-treated cells. In contrast, siAR-transfected cells were repressed for these same transcripts for cells treated with R1881 or CXCL12 alone or in combination, whereas siSRC-1-transfected cells expressed PSA and TMPRSS2 transcripts in response to R1881 treatment at levels similar to those of NT-transfected cells but were repressed for expression of these same transcripts in CXCL12-treated cells. Taken together, these studies suggest that the androgen-promoted transcription of the AR-regulated PSA and TMPRSS2 genes was AR-, but not SRC-1 dependent, whereas the CXCL12-promoted transcription of these same genes was both AR- and SRC-1 dependent.

Chromatin immunoprecipitation (ChIP) assays were next pursued to further test whether both AR and SRC-1 co-localized to the ARE regions of the promoters of the PSA and TMPRSS2 genes in response to CXCL12/CXCR4 axis activation in vivo. To accomplish this, LNCaP cells were grown in phenol red-free, SF RPMI then treated with vehicle, 1nM R1881 or 50pM CXCL12 for 48 hr. After cross-linking and shearing, the chromatin was immunoprecipitated using antibodies specific for AR or the isotypeIgG control, and the DNA amplified using primers specific for ARE-enriched promoter regions of PSA or TMPRSS2 genes. As seen in Figure 5D, minimal or no PCR product was observed in the mock reaction, reactions lacking input DNA or antibody, as well as samples immunoprecipitated using the isotype IgG antibody. In addition, when subjected to PCR, input DNA amplicon levels were equivalent in samples from cells treated with vehicle, R1881, or CXCL12. However, both R1881- and CXCL12-treated samples immunoprecipitated using the AR antibody demonstrated much higher levels of both PSA and TMPRSS2 gene promoter amplicon compared to vehicle-treated cells. Moreover, re-ChIP of these samples using the SRC-1 antibody demonstrated little or no PCR amplicon in IgG immunoprecipitated samples as well as vehicle- and R1881-treated cells, but high levels of amplicon in CXCL12-treated cells. Taken together, this data shows that activation of the CXCL12/CXCR4 axis specifically recruits both AR and SRC-1 to the ARE-rich regions of the promoters of the PSA and TMPRSS2 genes. The ChIP data is consistent with that observed in the siRNA experiments in showing that AR is required for both androgen- and CXCL12-mediated transcription, whereas SRC-1 is also required for CXCL12-mediated, but not androgen-mediated, transcription, of the PSA and TMPRSS2 genes.

3.6. CXCL12-Mediated AR Transactivation is Dependent upon Activation of MAPK and PI3K Signaling Pathways

Kinase mediated signal transduction pathways regulate the activity of nuclear receptors by reversible phosphorylation [Burnstein and Cidlowski, 1993; Orti et al., 1992]. AR phosphorylation is thought to play a survival role in prostate cancer by protecting cells from apoptosis [Lin et al., 2001]. We have previously shown that CXCL12/CXCR4 interactions induce intracellular kinases that the phosphorylation and activation of downstream signaling molecules [Begley et al., 2005, 2007; Kasina et al., 2009]. We therefore hypothesized that similar mechanisms might activate the AR and/or SRC-1 through phosphorylation.

In order to elucidate whether CXCL12/CXCR4 axis activation promoted phosphorylation of the AR, LNCaP cells were pre-treated with vehicle (DMSO) or the kinase inhibitors U0126 targeting MAPK MEK/ERK or wortmannin targeting PI3K pathways, followed by treatment with vehicle or CXCL12. Total cellular proteins were isolated and immunoprecipitated using an AR-specific antibody. Equivalent volumes of immunoprecipitated proteins were electrophoresed, then immunoblotted against antibodies that recognize phosphorylated serine (pSer) or tyrosine (pTyr) residues (Figure 6A). Densitometric analysis of the immunoblot demonstrated a 1.5–2.0 fold increase in serine-phosphorylated AR in response to treatment with CXCL12 which was ablated to 0.5-fold in cells pre-treated with wortmannin (Supplementary Figure S5). Immunoblots of protein lysates from vehicle- orCXCL12-treated cells failed to demonstrated a band of the appropriate molecular weight for AR when probed with the pTyr antibody either in the presence or absence of inhibitors (Figure 6A). These results suggested that activation of the CXCL12/CXCR4 axis mediated phosphorylation of serine, but not tyrosine, residues on the AR, and that this activity was PI3K-dependent.

Figure 6. CXCL12/CXCR4 Axis Activation Mediates AR and SRC-1 Phosphorylation.

A. LNCaP cells were pre-treated with vehicle (DMSO), 5uM U0126 (MEK/ERK inhibitor) or 1uM wortmannin (PI3K inhibitor) for 1 hr, followed by treatment with vehicle (V) or CXCL12 (C). Total cellular proteins were isolated and pre-cleared using unconjugated beads, then immunoprecipitated (IP) using antibody-coated beads specific for AR. Equivalent volumes of immunoprecipitated proteins were electrophoresed, then immunoblotted (IB) against antibodies that recognize phosphorylated serine (pSer) or tyrosine (pTyr) residues. CXCL12-treated cells not subjected to inhibitor pre-treatment demonstrated higher levels of pSer than vehicle-treated cells. CXCL12-treated cells pre-treated with wortmannin demonstrated both reduced AR and pSer-AR levels. Phosphorylation of AR Tyrosine residues was not observed under any of the experimental conditions used. Immunoblotting against an antibody specific for AR indicates AR protein levels in each lane.

B. LNCaP cells were treated as described in A., total cellular proteins were isolated and pre-cleared using unconjugated beads, then immunoprecipitated (IP) using antibody-coated beads specific for SRC-1. Equivalent volumes of immunoprecipitated proteins were electrophoresed, then immunoblotted (IB) against antibodies that recognize phosphorylated serine (pSer) or threonine (pThr) residues. CXCL12- treated cells not subjected to inhibitor pre-treatment demonstrated higher levels of pThr than Vehicle-treated cells. CXCL12-treated cells pre-treated with vehicle or wortmanin demonstrated higher levels of pThr-SRC-1, while those cells pre-treated with U0126 demonstrate reduced levels, compared to vehicle treated cells. A band of the appropriate molecular weight for SRC-1 was not detected, suggesting that activation of the CXCL12/CXCR4 axis did not promote SRC-1 phosphorylation at Serine residues. Immunoblotting against an antibody specific for SRC-1 indicates SRC-1 protein levels in each lane.

C. LNCaP cells were left un-pre-treated or were pre-treated with 5uM U0126 (MEK/ERK inhibitor), 1uM wortmannin (PI3K inhibitor) or vehicle (0.1% DMSO) for 1 hr, then treated with 50pM CXCL12 for 0, 10 or 60 minutes. The cellular proteins were subjected to electrophoresis and immunoblotting. Two immunoblots were generated; the first immunoblot was sequentially probed, stripped, and re-probed with antibodies specific for phospho-AR Ser210 (pAR), AR, GAPDH, and the second immunoblot was similarly assessed with antibodies specific for phospho-AktSer473 (pAkt), Akt, phospho-Erk 1 / 2 (pErk), Erk 1 / 2, and GAPDH. Phosphorylation of the AR at Ser210 was evident after 10 or 60 minutes treatment with CXCL12, and this was ablated upon pre-treatment with wortmannin, but not U0126. P13K signaling inhibition by wortmannin was confirmed by the observation that Akt was not phosphorylated in cells pre-treated with wortmannin but was evident in non-pretreated cells or those pre-treated with DMSO or U0126. Similarly, MAPK signaling inhibition by U0126 was confirmed by the observation that Erk 1 / 2 was not phosphorylated in cells pre-treated with U0126 but was evident in non-pretreated cells or those pre-treated with DMSO or wortmannin. The blots shown are representative of replicate experiments.

D. LNCaP cells were treated for 0, 5, 10, 15, 20 or 60 min with 50pM CXCL12. Protein lysates were prepared, electrophoresed, and immunoblotted against antibodies specific for Src family members or the same phosphorylated at residue Y418, or actin as loading control. Src phosphorylation in response to CXCL12 treatment is evident at 5, 10, 15 and 20 minutes. The blot shown is representative of replicate experiments; densitometric evaluation for this and 2 additional blots is provided in Supplementary Figure S7.

E. LNCaP cells were pre-treated with vehicle or the kinase inhibitors U0126 targeting activation of MAPK MEK/ERK signaling, wortmannin or LY294002 targeting PI3K signaling, BISI targeting PKC signaling, PP1 targeting Src family kinase signaling or the inactive analogue PP3, followed by treatment with vehicle or CXCL12, and the level of PSA secreted into the media was detected by ELISA. Cells treated with vehicle secreted baseline levels of PSA into the media, set at 1-fold. Pre-treatment of these cells with each of the various inhibitors did not significantly reduce the secretion of PSA into the media below baseline levels. Cells treated with CXCL12 secreted significantly higher (1.59-fold) levels of PSA into the media compared to vehicle-treated cells. CXCL12-treated cells demonstrated significantly lower levels of PSA secretion when pre-treated with each of the various inhibitors with the exception of PP3, the inactive analogue of PP1.

Previous studies by Rowan et al. (2000a) had demonstrated that SRC-1 phosphorylation was MAPK-dependent. Moreover, Ueda et al 2002 had elegantly demonstrated that phosphorylation of SRC-1 by MAPK was required for optimal ligand-independent activation of the AR by IL-6. Therefore, we determined whether CXCL12/CXCR4-mediated SRC-1 phosphorylation may be similarly MAPK-dependent. For these studies, LNCaP cells were treated as described above and proteins immunoprecipitated using an antibody against SRC-1. Equivalent volumes of immunoprecipitated proteins were electrophoresed, then immunoblotted against antibodies that recognize phosphorylated serine (pSer) or threonine (pThr) residues. These studies showed that CXCL12-treated cells pre-treated with vehicle in the absence of kinase inhibitor demonstrated higher levels of pThr-SRC-1 compared to vehicle-treated cells. CXCL12-treated cells pre-treated with U0126, but not wortmannin, demonstrated reduced pThr-SRC-1 levels compared to vehicle-treated cells (Supplementary Figure S6). No band of the appropriate molecular weight for SRC-1 was detected upon immunoblotting with a pan-phospho-Serine antibody, hence, no evidence of serine phosphorylation was detected in response to treatment with CXCL12 (Figure 6B). Taken together, these data suggest that activation of the CXCL12/CXCR4 axis promotes phosphorylation of SRC-1 at threonine, but not serine, residues, and that these events are MAPK-, but not PI3K-, dependent.

CXCL12/CXCR4-mediated activation of both the MAPK and PI3K pathways and AR phosphorylation at the specific Ser210/213 residue was also examined (Figure 6C). Phosphorylation of the AR at Ser210/213 was evident after 10 or 60 minutes treatment with CXCL12, and this was ablated upon pre-treatment with wortmannin, but not U0126. These data were consistent with those shown in Figure 6A demonstrating PI3K-dependent CXCL12/CXCR4 axis mediated phosphorylation of serine residues on the AR. P13K signaling inhibition by wortmannin was confirmed by the observation that Akt was not phosphorylated at the Ser473 residue in cells pre-treated with wortmannin, but was evident in non-pretreated cells or those pre-treated with U0126. Similarly, MAPK signaling inhibition by U0126 was confirmed by the observation that Erk 1 / 2 was not phosphorylated in cells pre-treated with U0126 but was evident in non-pretreated cells or those pre-treated with wortmannin. Because earlier studies from our laboratory had also implicated Src family kinase activation in response to CXCL12 treatment in PC3 and N15C6 prostate epithelial cells, we also assayed this activity in LNCaP cells. As shown in Figure 6D, phosphorylation of Src kinase at residue Y418 was demonstrated in LNCaP cells in response to short-term treatment with 50pM CXCL12 (densitometric evaluation provided in Supplementary Figure S7).

The studies just described suggest a role for kinase-mediated activation of AR and/or SRC-1 in response to activation of the CXCL12/CXCR4 axis. To test whether kinase activation was required for CXCL12/CXCR4 axis-mediated ligand-independent functional activation of the AR, LNCaP cells were assayed for PSA secretion in response to CXCL12 with or without pre-treatment with various kinase inhibitors. As previously observed, treatment with CXCL12 significantly increased PSA secretion. However, CXCL12/CXCR4-stimulated PSA secretion was significantly reduced upon pre-treatment with the MAPK MEK/ERK inhibitor, U0126, or the P13K inhibitors, wortmannin or LY294002 (Figure 6E). Notably, LY294002 inhibited basal secretion of PSA as well. The potential role of PKC and Src family kinases in mediating CXCL12/CXCR4 axis functional transactivation of AR were also determined by assaying cells for PSA secretion in response to CXCL12 with or without pre-treatment with the PKC inhibitor, BISI, or the Src family kinase inhibitor, PP1 and its inactive analog, PP3. As show in Figure 6E, CXCL12/CXCR4 axis-stimulated PSA secretion was significantly reduced upon pre-treatment with BISI or PP1 but not upon pre-treatment with the inactive analog, PP3. These data suggest that multiple kinase signaling pathways, including MAPK MEK/ERK, PI3K, Src kinase- and PKC-mediated pathways, are required for CXCL12/CXCR4-mediated functional AR transactivation.

4. DISCUSSION

This study was designed to determine whether CXCL12/CXCR4 interactions functionally transactivate the AR to mediate cellular proliferation in the absence of androgen, and to begin to elucidate which intracellular signaling events mediate this process. The results of these studies showed that activation of the CXCL12/CXCR4 axis promoted: The nuclear accumulation of both wild-type and mutant AR in several prostate epithelial cell lines; AR-dependent proliferative responses; nuclear accumulation of the AR co-regulator SRC-1 protein; SRC-1:AR protein:protein association; co-localization of AR and SRC-1 on the promoters of AR-regulated genes; AR- and SRC-1 dependent transcription of AR-regulated genes; AR-dependent secretion of the AR-regulated PSA protein; P13K-dependent phosphorylation of AR; MAPK-dependent phosphorylation of SRC-1, and both MAPK- and P13K-dependent secretion of the PSA protein, in the absence of androgen. Taken together, these observations are consistent with the identification of novel pathways through which activation of the CXCL12/CXCR4 axis promotes AR activity and function in the absence of steroid hormone.

Frigo et al. (2009) had previously reported that androgen works indirectly through up-regulation of the KLF5 transcription factor to increase the levels of both CXCR4 mRNA and functional protein in LNCaP cells. Seaton et al. (2008) reported that stimulation of the androgen-dependent cell lines, LNCaP and 22Rv1, with the CXC-type chemokine, CXCL8, increased AR gene expression at the messenger RNA and protein levels and AR transcriptional activity in both cell lines. Moreover, inhibition of CXCL8 binding to its cognate receptor, CXCR2, by small molecule antagonist, AZ10397767, attenuated CXCL8-induced increases in AR expression and transcriptional activity. The data presented in this manuscript shows that stimulation of LNCaP cells with either synthetic androgen or CXCL12 significantly increased AR and CXCR4 transcript levels compared to vehicle-treated cells in an AR-dependent manner (Figure 3A,B, and Figure 5C, Supplementary Figure S3). However, perhaps due to differences in experimental conditions, e.g., the use of lower levels of synthetic androgen, more stringent serum-free media growth conditions, and longer time points in the present studies compared to those used by Frigo et al. (2009), the observed increases in for AR and CXCR4 transcript levels were not apparent at the protein level. Taken together, the studies reported here and cited above suggest that some CXC-type chemokines, in particular, CXCL8 and CXCL12, may promote AR activation through increasing AR expression levels as well as through increasing AR nuclear accumulation and promoting AR-mediated gene transcription and proliferative responses.

McCall et al. (2008) recently reported that nuclear staining for serine-phosphorylated AR, as well as for cytoplasmic pAkt and PI3K, were significantly higher in hormone refractory compared to hormone sensitive human prostate tumors, and were significantly associated with poor disease prognosis and reduced patient survival. Other studies have demonstrated that poorly differentiated high Gleason grade prostate tumors demonstrate elevated levels of pAkt [Malik et al., 2002]. Elevated pAkt levels correlated with a high proliferative index (measured by Ki67 staining) in human prostate tumors [Ghosh et al., 2005], and were an excellent predictor of poor clinical outcome for prostate cancer patients [Kreisberg et al., 2004]. In vitro studies have shown that androgen ablation increases activation of the PI3K/Akt pathway, and thereby promotes prostate cancer cell growth and survival [Lin et al., 2001]. Moreover, activated Akt, but not PI3K, phosphorylates AR at Ser210 and Ser790 residues, which are Akt consensus phosphorylation sites [Lee et al., 2003]. Our studies show that CXCL12/CXCR4 interactions stimulated phosphorylation of AR at serine residues, and that inhibition of PI3K activation diminished CXCL12/CXCR4-mediated AR phosphorylation (Figure 6 A). These data suggest that one mechanism for CXCL12/CXCR4-stimulated AR transactivation may be through PI3K-mediated AR phosphorylation.

Another potential kinase-mediated mechanism through which activation of the CXCL12/CXCR4 axis could promote AR transactivation is through phosphorylation and activation of AR co-regulator proteins. Like other members of the steroid/nuclear receptor family, efficient transcriptional activation of AR also requires activation of one or more nuclear receptor co-activators. SRC-1 is the first cloned nuclear receptor co-activator, is known to interact with the amino-terminal domain of the AR, and is known to be activated by through MAPK-dependent phosphorylation [Alen et al., 1999; Berrevoets et al., 1998; Bevan et al., 1999; Ding et al, 1998; Ma et al., 1999; Onate et al., 1995; Rowan et al., 2000a]. SRC-1 interacts with the ligand-binding and the N- terminal domains of the AR, thereby promoting AR dimerization and activation [Edwards and Bartlett, 2005]. Phosphorylation of SRC-1 changes its affinity for specific nuclear receptors and modulates steroid receptor-dependent gene expression in a ligand independent manner [Rowan et al., 2000a, 2000b; Lopez et al., 2001]. MAPK-mediated phosphorylation of SRC-1 on Thr1179 and Ser1185 increases the affinity of SRC-1 for AR in prostate cancer cells, perhaps contributing to prostate cancer recurrence [Ueda et al., 2002; Gregory et al., 2004]. We have previously reported that activation of the CXCL12/CXCR4 axis stimulated MAPK pathways in prostate cancer cells, including LNCaP [Begley et al., 2005, 2007]. We now report that activation of the CXCL12/CXCR4 axis was associated with phosphorylation of the AR co-regulator, SRC-1, at threonine, but not serine, residues, and that this phosphorylation was diminished by MAPK MEK/ERK inhibition (Figure 6B). Taken together, these data suggest that, consistent with other ligand-independent mechanisms shown to functionally transactivate the AR, activation of the CXCL12/CXCR4 axis promotes MAPK MEK/ERK kinase-mediated SRC-1 phosphorylation.

The studies described here also demonstrate that inhibition of PKC and Src family kinase activation also repressed CXCL12/CXCR4 axis-mediated PSA secretion in LNCaP cells. Previous studies by our laboratory demonstrated that Src family kinases are activated immediately downstream of CXCL12/CXCR4 activation [Kasina et al., 2009]. Both Src family kinases and PKC can act ‘upstream’ of P13K and MAPK signaling pathways, and further studies should demonstrate whether CXCL12/CXCR4-mediated activation of Src family kinases and/or PKC functions in this manner to stimulate AR activation. Taken together, these studies suggest that CXCL12/CXCR4-mediated activation of multiple kinase pathways may mechanistically contribute to functional AR transactivation, and that these pathways mediate phosphorylation of AR as well as AR co-regulators, particularly SRC-1.

Three experiments conducted during the course of these studies suggest that CXCL12/CXCR4 axis-mediated functional transactivation of the AR was SRC-1- dependent. Immunoprecipitation studies suggested higher levels of SRC-1 nuclear accumulation and SRC-1:AR association in CXCL12-, but not androgen-, treated LNCaP cells (Figure 5A, Supplementary Figure S4). SRC-1 transcript silencing in LNCaP cells abrogated CXCL12/CXCR4-, but not androgen/AR-, mediated transcription of the AR-regulated PSA and TMPRSS2 genes (Figure 5C). Chromatin immunoprecipitation (ChIP) assays demonstrated an association of the AR with the promoters of both the PSA and TMPRSS2 genes in androgen- and CXCL12-treated cells. However, re-ChIP assays demonstrated a co-association of AR and SRC-1 on the PSA and TMPRSS2 gene promoters only in CXCL12-, and not androgen-, treated cells (Figure 5D). Previous studies by Ueda et al. (2002) clearly demonstrated that IL-6-mediated functional activation of the AR also involved the activation of SRC-1 and promoted AR:SRC-1 association. Based on these combined studies, it is likely that the SRC-1 co-regulator protein is a key facilitator of ligand-independent transactivation of the AR mediated by several different types of non-steroidal growth factors, including CXCL12 and IL-6.

This manuscript relates data showing that activation of the CXCL12/CXCR4 axis promotes AR-dependent gene transcription, cellular proliferation, and PSA secretion. However, as observed using multiple approaches, activation of the CXCL12/CXCR4 axis promotion of these AR-regulated activities was less robust than that observed for activation of the androgen/AR axis. For example, although low levels of CXCL12 and androgen worked in an additive manner to promote cellular proliferation, this additivity was no longer apparent at higher androgen levels (Figure 2). This speaks to the relatively weak nature of CXCL12/CXCR4-mediated AR activation, e.g., that it is easily superseded by androgen itself. However, it also speaks to the idea that diverse types of growth factors may work cooperatively with low levels of androgen to continue to maintain AR activity under hormone deprivation conditions. Several studies have suggested that hormone deprivation therapy does not completely ablate bio-available androgen in the prostate, and that the conversion of androgen precursors to active androgen, or the production of active androgens by the adrenal glands, can maintain the AR in an active and functional state in the prostate under these conditions [Geller, 1995; Gregory et al., 2004; Mohler, 2008; Mohler et al., 2011; Tan et al., 1997]. The study reported here raises the possibility that non-steroidal growth factors, such as interleukins and chemokines, may augment low levels of bio-available androgens to maintain the AR in an active and functional state under hormone ablation conditions in castration resistant prostate tumors. This concept is supported by other studies demonstrating that cytokine- and chemokine-type growth factors, notably, interleukin-6 [Ueda et al., 2002] and CXCL8 [Seaton et al., 2008], maintain AR activation in the absence of steroid hormone. Moreover, there is evidence that an activated CXCL12/CXCR4 axis augments estrogen receptor-mediated transcriptional activity and the expression of ER target genes in breast cancer cells [Esquenet et al, 1997]. Therefore, the CXCL12/CXCR4 axis may comprise one of many growth factor/growth factor receptor axes that enable prostate cancer progression towards a castration resistant phenotype. In addition, activation of the CXCL12/CXCR4 axis may comprise an important mechanism for the progression of endocrine cancers in general.

5. CONCLUSIONS

In summary, the data reported here identify the activated CXCL12/CXCR4 axis as a novel AR-dependent mechanism that promotes the growth of androgen-dependent prostate epithelial cells by functionally transactivating the AR in the absence of androgen. The implication of these findings is that therapeutically targeting the CXCL12/CXCR4 axis may be warranted to develop therapeutics to effectively halt the progression, and consequent mortality, of castration resistant prostate tumors.

Highlights.

The CXCL12/CXCR4 axis functionally activates the AR in the absence of androgen.

CXCL12/CXCR4 promotes nuclear accumulation of the AR and SRC-1 co-regulator.

CXCL12/CXCR4 stimulates AR-regulated gene transcription and cellular proliferation.

CXCL12/CXCR4 activates AR and SRC-1 viaPI3K- and MAPK-mediated phosphorylation.

CXCL12/CXCR4-mediated AR activation may provide a new therapeutic target.