Ku Is a Novel Transcriptional Recycling Coactivator of the Androgen Receptor in Prostate Cancer Cells*

The androgen receptor (AR) dynamically assembles and disassembles multicomponent receptor complexes in order to respond rapidly and reversibly to fluctuations in androgen levels. We are interested in identifying the basal factors that compose the AR aporeceptor and holoreceptor complexes and impact the transcriptional process. Using tandem mass spectroscopy analysis, we identified the trimeric DNA-dependent protein kinase (DNA-PK) complex as the major AR-interacting proteins. AR directly interacts with both Ku70 and Ku80 in vivo and in vitro, as shown by co-immunoprecipitation, glutathione S-transferase pull-down, and Sf9 cell/baculovirus expression. The interaction was localized to the androgen receptor ligand binding domain and is independent of DNA interactions. Ku interacts with AR in the cytoplasm and nucleus regardless of the presence or absence of androgen. Ku acts as a coactivator of AR activity in a luciferase reporter assay employing both Ku-defective cells and Ku small interfering RNA knock-down in a prostate cancer cell line. DNA-PK catalytic subunit (DNA-PKcs) also acts as a coactivator of androgen receptor activity in a luciferase reporter assay employing DNA-PKcs defective cells. AR nuclear translocation is not affected in Ku defective cells, implying Ku functionality may be mainly nuclear. Chromatin immunoprecipitation experiments demonstrated that both Ku70 and Ku80 interact with the prostate-specific antigen promoter in an androgen-dependant manner. Finally, in vitro transcription assays demonstrated Ku involvement in transcriptional recycling with androgen dependent promoters.

perfamily. The structure of the AR is consistent with members of this family, consisting of an amino-terminal activation function (AF-1), the DNA binding domain, the hinge region, a carboxyl-terminal ligand binding domain (LBD), and a second AF (AF-2) in the LBD. The AR is involved in growth, differentiation, and the progression of prostate cancer (for review, see Ref. 1).
For the AR to rapidly respond to fluctuations in hormone levels, it functions as a member of a multicomponent complex (2). Simplistically, the aporeceptor complex, primarily residing in the cytoplasm, translocates into the nucleus upon hormone binding. The AR-hormone complex binds the androgen response element (ARE), forming the holoreceptor complex. Both the aporeceptor and holoreceptor complexes act to facilitate, stabilize, and enhance the AR activity. One such complex consists of molecular chaperones that are involved in the stabilization of AR as well as its delivery to the nucleus. It is likely other complexes exist of which we have very little knowledge.
In an attempt to identify novel members of the AR aporeceptor and holoreceptor complexes, we used tandem mass spectroscopy to study proteins associated with AR. As expected, we identified multiple members of the heat shock protein family which have previously been shown to stabilize the unligated receptor in the cytoplasm, and to regulate ligand affinity (11,12). Additionally, we identified all members of the DNAdependent protein kinase (DNA-PK) heterotrimeric complex. DNA-PK is composed of the heterodimeric Ku and the DNA-PK catalytic subunit (DNA-PKcs) (13,14). Ku, the regulatory subunit, consists of proteins with approximate molecular masses of 70 and 80 kDa (Ku70 and Ku80, respectively). DNA-PKcs is a nuclear serine/threonine protein kinase, a member of the phosphatidylinositol 3-kinase family (15), with an approximate molecular mass of 470 kDa.
Ku was originally characterized by Mimori et al. (16) as an autoantigen recognized by the sera of patients with polymyositis-scleroderma overlap syndrome and now is often found in patients with other autoimmune diseases. Ku plays a role in a multitude of nuclear processes: that is, DNA repair, telomere maintenance, V(D)J recombination, and transcriptional regu-lation (for review, see Ref. 17). Ku is probably best known for its role in nonhomologous DNA-end-joining (NHEJ) repair, a process responsible for repairing a majority of DNA double strand breaks. The importance of Ku is demonstrated by the fact that cells deficient in either Ku are more sensitive to ionizing radiation (18 -20).
The critical function of Ku in DNA repair not withstanding, recently there has been an evolving role for Ku in transcription. Initial reports identified Ku as a transcription factor that directly bound to sequence-specific promoter elements (21)(22)(23). Other reports emerged that Ku was associated with RNA polymerase II sites (24), and Ku is also directly associated with the RNA polymerase II complex (25). The entire DNA-PK complex is also involved in transcriptional regulation. DNA-PK has been shown to interact with and/or phosphorylate a number of transcription factors including epidermal growth factor receptor (26), c-Myc (27), c-Jun (28), glucocorticoid receptor (22), progesterone receptor (29), and thyroid hormone receptor-binding protein (30). DNA-PK has also been shown to phosphorylate RNA polymerase II (31). Transcription is also affected in DNA-PK-deficient SCID (severe combined immunodeficiency) mice (32), DNA-PK-defective human glioma cells (33), and Ku80-defective XRS-5 cells (33,34). Ku by itself and as part of the DNA-PK complex appears to be integrally important for transcriptional control. Additionally, Ku has previously been identified as being involved in secondary initiation events (35,36). Cells deficient for Ku showed greatly reduced transcription, which has been attributed to defective reinitiation. Reinitiation is enhanced by the recycling of promoter and other transcription related proteins more efficiently (37).
Here we identify Ku as a transcriptional recycling coactivator of AR function. We show that 1) AR interacts directly with both Ku70 and Ku80 and indirectly with DNA-PKcs, 2) the interaction of Ku70 and Ku80 with AR is via the AR LBD, 3) both Ku and DNA-PKcs enhance AR activity in transactivation assays, 4) Ku is recruited to the prostate-specific antigen (PSA) promoter in an androgen-dependent manner, and 5) Ku enhances the transcriptional activity of AR through recycling of the transcriptional factors. These results define Ku as a recycle coactivator of AR.

EXPERIMENTAL PROCEDURES
Reagents and Antibodies-All cell culture reagents were from Invitrogen unless otherwise noted. FBS and charcoal dextran-treated FBS were from Omega Scientific. Antibodies to the androgen receptor were from Upstate and Neomarkers. Antibodies to Ku were from Neomarkers, antibodies to DNA-PKcs were from Upstate, and antibodies to tubulin were from Santa Cruz Biotechnology. siRNA oligos were from Dharmacon.
Expression Constructs and Protein Expression-All DNA-PK plasmids were generous gifts of Dr. David J. Chen (University of Texas Southwest Medical Center). AR fragments were constructed by PCR and ligated into pcDNA3.1. Insect cell viruses were generous gifts of Hong-Wu Chen (UC Davis). GST fusion proteins were expressed and purified according to the procedures provided by Amersham Biosciences.
GST Pull-down Assays-In vitro binding assays were performed by incubating 10 g of GST fusion protein or an equal amount of GST resin and 10 l of [ 35 S]methionine-labeled, in vitro translated proteins produced in rabbit reticulocyte lysate (Promega). Proteins were incubated at 4°C for 2 h in the binding buffer (50 mM Tris-HCl, pH 7.4, 100 mM NaCl, 5 mM MgCl 2 (for DNase activity), 1.0% Nonidet P-40, 10% glycerol, 1 mM dithiothreitol protease inhibitor). Bound proteins were washed three times with binding buffer and subjected to SDS-PAGE and autoradiography.
Mass Spectrometry Analysis-Proteins were visualized on SDS-PAGE by Sypro Ruby stain. Bands of interest were excised and subjected to trypsin digestion, microcapillary high pressure liquid chromatography nanoelectrospray tandem ion trap mass spectrometry, and mass spectrometry/mass spectrometry peptide sequence analysis using a reverse phase C18 Magic 2000 HPLC (Michrom) and LCQ Deca mass spectroscopy (Thermo). Protein identification was facilitated by the Sequest software.
Dual Luciferase Reporter Assays-All procedures were done according to the dual luciferase reporter assay system (Promega). The indicated cells were transiently transfected with pGL3 Basic containing either a 4ϫ repeat of the androgen response element (4XARE), the mouse mammary tumor virus promoter, the probasin (PRO) promoter, or the PSA full-length promoter and the pRL SV40 control plasmid. Cells lacking AR were also transfected with pcDNA:HA-AR containing full-length human AR. Cells were lysed using the passive lysis buffer (from the Promega assay kit mentioned above) and analyzed on a 96-well microplate luminometer (EG&G Berthold).
Chromatin Immunoprecipitation Assay-Chromatin immunoprecipitation assays were preformed with LNCaP cells treated as described above. Experiments were preformed exactly as described using the PCR primers described in Louie et al. (38).
siRNA Knockdown-PC-3(AR) cells were transiently transfected with siRNA oligos derived from Belenkov et al. (39) using Oligofectene (Invitrogen) as recommended by the manufacturer. Cells were treated for 72 h to achieve maximum knockdown.
In vitro Transcription-Lysates and assay were preformed essentially as described by Woodard et al. (35). Nuclear extracts were prepared from XRV15B and XRV15BϩKu (which lack endogenous AR) and from the same cells transiently transfected with AR. In vitro transcription used 25-50 g of extract in 50-l reaction volumes containing 500 ng of the androgen-dependent G-less reporter p(ARR 3 )LovTATA (40), 10 nM DHT, and where indicated 50 pmol of AR. Lysates were preincubated for 30 min before nucleotide addition. Heparin was added 5 min after initiation, and the reactions were allowed to proceed for a total of 45 min. The reactions were stopped, extracted, and precipitated before separation by 6% urea polyacrylamide gel electrophoresis at 250 V until the bromphenol blue ran off the bottom. Bands were quantified by phosphorimage analysis (Quantity One, Bio-Rad).

DNA-PK Complex Interacts with the Carboxyl-terminal Region of the Androgen Receptor-A great deal of attention has
been paid to identifying AR cofactors that modulate its activity in an androgen-dependent manner. Our laboratory has been interested in identifying basal factors that associate with AR in an androgen-independent manner. To facilitate identification of potential members of the multicomponent AR complex, the AR was divided into both amino-terminal (amino acids 1-478) and carboxyl-terminal (ARC, amino acids 479 -919) fragments to facilitate bacterial expression of the GST-tagged constructs. The GST-ARC fragment was expressed, partially purified on GST resin, and incubated with nuclear extracts of LNCaP cells (Figs. 1, A and B). LNCaP cells were either deprived of androgen (C) or stimulated with 10 nM DHT (D) to allow for any androgen-stimulated post-translational modifications for 2 h before lysis as described under "Experimental Procedures." Lysates were treated with DNase I, and complete digestion was confirmed by EtBr staining of extracts (data not shown). Interacting proteins were separated by SDS-PAGE.
Multiple bands in the region of 28 -400 kDa were observed. Only the predominant bands were isolated for further identi-fication by microsequence analysis. These proteins were isolated from the SDS-PAGE, and their identities were determined by proteolytic peptide sequencing. Three of the proteins analyzed belong to the DNA-PK complex; the 470-kDa DNA-PK catalytic subunit, the 70-kDa DNA-PK regulatory subunit Ku70, and the 80-kDa regulatory subunit Ku80. Other proteins identified were 100-kDa poly(ADP-ribose) polymerase, 90-kDa topoisomerase I, and 220-kDa RNA polymerase II polypeptide A, which have also been shown to be associated with hormone receptors (30). The stoichiometry of ARC and DNA-PK interaction is very high, indicating a higher affinity binding. The presence of multiple peptide fragments of each of the components of the DNA complex in tandem mass spectroscopy analysis gave us confidence in the assignment of these molecules as AR-interacting proteins. These interactions appear to be ligand-independent as the association is detected in both DHTtreated and untreated cell extracts. The in vitro interactions of DNA-PK complex with AR were confirmed by in vivo large scale immunoprecipitation of endogenous AR from LNCaP nuclear extracts followed by tandem mass spectroscopy. Table I shows the major co-immunoprecipitated proteins identified by tandem mass spectroscopy. In the molecular range of 70 -120 kDa, the major interacting complexes of the androgen receptor are the heat shock protein complex and the DNA-PK complex.
Ku70 and Ku 80 Directly Interact with AR in Vitro-The DNA-PK interaction was selected as the main focus due to the higher affinity of the complex, the increasing importance of DNA-PK as a nuclear hormone receptor transcriptional regulator (29,30), and the involvement of Ku in transcriptional recycling (35,36). To confirm the direct interaction of DNA-PK subunits with AR, GST pull-down assays were employed. GST-ARC was expressed in Escherichia coli and partially purified on GST resin. Ku70, Ku80, and DNA-PKcs were expressed and [ 35 S]methionine-labeled in the rabbit reticulocyte lysate system. Due to the large size of DNA-PKcs (ϳ470 kDa), which rendered the expression in E. coli impossible, DNA-PKcs was expressed in three separate overlapping constructs, the aminoterminal region of DNA-PKcs (N-DNAPK), middle region of DNA-PKcs (M-DNAPK), and carboxyl-terminal region of DNA-PKcs (C-DNAPK) ( Fig. 2A) GST-ARC interacted with both Ku70 and Ku80; however, the Ku70 interaction appears to be stronger (Fig. 2B). There appears to be no direct interaction between GST-ARC and any of the fragments of DNA-PKcs. We interpret this to mean that DNA-PKcs association with AR is via Ku70 and Ku80. Importantly, DNase treatment had no affect on the interaction of Ku70 and Ku80 with AR, suggesting the association is not due to contaminating DNA. No interaction was observed when GST resin alone was employed. Once a direct interaction between AR and Ku70 and Ku80 was confirmed, fragments of AR encompassing the DNA binding domain (DBD), ligand binding domain (LBD) and a larger fragment encompassing both the DNA binding domain and the LBD (DϩLBD) were constructed (Fig. 1B). The AR fragments along with full-length AR and green fluorescent protein were expressed and [ 35 S]methionine-labeled in rabbit reticulocyte lysate. GST-Ku70 and GST-Ku80 were expressed in E. coli and partially purified on GST resin. Both the AR LBD and DϩLBD fragments interacted with both Ku70 and Ku80 (Fig. 2B). The full-length AR had a weaker but noteworthy interaction with Ku70 and Ku80. Inclusion of 10 nM DHT did not enhance interaction with full-length AR. No interaction with GST resin alone was observed. The finding that the AR LBD but not the DNA binding domain is the interacting domain further strengthened the fact that the interaction is not DNA dependent.
Ku70 and Ku80 Directly Interact with AR in Vivo-Coimmunoprecipitation of endogenous AR and Ku70 from LNCaP cells again confirms the AR-DNA-PK interaction (Fig. 3A). LNCaP cells were deprived of androgen for 3 days then either left unstimulated or stimulated with 10 nM DHT for 2 h as described under "Experimental Procedures." Cells were subsequently lysed, fractionated into cytosolic and nuclear fractions, and immunoprecipitated with anti-AR or anti-Ku70 antibodies. Immunoprecipitates were separated by SDS-PAGE and subjected to immunoblot with anti-DNA-Pkcs, anti-AR, anti-Ku80, or anti-Ku70 antibodies. AR interacted strongly with Ku70 in the cytosol in an androgen-independent manner. AR interaction with Ku70 in the nucleus appeared to be enhanced by DHT, likely due to the increase of AR in the nucleus after DHT stimulation. AR interaction with DNA-PKcs appears to be predominately nuclear. This interaction has been confirmed by the colocalization of AR and DNA-PK in LNCaP cells using immunocytochemistry (data not shown).
Further confirmation of the interactions was obtained in Sf9 insect cells. Sf9 cells were infected individually and in combination with baculovirus carrying FLAG-AR, His-Ku70, and Ku80 for 36 -48 h. Cells were lysed and immunoprecipitated with anti-FLAG resin. Distinct interactions were observed between Ku70 and AR and between Ku80 and AR as well as between the Ku homodimer and AR (Fig. 3B). The formation of the AR-Ku complex in Sf9 cells is further confirmed by Sypro-Ruby staining of FLAG-immunoprecipitated AR from baculovirus-infected Sf9 cells (Fig. 3C).
Ku Homodimer as Well as DNA-PKcs Are Co-activators of Androgen Receptor Activity-To examine the influence of Ku on AR activity, full-length human AR together with PSA-Luc plasmids were transfected into XR-V15B, a cell line lacking Ku80 and Ku70, or its transfected variant, XR-V15B-Ku, which expresses both Ku70 and Ku80. The Renilla luciferase construct served as an internal transfection efficiency control. After transfection cells were treated with 10 nM DHT, and the luciferase levels were quantified. Ku-containing cells showed a substantial increase in AR activity for the 6.0-kilobase PSA reporter construct (Fig. 4A). Ku also enhanced AR activity for the 4ϫ ARE and mouse mammary tumor virus reporter constructs employed (data not shown). This suggests that Ku enhances AR-mediated transactivation.
Because AR plays a major role in prostate cancer, we sought to examine the effect of Ku knock-down on AR activity in the prostate cancer cell line. We chose PC-3(AR), a prostate cancer cell line expressing AR, because of its demonstrated androgen response and the relatively high transfection efficiency. To knock down Ku levels, a siRNA duplex against Ku80 was employed. Treatment of the prostate cancer cell line PC-3(AR) with Ku80 siRNA (39) resulted in a reduction of Ku80 expression versus a scrambled control (Fig. 4B). As previously reported, Ku70 expression was also reduced in the Ku80 knockdown cells (39) because the expression of Ku70 is tightly regulated by Ku80. If Ku is involved in AR-mediated transactivation, then endogenous AR activity should diminish subsequent to Ku siRNA transfection. Treatment with two concentrations of siRNA duplex both resulted in reductions in AR activity as measured by a 6.0-kilobase PSA promoter luciferase reporter versus the control (Fig. 4C). These data are all con-sistent with Ku being a co-activator of AR activity.
Having demonstrated the role of Ku in AR transactivation, we asked whether the associated DNA-PKcs is also important in this process. To examine the influence of DNA-PKcs on AR activity, full-length human AR construct was transiently transfected into CHO V3 (DNA-PKcs mutant) or the parental CHO AA8 (wild type) cells along with AR luciferase reporter constructs. We tested two reporter constructs, 4XARE-luciferase and mouse mammary tumor virus luciferase, suitable for CHO cells. "DNAPKϩ" cells showed considerable enhancement of AR activity over "DNAPKϪ" cells for both reporters (Fig. 4D), suggesting DNA-PKcs is also AR transactivator.
Ku Is Recruited to the PSA Promoter after DHT Stimulation of LNCaP Cells-If Ku is involved in the AR transcriptional process, one would expect that Ku would associate with the ARE in a manner similar to AR. To determine the association of Ku proteins with the PSA promoter after DHT treatment of LNCaP cells, we preformed chromatin immunoprecipitation experiments. Cells were cross-linked with formaldehyde, and after sonicating the chromatin to a length of about 500 base pairs, DNA was immunoprecipitated with antibodies that rec-  ognize AR, Ku70, Ku80, or DNA-PKcs. After immunoprecipitation, proteins were digested, and purified DNA was analyzed by PCR. Primers covering five regions of the PSA promoter were employed (Fig. 5A). After DHT stimulation, levels of AR, Ku70, and Ku80 associated with the enhancer and promoter regions of the PSA promoter were enhanced (Fig. 5B). Little association was found in the two control regions employed. As previously shown (38), the enhancer region showed the greatest association with AR. The PSA promoter region showed higher basal levels of Ku association, potentially due to the other functions of Ku. These data provide a further confirmation of the role of Ku in AR transcription.
Ku Is Involved in Transcriptional Recycling of AR-Ku has previously been shown to be involved in transcriptional recycling (35,36). The G-less cassette p(ARR 3 )LovTATA, containing three repeats of the androgen response regions of the rat probasin gene (40) immediately upstream of a 360-nucleotide segment lacking guanosine, was used to determine the effects of Ku in AR recycling. The G-less cassette allows for in vitro transcription to be terminated at the first G residue without linearizing the plasmid. Employing a circular, supercoiled plasmid is essential to more accurately replicating in vivo transcription and to eliminating the possibility of Ku binding to DNA ends. Nuclear extracts from XRV15B (Ku Ϫ ) and XRV15B-Ku (Ku ϩ ) cells were prepared as described under "Experimental Procedures." Approximately 50 pmol of purified AR was added to the transcription reactions where indicated. Transcription of p(ARR3)LovTATA was dependent upon both AR and DHT (Fig. 6A). Transcription reactions were preincubated for 30 min before the addition of nucleotides. Primary versus secondary initiation was distinguished by the addition of heparin in identified lanes (41). Heparin prevents the formation of new initiation complexes without disrupting actively transcribing complexes. Reactions were allowed to proceed an additional 45 min. Reactions were separated by 6% denaturing polyacrylamide gel electrophoresis (Fig. 6B). The ratio of the band in the absence of heparin to that in the presence indicates the efficiency reinitiation in this system. As shown in Fig. 6B, Ku enhanced AR-mediated transcription of the reporter constructs. The efficiency of reinitiation was much higher when Ku was present. To further confirm these results, XRV15B and Cells were formaldehyde-cross-linked and sonicated to shear DNA into ϳ500-bp fragments. Lysates were immunoprecipitated with AR, Ku70, and Ku80 as indicated above. Cross-linking was reversed and proteindigested, and DNA was amplified using primers identified previously. Antibodies used for immunoprecipitation are indicated above the panels, and primers used for amplification are indicated on the left. XRV15B-Ku cells were transiently transfected with AR (Ku Ϫ /AR and Ku ϩ /AR, respectively). Nuclear lysates were prepared as before, except with no addition of exogenous AR to the transcription reactions. Lysates were added to transcription reactions, and as previously observed the presence of Ku enhanced the ability of AR to perform multiple rounds of transcription (Fig. 6C). The increased transcription could be attributed to the ϳ2-fold greater reinitiation of the Ku-containing extracts. Average -fold increase over heparin-treated cells is noted beneath the lanes. Fold increase is the average of at least three independent experiments. As previously reported (35) Ku ϩ cell lines showed decrease transcription relative to Ku Ϫ in heparin-treated lanes. The greater difference for heparin treatment in the Ku ϩ cells is indicative of the involvement of Ku in transcriptional recycling. DISCUSSION We undertook a targeted discovery of potential factors that comprise the aporeceptor and holoreceptor complexes of the androgen receptor. We began by attempting to identify proteins that interact with the carboxyl termini of the androgen receptor. Mass spectroscopy analysis identified trimeric DNA-PK complex as the major bands interacting with the ARC. DNA-PK consists of the Ku70/Ku80 heterodimer and the catalytic subunit DNA-PKcs. Recently, Ku itself and as part of the DNA-PK complex has emerged as a part of the transcriptional machinery (21,23,35,36). Specifically DNA-PK has been identified as a co-regulator of several steroidal transcription factors (22,29,30).
We first confirmed the identified interactions by multiple methods. GST pull-down assays and Sf9 cell protein interaction studies indicated that both Ku70 and Ku80 directly and independently interacted with AR; however, no interaction was observed with DNA-PKcs. This could indicate that Ku functions as a docking protein for DNA-PKcs as has been shown for a variety DNA-PKcs substrates (42). Although it has been previously shown that Ku subunits do not independently interact with DNA (43,44), we felt it necessary, due to the fact both the Ku heterodimer and AR are DNA binding proteins, to DNase-treat interactions to exclude the possibility that the interaction was bridged by DNA. As expected the interactions were not impaired by DNase treatment. Subsequently the interaction was localized to the LBD of AR. It has previously been shown that LXXLL and FXXLF motifs are responsible for protein interactions with the AR LBD (45)(46)(47). Ku70 contains a LXXLL motif, and Ku80 contains the FXXLF motif. However, mutation of the canonical Ku70 LXXLL motif and Ku80 FXXLF motif does not affect their binding to AR, indicating that they are not important for AR binding (data not shown). This is not totally unexpected because both Ku70 and Ku80 bind AR in the cytoplasm in the absence of ligand, when the hydrophobic groove in the LBD is occupied by the FXXLF of the AR aminoterminal domain.
After confirmation of the direct interaction, we examined the role of Ku in AR activity. If Ku is a member of the AR macromolecular complex, Ku might regulate AR transcriptional activity. Luciferase reporter assays indicated that Ku was a positive regulator of AR activity for multiple promoters. Although the cell line employed in these experiments is only mutant for Ku80, Ku70 levels were also greatly reduced or eliminated (19), and expression of exogenous Ku80 rescued Ku70 expression, making it impossible to determine whether the presence of one subunit is sufficient for Ku co-activator activity. These results were further confirmed in the more relevant prostate cell line PC-3(AR), stably expressing exogenous human AR, employing Ku80 siRNA adapted from Belenkov et al. (39). As before, knock down of Ku80 also reduced Ku70 levels, thus complicat-ing interpretation of the data. Even in Sf9 cells Ku70 expression was heavily dependent upon Ku80 expression.
We next examined several potential mechanism of Ku coactivator activity. Co-immunoprecipitation assays indicated that Ku interacted with AR both in the cytoplasm and in the nucleus. Both Ku70 and Ku80 contain nuclear localization signals and are actively transported into the nucleus (24,48,49). Potentially Ku was involved in nuclear localization of AR after ligand binding. Green fluorescent protein-tagged AR localization was observed both in Ku mutant and Ku rescued cells. Lack of Ku did not appreciably affect AR localization either in the presence or absence of DHT (data not shown), indicating Ku is not involved in nuclear localization of AR. Next we examined the role of the other member of the DNA-PK complex, DNA-PKcs in AR activity. A cell line mutant for DNA-PKcs and the parental cell line were employed in the same AR activity assay used for Ku. DNA-PKcs showed similar co-activator activity to Ku. This indicated that DNA-PKcs may be involved in the coactivator activity of AR. It has been shown previously that Ku can bridge interactions between a protein and DNA-PKcs, allowing DNA-PKcs to phosphorylate the target protein (42). AR contains multiple consensus (S/T)Q DNA-PK phosphorylation motifs clustered mainly in or near the LBD and has also previously been shown that DNA-PK phosphorylates AR (50). Although it is likely that one of the functions of Ku is to bridge the AR-DNA-PKcs interaction, Ku has several other potential co-activator functions. Ku has been identified as a member of the RNA polymerase II complex (24,25), which also was identified by liquid chromatograph-tandem mass spectroscopy as an ARC-interacting protein. This interaction itself may account for some of the co-activator function. Additionally, Ku has been implicated in the transcriptional reinitiation process (35,36), which may account for co-activator function.
In an effort to further understand the role of Ku plays in enhancing AR transcription, chromatin immunoprecipitation assays were preformed. If Ku were involved in recycling AR, one would expect that Ku would colocalize with AR at an ARE in an androgen-dependent manor to release AR from the ARE after one round of transcription. Both Ku70 and Ku80 were found associated with the PSA promoter in an androgendependent manner. This association within the promoter would support the role of Ku in reinitiation. If the role of Ku was limited to DNA-PKcs recruitment, one might expect rapid disassociation after DNA-PKcs phosphorylation. The definitive test of AR recycling is the actual examination of Ku recycling abilities. Ku had previously been implicated in the transcriptional recycling employing several non-steroidal promoters (35,36). Utilizing an androgen receptor-dependent promoter system, we demonstrated the role of Ku in reinitiation. A Ku-deficient cell line underwent secondary rounds of transcription initiation at a much lower rate than the Kurescued cell lines. This was independent of the source of AR. Additionally, XRV15B-Ku86 cells showed a much higher level of transcription.
In addition to the DNA-PK subunits, several other interesting proteins were also found associated to ARC including poly-(ADP-ribose) polymerase and DNA topoisomerase I that have not yet been studied. These factors commonly appear in interactions with other transcription factors (30). Both poly(ADPribose) polymerase (51,52) and topoisomerase I (53, 54) are involved in "preparing" a gene for transcription. It appears the AR can recruit a "pre-initiation" complex of several of the factors required for transcriptional initiation. All these factors, Ku, DNA-PKcs, topoisomerase I, and poly(ADP-ribose) polymerase, have been implicated in cancer. However, by Western blot Ku levels are not significantly altered in several prostate cancer cell lines (CWR22R, DU145, LNCaP, PC-3) versus normal prostate epithelial cells (data not shown). Additionally there was no consistent microarray data indicating a substantial change in Ku or DNA-PKcs message levels from normal to various levels of cancerous prostate cells (55,56) However, this does not imply that DNA-PK does not play a role in prostate cancer progression. It has recently been shown that although there are no significant differences in gene expression for members of the non-homologous end-joining pathway between malignant and non-malignant prostate cells, malignant cells were defective for DNA repair (57). This has many interesting implications for the role of DNA damage in AR-dependent transcription and the role of the AR-DNA-PK interaction in DNA repair.
The continuous recycling hypothesis of steroid receptor regulation (2, 58) is an attractive mechanism for explaining the rapid and modulated response of AR to androgens. Ku meets all identified criteria for being a recycling factor. 1) Ku interacts with the androgen receptor ligand binding domain, 2) Ku is an ATPase, and 3) Ku is involved in transcriptional recycling. The model consists of three phase: signaling, assembly, and disassembly. The recycling factors are proposed to be involved in both the disassembly and signaling phases. Therefore, one would expect that AR and Ku would interact in the cytoplasm and the nucleus both in the presence and absence of androgen, which is exactly what was found. Finally, Ku is a coactivator of AR activity as would be expected for a recycling factor.