Structural and functional association of androgen receptor with telomeres in prostate cancer cells

Casodex induces telomere dysfunction via its effect on AR

We previously reported that the specific AR antagonist Casodex, added at a concentration of 100 μM to cells in 10% serum-containing medium (complete), causes a TIF response in AR-positive LNCaP cells [44]. Contrary to the relative effectiveness of 10 μM Casodex in blocking AR activity in steroid-depleted medium [45-47], a higher concentration of Casodex, 80-100 μM, is needed to similarly inhibit AR activity in complete medium (S. Murthy, U. Bai and G.P. Reddy; manuscript in preparation; [47]), likely because of a high concentration of steroids in fetal bovine serum, many of which can bind to the mutated AR in LNCaP cells. Notably, the steady-state serum level of Casodex in prostate cancer patients treated with 150 mg Casodex/day is reported to be approximately 27 μg/ml (90 μM) [48, 49]; this indicates that the concentration of Casodex we use in vitro is pharmacologically relevant in vivo.

Fig. 1A shows dose-dependence of the TIF response to Casodex (IC₅₀ ~40 μM), and Fig. 1B shows that the inhibition of PSA expression by Casodex has a similar dose-response. Notably, the IC₅₀ for Casodex-mediated inhibition of PSA expression is ~40 μM in FBS or CSS (S. Murthy, U. Bai and G.P. Reddy; manuscript in preparation). Thus, the TIF response to Casodex is associated with AR inactivation; this implicates a role for AR in telomere stability, but does not address the question whether AR transcriptional activity is required for this role.

Notably, the TIF response to AR inactivation is not unique to Casodex. Fig. 1C shows that MDV3100, a newly developed AR antagonist with 5-8 fold higher affinity for AR, is indeed more potent than Casodex in inducing a TIF response, as 3 μM MDV3100 had the same effect as 50 μM Casodex, and 10 μM MDV3100 had the same effect as 100 μM Casodex.

Since the only known target of Casodex is the AR, we interpreted the failure of Casodex to induce a TIF response in AR-negative PC3 prostate cancer cells as due to the absence of AR [44]. However, since the p53 status also differs between LNCaP (wild-type p53) vs. PC3 cells (inactive mutant p53) [50], we sought to rule out the possibility that the TIF response is mediated by or requires wild-type p53. Therefore, p53 was inactivated in LNCaP cells by expressing a dominant-negative p53 element (GSE-22, described by [51]). Whereas etoposide treatment of control vector-transfected LNCaP cells induced the expression of p21^Cip1 (a p53 target gene) (Fig. 2B, upper panel), GSE-22 prevented this effect of etoposide (Fig. 2B, lower panel), thereby demonstrating effective inactivation of p53. Notably, inactivation of p53 did not prevent the TIF response to Casodex (Fig. 2A); therefore, the TIF response to Casodex in LNCaP cells is independent of and not mediated by p53.

Since the LNCaP cell AR is mutated [52], we tested the effect of Casodex on LAPC4 prostate cancer cells, which have wild-type AR [53]. As shown in Fig. 2C, we observed a dramatic increase in the number of immunofluorescent γH2AX foci in Casodex-treated LAPC4 cells as compared to controls. These γH2AX foci in Casodex-treated cells colocalized with TIN2 (Fig. 2C and 2D), indicating the presence of dysfunctional telomeres. The TIF response of LAPC4 cells to Casodex treatment was similar to that of LNCaP cells (Fig. 2E). Therefore, Casodex induces a DNA damage response in prostate cancer cells with wild-type AR (LAPC4 cells) or mutant AR (LNCaP cells).

Androgen is required for telomere stability

Although Casodex and MDV3100 are known as specific AR antagonists [54-57], and the TIF response appears to be due to AR inactivation (Fig. 1B), we tested the effect of androgen depletion as an alternative approach to inactivate AR. Since Casodex binding to AR causes nuclear accumulation of AR and binding to target genes, though without activating them [47], we considered the possibility that the TIF response to Casodex might be ligand-independent, in which case androgen depletion might not cause a TIF response. As shown in Fig. 3, switching cells from complete medium with serum to medium with charcoal-stripped serum caused a TIF response, and this effect was abrogated by the addition of R1881, a synthetic androgen. Therefore, androgen is required for telomere stability. The TIF response to Casodex also was abrogated by the addition of R1881 (Fig. 3), confirming that the effect of Casodex was due to AR antagonism and not a nonspecific toxic effect.

AR is required for telomere stability

In order to provide direct evidence for a role of AR in telomere stability, we used RNA interference with AR-siRNA to inhibit AR production in LNCaP cells. Transfection of cells with AR-siRNA reduced the AR protein level by ~90%, compared to control transfection with scrambled (SC)-siRNA (Fig. 3B), and the decrease in AR was associated with a concomitant decrease in AR-dependent PSA protein and mRNA (Fig. 3B). As illustrated in Fig. 3C, AR-siRNA treatment caused a dramatic increase in the number of immunofluorescent γH2AX or 53BP1 foci, compared to SC-siRNA controls; this indicates that AR knockdown caused a DNA damage response. The colocalization of 53BP1 foci and γH2AX foci (Fig. 3C) indicates that γH2AX and 53BP1 represent the same DNA damage foci. The colocalization of γH2AX with TIN2 in AR-siRNA-treated cells (Figs. 3D and E) indicates that the DNA damage response is at telomeres. Quantitation of this response (Fig. 3F) indicates that AR knockdown with AR-siRNA caused a dramatic TIF response, compared to control treatment with SC-siRNA.

AR transcriptional activity is not required for telomere stability

Since the TIF response to Casodex is associated with inhibition of AR transcriptional activity (Fig. 1), we considered the possibility that the TIF response was due to the inhibition of AR transcriptional activity. Therefore, we treated cells with the general transcription inhibitor actinomycin D or the translation inhibitor cycloheximide in order to determine the role of de novo biosynthesis in telomere stability (Fig. 4). Actinomycin D decreased the mRNA level (Fig. 4A), and cycloheximide decreased the protein level (Fig. 4B), of AR target genes PSA and NKX3.1, but neither actinomycin D nor cycloheximide caused a TIF response (Fig. 4C). Thus, inhibition of de novo biosynthesis of mRNA and protein was not sufficient to induce a TIF response.

However, these data do not rule out the possibility that the TIF response to Casodex is due to inhibition of AR transcriptional activity, as AR transcriptional activity increases the expression of some genes, and decreases the expression of other genes [58, 59]; therefore, inhibition of AR transcriptional activity by Casodex would be expected to increase the expression of genes down-regulated by AR. Therefore, in order to determine whether the TIF response to Casodex is mediated by up-regulation of such genes, we co-treated cells with Casodex and actinomycin D to inhibit the expression of genes potentially up-regulated by Casodex, or with Casodex and cycloheximide (Fig. 4C). Notably, neither actinomycin D nor cycloheximide blocked the TIF response to Casodex (Fig. 4C); thus, the TIF response to Casodex is not mediated by an effect on AR transcriptional activity, and the role of AR in telomere stability does not require AR transcriptional activity.

AR is associated with telomeric chromatin

Co-localization of AR and TIN2 (a component of shelterin in telomeres) in untreated LNCaP cells, and co-immunoprecipitation of AR with TRF1 and TRF2 (telomere DNA binding proteins), indicate the presence of a subset of AR in telomeres [44]. In order to provide further evidence that AR is a component of telomeres, we prepared chromatin from formaldehyde-fixed LNCaP cells, used AR antibody to immunoprecipitate AR-bound chromatin (AR-ChIP), and probed this fraction for the presence of telomere DNA. As a positive control to confirm that AR-ChIP contains AR-bound chromatin, we demonstrated the presence of PSA gene androgen response elements (ARE) (Fig. 5A). Interestingly, Rap1-ChIP and, to a lesser extent, TRF2-ChIP also contained PSA ARE (Fig. 5A); this is not surprising, in light of reports that Rap1 and TRF2, besides being components of shelterin at telomeres, may also be associated with transcribed genes [60-63].

Notably, telomere DNA was detected in AR-ChIP, based on the use of a digoxigenin-labeled 10-mer telomere repeat; by contrast, a mutated DNA repeat did not yield a signal (Fig. 5B). Thus, AR appears to be physically associated with telomeric chromatin in LNCaP cells. As positive controls, we demonstrated the presence of telomere DNA in TRF2-ChIP and Rap1-ChIP, consistent with the presence of these shelterin components in telomeres, whereas the absence of telomere DNA in control IgG-ChIP served as a negative control (Fig. 5B). The telomere DNA signal in AR-ChIP was substantially lower than that in TRF2-ChIP or Rap1-ChIP (Fig. 5B), suggesting AR association with a subset of telomeres.

Having demonstrated the presence of telomere DNA in AR-ChIP, we sought to demonstrate the presence of AR in telomeric chromatin. A novel protocol to isolate telomeric chromatin (referred to as the PICh protocol) [64] has been described recently; it is based on the use of a biotinylated telomere DNA repeat as a probe to hybridize, under stringent conditions (70°C), with endogenous telomere DNA in chromatin fragments of ~600 bp length; streptavidin-labeled beads are then used to pull down the telomere DNA with its associated proteins [64]. The stringent features of this protocol minimize the isolation of non-telomeric chromatin, and a biotin-labeled, scrambled DNA sequence not found in the genome is used as a negative control in this approach [64]. We used this approach to isolate telomeric chromatin and probed it for the presence of AR (Fig. 5C-D). Using 293T cells to optimize the PICh protocol, we demonstrated that the telomere probe pulled down chromatin that contained shelterin components, based on immunoblot analysis with anti-bodies to Rap1, TRF2, and TIN2, whereas the control scrambled probe did not (Fig. 5C). Thus, it appears that the telomere probe can be used to enrich for telomeres. This fraction also contained histone 3 (Fig. 5C), as telomeric chromatin contains nucleosomes [64, 65]. We then used this protocol to isolate telomeres from LNCaP cells; a subset of AR in total chromatin was present in the fraction (as were TRF2 and Rap1) that was pulled down with the telomere-specific probe, but not in the fraction pulled down with the scrambled probe (Fig. 5D). This indicates that a subset of AR is associated with telomeric chromatin in LNCaP cells.

Casodex induces the formation of fragile telomeres

Since telomere dysfunction can lead to chromosome structural abnormalities, such as telomere end-to-end fusion [5], we investigated the consequence of telomere dysfunction caused by Casodex. Metaphase chromosome spreads were prepared from untreated and Casodex-treated LNCaP cells, and probed for the presence of telomere aberrations using fluorescence in situ hybridization of a labeled telomere probe (telomere-FISH) (Fig. 6). The telomere-FISH signal at individual chromatid ends is normally represented as a single signal with an intensity that is roughly equal to that of the sister chromatid end; this was the predominant pattern seen in control LNCaP cell metaphase spreads (Fig. 6A, left panel). A strikingly different pattern of telomere-FISH signals was seen in the metaphase spreads of Casodex-treated cells (Fig. 6A, right panel). Several chromosomes in Casodex-treated cells had multiple telomere signals (Fig. 6A right panel, white arrows; higher magnification shown in Fig. 6B, left panel). Of chromosomes with clearly discernible telomere signals, about 14% had multiple signals in Casodex-treated cells, compared to about 3% in control cells (Fig. 6B right panel).

The metaphase spreads of Casodex-treated cells (Fig. 6A right panel) also show some chromosomes with a single telomere signal, instead of two, one for each chromatid end. A single telomere signal at the end of a chromosome may represent the juxtaposition of two chromatid ends or the fusion of two sister chromatid telomeres; it is not possible to distinguish between these possibilities when the single signal is seen on the short arm of the chromosome. Therefore, we did not quantitate single signals at the short arm end of a chromosome. By contrast, a single telomere signal (instead of 2 signals) on the long arm of a chromosome with clearly separated long arms is likely to represent telomere fusion [yellow arrowheads in Fig. 6A and Fig. 6C (left panel)] [8, 10]; therefore, we quantitated the percentage of chromosomes with long arm sister telomere fusion (Fig. 6C). This phenotype was more common in Casodex-treated cells (Casodex, 4.5%; control, 0.8%).

Notably, a similar pattern of multiple telomere signals and sister telomere association in the long arms of telomeres has been found following knockdown of shelterin component TRF1, which leads to inhibition of telomere DNA replication and subsequently, telomere DNA breakage [10, 36]; the pattern of multiple telomere DNA signals is referred to as fragile telomeres [10]. It is known that telomeres are difficult to replicate because their repetitive arrays of guanosine-rich DNA sequences can form G quadruplexes that hamper progression of the DNA replication machinery [10, 66]. BLM, a TRF1-associated protein, and RTel have also been shown to play a role in telomere DNA replication [66], and their knockdown also causes an increase in fragile telomeres. Thus, our data suggest that the role of AR in telomere stability may occur via a role in telomere DNA replication.

The metaphase chromosomes of both Casodex-treated and control cells (Fig. 6A) also show chromatids with a weak telomere signal or no detectable telomere signal. Because the intensity of a telomere signal is proportional to telomere length, a weak or absent telomere signal may represent a shortened telomere, relative to other telomeres. Interestingly, telomere length of individual human chromosomes is heterogeneous [67], and the shortened telomeres of human cancer cells undergo dynamic shortening and lengthening [35]. Notably, the presence of weak or undetectable telomere signals in untreated human cancer cells has not been previously discussed.

Cell culture

LNCaP (ATCC), LAPC4 (a generous gift from Drs. Robert Reiter and Charles Sawyers) or 293T cells (ATCC) were grown in RPMI (Gibco BRL), Iscove (Gibco BRL), or D-MEM medium (Gibco BRL), respectively, each containing 10% fetal bovine serum (FBS), 2.5 mM glutamine, 100 μg/ml streptomycin and 100 U/ml penicillin (complete medium). Exponentially growing cells, in complete medium, were treated with or without 20 μg/ml cycloheximide (Sigma), 0.5 μg/ml actinomycin D (Sigma), 100 μM bicalutamide (Casodex from LKT Laboratories, MN) or 10 μM MDV3100 (Selleckchem, TX). Steroid-depleted medium was phenol red-free RPMI medium (Gibco BRL) containing 10% charcoal-stripped serum (CSS, from Gibco BRL), 2.5 mM glutamine, 100 μg/ml streptomycin and 100 U/ml penicillin.

Measurement of a TIF response

Telomere disruption leads to a DNA damage response, which includes the recruitment of 53BP1 to, and phosphorylation of H2AX at, telomeres; this process can be visualized by the presence of foci, referred to as telomere-dysfunction-induced foci (TIF), that represent colocalization of 53BP1 or γH2AX with shelterin components such as TRF2 or TIN2 [83, 84]. This method of monitoring telomere disruption is referred to as a TIF response [83]. By contrast, DNA double strand breaks at non-telomeric sites in the genome, such as occur following treatment with etoposide, also lead to a DNA damage response that includes recruitment of 53BP1 and phosphorylation of H2AX, however, these foci do not co-localize with telomere-associated proteins [44]. In untreated control cells grown in complete medium (FBS), 53BP1 foci rarely colocalize with TRF2 [44]. 53BP1 foci are counted and data are expressed as the percentage of cells with a specified number of foci/cell. For LNCaP cells, the cutoff is 5, as about 80% of untreated LNCaP cells have ≤5 53BP1 foci/cell [44]; for LAPC4 cells, the cutoff was 10, suggesting that these cells have a higher level of DNA damage, though at non-telomeric sites.

Indirect immunofluorescence

The immunofluorescent staining of cells grown on glass slides was performed as described [85]. Cells were fixed with 4% paraformaldehyde, permeabilized with 0.5 % Triton X-100 and incubated at 4°C overnight with antibodies against TIN2 [85], AR (AR-N20, AR-441, Santa Cruz), 53BP1 (Abcam), γH2AX (i.e., phosphorylated-H2AX) (Upstate), or TRF2 (IMG-124A, Imgenex). After washing, cells were stained with goat-anti-rabbit-FITC and/or goat-anti-mouse-Texas Red (Molecular probes) secondary antibodies. Images of cells were acquired on an LSM-410 confocal microscope (Zeiss).

p53 knockdown

LNCaP cells were infected with a dominant-negative p53 (GSE-22) [51] or control retroviral expression vector (pBabe), and selected with 0.5 μg/ml puromycin for 2-3 d as described in Kim et. al. [9]. Loss of p53 activity was confirmed by the absence of detectable p21 (BD Pharmingen) immunostaining in GSE-22-expressing cells treated with 10 μg/ml etoposide (Sigma) for 1 hr.

AR knockdown

Exponentially growing LNCaP cells (1.0 −2.0 × 10⁵ cells/wellof a six-well plate) were transfected with 200 pmol of AR-siRNA (Sc-29204, Santa Cruz) or scrambled-siRNA (Santa Cruz), using Lipofectamine 2000 (Invitrogen) following the manufacturer's instructions. Cells were processed48 h later for immunofluorescence staining or Western blotting.

Cell extracts and Western blot analysis

Cells were digested with trypsin, washed with PBS and suspended in Buffer A (50 mM Tris-HCl, pH 7.4, 250 mM NaCl, 0.1% Triton X-100, 5 mM EDTA, 50 mM NaF, and 0.1 mM Na₃VO₄) supplemented with protease inhibitor mixture (P-8340, Sigma) as described [86]. Cells were then subjected twice to 30 pulses of sonication with a Branson Sonifier 250 set at output control 2 and duty cycle 20, with intermittent cooling on ice. The sonicated cell extract was cleared by centrifugation in an Eppendorf centrifuge at 12,500 rpm for 10 min. For Western blot analysis, membranes were probed with antibodies against AR (AR-N20, Santa Cruz), TRF2 (IMG-124A, Imgenex), TIN2 [85], TPP1 (A303-069A, Bethyl Laboratories), Rap1 (A300-306A, Bethyl Laboratories), HIS 3 (1791, Abcam), PSA (C-19, Santa Cruz), NKX3.1 (H-50, Santa Cruz), actin (I-19, Santa Cruz) or GAPDH (AB2302, Millipore). Immunoreactive bands were developed using horseradish peroxidase-conjugated secondary antibodies and Super-Signal WestPico chemiluminescent substrate (Pierce), and visualized using X-ray film.

RT-PCR analysis

Total RNA was prepared as described [86]. RNA was reverse transcribed using random hexamers and oligo (dT) primer and Transcriptor Reverse Transcriptase (Roche Applied Science) according to the manufacturer's instructions. PCR of cDNA was carried out using Platinum PCR SuperMix (Invitrogen). PCR primers for PSA were 5’-gcacccggagagctgtgt (forward) and 5-gatcacgcttttgttcctgat (reverse), for NKX3.1 were 5’- gtacctgtcggcccctgaacg (forward) and 5’- gctgttatac acggagaccagg (reverse), and for GAPDH were 5’-gagatccctccaaaatcaagtg (forward) and 5’ ccttccacgatac caaagttgt (reverse). Cycle parameters were 94°C for 2 min, 94°C for 30 sec, 55°C for 30 sec and 68°C for 1 min. PSA was amplified for 25 cycles, NKX3.1 for 30 cycles and GAPDH for 25 cycles.

Chromatin immunoprecipitation (ChIP) analysis

Cells were fixed in 1% formaldehyde in PBS for 60 min at room temperature, scraped, washed with PBS, and lysed in 1% SDS, 50 mM Tris-HCl, pH 8.0, 10 mM EDTA at a density of 10⁷ cells/ml, as described by Loayza and de Lange [87]. The lysate was sonicated using a Branson Sonifier 250 under the following conditions: output control of 5.5 and duty cycle of 50% with six cycles of 20 sec with intermittent cooling on ice. Lysate (0.2 ml) was diluted with 1.2 ml Buffer (0.01% SDS, 1.1% Triton X-100, 1.2 mM EDTA, 16.7 mM Tris-HCl, pH 8.0, and 150 mM NaCl), incubated at 4°C overnight with antibody (5 μg of AR N-20; 5 μg of Rabbit IgG; 5 μg of Goat TRF2; 5 μg of Rabbit Rap1), and then antibody-bound material was precipitated with 30 μl protein-G Sepharose beads (Invitrogen) that had been pre-equilibrated with 30 μg bovine serum albumin (BSA) and 5 μg sheared Escherichia coli DNA for 30 min at 4°C. In order to isolate DNA from the ChIP pellet (ChIP DNA), cross-linking was reversed at 65°C for 4 h, treated with RNAase A and proteinase K at 37 °C, and extracted with 0.5 ml phenol/chloroform/iso- amylalcohol. In order to probe for telomere DNA, equal amounts of ChIP DNA were applied as dots to a nylon membrane (Invitrogen) using a dot-blot apparatus (Bio-Rad). The membrane was then hybridized with a digoxigenin (DIG)-labeled telomere DNA (gggtaa)₁₀ probe at 42°C overnight. DIG labeling and detection were performed using the DIG Oligonucleotide 3-End Labeling and Detection Kit (Roche). After capturing the signals produced by this hybridization, the membrane was stripped by washing twice with 0.2 M NaOH and 0.1% SDS for 30 min at 52°C and then re-hybridized with a DIG-labeled mutant telomeric (gcctaa)₁₀ probe. In order to probe ChIP DNA for AR binding sites in the PSA gene, ChIP DNA was subjected to PCR using primers for PSA ARE III: 5’- cttctagggtgaccagagcag (forward) and 5’- gcaggcatccttg-caagatg (reverse). Cycle parameters were 94°C for 2 min, 94°C for 30 sec, 55°C for 30 sec and 68°C for 1 min for 30 cycles.

PICh protocol to isolate telomeric chromatin

The PICh protocol was performed essentially as described by Dejardin and Kingston {[64] and http://genetics.mgh.harvard.edu/kingstonweb/labmembers1.html}. Cells were treated with 3% formaldehyde in PBS for 30 min at room temperature (RT) prior to harvesting. Nuclei were isolated by homogenization and digested with micrococcal nuclease (MN) for 30 min at 37°C. MN-treated nuclei were sonicated using a Branson Sonifier 250 (output 5.5 with 28 cycles of 15 sec/cycle) with intermittent cooling on ice to further reduce the length of chromatin fragments to ~600 bp. Fragmented chromatin was then hybridized to a biotinylated telomere DNA probe [desthiobiotin-108 carbons-5′TtAgGgTtAgGgTtAgGgTtAgGgt-3′; capital letters indicate locked nucleic acid (LNA) residues, which increase the stability of probe-chromatin interactions], or to a biotinylated scrambled DNA probe (desthiobiotin-108 carbons-5’ GaTgTgTgGaTgTggAt-GtGgAtgTgg-3’). Hybridization conditions were 25°C for 3 min, 70°C for 6 min (3 cycles), 38°C for 60 min, 60°C for 2 min, 38°C for 60 min, 60°C for 2 min, 38°C for 120 min, and 25°C final temperature, using an MJ PTC-200 PCR machine. The probes contain a long 108-carbon spacer between the immobilization tag (desthiobiotin) and the LNA probe; the spacer was designed to minimize steric hindrance [88]. These probes were synthesized by Fidelity Systems (Gaithersburg, MD) and Exiqon (Woburn, MA). The hybridized chromatin was captured onto magnetic streptavidin beads. The beads were washed extensively under stringent conditions at 42°C, and then protein was eluted using sample buffer for Western blot analysis of AR (AR antibody N-20, Santa Cruz) and shelterin components {antibodies against TRF2 (Imgenex), Rap1(Bethyl laboratories), and TIN2 [9, 44]}. Histone 3 (Abcam) signal intensity served as an indicator of sample loading.

Fluorescence in situ hybridization (FISH) analysis of telomeres

Telomeres were visualized by fluorescence in situ hybridization (FISH) on metaphase spreads using a telomeric protein nucleic acid (PNA) probe, as described [89]. LNCaP cells were treated with 100 μM Casodex for 24 hr, then washed in medium without Casodex, and incubated overnight at 37°C in medium containing colcemid (0.1 μg/ml, Sigma). Cells were trypsinized and collected at 1000 × g (8 min). After hypotonic swelling in 0.075 mM KCl for 25 min at 37^o C, cells were fixed in methanol:acetic acid (3:1). Telomere FISH was performed as described [90], using a Cy3-OO-(CCCTAA)₃-peptide nucleic acid (PNA) probe (Panagene).