Par-4 transcriptionally regulates Bcl-2 through a WT1-binding site on the bcl-2 promoter.

Elevated expression levels of the bcl-2 proto-oncogene have been extensively correlated with the appearance of androgen independence in prostate cancer. Although bcl-2 was first cloned as the t(14:18) translocation breakpoint from human follicular B cell lymphoma, the mechanism of overexpression of bcl-2 is largely undefined for advanced prostate cancer because there are no gross alterations in the gene structure. We investigated the role of the product of the prostate apoptosis response gene-4 (Par-4) and the product of the Wilms' tumor 1 gene (WT1) in the regulation of Bcl-2 expression in prostate cancer cell lines. We observed growth arrest and apoptosis, upon decreasing Bcl-2 protein and transcript in the high Bcl-2-expressing, androgen-independent prostate cancer cell line, by all-trans-retinoic acid treatment (ATRA), but this did not occur in the androgen-dependent cell line expressing low levels of Bcl-2. The decrease in the Bcl-2 protein and transcript following all-trans-retinoic acid treatment was accompanied by changes in localization of Par-4 and an induction in the expression of WT1 protein. In stable clones expressing ectopic Par-4 and in ATRA-treated cells, we observed decreased Bcl-2 protein and transcript. This was accompanied by an induction in WT1 expression. The involvement of WT1 in the Par-4-mediated down-modulation of Bcl-2 was further defined by blocking endogenous WT1 expression, which resulted in an increase in Bcl-2 expression. Finally, we detected Par-4 and WT1 proteins binding to a previously identified WT1-binding site on the bcl-2 promoter both in vitro and in vivo leading to a decrease in transcription from the bcl-2 promoter. We conclude that Par-4 regulates Bcl-2 through a WT1-binding site on the bcl-2 promoter. These data also identify Par-4 nuclear localization as a novel mechanism for ATRA-mediated bcl-2 regulation.

Overexpression of Bcl-2 in prostate cancer is a hallmark of advanced hormone-refractory disease and may account for the resistance to apoptosis that is characteristic of the late stages (1,2). This correlation is strengthened by in vitro studies that demonstrate an increased Bcl-2 expression upon androgen withdrawal in androgen-sensitive prostate cancer cells (3)(4)(5).
Previously, we have shown that high Bcl-2 expression in androgen-independent prostate cancer cells is associated with cytoplasmic sequestration of the p53 protein, thereby impairing the nuclear transport of p53 in response to an apoptotic signal (6). Overexpression of Bcl-2 also correlates with poor prognosis for patients undergoing radical prostatectomy or hormonal ablation. In addition, high levels of Bcl-2 confer resistance to chemotherapy in prostate cancer cell lines, and current clinical efforts are aimed at modulating the expression of Bcl-2 (3). Furthermore, in vivo studies have demonstrated that increased expression of Bcl-2 accelerates the appearance of the androgen-independent phenotype in prostate cancer and that decreasing Bcl-2 expression delays progression to androgen independence along with enhancing chemosensitivity of these tumors (7)(8)(9).
The bcl-2 gene is a key determinant of neoplastic cell expansion whose oncogenic activity has been ascribed primarily to its ability to promote cell survival (10). It was first cloned as the t(14:18) translocation breakpoint from follicular lymphoma (11,12). This translocation results in the juxtaposition of the bcl-2 gene in front of the immunoglobulin heavy-chain enhancer that leads to aberrant expression. In addition to lymphomas with t(14:18) translocations, high levels of Bcl-2 protein, aberrant patterns of Bcl-2 protein production, or both have been observed in a variety of solid tumors (13,14). However, in contrast to lymphomas, no evidence for gross alterations in the bcl-2 gene structure has been obtained for these other types of cancer, suggesting that expression is controlled through transcription initiation at the bcl-2 promoter (15). Two promoter regions have been identified in the 5Ј-regulatory region of the bcl-2 gene. P1, the predominant promoter in B cells, is a TATAless, GC-rich region containing multiple initiation sites (16 -18). The second promoter, P2, is located 1.3 kb downstream of P1, contains a CCAAT box, an octamer motif, and a TATA element. This promoter is active in neuronal cells and some pre-B cells (18,19). bcl-2 promoter usage in prostate cancer cells and the identity of transcription factors that regulate its expression in these cells remain uncharacterized. Recently, nuclear factor-B has been identified to bind to the bcl-2 P2 promoter via B sites and activate its transcription in prostate cancer cells in response to tumor necrosis factor-␣ treatment (20). However, the mechanisms of Bcl-2 overexpression in androgen-independent prostate cancer are undefined.
The product of the prostate apoptosis response gene-4 (Par-4), although not a classical transcription factor, is co-expressed along with Bcl-2 in the normal prostate basal epithelia (23). Following differentiation of these cells to luminal/secretory epithelia, Par-4 expression is decreased and mainly restricted to the nucleus, and Bcl-2 expression is lost (23). Previously, ectopic expression of Par-4 led to the down-modulation of Bcl-2 * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. protein expression in NIH3T3 mouse fibroblasts and in PC3 prostate cancer cell lines (24). An inverse correlation in the expression of Bcl-2 and Par-4 also has been reported in the xenografts of human androgen-dependent and androgen-independent prostate cancer developed in mice as well as in acute lymphocytic leukemia (21,22,24).
Par-4 is approximately a 39-kDa protein, first identified by differential screening for early response genes up-regulated during induction of apoptosis in prostate cells (25). The nucleotide sequence of Par-4 predicts a leucine zipper protein with a nuclear localization signal, although no DNA binding has been demonstrated for this molecule (26,27). A stretch of 70 amino acids at the COOH terminus of Par-4 shows homology to a region referred to as the death domain, which is also found in other pro-apoptotic proteins such as Fas, FADD, and TNFR-1 (26,28). Par-4 seems to be functionally required but not sufficient for apoptosis (26,28,29). Expression of this protein was found to be ubiquitous and not restricted to any specific organ (23).
Par-4 is known to interact with the following three proteins: the product of the Wilms' tumor gene (WT1), 1 the atypical protein kinase C-, and the death-activated protein-like kinase (27,29,30). WT1 is a 52-54-kDa protein that possesses many characteristics of a transcription factor including a glutamine/ proline-rich NH 2 terminus, a nuclear localization domain, and four Cys 2 -His 2 zinc finger motifs (31). Three COOH-terminal zinc fingers share 64% homology to the zinc fingers of the early growth response gene 1 (32). WT1 is involved in the transcriptional regulation of the bcl-2 gene (33)(34)(35). WT1 has been shown previously (34,35) to bind to two different sites on the bcl-2 P1 promoter to both repress and activate the transcription of this gene. Par-4 binds to two different domains of WT1 and regulates the transcriptional activation and repression potential of WT1 (27,36). In view of the correlation of Bcl-2 overexpression and progression of prostate cancer from androgen dependence to independence, the observation that ectopic expression of Par-4 can decrease Bcl-2 expression and that Par-4 binds to WT1, we set out to examine whether Par-4 utilizes WT1 to bind to the bcl-2 promoter to transcriptionally regulate Bcl-2 expression in an androgen-independent prostate cancer cell line. We report here evidence that Par-4 can transcriptionally modulate Bcl-2 expression by associating with a previously defined WT1-binding site on the bcl-2 P1 promoter.

EXPERIMENTAL PROCEDURES
All-trans-Retinoic Acid-All-trans-retinoic acid (ATRA) has been used previously (37,38) to down-modulate the expression of Bcl-2. Because ATRA is a hydrophobic drug and to prevent its rapid clearance due to enhanced metabolism, we used liposomes to deliver the drug (39). L-ATRA, the liposomal form of ATRA, was prepared in bottles from lyophilized powder in bottles containing ATRA (100 mg) in a 1:15 drug:lipid formulation that consisted of soy oil and dimyristoylphosphatidylcholine (kindly supplied by Aronex Pharmaceuticals, Inc., The Woodlands, TX, as Atragen R ). L-ATRA was stored in the dark at 4°C, reconstituted by adding 50 ml of normal saline to each bottle, and vortexed for 2 min immediately before use.
Cell Lines, Antibodies, and Reagent-The androgen-sensitive prostate cancer cell line LNCaP was obtained from American Type Culture Collection (Manassas, VA) (40). The androgen-insensitive prostate cancer cell line LNln3, which is an isogenic derivative of LNCaP, was obtained from Dr. Curtis Pettaway, the University of Texas M. D. Anderson Cancer Center (41). Dr. Pettaway also provided the LNPro5 cell line, an isogenic derivative of LNCaP derived from the fraction of the cells that have the propensity to stay in the mouse prostate upon orthotopic implantation of LNCaP in nude mice (41). All cell lines were routinely cultured in RPMI 1640 supplemented with 10% fetal bovine serum, 2 mM L-glutamine, and antibiotics. The following antibodies were used in the immunoblotting experiments: (a) anti-Bcl-2, a mouse monoclonal antibody; (b) anti-Par-4, a rabbit polyclonal antibody; and (c) anti-WT1, a mouse monoclonal antibody (Santa Cruz Biotechnology, Santa Cruz, CA). The secondary antibody was either anti-rabbit or anti-mouse IgG conjugated to horseradish peroxidase (Amersham Biosciences). Enhanced chemiluminescence reagents were purchased from Kirkegaard & Perry Laboratories (Gaithersburg, MD).
Cell Growth Assay-Cell survival was determined using the cell titer MTS/phenazine methosulfate cell proliferation assay (Promega, Madison, WI). Briefly, log phase growing prostate cancer cells were cultured in flat-bottomed 96-well plates in RPMI 1640, 10% fetal bovine serum medium overnight. Cells were plated at a density of 2 ϫ 10 3 per well for LNCaP and LNln3 and incubated with L-ATRA at concentrations ranging from 1 nM to 10 M. The effects of L-ATRA on growth were compared with untreated cells, which were taken to represent 100% growth.
The well contents were gently washed, and treatment was added in 100 l of medium. The number of live cells 96 h thereafter was determined by the uptake of the MTS/phenazine methosulfate dye and measured on a max Kinetics microplate reader (Molecular Devices, Sunnyvale, CA) at A 490 nm . For each treatment, triplicates were run, and each experiment was performed three times.
Western Blot Analysis-LNCaP and LNln3 cells were plated in T75 flasks and were either treated with L-ATRA (10 M) for 96 h or left untreated. Following this, the cells were collected by trypsinization and washed in PBS. Then cell pellets were lysed in 1% Nonidet P-40 lysis buffer (150 mM NaCl, 1% Nonidet P-40, 0.5% sodium deoxycholate, 0.1% SDS, 50 mM Tris, pH 8.0) to which 2 mM phenylmethylsulfonyl fluoride, 3.3 g/ml leupeptin, and 5 g/ml pepstatin were added fresh before use. Lysis was carried out at 4°C for 30 min. Cells were then centrifuged at 14,000 rpm for 10 min, and the supernatants were collected. Fifty micrograms of protein from each cell line was loaded on precast SDS/ Tris/glycine gels (Bio-Rad), and after electrophoresis, proteins were transferred to nitrocellulose membranes, which were blocked in 5% nonfat dry milk and blotted with the appropriate primary antibody. Membranes were then incubated with the appropriate secondary antibody linked to horseradish peroxidase and developed using the ECL kit according to the manufacturers' protocols.
Fluorescence-activated Cell Sorter Analysis-LNCaP and LNln3 cells were cultured in T75 flasks and either treated with 10 M L-ATRA or left untreated. Cells were collected by trypsinization 96 h after treatment and washed in PBS. Subsequently, cells were fixed in 70% ethanol overnight at Ϫ20°C. After removal of the ethanol by centrifugation, cells were incubated with propidium iodide containing RNase A (1 mg/ml) at 37°C for 15 min. DNA fluorescence of propidium iodidestained cells was measured by flow cytometric analysis (Coulter Epics Profile, Coulter Corp., Miami, FL).
Immunofluorescence Microscopy-LNCaP and LNln3 cells were cultured in 2-chambered Falcon culture slides in the presence or absence of 10 M all-trans-retinoic acid for 96 h. The cells were fixed for 10 min in methanol:acetone (50:50) at Ϫ20°C. Following permeabilization with 0.1% Triton X-100 in PBS, the cells were incubated with rabbit anti-Par-4 polyclonal antibody (Santa Cruz Biotechnology, Santa Cruz, CA) in PBS containing 2% fetal calf serum, overnight at 4°C. Subsequently, the cells were incubated with anti-rabbit rhodamine antibody (Santa Cruz Biotechnology, Santa Cruz, CA), mounted in anti-bleaching solution, and analyzed by laser confocal microscopy Olympus Fluoview FV500.
Real Time Quantitative PCR-Duplicate samples of 1 l of each cDNA were used as a template. Real time quantitative PCR was performed using the ABI PRISM 7700 sequence detection system (PE Applied Biosystems, Wellesley, MA). We used primers and a probe generated by the program Primer Express (PerkinElmer Life Sciences; kindly provided by Dr. Michael Andreeff, the University of Texas M. D. Anderson Cancer Center). Bcl-2 primers were F (forward) 5Ј-GAG-GAGCTCTTCAGGGACGG-3Ј and Bcl-2 R (reverse) 5Ј-GGTGCCGGT-GCAGGTACTCA-3Ј. TaqMan probe (FAM-labeled 5Ј-AGGATTGTGGC-CTTCTTTGAGTTCGGTG-3Ј) was added to the PCR mixture to a final concentration of 200 nM. GAPDH primers were F (forward) 5Ј-GAAG-GTGAAGGTCGGAGTC-3Ј and GAPDH R (reverse) 5Ј-GAAGATGGT-GATGGGATTC-3Ј and GAPDH TaqMan probe fluorescence labeled 5ЈF 5Ј-CAAGCTTCCCGTTCTCAGCC-3Ј (PerkinElmer Life Sciences) were used as internal controls. Amplification conditions were as follows: 50°C for 2 min, 95°C for 10 min, and then 40 cycles at 95°C for 15 s and 60°C for 60 s. GAPDH was co-amplified with Bcl-2.
Northern Blot Analysis-Isolation of total RNA was accomplished using the TRIzol reagent (Invitrogen). The probe for Bcl-2 was generated with the 8.0-kb cDNA fragment from the Bcl-2 expression plasmid p442 (gift from Dr. T. J. McDonnell, the University of Texas M. D. Anderson Cancer Center) by cutting it with EcoRI. The probe thus generated was random primer-labeled with 32 P using the Megaprime TM DNA-labeling system (Amersham Biosciences). The GAPDH probe has been described previously (11) and was used as a loading control. Prehybridization and hybridization were performed at 62°C for 30 min overnight. The blot was subjected to low and high stringency washes using the buffers provided in the NorthernMax TM kit for Northern blots (Ambion Inc.; Austin, TX).
Transfection and Generation of Stable Clones Expressing Par-4 -LNln3 prostate cancer cell line was stably transfected with cytomegalovirus Par-4 expression construct expressing the full-length Par-4 using the calcium phosphate method (26). Stable clones expressing Par-4 were selected using the G418 antibiotic. The clones were analyzed for Bcl-2 and Par-4 expression by immunoblotting and were analyzed for the transcript of Bcl-2 by Northern blot and quantitative real time PCR.
WT1 Antisense-For inhibition of WT1 expression, we used antisense (AS) deoxyoligonucleotides (5Ј-GTCGGAGCCCATTTGCTG-3Ј) that were complementary to the region spanning the initiation codon of the human WT1 (42). The P-ethoxy-modified oligonucleotide was incorporated into a dioleoylphosphatidylcholine liposome formulation as described previously (43). Briefly, LNln3 cells were plated in 6-well plates (1 ϫ 10 5 per well), and 10 M antisense was added. Cells treated were harvested after 3 days of WT1 antisense incubation. Expression of Bcl-2 and WT1 proteins was determined by Western blotting.
Isolation of Nuclear Proteins Binding to Biotinylated Probes-Oligonucleotides representing the WT1-binding site on the bcl-2 promoter sequence were synthesized, and 5 g of the sense strand was 3Ј-labeled using the terminal deoxynucleotidyltransferase (Promega) and biotin-14-dATP (Invitrogen) (34,44). The biotin-labeled sense strand was annealed to its complementary antisense strand and purified over a Sephadex column. The concentration of purified oligonucleotide was measured by absorbance at 260 nm, and equal amounts of annealed oligonucleotide were incubated with streptavidin-coated magnetic beads (Promega) for 30 min at room temperature. Coupling of the oligonucleotides to the beads was measured by absorbance. For the binding reaction, 50 g of nuclear extract alone or preincubated with anti-Par-4 antibody (R334) was used. The beads were captured by a magnet, washed three times with a high salt buffer, and resuspended in Laemmli buffer. The samples were heated at 95°C for 5 min to elute all the proteins, loaded onto a 12% SDS-PAGE, electrophoresed, and transferred to a nitrocellulose membrane. Par-4 was detected using the Par-4 antibody (R334) by immunoblotting as described earlier.
Chromatin Immunoprecipitation-Chromatin immunoprecipitation was done as described earlier with little modification (46). LNln3 cells and LNln3 clones stably expressing Par-4 were plated in a 100-mm dish. About 70% confluent dishes were treated with formaldehyde (1%) for 10 min at 37°C to cross-link proteins to DNA. The cells were washed twice with PBS, pH 7.4, containing protease/inhibitor mixture (Roche Applied Science) and then lysed with lysis buffer containing 1% SDS. Sonication of cross-linked chromatin was performed at 30% of maximum power with two rounds of 10-s pulses so that chromatin fragments thus obtained ranged from 500 to 2000 bp in size. Soluble chromatin was subjected to overnight immunoprecipitation with either anti-WT1 antibody or anti-Par-4 antibody. A portion of the chromatin solution was kept to check the amount of input DNA in different samples before immunoprecipitation. Following immunoprecipitation and elution, the eluent was heated to 65°C for 6 h to reverse the cross-link. Phenol/ chloroform extraction was performed, and the supernatant was ethanol-precipitated. DNA thus obtained was resuspended in 10 mM Tris, 1 mM EDTA, pH 8.0, and subjected to PCR. LNCaP and LNln3 cells cultured in the presence or absence of 10 M of all-trans-retinoic acid for 24 h were also subjected to chromatin immunoprecipitation with the anti-Par-4 antibody. Forward and reverse primers selected for the bcl-2 promoter are as follows: bcl-2 5Ј-AAAACTGTCTGCTTGCGCGGTA; bcl-2 3Ј-GTAGCTTCCTAGGTGTCCGCG; GenBank TM accession number AC021803 (Homo sapiens).
Due to high GC content (36,182-36,560), PCR of the immunoprecipitated chromatin sample has been performed a little upstream (about 900 bases) adjacent to the predicted WT1-binding site (37,426 -37,443). For that matter the modified sonication was done to have the fragments in the range of 500 -2000 bp.
Forward and reverse primers representing a site upstream of the WT1-binding site, used as negative controls, are as follows: bcl-2 5Ј-AGGTTGGGGCCATGGTTTACT; bcl-2 3Ј CAGCCTGGGTGACAGACT-GATAC. The negative primer was designed targeting the site further upstream and is about 3664 bases away from the known WT1-binding site on the bcl-2 promoter. The PCR products were resolved on a 1.5% agarose gel and stained with ethidium bromide.
Plasmid Constructs-The bcl-2 promoter constructs are a generous gift from Linda Boxer, Division of Hematology, Stanford University School of Medicine, Stanford, CA. The bcl-2 5Ј promoter (P1) deletions were inserted into pBluescript II KS-plasmid from Stratagene. These deletions all end Ϫ1287 bp from the bcl-2 translation start site. The SacI site at Ϫ1287 has been changed to a PstI site, which is followed by a sequence of multiple cloning site of pBluescript II KSϪ and then the luciferase gene sequence. The LB124 is a P1 promoter deletion containing the previously identified WT1-binding sites. The LB360 is also a P1 promoter deletion lacking the WT1-binding sites. These constructs have been described previously (34,47). Full-length Par-4 expression plasmid is a 1.4-kb EcoRI fragment of Par-4 cDNA subcloned in the EcoRI site of pCB6 driven by cytomegalovirus promoter. The Par-4 plasmids are a generous gift from Vivek Rangnekar (26).
Transient Transfection and Luciferase Assays-Transfections were performed in 6-well plates using FuGENE 6 (Roche Applied Science) according to the manufacturer's direction. The transfected DNA mixture consisted of 4 g of the full-length construct of human bcl-2 gene 5Ј-promoter region cloned upstream of the firefly reporter gene and 400 ng of pRL-Tk (Promega), which served as an internal control for transfection efficiency. pRL-Tk directs expression of the Renilla luciferase (RL) using the herpes simplex virus thymidine kinase promoter. A minimal bcl-2 P1 promoter lacking the WT1 sites described previously was also used. Where appropriate the transfection mixture also contained either the full-length Par-4 expression plasmid or an empty expression vector (2-4 g). Cell extracts were prepared 24 h posttransfection and assayed for firefly and Renilla luciferase activity using the stop and glo kit (Promega) as directed by the manufacturer.

RESULTS
The Vitamin A Derivative, All-trans-retinoic Acid, Causes a Down-modulation of the Bcl-2 Protein and Transcript in the Androgen-independent Prostate Cancer Cell Lines-Our initial

Par-4 Regulates Bcl-2 Expression
investigations of the regulation of Bcl-2 expression in prostate cancer have focused on signals that are known to down-modulate Bcl-2. One known biological agent that has been shown to down-modulate Bcl-2 in various systems is ATRA (37,38).
Retinoids are known to induce and repress numerous genes that modulate cell growth and differentiation and to exert their anti-cancer action by inducing cells to undergo apoptosis (48). The relatively slow progression of prostate cancer and its association with vitamin A deficiency set precedence for use of ATRA as a therapeutic agent (49). It has been shown previously (37) that L-ATRA, but not free ATRA, induces apoptosis in follicular lymphoma cell lines. This is accompanied by a decrease in the expression of Bcl-2 and an increase in the expression of Bax, a pro-apoptotic molecule. In prostate cancer, ATRA in conjunction with liarazole, a cytochrome p450 inhibitor, induces apoptosis accompanied by down-modulation of Bcl-2 (38). Prostate cancer cells are known to display an increased rate of ATRA clearance and metabolism. In order to circumvent the problem of increased clearance and metabolism of ATRA, we used L-ATRA for our studies.
In an attempt to modulate Bcl-2 expression, we treated the prostate cancer cell lines with ATRA. We first examined the effects of L-ATRA on the proliferation of prostate cancer cell lines. L-ATRA induced a dose-dependent growth inhibition in the androgen-insensitive line, LNln3, but not in the androgensensitive line, LNCaP (Fig. 1A).
Microscopic examination of LNln3 cells treated with 10 M L-ATRA for 96 h revealed many flattened cells with membrane blebbing and condensed nuclei, which are changes typical of cells undergoing apoptosis. No such changes were observed for the LNCaP cell line following treatment with L-ATRA (Fig. 1B).
L-ATRA induced a 50% decrease of the Bcl-2 protein in LNln3 cells, as determined by densitometric scanning. No changes in Bcl-2 expression were observed in the LNCaP cell line ( Fig. 2A).
The Bcl-2 message was also down-modulated in LNln3 cells corresponding to the observed decrease of the Bcl-2 protein, following L-ATRA treatment. No decrease in the Bcl-2 transcript was observed in the LNCaP cell line (Fig. 2B and Table I).

Decrease in Bcl-2 Expression Correlates with G 1 Arrest and Apoptosis in Androgen-independent Prostate Cancer Cell
Lines-Previous studies (37) in follicular lymphoma have demonstrated that decreasing Bcl-2 leads to apoptosis in these cells due to alteration of the Bcl-2:Bax ratio. In solid tumors, including prostate cancer, decreased Bcl-2 is associated with enhanced sensitivity to chemotherapeutic drugs (7,8). Hence, we next determined whether the decrease observed in Bcl-2, at the protein and transcript level, exerted an effect on the cell cycle profile of the prostate cancer cells. The cell cycle profile of the LNCaP cells remained unaltered following 96 h of treatment with L-ATRA (Fig. 3A). A significant G 1 arrest (19%) was seen for the LNln3 cell line and was accompanied by a decrease in the S phase (14.3%) and by the appearance of a sub-G 0 /G 1 population (15.5%), as compared with none seen for the untreated cells (Fig. 3B). These data demonstrate that downmodulating Bcl-2 in the androgen-independent prostate cancer cells produces an effect in terms of growth arrest and apoptosis in these cells. These data are in agreement with previous reports (8,9) that demonstrated in vivo that decreasing Bcl-2 in prostate cancer is of significance to the progression and pathogenesis of this disease.
Altered Par-4 Localization and Induction of WT1 Expression Accompany Decreased Bcl-2 Expression-Because the downmodulation of Bcl-2 in response to L-ATRA treatment was a late event (occurring 96 h after treatment) and because the bcl-2 promoter does not have any retinoic acid response element, we next examined the potential involvement of Par-4. The rationale for investigating this is based on the strong correlation observed for the expression of Bcl-2 and Par-4 in normal prostate basal epithelia. In addition, a previous study documented that ectopic expression of Par-4 results in downmodulation of Bcl-2 in the mouse fibroblast cell line NIH3T3 and in the prostate cancer cell line PC3 (24). Kinetics done for Par-4 expression following treatment with 10 M ATRA did not reveal any change in Par-4 expression at various time points tested in the two cell lines (data not shown).
However, Par-4 has a bipartite localization that may contribute to the different interactions and functionality of this protein (26,27,29,30). The localization of Par-4 following L-ATRA treatment was analyzed in the LNCaP and LNln3 cells using immunofluorescence microscopy. In LNCaP cells Par-4 was found to be cytoplasmic and nuclear in its localization. Following treatment with L-ATRA, Par-4 remains in the cytoplasm (Fig. 4, A and D). Par-4 is observed to be mainly cytoplasmic in the LNln3 cells; however, following treatment with L-ATRA for 96 h, a rise in concentration of Par-4 in the nucleus is observed, which correlated with the decreased Bcl-2 expression and transcript in these cells (Fig. 4, E and H). The concentration of Par-4 in the nucleus accompanied by a decrease in Bcl-2 expression correlates well with the nuclear localization of Par-4 in the luminal/secretory epithelial cells of the normal prostate that lack Bcl-2 expression.
Par-4 via its leucine zipper is capable of interacting with two different domains of WT1. Upon binding to the exon 5 domain of WT1, Par-4 acts as a co-activator of WT1-mediated transcriptional activation, whereas upon binding to the zinc finger domain of WT1, Par-4 augments the transcriptional repression and inhibits the transcriptional activation being mediated by WT1 (27,36). WT1 is known to bind to two different sites on the bcl-2 promoter to activate as well as repress transcription of this promoter (34,35). Hence, we next examined the effects of L-ATRA treatment on WT1 expression. WT1 expression was increased in LNCaP and LNln3 cell lines following L-ATRA treatment (Fig. 5). However, LNCaP cells did not show any down-modulation in the Bcl-2 expression, which may be attributed to the unchanged Par-4 localization in this cell line following L-ATRA treatment, as compared with LNln3 cells. Data from this set of experiments suggest that altered localization of Par-4 and induction of WT1 expression may be the two events required for successful down-modulation of Bcl-2 in the prostate cancer cells.
Ectopic Expression of Par-4 Down-modulates Bcl-2 Protein and Transcript-Based on the strong correlation observed between Par-4, WT1, and Bcl-2 from our L-ATRA study, we next examined the relationship between these molecules. Previous studies (24) have shown an inverse correlation between Par-4 and Bcl-2 expression. We generated stable clones expressing Par-4 in the LNln3 cell line, and we examined the effect of this on the Bcl-2 protein and message. Bcl-2 protein expression was found to be down-modulated in LNln3 clones stably overexpressing Par-4 (Fig. 6A). The Bcl-2 transcript was assessed by Northern blotting (Fig. 6B) and real time quantitative PCR (Fig. 6C) and was found to be decreased, suggesting a transcriptional mechanism for the Par-4-mediated down-modulation of the Bcl-2 protein that was previously not defined. By using Western blotting, we also assessed the expression of WT1 in the stable clones expressing Par-4. WT1 expression was increased in the clones stably overexpressing Par-4 (Fig. 6D). The induction in WT1 expression observed following ectopic expression of Par-4 corresponds to that accompanying L-ATRAinduced decrease in Bcl-2 in the same cell line. Taken together, these data suggest that WT1 expression needs to be induced above basal levels in the LNln3 cell line for Bcl-2 to be down-modulated.
Blocking Endogenous WT1 Results in an Increase in Bcl-2 Protein-To characterize the involvement of WT1 in the regulation of Bcl-2 in the LNln3 cell line, we blocked the expression of the endogenous WT1 protein using antisense (Fig. 7). We then determined the effect of WT1 inhibition on Bcl-2 expression. Bcl-2 expression was increased following inhibition of WT1 expression (Fig. 7). These data indicate that endogenous Par-4 requires high levels of WT1 to cause a decrease in Bcl-2 expression.
Par-4 Is Part of a Protein Complex Interacting with the WT1binding Site on the bcl-2 Promoter-To characterize further the relationship between Par-4, WT1, and Bcl-2, we used a previously identified WT1-binding site (Ϫ1460 relative to ATG) on the bcl-2 P1 promoter as bait to pull out potential interacting molecules (34). This bcl-2 promoter sequence was labeled with biotin using terminal deoxynucleotidyltransferase and incubated with nuclear extracts from the LNln3 cell line. Following this, the samples were subjected to high salt washes to remove any nonspecific binding. The washed complex was pulled out, using streptavidin conjugated to magnetic beads, with the help of a magnet. The proteins binding to the promoter were eluted by boiling in Laemmli's buffer and subjected to SDS-PAGE, and Western blotting for Par-4 was performed using anti-Par-4 antibody. Indeed, we observed that Par-4 is present in a complex interacting with the WT1-binding site on the bcl-2 promoter (Fig. 8). The Par-4-specific antibody (R334) has been shown previously (36) to disrupt the interaction between Par-4 and WT1. We could not detect Par-4 in nuclear extracts preincubated with the R334 antibody and then incubated with the biotin-labeled WT1-binding site on the bcl-2 promoter. Hence, these data would suggest that Par-4 interacts with the previously identified WT1-binding site on the Bcl-2 promoter, possibly through WT1. These results also show the presence of Par-4 on the bcl-2 promoter, reinforcing transcriptional control of the bcl-2 gene by Par-4.

Par-4 and WT1 Proteins in LNln3 Cells
Bind to the bcl-2 (Ϫ1460) WT1 Site in Vitro-To ascertain whether both Par-4 and WT1 interact with the bcl-2 promoter, previously defined WT1-binding sites on the bcl-2 P1 promoter (Ϫ1460 and Ϫ1807) were used as probes in EMSA experiments (34,35). The bcl-2 (Ϫ1807) oligonucleotide site did not show any complex formation upon incubation with nuclear extracts from LNln3 cells (Fig. 9A, lane 6). Two complexes were formed with the bcl-2 (Ϫ1460) oligonucleotide (Fig. 9A, lane 8). Competition with increasing amounts of cold competitor demonstrated that both complexes were specific (Fig. 9A, lanes 9 and 10). To determine the presence of Par-4 and WT1 in the complexes formed, nuclear extracts from LNln3 cells were incubated with the Ϫ1460 probe along with antibodies directed against Par-4 and WT1. The anti-Par-4 antibody is directed against the COOH terminus of Par-4 and has been demonstrated previously (36) as a blocking antibody. This antibody caused disappearance of the faster migrating complex and reduction in intensity of the slower migrating complex, thus indicating the presence of Par-4 in both the complexes (Fig. 9A, lane 11). The anti-WT1 antibody caused a reduction in intensity of the slower migrating complex indicating the presence of WT1 (integrated density value for this complex is 76,874 in comparison to the integrated density value for the slower migrating complex in lane 8, which is 100,096, as revealed by spot densitometric analysis). The anti-WT1 antibody is also directed against the COOH terminus of WT1 and has been shown previously (34) to cause reduction in intensity of specific complex containing the WT1 protein. The mutated bcl-2 (Ϫ1460) site showed very weak complex formation (Fig. 9A, lane 4). The IGF-II promoter, P3, containing the WT1-binding site was used as a positive control and demonstrated the formation of a single complex (Fig. 9A, lane 2) (34). These results support the participation of Par-4 and WT1 in the DNA-protein complex formed at the Ϫ1460 site on the bcl-2 promoter.
The Ϫ1460 WT1-binding site was also used as a probe to perform EMSA with nuclear extracts from LNCaP and LNPro5 cell lines (Fig. 9B). The Ϫ1460 site, on the bcl-2 P1 promoter, was not bound by any proteins when LNPro5 nuclear extracts were used (Fig. 9B, 13th lane). The lack of any retarded complex formation with the nuclear extracts of LNPro5 correlates very well with the barely detectable levels of Bcl-2 and Par-4 in this cell line. However, the same nuclear extracts do show formation of DNA-protein complex when the WT1-binding site on the IGF promoter was used as a probe (Fig. 9B, 9th lane). No complex formation was observed upon using a mutated WT1- binding site as a probe (Fig. 9B, 11th lane). DNA-protein complexes were formed upon using nuclear extracts from the LNCaP cell line; however, these were much lower in intensity as compared with the LNln3 cell line and were competed-out upon using 50-fold excess of the cold probe (Fig. 9B, 6th and 7th  lanes). Taken together, these results suggest that Par-4 and WT1 are able to bind the bcl-2 promoter at position Ϫ1460 forming a multiprotein complex, which suggests their cooperation in driving the expression of the bcl-2 gene in the LNln3 and in the lack of the same in the LNPro5 prostate cancer cell line.
Par-4 and WT1 Proteins in LNln3 Cells Bind to the bcl-2 (Ϫ1460) WT1 Site in Vivo-To assess directly the potential significance of physical interaction between Par-4, WT1, and the bcl-2 promoter, we investigated whether WT1 and Par-4 associate on the chromatin of endogenous bcl-2 promoter using the chromatin immunoprecipitation assay. We immunoprecipitated chromatin from the LNCaP, LNln3 cells and the clones of LNln3 stably overexpressing Par-4, using specific antibodies against WT1 and Par-4 or no antibody at all as a control. Genomic DNA fragments bound to WT1 or Par-4 were analyzed by PCR using primers upstream of the Ϫ1460 WT1 site on the bcl-2 promoter. Analysis of genomic DNA immunoprecipitated with anti-WT1 revealed bcl-2 promoter in both the parental LNln3 cells and Par-4 clones (Fig. 10A, 2nd and 3rd lanes,  middle panel). Immunoprecipitates from anti-Par-4 antibody revealed the presence of the bcl-2 promoter in both the parental LNln3 cells and the clone with greater amounts of bcl-2 promoter being associated with Par-4 in the clone as compared with the parental cell line (Fig. 10, 2nd and 3rd lanes, top  panel). The samples subjected to the immunoprecipitation procedure minus the antibody as expected did not pull out any genomic DNA (Fig. 10A, bottom panel). Chromatin immunoprecipitated with anti-Par-4 antibody or anti-WT1 antibody, subjected to PCR using primers representing a non-WT1-binding site on the bcl-2 promoter, did not amplify any DNA, thus demonstrating that Par-4 and WT1 specifically bind to the WT1-binding site on the bcl-2 promoter. Together, these findings strongly support the idea that Par-4 and WT1 interact with a previously defined WT1-binding site on the bcl-2 promoter in the LNln3 cells and that the ectopic expression of Par-4 causes an enhanced binding of the two molecules Par-4 and WT1 at this site. Neither Par-4 nor WT1 were found to be present on the WT1-binding site on the bcl-2 promoter in the low Bcl-2-expressing androgen-sensitive LNCaP cells (Fig.  10A, 1st lane top and middle panels, respectively).
Upon examining LNln3 cells treated with 10 M all-transretinoic acid for 96 h, we observed greater amounts of the bcl-2 promoter pulled down with the chromatin immunoprecipitated with the anti-Par-4 antibody as compared with the untreated cells, indicating enhanced amounts of Par-4 at the WT1-binding site on the bcl-2 promoter that correlated with decreased Bcl-2 protein and transcript observed in these cells following all-trans-retinoic acid treatment (Fig. 10B, 1st and 2nd lanes). Chromatin isolated from the same cells (untreated and treated with ATRA) immunoprecipitated with the anti-Par-4 antibody, and PCR amplified using primers representing non-WT1-binding site on the bcl-2 promoter, failed to amplify any DNA thereby demonstrating that the WT1-binding site on the bcl-2 promoter is required by Par-4 to bind the promoter (Fig. 10B,  5th and 6th lanes).
Transrepression of Transcription by Par-4 from the bcl-2 P1 Promoter in the LNln3 Cell Line-The physical data presented above demonstrate that Par-4 can bind to the bcl-2 promoter via a WT1-binding site. However, to determine whether the WT1-binding site on the bcl-2 P1 promoter had any functional activity in presence of Par-4 in LNln3 cells, transient transfection experiments were performed. The constructs for the transient transfection experiment are illustrated in Fig. 11A. We demonstrated that co-transfection with increasing amounts of full-length Par-4 resulted in decreased activity of the bcl-2 promoter in a dose-dependent manner (Fig. 11B). Par-4 does not mediate its repression when co-transfected with a construct of bcl-2 promoter that lacks the WT1-binding site (Fig. 11B), thus indicating that the WT1-binding site is required by Par-4 to transcriptionally down-modulate Bcl-2 expression in the LNLn3 cell line. DISCUSSION We demonstrated that Par-4 regulates the transcription of the bcl-2 gene by interacting with a previously defined WT1binding site on the bcl-2 promoter. We provide evidence that Par-4 and WT1 are involved in regulating Bcl-2 expression in prostate cancer cell lines and that ectopically expressed Par-4 is involved in the transcriptional repression of Bcl-2 in the androgen-insensitive prostate cancer cell line LNln3. Previous studies have reported an inverse correlation between Par-4 and Bcl-2; however, here we demonstrate the presence of Par-4 at the promoter of a bcl-2 both in vitro and in vivo. We have also demonstrated the presence of Par-4 at the bcl-2 promoter following all-trans-retinoic acid treatment, which correlates with the decreased Bcl-2 protein and transcript, thus defining a mechanism for retinoic acid-mediated decrease of Bcl-2 in prostate cancer cells. So far the regulation of Bcl-2 expression in prostate cancer has not been extensively defined. A recent study (20) has demonstrated the transcriptional activation of Bcl-2 in prostate cancer cell lines by NF-B through B sites on the P2 promoter upon treatment with TNF-␣. We have also demonstrated that Par-4 requires the WT1-binding site on the bcl-2 promoter to mediate repression of bcl-2 gene transcription.
Our data also indicate that an induction in the expression of WT1, over and above endogenous levels, is required for Par-4 to successfully down-modulate Bcl-2. Our study has defined a mechanism of regulating Bcl-2 expression in androgen-insen-sitive prostate cancer cells. The significance of this comes from the fact that previous studies (7)(8)(9) have clearly demonstrated that decreased Bcl-2 levels sensitize prostate cancer cells to chemotherapeutic agents and also delay progression to androgen independence. As such, our data implicate Par-4 and its interacting partner WT1 in the transcriptional regulation of bcl-2 via a previously defined WT1-binding site on the bcl-2 P1 promoter.
In normal prostate epithelia, Par-4 and Bcl-2 are co-expressed in the basal epithelial cells, which do not require androgens to survive. As these cells differentiate to luminal or secretory epithelia, the growth and differentiation of these cells is tightly regulated by androgens, making these cells sensitive to apoptosis upon androgen ablation. The secretory epithelium expresses no Bcl-2 and lower levels of Par-4, which is inducible upon hormonal ablation (4,23,24). An inverse correlation has been reported for Par-4 and Bcl-2 expression in androgen-dependent and androgen-independent prostatic tissues from human prostate cancer xenografts in the rat (the CWR22 and the CWR22R models) (4,23,24). It has also been demonstrated that exogenous apoptotic insults, such as androgen withdrawal from prostate cancer cells, cause a decrease in Bcl-2 expression (4). Androgen withdrawal also results in induction of Par-4 in androgen-sensitive prostatic tissue (25). Previously, ectopic ex- FIG. 9. EMSA of the bcl-2 WT1 sites (؊1460 relative to ATG). A, EMSA of the WT1 site on the IGF promoter (lanes 1 and 2), the mutated WT1-binding site (Ϫ1460) on the bcl-2 promoter (lanes 3 and 4), the WT1-binding site Ϫ1807 on the bcl-2 promoter (lanes 5 and 6), and the WT1-binding site Ϫ1460 on the bcl-2 promoter P1 (lanes 7-12). Each probe is incubated (from left to right) with no nuclear extracts (lanes 1, 3, 5, and 7), nuclear extracts from LNln3 cells (lanes 2, 4, 6, and 8 -12), nuclear extracts from LNln3 cells incubated with 10-and 30-fold excess of the cold competitor (lanes 9 and 10, respectively), nuclear extracts from LNln3 cells incubated with the anti-Par-4 antibody R334 (400 ng) (lane 11), and nuclear extracts from LNln3 cells incubated with the anti-WT1 antibody C-19 (400 ng) (lane 12). B, EMSA of the WT1 site on the IGF promoter (1st and 2nd and 8th and 9th lanes), the mutated WT1-binding site (Ϫ1460) on the bcl-2 promoter (3rd and 4th and 10th and 11th lanes), and the WT1-binding site Ϫ1460 on the bcl-2 promoter P1 (5th to 7th and 12th to 14th lanes). Each probe is incubated (from left to right) with no nuclear extracts (1st, 3rd, 5th, 8th, 10th, and 12th lanes), nuclear extracts from LNCaP cells (2nd, 4th, 6th, and 7th lanes), and nuclear extracts from LNPro5 cells (9th, 11th, 13th, and 14th  lanes). Unlabeled probe representing the Ϫ1460 WT1-binding site on the bcl-2 P1 promoter was added in 10-fold excess for the LNCaP and the LNPro5 nuclear extracts, respectively (7th and 14th lanes). , immunoprecipitated with antibody against Par-4, and the resulting DNA is amplified using the "negative control" primers as described above. The last two lanes represent total input genomic DNA, amplified using the "negative control" primers.
pression of Par-4 in the NIH3T3 mouse fibroblast and in the PC3 prostate cancer cell line has been shown to down-modulate Bcl-2 expression (24). Furthermore, an inverse correlation between Bcl-2 and Par-4 has also been reported in acute lymphocytic leukemia. These observations suggest that Par-4 is an important component of the pathway regulating Bcl-2 expression. However, the molecular pathway(s) leading to Bcl-2 regulation by Par-4 have not been defined.
Apart from translocation and amplification, the bcl-2 gene is also subject to transcriptional, post-transcriptional, translational, and post-translational regulation, which could result in its deregulated expression. We were able to demonstrate a decrease in the bcl-2 transcript corresponding to decreased protein following ectopic expression of Par-4, suggesting that Par-4-mediated down-modulation of Bcl-2 expression occurs at the transcriptional level. Par-4 is a leucine zipper molecule possessing two nuclear localization signals; however, no DNA binding for Par-4 has been demonstrated so far. Par-4 has been identified as a co-activator and a repressor for WT1-mediated transcription. WT1, a zinc finger transcription factor, is known to bind to the bcl-2 promoter and modulate its transcription (27,36). This led us to investigate the role of WT1 in the Par-4-mediated regulation of Bcl-2. We observed an induction in WT1 expression accompanying L-ATRA-mediated downmodulation of Bcl-2 in the LNln3 cell line. Furthermore, the ectopic expression of Par-4 in the LNln3 cell line also caused an induction in the expression of WT1, which led us to believe that Par-4 may require high levels of WT1 protein for the successful down-modulation of Bcl-2. This inference was confirmed when we blocked endogenous WT1 and observed an increase in endogenous Bcl-2 expression indicating that basal levels of WT1 are also involved in repression of Bcl-2 to some extent; how-ever, increasing endogenous WT1 levels over and above the basal magnifies this effect.
Par-4 is known to regulate the transcriptional activity of WT1 depending on which WT1 domain to which it binds (27,36). By binding to the zinc fingers of WT1, Par-4 can augment the transcriptional repression and inhibit the transcriptional activation being mediated by WT1. Upon binding to the exon 5-encoded domain of WT1, Par-4 has been shown to act as a co-activator of WT1-mediated transcription and is involved in rescuing cells from apoptosis. WT1 binds to two different GCrich sites on the bcl-2 P1 promoter, Ϫ1460 and Ϫ1807, relative to the translation start site, and can both repress and activate transcription of the bcl-2 gene (33)(34)(35).
We have shown both in vitro and in vivo that both the Par-4 and the WT1 proteins are a part of the complex that binds to the Ϫ1460 site on the bcl-2 P1 promoter, thus suggesting that Par-4 mediates transcriptional control of the bcl-2 gene. The Ϫ1460 site on the bcl-2 P1 promoter was identified as a site where WT1 binds and represses transcription of the normal bcl-2 allele in the t(14:18) follicular lymphoma. The same site seems to be driving the expression of the bcl-2 gene in the prostate cancer cell lines examined because this site was bound up in the high Bcl-2-expressing cell line LNln3 and was not bound in the LNPro5 cell line, which has barely detectable levels of the Bcl-2 or Par-4 protein and also showed a weak complex formation in the LNCaP cell line, which has lower levels of the Bcl-2 protein than the LNln3 cell line. However, increased levels of WT1 protein, achieved either by L-ATRA treatment or by ectopic expression of Par-4, change the dynamics of the complex at the Ϫ1460 site leading to increased amounts of Par-4 at this site and decreased levels of Bcl-2 in the LNln3 cell line. We have demonstrated in vivo that Par-4 and WT1 are present at the Ϫ1460 WT1-binding site on the bcl-2 promoter in the LNln3 cells, whereas these proteins are barely detectable in the LNCaP cells at the same site. Ectopic expression of Par-4 resulted in an increased amount of Par-4 and WT1 at the Ϫ1460 WT1-binding site on the bcl-2 promoter in vivo. Treatment with all-trans-retinoic acid also caused an increased amount of Par-4 at the Ϫ1460 WT1-binding site on the bcl-2 promoter. By our promoter assays we have demonstrated that increased binding of Par-4 and WT1 at the Ϫ1460 site on the bcl-2 promoter translates into decreased transcription of the bcl-2 gene which requires the WT1-binding site on the bcl-2 promoter.
Hence our results suggest that in the presence of basal or endogenous levels of WT1, Par-4 activates bcl-2 transcription via the Ϫ1460 WT1-binding site of the bcl-2 promoter. However, in the presence of higher levels of WT1, Par-4 represses bcl-2 transcription via the same binding site. It can be speculated based on our data and previous studies (27,36) that in the presence of endogenous levels of Par-4, this protein binds the exon 5 domain of WT1 and behaves as a co-activator of WT1mediated transcription of the bcl-2 gene. However, when the levels of Par-4 in the nucleus are increased, the excess Par-4 may now bind the zinc fingers of WT1, the exon 5 domain being already bound, and inhibit transcriptional activation of bcl-2 thus changing this from an activator complex to a repressor complex. In addition to this, WT1 is known to have numerous isoforms, the functions of which are not totally characterized (31). Hence, there is a need to explore isoforms that may be involved, and whether upon induction of WT1 by L-ATRA or ectopic Par-4 there is a change in isoforms of WT1 that changes the Par-4-WT1 complex at the bcl-2 promoter from an activator complex to a repressor complex. Currently, we are in the process of characterizing these interactions, and future experiments should yield valuable information regarding this effect. FIG. 11. Par-4 trans-represses the bcl-2 promoter activity via a WT1-binding site. A, schematic representation of Bcl-2 luciferase reporter constructs. B, LNln3 prostate cancer cells were co-transfected with bcl-2 reporter constructs (P1 containing both the WT1-binding sites or P1 without the WT1-binding sites) and the pRL-TK vector as an internal control for transfection efficiency. At the same time the cells were also transfected with either an empty vector or increasing amounts of full-length Par-4 (2 or 4 g). Luciferase activity was determined 24 h after the transfection. Non-transfected cells were used as the background. All values were normalized for expression of Renilla luciferase and expressed as X-fold induction relative to the activity of the P1 promoter co-transfected with the empty expression vector control (equal to 100). Shown is a representative experiment of the two that were performed.