Androgen receptor activity is inhibited in response to genotoxic agents in a p53-independent manner

Abstract

The androgen receptor (AR) is fundamental to androgen signalling within the prostate gland, and deregulation of its activity is frequently linked to the development of prostate cancer. Advanced prostate cancer is often treated with chemotherapy and most of these drugs exert their function by generating genotoxic stress such as DNA damage. We have investigated here the effects of genotoxic agents used in chemotherapeutic regimens on AR function and expression. We have discovered that endogenous AR activity in LNCaP cells is inhibited in response to the chemotherapeutic agents etoposide and cisplatin. This loss of AR activity is not caused by a change in cell cycle distribution, a change in subcellular localisation of the AR nor by induction of apoptosis. In addition, we found that inhibition of AR activity in response to genotoxic stress is independent of p53 function. Interestingly, our studies revealed that genotoxic stress inhibits the hormone-stimulated recruitment of AR to androgen response elements. Thus, we report for the first time a mechanism by which the AR activity is inhibited in response to different chemotherapeutic agents.

Introduction

Androgens are the key male hormones involved in the development of the prostate and the male phenotype. The functions of androgens are mediated through the androgen receptor (AR), a member of the steroid hormone superfamily of nuclear receptors, which function as ligand-activated transcription factors (Feldman and Feldman, 2001). When testosterone, the main circulating androgen, enters prostate cells, it is converted into the more active hormone, dihydrotestosterone (DHT, 5α-androstan-17βol-3-one), that binds to and activates the AR (Bruchovsky and Wilson, 1968; Feldman and Feldman, 2001). The activated AR is translocated to the nucleus where it binds to androgen response elements (AREs) in promoter regions of androgen-responsive genes (Riegman et al., 1991; Cleutjens et al., 1997; Tyagi et al., 2000). Recruitment of cofactors enables the AR to interact with the general transcription apparatus and through this either activate or repress the transcription of target genes (Rosenfeld and Glass, 2001; Nelson et al., 2002).

Deregulation of AR function or expression is often seen in prostate cancer, which is one of the most common causes of cancer-related deaths in men in Western countries (Denmeade and Isaacs, 2004). Treatments for hormone-dependent prostate cancer include androgen ablation, prostatectomy and radiotherapy, whereas for the hormone-independent later stages and recurrent cancers, the treatment mainly involves chemotherapy (Pilat et al., 1998; Linton and Hamdy, 2003; Parker and Dearnaley, 2003). The frequency of AR alterations such as AR mutations and AR gene amplification is often increased during the progression of early-stage prostate cancer to the late metastatic disease. High expression levels of AR are also found in many late-stage hormone-refractory prostate cancers (Marcelli et al., 2000; Edwards et al., 2003). Additionally, most forms of prostate cancers rely on AR activity to promote cell proliferation, albeit in the case of hormone-refractory tumours, this is without the necessity of physiological levels of ligand (Litvinov et al., 2003; Chen et al., 2004). Consequently, a better understanding of how the AR is regulated at the molecular level could provide the basis for improving existing therapies.

Most drugs used in chemotherapeutic regimens are genotoxic and exert their function through interaction directly with DNA or with DNA-binding proteins, causing DNA lesions (Espinosa et al., 2003). Cellular responses to genotoxic stress are mediated through signalling pathways, where the registration of DNA damage by sensor proteins such as ATM and ATR is passed on to transducer proteins including checkpoint kinases and the tumour suppressor p53. The transducer proteins pass these signals on to effector proteins, resulting in cell cycle arrest and the initiation of DNA repair or, if the damage is too severe, the induction of apoptosis (Yang et al., 2003; Kastan and Bartek, 2004). Cancer cells often have defects in DNA damage responses or DNA repair pathways and are therefore more susceptible to death from treatment with genotoxic agents than normal cells. Although many cellular responses to DNA damage have been clarified over the past years, the effects of genotoxic stress and DNA damage on AR function remains to be elucidated.

In this study, we investigated the effects of genotoxic stress caused by two chemotherapeutic agents on endogenous AR function in the prostate cancer cell line LNCaP. We have chosen to study the effects of etoposide and cisplatin, as both have been in clinical trials for prostate cancer treatment (Sciuto et al., 2002; Steineck et al., 2002; Hoshi et al., 2003; Vaishampayan et al., 2004). Etoposide exerts its function through blocking topoisomerase II in the religation step, which results in single- and double-stranded DNA breaks (Hande, 1998). Cisplatin, on the other hand, is a DNA cross-linking agent generating inter- and intra-strand adducts that are difficult to repair, cause replication fork obstruction and ultimately DNA strand breaks (Jordan and Carmo-Fonseca, 2000). Here, we show that AR activity is inhibited in response to the genotoxic stress induced by these chemotherapeutic agents. This loss of AR activity is not caused by a change in cell cycle distribution, a change in subcellular localisation of the AR nor induction of apoptosis, as none of these effects occur within the time frame in which AR activity is altered. Furthermore, studies with short hairpin RNA (shRNA) suggest that this loss of AR activity is independent of p53 function. Here, we provide evidence that a decreased recruitment of AR to AREs is a mechanism by which genotoxic stress leads to the inhibition of AR activity.

Results

Genotoxic stress-induced inhibition of AR activity

To ensure that we compared the effects of the different chemotherapeutic agents at equitoxic concentrations, we determined the IC₅₀ value for the growth inhibition of the androgen-sensitive and AR-expressing prostate cancer cell line LNCaP by the two drugs etoposide and cisplatin. This was carried out in a 96 h growth assay with continuous exposure to the drugs and the IC₅₀ values were 0.9±0.03 μ M for etoposide and 16±0.35 μ M for cisplatin (Figure 1a). To measure endogenous AR activity, LNCaP cells were transfected with a luciferase reporter construct containing either the prostate-specific antigen (PSA) promoter (PSA61luc) or the Probasin promoter (pGL3PB), both of which contain AREs and are known to be regulated by AR (Rennie et al., 1993; Cleutjens et al., 1997). A titration of the two drugs showed that after a 16 h treatment, AR activity was inhibited in a dose-dependent manner in response to each of these agents (Figure 1b–e). When AR activity was measured in PC3-wtAR cells, which is the prostate cancer cell line PC3 stably transfected with wild-type AR, we found that it was likewise inhibited in response to genotoxic stress induced by etoposide (Figure 1f). The antiandrogen bicalutamide (Casodex) was included in all the experiments as a positive control for the inhibition of AR activity. Cisplatin turned out to be a more potent inhibitor of AR activity than etoposide in relation to the IC₅₀ value.

Figure 1

Genotoxic agents inhibit AR activity in LNCaP and PC3-wtAR cells. (a) Table of IC₅₀ values for two chemotherapeutic agents for LNCaP cells measured in a 96 h growth assay using the colorimetric dye AlamarBlue. Values are means of eight determinations repeated three times. (b–e) LNCaP cells were grown in hormone-depleted serum for 48 h before transfection with (b, c) PSA61luc or (d, e) pGL3PB. Twenty-four hours post-transfection, the cells were treated with different multiplications of the IC₅₀ value for (b, d) etoposide or (c, e) cisplatin. Thirty minutes after drug addition, cells were stimulated with DHT (10 nM) or DHT and bicalutamide (10 μ M) for 16 h, after which cells were harvested and luciferase activity measured. (f) PC3-wtAR cells were likewise hormone deprived for 48 h before transfection with PSA61luc, which 24 h later was followed by the addition of etoposide (90 μ M) 30 min before DHT (10 nM) stimulation for 16 h. Bicalutamide (10 μ M) was added at the same time as DHT. In all experiments, a β-gal expression vector was co-transfected with the luciferase reporter and used to normalize the luciferase activity against. Relative luciferase activities were expressed as the per cent of DHT control. Bical: Bicalutamide. Experiments were carried out in triplicate and repeated twice. Error bars denote standard error.

Inhibition of AR activity is p53 independent

Exposure of cells to stress such as DNA damage leads to the activation of the tumour suppressor p53, which involves an increase in the protein half-life leading to an accumulation of p53 protein in the cell (Meek, 2004). Moreover, p53 has been shown to interact with the AR and to be involved in negative regulation of AR activity (Shenk et al., 2001; Cronauer et al., 2004). We therefore examined the involvement of p53 signalling in the inhibition of AR activity in response to drug treatment in the p53^+/+ LNCaP cell line. By transfecting cells with a construct expressing an shRNA against p53, it was possible to knock down (KD) the expression of p53 compared to control-transfected cells (Figure 2a). The activation of p53 seen in control cells in response to etoposide was substantially reduced in p53 KD cells (Figure 2a). Likewise, the expression of p21, a downstream target of p53, was greatly reduced in p53 KD cells compared to control cells. Importantly, the loss of p53 did not affect AR expression. The overall AR activity was found to be higher in p53 KD cells than in control cells; however, both the p53 KD and the control cells showed the inhibition of AR activity to the same extent in response to drug treatment (Figure 2b). This result suggests that a p53-independent mechanism is involved in the inhibition of AR activity in response to genotoxic stress.

Figure 2

LNCaP cells were electroporated with pSUPER-p53 or the control vector pSUPER-NS and 3 days later transfected with reporter constructs by means of Effectene transfection reagent (see Materials and methods). Cells were stimulated with DHT (10 nM) alone or in combination with etoposide (90 μ M) or cisplatin (160 μ M). Both etoposide and cisplatin was added 30 min before DHT stimulation began. (a) Western blot analysis of p53 expression in pSUPER-NS- (control) and pSUPER-p53-transfected cells (p53 KD) treated with etoposide and DHT for the indicated time. The Western blot is representative of two independent experiments. (b) AR activity was measured in a luciferase assay 8 h after stimulation with DHT in the presence or absence of etoposide and cisplatin treatment (as described above). Tables contain the relative luciferase activities calculated as the per cent of DHT controls. The luciferase assay was carried out in triplicate and repeated twice. Error bars denote standard error.

Cell cycle distribution

AR activity has previously been shown to be considerably reduced at the G1/S boundary in synchronised mouse cells compared to G0 and S phase (Martinez and Danielsen, 2002). We therefore investigated if LNCaP cells underwent an alteration in cell cycle distribution in response to the chemotherapeutic agents in a manner that could explain the loss of AR activity. The cell cycle distribution was analysed by flow cytometry after 16 h of exposure to etoposide or cisplatin. Etoposide and cisplatin caused a slight accumulation of cells in G2 and S phase, respectively, an outcome that was not affected notably by the presence of DHT (Figure 3a and b). As we did not find a significant accumulation of cells at the G1/S boundary, the loss of AR activity seen in response to the chemotherapeutic agents cannot be explained by such a change in cell cycle distribution.

Figure 3

(a) Cell cycle distribution of LNCaP in response to etoposide or cisplatin treatment. Cells grown for 3 days in hormone-depleted serum (CSS) were stimulated with etoposide or cisplatin alone or in combination with DHT, and harvested 16 h later for FACS analysis. FCS: fetal calf serum; CSS: charcoal-stripped FCS; DHT:dihydrotestosterone (10 nM); E: etoposide (90 μ M); C: cisplatin (160 μ M). The cell cycle distributions are represented in a diagram and are representative of two independent experiments. (b) Diagram of cell cycle profiles in (a). (c) LNCaP cells were subjected to cytoplasmic and nuclear extraction. Cells were hormone depleted before treatment with etoposide (90 μ M) or cisplatin (160 μ M) alone or 30 min before a 2 h DHT (10 nM) stimulation. PARP and β-Tubulin are nuclear and cytoplasmic proteins, respectively, and serve here as controls for segregation of the nuclear and cytoplasmic proteins. The Western blots are representative of two independent experiments. C: cytoplasmic fraction; N: nuclear fraction.

Subcellular localisation of the AR in response to genotoxic stress

In the absence of DHT, the AR is sequestered in the cytoplasm, and upon activation, it undergoes several modifications including conformational change, phosphorylation and translocation to the nucleus (Feldman and Feldman, 2001). To determine if genotoxic stress has any influence on the subcellular localisation in response to hormone stimulation, we performed a cytoplasmic and nuclear extraction of cells treated with etoposide or cisplatin in the presence or absence of a 2 h DHT stimulation (Figure 3c). This experiment showed that the AR is translocated to the nucleus in response to DHT stimulation. Neither etoposide nor cisplatin treatment prevents or accelerates this nuclear translocation of the AR (Figure 3c). A similar experiment was performed where the cells were DHT stimulated for 8 h, and again, no changes in the distribution of nuclear and cytoplasmic AR was observed in response to genotoxic stress (data not shown). Hence, loss of AR activity in response to etoposide and cisplatin cannot be explained by an inhibition of the nuclear translocation of the AR.

Genotoxic stress affects the hormone binding of the AR

Heat-shock protein 90 (Hsp90) is an abundant molecular chaperone known to regulate the stability and function of a wide spectrum of proteins including steroid receptors (Whitesell and Lindquist, 2005). In the absence of a ligand, the AR is associated with Hsp90 in the cytoplasm and kept in a high-affinity hormone-binding conformation necessary for ligand binding. As a consequence, inhibition of the chaperone function of Hsp90 impairs the AR ligand-binding ability (Fang et al., 1996; Georget et al., 2002) With the aim of elucidating the mechanism behind the genotoxic stress-induced loss of AR activity, the ligand-binding ability of the AR in the presence of etoposide and cisplatin was investigated. Hormone-deprived LNCaP cells were stimulated with radiolabeled DHT for 1 h in the presence or absence of etoposide and cisplatin. As a positive control for the inhibition of AR ligand binding, the Hsp90 inhibitor 17AAG was included and used at 250 nM, which is 10 times the IC₅₀ value for growth inhibition in LNCaP cells (Solit et al., 2002). With the ligand-binding assay, we demonstrated that 17AAG inhibited the AR ligand binding as expected (Figure 4). Surprisingly, etoposide turned out to inhibit AR ligand binding to a similar extent as 17AAG, whereas cisplatin had little effect. As the nuclear translocation of the AR is not inhibited by etoposide, these results suggest that one of the cellular effects of etoposide treatment is to cause a reduced ability of the AR to remain associated with its ligand, whereas cisplatin does not cause such an effect.

Figure 4

Etoposide is a more efficient inhibitor of AR ligand binding than cisplatin. Hormone-deprived LNCaP cells were stimulated with [³H]-DHT for 1 h in the presence or absence of 17AAG (250 nM), etoposide (90 μ M) or cisplatin (160 μ M). The radiolabeled DHT was added to the cells 30 min after 17AAG, etoposide or cisplatin. The diagram represent the mean value of three independent experiments each performed in triplicate. Error bars denote standard error.

AR expression in response to genotoxic stress

We next investigated whether the loss of AR activity was a consequence of etoposide- or cisplatin-induced apoptosis. A marker for the induction of apoptosis is the cleavage of the nuclear enzyme poly(ADP-ribose) polymerase (PARP), which is an early event in apoptosis resulting from increased caspase activity (Cohen, 1997). Apoptosis was not induced in cells treated with 90 μ M etoposide (100 times the IC₅₀ value) for up to 24 h (Figure 5b). However, cells treated with 160 μ M cisplatin (10 times the IC₅₀ value) began to show PARP cleavage around 16–24 h of exposure (Figure 5c). Control cells induced with DHT alone did not show any PARP cleavage (Figure 5a). LNCaP cells treated with okadaic acid underwent apoptosis and were included as a positive control for PARP cleavage. The accumulation of p53 as a result of its activation after etoposide and cisplatin treatment confirmed that both drugs were able to induce a DNA damage response. This experiment also showed that etoposide treatment had no effect on AR expression in contrast to cisplatin, which caused a slight reduction in AR expression after 24 h of treatment (Figure 5b and c). The initial PARP cleavage induced by cisplatin is seen after 16–24 h (Figure 5c), and this only marginally overlaps with the 16 h time frame in which we have observed a loss of AR activity (Figure 1c and e). It is therefore unlikely that induction of apoptosis or the loss of AR protein expression is the main mechanism by which AR activity is inhibited in response to etoposide and cisplatin treatment.

Figure 5

Western blot analysis of hormone-depleted LNCaP cells stimulated with (a) DHT (10 nM), (b) DHT+etoposide (90 μ M) or (c) DHT+cisplatin (160 μ M) for the indicated periods of time. Cells were stimulated with DHT 30 min after drug addition and harvested at the given time points after DHT addition. Cells treated with okadaic acid (50 nM) were included as a positive control for PARP cleavage. α-Tubulin serves as a loading control. Tables contain the ratio between quantified AR and α-Tubulin blots. The Western blots shown are representative of three independent experiments.

Ser81 phosphorylation of the AR

The AR is a 112 kDa protein and is reported to contain at least 10 different phosphorylation sites (Lin et al., 2001; Gioeli et al., 2002; Wong et al., 2004). Although the knowledge of the functional significance of these sites is currently limited, it is well established that phosphorylation of serine 81 (Ser81) is hormone dependent (Gioeli et al., 2002; Black et al., 2004). Accordingly, we observed an increase in phosphorylation of Ser81 when hormone-depleted cells were stimulated with DHT (Figure 6). Interestingly, in the presence of cisplatin, a reduction in Ser81 phosphorylation was observed after 8 h of exposure (Figure 6b), whereas etoposide did not cause such an effect (Figure 6a). This suggests that the loss of phosphorylation on this residue is a cisplatin-induced response.

Figure 6

Phosphorylation of the AR at Ser81. LNCaP cells were hormone depleted (CSS) before stimulation with DHT (10 nM) alone or DHT in combination with either (a) etoposide (90 μ M) or (b) cisplatin (160 μ M). DHT was added to the cells 30 min after the drugs and cells were harvested at the given time points thereafter. The p53 blot serves as a control for an induced DNA damage response and β-Tubulin as a loading control. Tables contain the ratio between quantified phospho-Ser81 AR (P-81) and total AR blots. The Western blots shown are representative of two independent experiments.

Recruitment of AR to AREIII in response to etoposide and cisplatin

Our localisation studies suggest that a nuclear mechanism is involved in the inhibition of AR function in response to genotoxic stress. Consequently, we studied the AR function by its ability to bind AREs in the presence of the etoposide and cisplatin. PSA is an AR-regulated gene with a promoter containing two AREs (AREI and AREII) and an enhancer containing one ARE (AREIII) (Figure 7a). Of the three AREs in the PSA promoter/enhancer, the AREI has been reported to be the most active followed by AREIII (Cleutjens et al., 1997). By use of chromatin immunoprecipitation (ChIP) assays, we have investigated the pattern of AR binding to the PSA enhancer in response to etoposide and cisplatin treatment. Our studies showed a recruitment of AR to both the AREI and AREIII in response to DHT stimulation (Figure 7b). This recruitment was specific to AREs as AR was not bound to DNA regions, which lack AREs, such as the middle part of the PSA promoter and the Hsp70 promoter (Kang et al., 2002). DHT-induced recruitment of AR to the AREIII was more pronounced than that to the AREI. This observation is in accordance with previous reports showing that more AR is recruited to the AREIII than to the AREI in response to androgen stimulation (Louie et al., 2003; Kang et al., 2004). For this reason, we decided to focus on AR recruitment to AREIII. We found that cells treated with either etoposide or cisplatin showed a marked reduction in the amount of AR recruited to the AREIII compared to cells stimulated with DHT alone (Figure 7c and d). This was a rapid response seen within the first 30–60 min. The semiquantitative PCR results were backed up by quantitative PCR (Figure 7c and d, bottom). It is possible that DNA strand breaks or cross-links would cause a reduced AR recruitment to the AREIII region. To test the effects of DNA strand breaks, ChIP assay input samples were digested with the restriction enzyme BlpI, which cuts specifically in the AREIII region but not in the AREI region of the PSA promoter. This resulted in a reduced PCR amplification of the AREIII region but did not affect the AREI region. To induce DNA cross-linking, input samples were incubated in vitro with cisplatin. This also led to a decrease in the amount of DNA amplified by PCR (data not shown). Thus, DNA strand breaks and cross-linking can cause a decrease in the PCR amplification; however, all input fractions in the ChIP assays were amplified to comparable levels, which suggests that neither of these were present in the AREIII region in these samples. Therefore, the loss of AR recruitment in response to etoposide or cisplatin treatment is not likely to be the result of strand breaks or cross-linking in the AREIII region (Figure 7c and d). Together, these results imply that a rapid genotoxic stress-induced mechanism different from physical DNA damage to the AREIII is involved in reducing recruitment of ligand-activated AR to the PSA enhancer.

Figure 7

Less AR is recruited to the AREIII in the PSA enhancer in response to etoposide and cisplatin treatment. (a) Schematic representation of the PSA promoter/enhancer. Arrows indicate primers used for the amplification of ChIP samples. (b) ChIP assay showing that AR is recruited to AREI and AREIII in response to DHT treatment (10 nM DHT for 1 h). (c, d) Recruitment of AR to the AREIII in the PSA enhancer in response to DHT treatment is inhibited when cells are treated with a combination of DHT (10 nM) and (c) etoposide (90 μ M) or (d) cisplatin (160 μ M). The ChIP assay was analysed by conventional PCR (top) and real-time PCR (bottom). The data shown are representative of three independent experiments.

Discussion

Deregulation of the AR function leading to different hard wiring of AR signalling pathways, as well as overexpression of the AR, have emerged as major mechanisms in the development of hormone-refractory prostate cancer (Litvinov et al., 2003). Therefore, a better understanding of how the AR activity is regulated will have implications for the future improvement of prostate cancer treatment. Here, we report that AR activity is inhibited in response to genotoxic stress induced by two chemotherapeutic agents. This cellular effect was seen with two different classes of drugs, the topoisomerase II inhibitor etoposide and the DNA cross-linking agent cisplatin. Both drugs inhibited AR activity in a dose-dependent manner and the DNA cross-linking agent cisplatin was the most potent inhibitor of AR activity in relation to its IC₅₀ for cell growth. In our studies, we used two different AR-regulated luciferase reporter constructs containing either the PSA promoter/enhancer region or the Probasin promoter (Rennie et al., 1993; Cleutjens et al., 1997). AR activity measured as a read out from each of the reporters was reduced in response to genotoxic stress, suggesting that the effect seen was not promoter specific but rather a general effect on the AR. We have chosen to focus our studies on endogenously expressed AR, and owing to the shortage of prostate cancer cell lines expressing a functional androgen-responsive AR endogenously, we have performed our studies in LNCaP cells. While LNCaP cells remain androgen sensitive, they do contain a point mutation (T877A) in the ligand-binding domain of the AR. This renders the receptor promiscuous towards other ligands such as estrogens, progestagens and some antiandrogens (Veldscholte et al., 1990; Sack et al., 2001). One issue was whether the loss of AR activity in response to genotoxic agents is T877A mutation dependent, even though this mutation is only reported to alter the profile of ligands that bind to it. We found that the activity of wild-type AR in PC3-wtAR cells was also inhibited in response to etoposide (Figure 1f). This implies that the loss of AR activity in response to genotoxic stress observed in LNCaP cells cannot be explained by the T877A mutation of the AR. Given that all our experiments have been performed in hormone-deprived cells stimulated with DHT, the promiscuity of the AR in LNCaP was accordingly evaluated to have a minimal influence on the results.

The p53 tumour suppressor is a central component of the signalling pathways leading to various genotoxic stress and DNA damage responses, and we therefore investigated if p53 signalling was involved in the inhibition of AR activity in response to etoposide and cisplatin treatment. Our results suggest that genotoxic stress-induced inhibition of AR activity occurs through a p53-independent mechanism (Figure 2). This is in accordance with our observation that AR activity is inhibited in response to etoposide induced DNA damage in the p53^−/− PC3-wtAR cell line (Figure 1f). Moreover, we noted that in the absence of genotoxic stress, AR activity was increased in p53 KD cells compared to control cells (Figure 2b). This is in contrast to a recent report where inhibition of p53 function in LNCaP cells by means of the p53 inhibitor, Pifithrin-1α, resulted in a decrease in AR activity (Cronauer et al., 2004). The reason for this discrepancy is not clear, but a possible explanation could be the different techniques used to inhibit p53 function. Despite being a p53 inhibitor, Pifithrin-1α also has toxic effects unrelated to p53 including inhibition of firefly luciferase activity, which was the reporter system used by Cronauer and colleagues (Komarova et al., 2003; Rocha et al., 2003).

Expression and activity of many proteins vary according to cell cycle phase, and recently, the AR was proposed to be among these proteins (Martinez and Danielsen, 2002). AR activity was shown to be reduced considerably at the G1/S boundary in synchronized L292 cells compared to the activity in G0 and S phases of the cell cycle (Martinez and Danielsen, 2002). In our experiments, however, loss of AR activity in response to genotoxic stress could not be linked to an accumulation of cells at the G1/S boundary, as we only saw a minimal cell cycle effect after 16 h of drug exposure and hormone stimulation (Figure 3a and b). This was anticipated firstly because the doubling time of the LNCaP cells used in our laboratory is 2.5 days and secondly because we used asynchronously growing cells. We therefore concluded that the cell cycle effect we observed was unlikely to be sufficient to cause the inhibition of AR activity observed.

A change in the nuclear translocation of the AR was not observed following DHT stimulation and treatment with genotoxic agents. This suggests that the mechanism involved in the inhibition of the AR is functioning within the nucleus. The partial nuclear translocation of the AR we observed is in contrast to a complete nuclear translocation previously reported by immunofluorescence studies in LNCaP cells (Tyagi et al., 2000). The reason for this difference is not clear but is most likely to be explained by the use of different techniques.

The two classes of chemotherapeutic agents studied here have different mechanisms of action in terms of generating genotoxic stress and DNA damage, and our results suggest that at least two of the mechanisms through which they act on the AR may also be distinct. One possible mechanism was observed in the ligand-binding assay, where etoposide turned out to be a potent inhibitor of AR ligand binding in contrast to cisplatin that only slightly decreased the ligand binding (Figure 4). According to our cellular fractionation studies, the fraction of AR translocated to the nucleus in the presence of etoposide is of similar magnitude to what is seen in its absence (Figure 3c). This could suggest that the initial ligand binding of the AR in response to DHT stimulation is not prevented. However, in conjunction with the etoposide-induced reduction of DHT bound in the ligand-binding assay, it can be speculated that the ligand binding of the AR is more unstable in the presence of etoposide, which could lead to a dissociation of the ligand from the AR after the nuclear translocation. It has been reported that AR is translocated to the nucleus within an hour of hormone stimulation, but when the ligand is washed off the cells, it takes 12 h for the AR to be relocalised to the cytoplasm (Tyagi et al., 2000). It is therefore possible that in the presence of etoposide, an equilibrium of ligand-bound and -unbound AR is found in the nucleus. It is, however, important to bear in mind that a ligand-binding assay only provides a measure of the amount of DHT bound in the cells rather than a direct measure of ligand bound to the AR. This is even despite the fact that nonspecific binding is subtracted from the total binding measured. It can therefore not be precluded that the decrease in DHT measured in the cells in response to etoposide is caused by reduced amount of DHT bound to subcellular compartments or other proteins rather than to AR itself.

The second mechanism by which etoposide and cisplatin act differently on the AR was observed when cisplatin but not etoposide caused a loss of Ser81 phosphorylation after 8 h of drug exposure (Figure 6). Although AR phosphorylation at residue Ser81 is a DHT-stimulated event (Gioeli et al., 2002; Black et al., 2004), the cellular function of this modification is not known. As the half-life of the AR is increased in response to androgens (Gregory et al., 2001), it is possible that phosphorylation of Ser81 is involved in this stabilization of the AR. Although still a hypothesis, this could mean that loss of Ser81 phosphorylation in response to cisplatin treatment is involved in destabilization of the AR leading to its degradation at later time points. It should be noted that our cell fractionation experiment (Figure 3c) did not reveal a subcellular redistribution of Ser81 phosphorylated AR at early time points in response to genotoxic stress (data not shown). The activity of exogenously expressed AR and glucocorticoid receptor (GR) in neuroblastoma cells has been demonstrated to be inhibited in response to cisplatin, and for the GR, this was linked to the cisplatin-induced inhibition of the molecular chaperone Hsp90 (Rosenhagen et al., 2003). As the AR is an Hsp90 client protein (Georget et al., 2002), it is possible that loss of AR activity and loss of protein expression at later time points, in response to cisplatin, could be related to a cisplatin-induced inhibition of Hsp90 in LNCaP cells.

The ligand-binding ability of exogenously expressed GR was shown to be significantly inhibited by cisplatin, which led to the hypothesis that cisplatin-induced inhibition of Hsp90 only affects a fraction of Hsp90 client proteins, for example, steroid receptors (Rosenhagen et al., 2003). Our ligand-binding assay performed in LNCaP cells with endogenous expression of AR showed that cisplatin only caused a slight reduction in the ligand-binding activity of this receptor. This suggests that a cisplatin-induced inhibition of Hsp90 may not cause a general inhibition of the ligand-binding ability of steroid receptors, but may instead cause a specific response with some steroid receptors, which in this case does not include the AR.

As our results suggest a p53-independent mechanism, localised in the nucleus, to be involved in the inhibition of AR activity, we subsequently studied AR function by ChIP assays, as this allows the analysis of AR function within much shorter time frames than the ones studied with luciferase reporter assays. We demonstrated that upon DHT stimulation less AR is recruited to the AREIII of the PSA enhancer in response to both cisplatin and etoposide. Surprisingly, this was a rapid response detectable 30–60 min after DHT stimulation of drug-treated cells (Figure 7). From these results, it seems reasonable to suggest that reduction in AR recruitment to the AREs is one mechanism by which AR activity is inhibited in response to genotoxic stress. The reduced recruitment of AR to the AREIII upon drug treatment is very rapid and does, for etoposide, correlate with the loss of ligand binding observed after 1 h. In contrast, the cisplatin-induced effects seen on Ser81 phosphorylation and total AR levels are late events compared to the loss of AR recruitment to the AREIII. The reason for this is not known, but it could be speculated that the initial fast response may result from the activation of a signalling pathway, which interrupts AR function by blocking or preventing its recruitment to AREs. Later, cisplatin possibly acts by affecting the AR phosphorylation and stability. However, this hypothesis remains to be tested experimentally.

Taken together, our results suggest that the inhibition of AR activity in response to genotoxic stress caused by etoposide or cisplatin treatment is mediated by multiple early and later mechanisms. In the context of the PSA enhancer, an early pathway common to both is the reduced recruitment of AR to the AREIII. Upstream of this, the actions of etoposide and cisplatin appear to be distinct as etoposide inhibits a sustained ligand binding, although the mechanism(s) by which cisplatin prevents AR recruitment to the AREs remain to be elucidated. Moreover, the data presented here suggest that DNA cross-linking agents are potentially more efficient inhibitors of AR activity than topoisomerase II inhibitors. This knowledge combined with an elucidation of the signalling pathway(s) involved in the loss of AR activity could prove valuable for future prostate cancer drug discovery.

Materials and methods

Cell culture and drugs

Human prostate carcinoma LNCaP cells (purchased from ATCC) were maintained in RPMI 1640 medium containing antibiotics supplemented with 10% fetal calf serum (FCS) in an incubator at 37°C with a humidified atmosphere containing 5% CO₂. PC3-wtAR cells (kindly provided by Professor Andrew Cato, Institute of Toxicology and Genetics, Karlsruhe, Germany) were cultured in Dulbecco's modified Eagle's medium supplemented with 10% FCS and G418 (400 μg/ml). For experiments, the cells were grown in phenol red-free medium containing antibiotics supplemented with 10% charcoal-stripped FCS (CSS) for 3 days, unless stated otherwise. Hereafter, the cells were stimulated with DHT for the times indicated. DHT was dissolved in 100% ethanol and then diluted in phenol red-free RPMI 1640 with 10% CSS, so that a final concentration of 10 nM DHT in 0.00001% ethanol was added to the cells. Etoposide (Sigma), okadaic acid (Calbiochem) and bicalutamide (Casodex, AstraZeneca) were dissolved in dimethyl sulfoxide (DMSO) and cisplatin (Sigma) directly into medium.

96 h growth assay

LNCaP cells were seeded into 96-well plates in RPMI 1640 with 10% FCS 48 h before drug addition. Drugs were left on cells for 96 h, after which AlamarBlue (Biosource) was added to a final concentration of 10%. Plates were incubated for 4 h at 37°C, and absorbance was read at 570 and 620 nm in a plate reading spectrophotometer (Victor2). The percentage-reduced dye, a measure for cell viability, was calculated according to the manufacturer's suggestion.

Transfection assays

Reporter constructs (described below) were transfected into LNCaP cells grown in CSS for 2 days by means of Effectene Transfection Reagent (Qiagen). The transfection mix was left on the cells for 24 h. Transfection of LNCaP cells with pSUPER-eCFP-p53 or pSUPER-eCFP-NS (kindly provided by Professor Alan Ashworth, Institute of Cancer Research, UK; Gudmundsdottir et al., 2004) was performed by electroporation. Cells were trypsinized, washed twice in cold phosphate-buffered saline (PBS), diluted in 200 μl PBS and mixed with 15 μg of DNA. The cells were transferred to a 4-mm electroporation cuvette (EquiBio), incubated on ice for 10 min and electroporated twice with a BioRad Gene Pulser apparatus set at 0.25 kW and the capacitance set at 125 μF. Cells were plated and left growing for 2 days in FCS-containing medium and 1 day in stripped serum before being transfected with the reporter constructs PSA61luc and pCMV β-galactosidase (β-gal) using Effectene.

Luciferase assay

The luciferase reporter construct, PSA61luc, contains 6.1 kb of the PSA promoter, and was kindly provided by Dr Trapman (Erasmus University, The Netherlands). The luciferase reporter construct, pGL3PB, contains the first 470 bases of the Probasin promoter. Dr Andy Feber (Institute of Cancer Research, UK) kindly provided the pCI-neo vector Probasin promoter construct, which was subsequently subcloned into the pGL3 basic vector (Promega). As a transfection control, a β-gal construct, pCMV β-gal, was used, which had kindly been provided by Dr Julia Bardos (Institute of Cancer Research, UK). Twenty-four hours after transfection with reporter constructs, LNCaP cells were stimulated with drugs and 10 nM DHT (Sigma) or DHT alone. Chemotherapeutic agents were always added 30 min before DHT. Plates were washed once in PBS, frozen at −80°C, thawed on ice and lysed in Passive Lysis Buffer (Promega) according to the manufacturer's instructions. Twenty microlitre of the cell lysis suspension was aliquoted into a 96-well microtitre plate (Dynex), and the luciferase activity was read on a luminometer (MLX, Microtitre Plate Luminometer, Dynex), automatically injecting 100 μl Luciferase Assay Substrate (Promega) to one well at a time, and reading the luciferase activity over a period of 24 s beginning 2 s after injection.

The β-gal assay was performed by mixing 20 μl of the cell lysis suspension with 1 ml of CPRG (chlorophenol red-β-D-galactopyranoside) mix (2.5 mM CPRG(Roche-Applied Science) and 1.25 mM MgCl₂), incubating overnight at 37°C and reading absorbance at 574 nm in a spectrophotometer. Relative luciferase activities were calculated by dividing the actual luciferase activity measured with their corresponding β-gal values. Relative luciferase activities were in some cases expressed as the per cent of DHT control. Conversion of s.e. to per cent of control was calculated as: (1/y)√(σx²+(x/y)²σy²), where y is the sample set to 100%, x is the sample calculated relative to y, σy the s.e. of y and σx the s.e. of x.

FACS

Cells were harvested for FACS analysis by trypsination, washed once in PBS and fixed by drop-wise addition of ice-cold ethanol while vortexing the cells. Samples were incubated 15 min on ice and stained overnight with propidium iodide (PI) (PBS containing PI (40 μg/ml) and RNase (250 μg/ml)) for 30 min at 37°C and hereafter left overnight at 4°C. Samples were analysed on an Elite ESP Cell Sorter.

Cell fractionation

Nuclear and cytoplasmic fractions from LNCaP cells were isolated according to the manufacturer's instructions using the NE-NER Nuclear and Cytoplasmic Extraction Reagents (Pierce). In addition to the manufacturer's instructions, a supplementary washing step was included after isolation of the cytoplasmic fraction in order to avoid cytoplasmic contamination of the nuclear fraction.

Immunoblotting

Cells were lysed for 30 min in protein extraction buffer (50 mM HEPES, pH 7.4, 250 mM NaCl, 1 mM EDTA, 1% NP-40) containing protease inhibitors (protease inhibitor cocktail tablets; Boehringer/Roche), phosphatase inhibitors (10 mM β-glycerophosphate, 1 mM NaF, 0.1 mM Na₃VO₄) and 1 mM DTT. Protein extracts were resolved on 8 or 10% SDS–PAGE gels and transferred onto an Immobilon-P membrane (Millipore), which was blocked in 5% low-fat milk diluted in TNT (50 mM Tris, pH 8.0, 150 mM NaCl, 0.1% Tween-20). Membranes were incubated with primary antibody overnight and secondary antibody (peroxidase-conjugated goat anti-rabbit/mouse antibody; BioRad) for 1 h. Blots were developed using ECL Western Blotting Detection Reagents and Hyperfilm both from Amersham Biosciences. Primary antibodies were mouse monoclonals to AR (DAKO), p53 (Ab6; Oncogene), PARP (BD Pharmingen), p21 (Upstate), GAPDH (BD Pharmingen), α-Tubulin (Sigma), β-Tubulin (Ab3; NeoMarkers) and rabbit polyclonal to P-Ser81 AR (Upstate) (Black et al., 2004). Blots were quantified using ImageJ software downloaded from http://rsb.info.nih.gov/ij/.

Chromatin immunoprecipitation

LNCaP cells were cross-linked in 1.5% formaldehyde, neutralized in glycine (125 mM) and harvested by scraping into Cell Collection Buffer (100 mM Tris-HCl, pH 9.4, 10 mM DTT and protease inhibitors). Cells were washed once in Nucleus/Chromatin Preparation (NCP) Buffer I (10 mM EDTA, 0.5 mM EGTA, 10 mM HEPES, pH 6.5 and 0.25% Triton X-100), once in NCP Buffer II (1 mM EDTA, 0.5 mM EGTA, 10 mM HEPES, pH 6.5 and 200 mM NaCl) and resuspended in Lysis Buffer (10 mM EDTA, 50 mM Tris-HCl, pH 8.1, 0.5% Empigen BB and 1% SDS). Lysates were sonicated three times for 10 s each time in an MSE Soniprep 150 sonicator and spun for 10 min at 10 000 g. One-tenth of the lysate was removed and kept as the input fraction while the remaining was diluted in IP buffer (2 mM EDTA, 150 mM NaCl, 20 mM Tris-HCl, pH 8.1, 1% Triton X-100 and protease inhibitors) to a final volume of 1 ml. Lysates were precleared for 2 h at 4°C with 2 μg of salmon testis DNA (Sigma), 10 μl of preimmune serum (Sigma) and Protein A–Sepharose (50 μl of 50% beads slurry in 10 mM Tris-HCl, pH 8.1 and 1 mM EDTA). Beads were removed and lysates subjected to immunoprecipitation with an anti-AR (C-19; Santa Cruz) antibody overnight. Immunocomplexes were collected by absorption onto Protein A–Sepharose beads. Beads were washed sequentially in Washing Buffer I (2 mM EDTA, 20 mM Tris-HCl, pH 8.1, 1% SDS, 1% Triton X-100 and 150 mM NaCl), in Washing Buffer II (2 mM EDTA, 20 mM Tris-HCl, pH 8.1, 1% SDS, 1% Triton X-100 and 500 mM NaCl), in Washing Buffer III (10 mM Tris-HCl, pH 8.1, 1 mM EDTA, 250 mM LiCl, 1% sodium deoxycholate and 1% NP-40) and finally three times in Washing Buffer IV (10 mM Tris-HCl, pH 8.1 and 1 mM EDTA). Specific DNA/protein complexes were extracted in Extraction Buffer (1% SDS and 0.1 M NaHCO₃), cross-linking reversed at 65°C overnight and DNA purified with Qiaquick columns (Qiagen). Primer sequences used for PCR amplification: AREI forward: ggtgcatccagggtgatcta, AREI reverse: cccaggagccctataaaacct; AREIII forward: ccactggtgagaaacctgag, AREIII reverse: ctctctcagatccaggcttg; nonspecific ARE on PSA promoter forward: ggtcaggttttggttgagga, nonspecific ARE on PSA promoter reverse: caagcacagtgagggagaca; HSP70 primers are described in Nissen and Yamamoto (2000). PCR products were analysed by agarose gel electrophoresis.

Real-time PCR

Real-time PCR was performed with the DNA Engine Opticon2 System (MJ BioWorks) according to the manufacturer's instructions. ChIP samples were mixed with DyNAmo SYBR Green qPCR kit (Finnzymes) and primers (described above) and amplified using the following conditions: 10 min initial denaturation followed by 35 cycles of 20 s at 94°C, 20 s at 60°C and 20 s at 72°C. Correct PCR products were confirmed by agarose gel electrophoresis and melting curve analysis.

Ligand-binding assay

The hormone binding of the AR was determined by stimulating LNCaP cells grown in 10% CSS for 3 days with 10 nM [³H]-DHT (Amersham). 17AAG (250 nM; Alexis Biochemicals), etoposide (90 μ M) and cisplatin (160 μ M) were added to the cells 30 min before [³H]-DHT stimulation. Nonspecific hormone binding was determined by adding 50 μ M of bicalutamide to parallel samples treated with 10 nM [³H]-DHT in the presence or absence of drugs. After 1 h of [³H]-DHT treatment, cells were harvested by trypsination, washed three times in cold PBS and resuspended in 100 μl protein extraction buffer (see Immunoblotting). Samples were sonicated in a water bath (Transsonic T310; Camlab) for 1 min, after which 95 μl were mixed with 3 ml of scintillation fluid (Ultima Gold, Packard BioSciences) and counted. The remaining sample was used for protein determination by Bradford assay according to the manufacturer's instructions (BioRad). For each sample, the scintillation counts were normalized to the protein concentration.