Effects of Sorafenib on C-Terminally Truncated Androgen Receptor Variants in Human Prostate Cancer Cells

Recent evidence suggests that the development of castration resistant prostate cancer (CRPCa) is commonly associated with an aberrant, ligand-independent activation of the androgen receptor (AR). A putative mechanism allowing prostate cancer (PCa) cells to grow under low levels of androgens, is the expression of constitutively active, C-terminally truncated AR lacking the AR-ligand binding domain (LBD). Due to the absence of a LBD, these receptors, termed ARΔLBD, are unable to respond to any form of anti-hormonal therapies. In this study we demonstrate that the multikinase inhibitor sorafenib inhibits AR as well as ARΔLBD-signalling in CRPCa cells. This inhibition was paralleled by proteasomal degradation of the AR- and ARΔLBD-molecules. In line with these observations, maximal antiproliferative effects of sorafenib were achieved in AR and ARΔLBD-positive PCa cells. The present findings warrant further investigations on sorafenib as an option for the treatment of advanced AR-positive PCa.

cells via inhibition of the canonical AR-signalling pathway. Inhibition of AR-signalling was paralleled by a downregulation of AR protein-levels [15]. As most AR phosphorylation sites are located at the N-terminus [13], a region shared by both, full length AR and ARΔLBD, we hypothesized that sorafenib might also affect ARΔLBD-function. Therefore, we investigated the effect of sorafenib on ARΔLBD-signalling using AR-negative PC-3 cells transiently transfected with the AR∆LBD-variant AR Q640X as well as the AR/AR-V expressing PCa cell line 22Rv1 as experimental models.

Effect of Sorafenib on Constitutively Active, C-Terminally Truncated AR-Mutant Q640X
Posttranslational modifications like phosphorylation on serine, threonine or tyrosine residues are involved in a large variety of steroid receptor functions [13,14]. Based on recent findings by Oh et al. it is tempting to speculate that the multikinase-inhibitor sorafenib targets the AR phosphorylation via blockade of a yet undefined kinase [15]. As the majority of AR phosphorylation sites are located at the N-terminus of the receptor molecule, we hypothesized that sorafenib might also affect phosphorylation of the ARΔLBD-N-terminus.
As an experimental model we transfected PC-3 cells with the AR-mutant AR Q640X , the product of a nonsense mutation leading to a stop codon in the hinge region adjacent to the LBD of the AR [6]. Transactivational activity of AR Q640X was shown to be very strong on artificial androgen-responsive promoters (ARE(2x)) but was very weak on the PSA promoter [17,18]. In contrast to the wild type AR many ARΔLBD are unable to activate the full panel of androgen-dependent genes [17][18][19]. As seen in Figure 2 sorafenib was able to inhibit transactivation of the constitutively active, C-terminally truncated AR Q640X in a dose-dependent manner. Inhibition was significant at a concentration of 5 μM (downregulation versus untreated controls: 30% ± 7%, p = 0.012), reaching its maximum at 10 μM (downregulation versus untreated controls: 46% ± 8%, p = 0.003). The data suggest that sorafenib affects wild type AR and ARΔLBD signalling in a similar way.

Figure 2.
Sorafenib inhibits AR Q640X -signalling in PC-3 cells. AR negative PC-3 cells were cotransfected with an AR Q640X construct (AR with point mutation in the hinge region, 640 amino acids) together with an ARE(2x)-reporter plasmid. pRL-tk-LUC was co-transfected as an internal control for transfection efficiency. Reportergene activity after sorafenib treatment (SORA) was measured using a Dual-Luciferase Reporter Assay as recently described [17]. Results are expressed in percent transactivation of untreated controls which were set at 100%; * p < 0.05; ** p < 0.01.

Sorafenib Induces Proteasomal Degradation of AR and AR-V Splice Variants in 22Rv1 Cells
There is experimental evidence that kinase inhibitors directed against p42/p44 MAPK, GSK-3β or CDK1 are able to trigger AR-signalling by modulating AR-protein levels [20][21][22]. Recently, the multikinase inhibitor sorafenib was shown to diminish PSA-levels in LNCaP and its bicalutamide resistant subline LNCaP-Bic. The reduction in PSA-levels was paralleled by a decrease of full length AR [15]. The observation that sorafenib is able to downregulate intracellular AR-levels prompted us to analyze its effects on intracellular ARΔLBD levels. Although transient expression of AR Q640X is largely sufficient to perform reportergene assays (Figure 2), the expression levels of the AR Q640X protein transfected into PC-3 cells are too low to perform a western blot analysis. Therefore we tested the effects of sorafenib on ARΔLBD-levels in 22Rv1 cells, known to express large amounts of the AR-splicing variant AR-V7 [23]. Although AR Q640X and AR-V7 are generated by different mechanisms, both ARΔLBD receptor forms share several common features like identical transactivation and DNA-binding domains, receptor size (AR-V7: 642 amino acids; AR Q640X : 640 amino acids), predominant nuclear localization in the absence of androgens and constitutive activity as shown by activation of PSAP1 luciferase reporter plasmid [23].
Interestingly, sorafenib was able to diminish both full length AR as well as AR-V in 22Rv1 cells. Downregulation of AR and AR-V protein levels following sorafenib treatment could be rescued by the proteasome inhibitor MG132, the latter suggesting that sorafenib induces a proteasomal degradation of AR-and ARΔLBD molecules in PCa cells ( Figure 3).

Figure 3.
Downmodulation of AR and AR-V in 22Rv1 cells is due to sorafenib induced proteasomal degradation. 22Rv1 cells were incubated with the proteasome inhibitor MG132 (5 µM) for 60 min followed by treatment with sorafenib (5 µM) for 18 h. Subsequently cell extracts were analyzed by Western blot analysis (AR: androgen receptor; AR-V, ARΔLBD generated by alternative splicing; β-actin: loading control; ctrl: untreated control; SORA: sorafenib; MG132: proteasome inhibitor). AR, AR-V and β-actin levels were quantified by densitometry and expressed as fold-change of AR/β-actin or AR-V/β-actin control (ctrl) levels which were set at 1.00.

Sorafenib Does not Modulate the Subcellular Distribution of AR and AR∆LBD in PCa Cells
Besides its effects on AR-stability, various kinase inhibitors were shown to modulate the intracellular localization of the AR [21,[24][25][26]. In consequence we wondered whether the multikinase inhibitor sorafenib is also able to modulate subcellular distribution of AR-molecules. Therefore, we transfected PC-3 cells with expression plasmids coding for green fluorescent AR-and AR Q640X -fusion proteins. As seen in Figure 4, sorafenib was unable to influence the subcellular distribution of AR as well as its c-terminally truncated ARΔLBD-counterpart AR Q640X .

Inhibition of Cell Proliferation after Sorafenib-Treatment in PCa Cell Lines
Based on the observation that sorafenib is able to inhibit AR as well as ARΔLBD-signalling we further investigated the antiproliferative effects of the compound on the androgen sensitive LNCaP (AR+) cells, the castration resistant 22Rv1 (AR+, AR-V+) cells as well as the androgen insensitive PC-3 (AR−) and DU-145 (AR−) cells using a MTT cell viability assay [27]. As depicted in Figure 5, the antiproliferative effects of sorafenib were more pronounced in AR-positive or AR/AR-V-positive prostate cancer cells as compared to those lacking the androgen receptor. Differences between AR+ and AR-cells were statistically significant at a sorafenib concentration of 2.5 μM (proliferation rate LNCaP: 60% ± 5% and 22Rv1: 76% ± 3% versus PC-3: 82% ± 3%, p = 0.002 and p = 0.036, respectively).

Plasmids and Chemicals
pSG5-AR encoding a wild-type full-length AR (919 amino acids) was supplied by Dr. H. Klocker (Innsbruck, Austria). pAR-t1EosFP coding for a green fluorescent Eos-AR-fusion protein was a generous gift from Dr. F. Oswald (Ulm, Germany). pCruz-ARQ640X and pEGFP-ARQ640X coding for the C-terminally truncated ARQ640X (aa 1-640) were provided by and Dr. J. Céraline (Strasbourg, France). The PSA reporter plasmid pPSA-61luc under control of a 6kb-fragment of the human PSA-promoter was a generous gift of Dr. J. Trapmann (Rotterdam, The Netherlands). The artificial ARE(2x) reporter plasmid pLC0548 (pARE(2x)-luc) under control of a synthetic ARE-promoter was created by H. Lebedur and provided by Dr. A. Allera (Bonn, Germany). Renilla reniformis luciferase reporter plasmid (pRL-TK) was purchased from Promega (Mannheim, Germany). Dihydrotestosterone (DHT) and the proteasome inhibitor MG132 were purchased from Sigma-Aldrich GmbH (Taufkirchen, Germany). Sorafenib was a product of LKT Laboratories Inc. (St. Paul, MN, USA). All other chemicals, if not specified, were products of Sigma-Aldrich GmbH (Taufkirchen, Germany).

Cell Culture
PC-3, DU-145, 22Rv1 and LNCaP cells were purchased from the American Type Culture Collection (Manassas, VA, USA). Cell culture was performed as recently described [21].

Reporter Gene Assays
PC-3 cells were transiently cotransfected in 24-well plates with AR-expression plasmids (pSG5-AR; pCruz-Q640X) and different reporter gene constructs (pPSA-61luc, pARE(2x)-luc) using the transfection reagent Polyfect (Qiagen, Hilden, Germany). pRL-TK was cotransfected as an internal control for transfection efficiency. Subsequently, cells were treated with/without 1 nM DHT. After 24 h, reporter gene activity was measured using the Dual-Luciferase Reporter Assay (Promega GmbH, Mannheim, Germany). In this experimental set-up, the PSA-and ARE-reporters are correlated with the effects of the specific experimental conditions, while the activity of the co-transfected constitutive pRL-TK reporter provides an internal control that serves as the baseline response. Normalizing the activity of the experimental PSA-and ARE-reporters to the activity of the internal control minimizes experimental variability caused by differences in cell viability, transfection efficiency, general effects on transcription, translation or protein stability. All experiments were performed as recently described [17].

Nuclear Translocation Assay
PC-3-cells were seeded in 24-well plates and grown in the absence of DHT for 24 h. Subsequently, cells were transfected with pAR-t1EosFP and pEGFP-ARQ640X. After 24 h, cells were treated with ethanol (solvent control), 10 nM DHT, ethanol + 5 µM Sorafenib or 10 nM DHT + 5 µM Sorafenib for 2 h. The fluorescent cells were subsequently counted using a fluorescent microscope as recently described [17,28].

Proliferation Assay
Cell viability was determined by means of a colorimetric MTT assay. This assay is based on the the reduction of tetrazolium salts to formazan derivatives by functional mitochondria. The assay was performed as described by Mosmann [27].

Western Blot Analysis and Immunodetection of AR and ARΔLBD
Total proteins were extracted from cells using RIPA buffer. 40 μg of lysate were separated by Sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-Page). Subsequently proteins were transferred onto a PVDF membrane (Invitrolon™ PVDF, Invitrogen, Carlsbad, CA, USA) by semi-dry blotting. AR and AR∆LBD were detected using the monoclonal antibody AR441 (Dako Deutschland GmbH, Hamburg, Germany) at a dilution of 1:2.000 in Tris buffered saline, 0.1% Tween20 (v/v) (TBS-T). Beta-actin served as a loading control using a mouse monoclonal antibody directed against beta-actin (ab8224, Abcam, Cambridge, UK). Immunoreactive bands were detected using a 1:20.000 dilution of peroxidase-coupled secondary goat anti-mouse IgG (sc-2031, SantaCruz Biotechnology Inc., Santa Cruz, CA, USA) in TBS-T. Signals were visualized by the ECL Plus Chemiluminescent Substrate from Pierce (Rockford, IL, USA).

Statistical Analysis
Data are reported as means ± standard deviations. Statistical significance was determined by an analysis of variance (ANOVA). One-way ANOVA results were confirmed with Bonferroni's multiple comparison tests. All analyses were performed with the use of SPSS Statistics (version 19.0; SPSS Software, Chicago, IL, USA, 2010).

Conclusions
Only recently, the efficacy of various tyrosine kinase inhibitors has been evaluated in the treatment of advanced PCa [29][30][31][32][33]. So far, several clinical phase II trials analyzing the impact of a sorafenib-monotherapy on CRPCa reported only moderate therapeutic effects [34][35][36][37]. Unfortunately, interpretation of the clinical data generated from these studies is hampered by the relatively low number of patients enrolled in these trials and a limited knowledge about the molecular changes in CRPCa cells following sorafenib treatment.
In vitro studies analyzing the precise molecular effects of the multikinase inhibitor sorafenib on CRPCa cells are sparse [38,39]. Recently, Oh et al. showed that sorafenib affects AR-signalling in PCa cell lines grown in presence of the synthetic androgen R1881. Inhibition of the canonical AR-signalling pathway after sorafenib treatment was due to a downregulation of AR-levels by an unknown mechanism [15]. To our knowledge, the present study is the first to demonstrate that sorafenib is able to inhibit signalling of C-terminally truncated, constitutively active AR variants (ARΔLBD), the latter are thought to be key players in the development of CRPCa. Interestingly, inhibition of AR-as well as ARΔLBD-signalling was paralleled by a sorafenib-induced proteasomal degradation of both receptor types. Consistently, the antiproliferative effects of sorafenib in our experimental model were the most pronounced in AR-or AR/ARΔLBD-positive PCa cells as compared to their AR-negative counterparts (proliferation rate: 22Rv1/LNCaP < PC-3/DU-145). Our finding of a dose-dependent decrease in PCa cell growth after sorafenib treatment is in accordance with previous findings by Oh et al. [15]. However, in our study the sorafenib concentrations necessary to induce significant antiproliferative effects were higher than those described by Oh et al. At least partially, this may be due to the varying experimental settings used in both studies. Oh and colleagues determined the antiproliferative effect of sorafenib treatment by measuring 3 H-thymidine incorporation after 48 h in PCa cells that were grown and treated in serum free HITES medium (containing hydrocortisone, insulin, transferrin, estrogen, and selenium). By contrast, in our experiments PCa cells were grown in RPMI-1640 under standard conditions (10% fetal bovine serum, antibiotics), cell viability following sorafenib treatment was determined after 72 h using an MTT-assay which reflects mitochondrial activity by measuring the reduction of tetrazolium salts to formazan derivatives. Therefore it is conceivable that the concentration dependencies may differ in both studies.
In line with our observations, Beardsley et al. presented encouraging results using a combination therapy of sorafenib with the nonsteroidal anti-androgen bicalutamide in 39 chemotherapy naïve CRPCa patients. PSA declines or stable disease (≥6 months) were observed in 47% of patients including 26% of patients previously progressing on bicalutamide monotherapy [40].
In summary, the present results suggest that a subset of advanced CRPCa patients, especially those who are expressing AR and/or ARΔLBD, might benefit from a sorafenib treatment. Moreover, strategies to combine multi-targeted kinase inhibitors like sorafenib with hormonal therapies warrant further experimental studies in CRPCa.