Regions of Prostate-specific Antigen (PSA) Promoter Confer Androgen-independent Expression of PSA in Prostate Cancer Cells*

Prostate-specific antigen (PSA) is expressed primarily by both normal prostate epithelium and the vast majority of prostate cancers. Increases in serum PSA during endocrine therapy are generally considered as evidence for prostate cancer recurrence or progression to androgen independence. The mechanisms by which PSA up-regulation occurs in androgen-refractory prostate cancer cells are unknown. In this study, by using LNCaP and its lineage-derived androgen-independent PSA-producing subline, C4-2, we identified two cis-elements within the 5.8-kilobase pair PSA promoter that are essential for the androgen-independent activity of PSA promoter in prostate cancer cells. First, a previously reported 440-bpandrogen-responsive element enhancer core (AREc) was found to be important for the high basal PSA promoter activity in C4-2 cells. Both mutation analysis and supershift experiments demonstrated that androgen receptor (AR) binds to the AREs within the AREc and activate the basal PSA promoter activity in C4-2 cells under androgen-deprived conditions. Second, a 150-bp pN/H region was demonstrated to be a strong AR-independent positive-regulatory element of the PSA promoter in both LNCaP and C4-2 cells. Through DNase I footprinting and linker scan mutagenesis, a 17-bp RI site was identified as the key cis-element within the pN/H region. Data from electrophoretic mobility shift analysis and UV cross-linking experiments further indicated that a 45-kDa (p45) cell-specific transcription factor associates with RI in prostate cancer cells and may be responsible for driving the PSA promoter activity independent of androgen and AR. Furthermore, by juxtaposing AREc and pN/H, we produced a chimeric PSA promoter (supra-PSA) that exhibits 2–3-fold higher activity than the wild type PSA promoter in both LNCaP and C4-2 cells.

Prostate-specific antigen (PSA) 1 is a chymotrypsin-like serine protease synthesized primarily by normal, hyperplastic, and malignant prostatic (1)(2)(3). PSA expression is tightly regulated by androgen through the action of androgen receptor (AR) (2,4,5). Upon binding to androgen, AR translocates into the nucleus and binds to the androgen response elements (AREs) on the PSA promoter, where it interacts with other transcription factors and activates PSA gene transcription. Cleutjens et al. (6,7) have identified three AREs within the 5.8-kb PSA promoter. ARE-I and -II are located within the proximal region of the promoter, whereas ARE-III is contained within a 440-bp strong enhancer element core (AREc) located at Ϫ4.2 kb of the promoter (7)(8)(9). Recently, additional non-consensus AREs have been identified within the AREc enhancer element. This study proposes that androgen regulation of the AREc in prostate cells might involve the formation of AR and prostate-specific factor nucleoprotein complexes on the multiple non-consensus AREs in the enhancer region (10).
PSA is a serum marker for prostate cancer. It has been shown that serum PSA is proportional to tumor volume and correlates positively with the clinical stage of the disease (11). Progression of prostate cancer to androgen independence is commonly associated with a rebound of serum PSA (12). PSA elevation in hormone-refractory prostate tumors has been attributed to: 1) mutations and/or amplifications of AR (13)(14)(15)(16)(17)(18) that broaden its ligand specificity and/or enhance tumor cells' responsiveness to androgen, respectively (19); 2) androgenindependent (AI) activation of the AR by growth factor signaling pathways like insulin-like growth factor-1 and keratinocyte growth factor, which would elicit AR-mediated transcriptional activation (20,21). It has also been demonstrated that AR could "cross-talk" with protein kinase A (PKA) and/or protein kinase C signaling pathways (22)(23)(24)(25), and/or 3) the direct stimulatory action by soluble prostate-specific autocrine factor(s) (PSAF) secreted by hormone-refractory prostate cancer cells (26).
To understand the molecular pathways that may regulate PSA expression by androgen in normal prostate epithelium, and dysregulation during AI progression, much effort has focused on delineating the activities of AR in mediating androgen regulation of PSA expression in prostate cancer cells (2, 6 -8, 27, 28). Due to the limited availability of AI yet PSA-producing prostate cancer cell lines, little is known about the androgenindependent regulation of PSA expression in hormone-refractory prostate cancer cells. The development of the androgenindependent PSA-producing C4-2 cell line from androgendependent parental LNCaP cells (29) has allowed us to study how PSA expression is regulated in an androgen-and growth factor-deprived environment. Like hormone-refractory prostate cancer, C4-2 was shown to be androgen-independent, de-fined here as cells that, when inoculated alone subcutaneously in immunocompromised mice without supporting extracellular matrices or stromal cells, form PSA-secreting tumors in castrated animals (30 -32). PSA expression in C4-2 cells also mimics clinical hormone-refractory prostate cancer in that the cells have acquired the ability to turn on the PSA promoter independent of androgen, thus synthesizing and secreting high amounts of PSA in the absence of androgen stimulus in vitro (32).
In this report, we have taken advantage of the remarkable difference in basal PSA expression between two lineage-related LNCaP cell lines, LNCaP and C4-2, for the analysis of androgen-independent regulation of PSA promoter in prostate cancer cells. The molecular insights gained from this study will lead to the understanding of PSA rebound in hormonal-refractory prostate tumors. Through deletion analysis on PSA promoter, we provide evidence that AREc and a pN/H element within the proximal promoter region are responsible for conferring the higher AI PSA promoter activity in C4-2 cells. An AR-dependent but androgen-independent pathway appears to be involved in AREc activation, whereas a 17-bp RI site within the pN/H was shown to associate with cell-specific transcription factor and regulates the PSA promoter activity independent of AR or androgen.

MATERIALS AND METHODS
Cell Culture and Transfection-LNCaP and C4-2 were cultured in T-medium (33) supplemented with 5% fetal bovine serum. For transfection, cells were grown in phenol red-free and serum-free RPMI 1640 (Life Technologies, Inc.). LNCaP and C4-2 were plated at 3.3 ϫ 10 5 cells/well in six-well plates 2 days before transfection. Plasmid DNAs were introduced into cells by complexing with DOTAP (Roche Molecular Biochemicals). Briefly, 2.5 g of the tested DNA constructs and 0.5 g of the internal control CMV/␤-galactosidase DNA were mixed with 27 l of 20 mM HEPES (pH 7.4) and added to separate tube containing 8 l of DOTAP in 16 l of HEPES with gentle mixing. DNA-lipid complexes were allowed to form for 15 min at room temperature prior to their addition to each well containing 1 ml of serum-free and phenol red-free RPMI 1640 medium. The cells were incubated with the complexes at 5% CO 2 , 37°C for 5 h. DNA-lipid containing medium was then replaced with fresh serum-free and phenol red-free RPMI 1640 medium. Cells were collected after 36 -48 h of additional incubation. All the transfections were carried out in the above serum-free conditions, unless specified otherwise.
Luciferase Assay-Cells were washed with 1 ml of phosphate-buffered saline/well and lysed in 300 l of 1ϫ lysis buffer (Promega, Madison, WI). Cell lysates were vortexed for a few seconds and spun for 3 min. For luciferase activity detection, 20 l of the supernatant was mixed with 100 l of luciferase substrate (Promega) and measured by a luminometer (Monolight 2010, Analytical Luminescence Laboratory, Sparks, MD). For ␤-galactosidase activity detection, 100 l of the supernatant was mixed with an equal volume of 2ϫ ␤-galactosidase substrate (Promega) and incubated at 37°C for 15-30 min. The ␤-galactosidase activity was determined by plate reader at 405 nm wavelength. Data are expressed as relative luciferase activity (RLA), which is defined as luciferase activity normalized to internal control CMV/␤galactosidase activity for transfection efficiency. RLA is expressed as the mean Ϯ standard error of the mean of at least three independent experiments.
Western Blot and PSA Immunoassay-Immunoblotting was performed as described in Sambrook et al. (34). Briefly, proteins were separated on a 7.5% SDS-polyacrylamide gel and transferred to a 0.42-m nitrocellulose membrane. Nonspecific binding was blocked with 5% nonfat milk in phosphate-buffered saline for 30 min. Antibody (PG-21-21A) against the amino terminus of AR was used as the primary antibody. Secondary antibody against the anti-AR IgG (horseradish peroxidase anti-rabbit antibody) (Amersham Pharmacia Biotech) was used in a 1:2000 dilution. Detection was performed with ECL (Amersham Pharmacia Biotech). PSA proteins were detected by the IMX (Abbott Laboratory, Chicago, IL) immunoassay method as described previously (35).
Linker-scanning Mutagenesis-Both L4 and A3a were generated by replacing the respective AREs with the GAL4 sequence (cggagtactgtcctccg). Briefly, two primers (L4, gtcctccgttgtcttgacagtaaacaaa, agtactccgtccagagtaggtctgttttc; A3a, gtcctccgttgcaaggatgcctgctttac, agtactccgatccaggcttgcttactgtc) each with half of the GAL4 site were designed to run PCR with the uncut AREc/TATA. P2 was generated similarly, except that pN/H was used as the template instead. The primers used for P2 mutants are 5Ј-agtactccgtggcacccagaggctgacca-3Ј and 5Ј-gtcctccgtcctggggaatgaaggtttta-3Ј. Platinum Pfx enzyme (Life Technologies, Inc.) was used in the experiments to generate blunt-end PCR products. The PCR products were purified from 0.8% agarose gel and ligated at room temperature with a rapid ligation kit (Roche Molecular Biochemicals). Clones were screened by PCR with a GAL4 primer (cggagtactgtcctccg) and GLprimer2 (ctttatgtttttggcgtcttcc) of the pGL3-basic vector. All clones were confirmed by DNA sequence.
Gel Retardation Assay-Polyacrylamide gel electrophoresis-purified oligos (Sigma-Genosys, Woodlands, TX) were annealed by heating up to 95°C and slowly cooled down to room temperature. The oligo sequences used as probes or competitors are as follows: ARE-III, 5Ј-tcgacgaggaacatattgtatcgagtcga-3Ј (7); A3a, 5Ј-tcgacacagtaagcaagcctgggtcga-3Ј (10); RI, 5Ј-ggtgccagcagggcaggg-3Ј; SP-1, 5Ј-attcgatcggggcggggcgagc-3Ј; AP-1, 5Ј-cgcttgatgactcagccggaa-3Ј. The double-stranded probes were end-labeled with [␥-32 P]ATP by using T4 polynucleotide kinase (New England Biolabs, Beverly MA). Nuclear extracts were prepared from cells growing under 3-day complete serum-free condition (36). 100,000 cpm of labeled probe and 5.6 -10 g of nuclear extracts were incubated with binding buffer containing 10 mM Tris-HCl (pH 7.5), 50 mM NaCl, 0.5 mM EDTA, 0.5 mM dithiothreitol, 4% glycerol, 1 g of poly(dI-dC) (Amersham Pharmacia Biotech), and 1 mM KCl at 30°C for 30 min. The samples were subjected to electrophoresis at room temperate on a 4% nondenaturing polyacrylamide gel in 0.5% TBE at 35 mA for 2 h. For experiments using SP-3 antibody (Santa Cruz Biotechnology, Santa Cruz, CA), 4 g of antibody was added to the reaction mixture for 30 min after the incubation period of the probe and nuclear extracts; for experiments using AR antibody (CW2), nuclear extract and 2 g of antibody were pre-incubated at 37°C for 1 h before the addition of probe. In competition experiments, competitor oligos were incubated with nuclear extracts for 30 min at 30°C before the addition of the probe.
DNase I Protection Assay-pN/H was labeled by Klenow (New England Biolabs) according to Current Protocols (36). The binding reactions for DNase I footprinting were described previously (37,38). 20,000 cpm of labeled probe and 10 g of nuclear extracts from LNCaP or C4-2 cells were incubated in 13 l of buffer containing 12.5 mM HEPES, pH 7.9, 12.5% glycerol, 5 mM MgCl 2 , 70 mM KCl, 0.2 mM EDTA, 60 mM mercaptoethanol, 0.5 mg/ml bovine serum albumin, and 200 ng of poly(dI-dC). After 30 min at 30°C, 2 l of DNase I (Promega) was added to the reactions. The cleavage reactions were terminated after 1 min by addition of 100 l of stop buffer containing 400 mM sodium acetate (pH 5.2), 0.2% SDS, 10 mM EDTA, 50 g/ml yeast tRNA, and 100 g/ml proteinase K (Roche Molecular Biochemicals). The mixtures were incubated at 50°C for 15 min, extracted with phenol/chloroform, and precipitated with ethanol. Precipitates were dissolved in formamide loading buffer (Amersham Pharmacia Biotech) and analyzed on 6% sequencing gels.
UV Cross-linking-Bromodeoxyuridine-substituted probe was prepared according to the Current Protocols (36). Briefly, a 50-bp synthetic oligo (containing the RI site) was annealed to a complementary 15-bp synthetic oligo and filled in with Klenow fragment (New England Biolabs). The binding reaction was set up in 96-well plate, containing 100,000 cpm labeled probe, 16 g of C4-2 serum-free nuclear extract, and 2 g of poly(dI-dC). The reaction was then incubated at 30°C for 30 min, followed by UV irradiation (ϳ305 nm) at 4°C for 30 min. The samples were subjected to electrophoresis on a 4 -12% SDS-gradient gel (Invitrogen) at room temperature for 1 h, fixed with 30% methanol, 3% glycerol solution for 20 min, and then dried. For competition experiments, 400 ng of the competitor was incubated with nuclear extracts for 30 min at 30°C prior to the addition of probe.

RESULTS
Up-regulation of PSA Gene Expression in an Androgen-independent Prostate Cancer Cell Line, C4-2-C4-2, a lineage-derived LNCaP subline, was shown previously to be able to grow in castrated hosts and exhibit metastatic potential in vivo (30 -32). This cell line synthesizes and secretes a higher basal level of PSA than LNCaP cells in the absence of androgen stimulus in vitro (32). This unique feature of C4-2 cells provides an opportunity to study PSA promoter regulation in prostate cancer cells with characteristics mimicking hormone-refractory status. To compare PSA secretion between parental LNCaP and its C4-2 subline in vitro, cells were first cultured in Tmedium supplemented with 5% fetal bovine serum and switched to serum-free and phenol red-free RPMI 1640 medium when they reached 80% confluence. After a 3-day incubation period, cells were counted and medium was collected for immunoassay of the PSA protein. C4-2 cells consistently secreted 4 -5-fold (14.9 Ϯ 1.12 versus 2.89 Ϯ 0.22 ng/ml/10 6 cells) more PSA protein than LNCaP cells into the medium in the complete absence of exogenous androgen and growth factor stimulus over a 3-day period. To investigate whether the upregulation of PSA protein in C4-2 cells occurs at the transcriptional level, a 5.8-kb PSA promoter was inserted upstream to a luciferase reporter gene (p61/pGL3) and transfected into LN-CaP and C4-2 cells for transient expression analysis. In parallel with PSA secretion results, the basal PSA promoter activity in C4-2 is 18 Ϯ 3.7-fold higher than that observed in LNCaP cells. Thus, up-regulation of PSA protein expression in C4-2 is due to a high intrinsic basal PSA promoter activity.
Since the PSA promoter contains multiple AREs, the upregulation of PSA promoter activity in C4-2 may be explained by an elevated AR protein level in this AI cell line. A Western blot analysis of AR was performed with the total cell lysate of LNCaP and C4-2 ( Fig. 1). Consistent with our previously published results (32), AR protein was found to be expressed by these two cell lines at comparable levels. Moreover, we have sequenced both the ligand-binding domain and the DNA-binding domain of AR from LNCaP and C4-2 cell lines and have identified the reported single point mutation in the ligandbinding domain (39) in both of these cell lines without additional mutations. 2  pacity between LNCaP and C4-2 cells revealed that both of these cell lines contain high affinity and capacity AR (32). Thus, neither the steady-state level of AR expression nor additional AR mutations play a role in the up-regulation of the PSA promoter activity in C4-2 cells.
Identification of AREc as an Essential Element for the Basal PSA Promoter Activity in C4-2 Cells-Deletion analysis was performed to dissect out the cis-element(s) essential for conferring high PSA promoter activity in C4-2 cells under androgendeprived conditions. Various deletion constructs were generated by restriction enzyme digestion (Fig. 2A). The P61-2 construct containing a deletion between ARE-II and ARE-III retains all of the wild-type promoter activity. In contrast, a 1.5-kb terminal deletion immediately proximal to AREc (p61-5) caused a 50% drop in promoter activity only in C4-2, and not in LNCaP cells (Fig. 2A). It is possible that the deletion has partially destroyed the 5Ј end of the AREc element resulting in a sharp drop of PSA promoter activity in C4-2 cells. A further decrease of promoter activity was observed in C4-2 cells when the terminal deletion extended downstream to include the AREc (p61-4). The 500-bp AREc region was then removed in p61-3 to demonstrate the effect of AREc deletion in PSA promoter activity. As indicated in Fig. 2A, p61-3 has a similar activity to p61-4. The AREc deletion reduced the promoter activity to 14% of the 5.8-kb promoter in C4-2 cells, but again without appreciably affecting the basal PSA promoter activity in LNCaP cells. Hence, AREc is a crucial cis-element for maintaining the androgen-independent PSA promoter activity in C4-2 cells.
AREc was shown to be important for androgen induction of PSA promoter activity in LNCaP cells (7,8). However, AREc does not seem to have an important role in regulating the PSA basal promoter activity in LNCaP cells ( Fig. 2A). On the contrary AREc is indispensable for the maintenance of high basal PSA promoter activity in C4-2 cells (Fig. 2A). The differential activity of AREc between the two cell lines was demonstrated by inserting AREc upstream to an artificial TATA box (AREc/ TATA). We chose a simple TATA box promoter because transfection experiments indicated that this promoter alone did not respond to AR or androgen in the absence of AREs in prostate cancer cells. Consistent with the above observations, the AREc element has about 10 times higher activity in C4-2 than LN- FIG. 3. The role of ARE on androgen-independent regulation of the PSA promoter. A, mutation of AREs in AREc cause AREc activity decrease in C4-2 cells. AREc/TATA was used as the template for linker-scanning mutagenesis experiment (see "Materials and Methods"), ARE-III or A3a was mutated into a GAL4 sequence (solid box) in L4/TATA and A3a/TATA, respectively. The RLA of the mutant constructs were presented as relative activity to the wild type AREc/TATA, which was set to be 100% in LNCaP and C4-2 cells, respectively. B, mutation of ARE-III has an effect similar to that of AREc deletion on the PSA promoter activity. The effects of ARE-III multiple point mutations (indicated by an asterisk in AREmIII), and AREc deletion (p61-3) was assessed in the context of the full-length PSA promoter. C, EMSA shows ARs bind to ARE-III and A3a in C4-2 cells. AR-DNA complex is indicated by the arrow, and the supershifted complexes are indicated by the asterisks. Both ARE-III and A3a gel shift reactions were run on the same gel, but with different exposure time.
CaP cells (Fig. 2B). The 440-bp AREc was shown to be a strong tissue-specific enhancer element that exhibits high androgen responsiveness only in PSA-positive cells. In addition to containing multiple AREs, the AREc potentially contains other prostate-specific regulatory elements (7,8). It has been hypothesized that liganded AR binds co-operatively to the cluster of non-consensus AREs in the enhancer core, where it assembles into a nucleoprotein complex with prostate-specific factor(s) FIG. 4. The activity of a short proximal PSA promoter in prostate cancer cells. A, a 150-bp pN/H element confers high activity in LNCaP and C4-2 cells. The activities of several p61/pGL3 deletion constructs were shown here as relative activities to the wild-type p61/pGL3. The p61/pGL3 activity in LNCaP was set to be 100%. B, pN/H activity is independent of androgen. 1 nM R1881 (ϩ group) was added to the transfected cells for 36 h. The basal activity of pN/H was set to be 100% in LNCaP with approximately 4500 -5500 RLA. C, AREc and pN/H could co-operate synergistically in activating PSA promoter in prostate cancer cells. In sPSA, the 440-bp AREc was inserted directly upstream to the pN/H region of pN/H construct. Experiments were carried out as described under "Materials and Methods." and acts synergistically to activate PSA enhancer activity in PSA-producing prostate cancer cells (10,40).
Functional AREs Are Required for the High Basal AREc Activity in C4-2 Cells-To address the role of AREs in the differential regulation of AREc/TATA activity in LNCaP and C4-2 cells, two ARE mutants were generated by replacing specific regions of ARE in the AREc enhancer core with a 17-bp GAL4 sequence. Mutant L4/TATA has the ARE-III sequence replaced, while mutant A3a/TATA has the A3a sequence (38) replaced. Fig. 3A demonstrates that mutation of ARE-III caused a 70% decrease in the basal AREc/TATA activity, while mutation of A3a resulted in an 85% decrease in the basal AREc/TATA activity in C4-2 cells. None of the AREs mutations appreciably affect the basal AREc/TATA activity in LNCaP cells. These results were further confirmed when multiple point mutations at ARE-III in p61/pGL3 (AREmIII) mimicked the effect of AREc deletion (p61-3) on the basal PSA promoter activity in C4-2 cells (Fig. 3B). Therefore, we concluded that the AREs within the enhancer core are essential for its activity in C4-2 cells.
We employed two approaches to investigate whether AR is involved in activating the AREc in C4-2 by binding to the AREs. First, anti-androgen Casodex (23) was used to block the AR transcriptional activity in C4-2 cells. Casodex was able to bring AREc/TATA activity down to 11.78 Ϯ 1.75% of the control level. Second, by using AR antibody in EMSA, we demonstrated that AR indeed is the transcription factor in C4-2 cells that binds to ARE-III and A3a sites. When radiolabeled ARE-III and A3a were used to perform EMSA with C4-2 nuclear extracts (Fig. 3C), DNA-protein complexes were observed that could be supershifted by anti-AR antibody (CW2) into two higher molecular weight species. Although two bands were observed with ARE-III oligo, the bottom band appears to be nonspecific because it also diminished with control SP-3 antibody. Together these results showed that the AR in C4-2 cells is highly active and the co-operative binding of AR to the multiple AREs on the enhancer core allows the core to be transcriptionally active even in the absence of androgen.
The PSA Proximal Promoter Contains a 150-bp Positive Regulatory Element-The fact that p61-3 has substantial activity in C4-2 cells and still displays differential activity between LNCaP and C4-2 cells (Fig. 4A) implies that AREc is not the only cis-element that supports the AI PSA promoter activity in C4-2 cells. Detailed analysis of the deletion construct data led us to look for additional positive regulatory element(s) within the 600-bp proximal PSA promoter region. To test this possibility, we first compared the activity of the short PSA promoter-pPA8 (6), which contains the 632-bp proximal PSA promoter region including ARE-I and ARE-II, between LNCaP and C4-2 cells. As shown in Fig. 4A, pPA8 exhibits 6-fold higher basal activity in C4-2 than in LNCaP cells. Through terminally deleting the pPA8 until 150-bp upstream of the TATA box (just beyond the ARE-I), pN/H was generated. In comparison to the full-length PSA promoter (p61/pGL3), pN/H was able to retain a substantial basal PSA promoter activity in C4-2 cells (ϳ40% that of the full-length PSA promoter). pN/H construct also demonstrated differential activity (Fig. 4B) between LNCaP and C4-2 cells (ϳ5-fold higher activity in C4-2 than LNCaP cells). These results strongly suggest that like AREc, pN/H is a positive regulatory cis-element that contributes to the AI PSA promoter activity in C4-2 cells. In contrast to AREc, the activity of pN/H is not regulated by androgen, for the addition of R1881 did not further enhance pN/H activity in either LNCaP or C4-2 cell lines (Fig. 4B). By juxtaposing these two positive-regulatory elements together, we not only reconstituted the activity of the full-length promoter, we have created a 590-bp sPSA promoter that exhibits 2-3-fold higher activity than the wild-type promoter (Fig. 4C). The superior activity of sPSA implies that AREc could co-operate with pN/H in maintaining the high steady state basal PSA promoter activity in C4-2 cells.
The Identification of RI Site-To locate the crucial cis-element within the pN/H contributing to its high basal activity in C4-2 cells, we first performed linker-scanning mutagenesis. Briefly, a 17-bp fragment in the pN/H was replaced systematically with a GAL4 binding site, a panel of eight mutant constructs (P1-P8) were generated, with P1 being the closest to the TATA box and P8 being the most upstream to the TATA box. Among the eight mutant constructs, only P1 and P2 showed significant activity decrease in LNCaP and C4-2 cells. P1 has about 50% of the wild type pN/H activity (data not shown), while P2 has only 2-7% of the pN/H activity (Fig. 5A). In addition, when a single copy of the P2 element was inserted upstream to a simple TATA box (P2/TATA), it was able to exhibit significant activity in both LNCaP and C4-2 cells, with 5-6-fold higher activity in C4-2 than in LNCaP cells (Fig. 5B). However, P2/TATA showed no activity in the PSA-negative prostate cancer cell line (PC3) when compared with the PSApositive LNCaP and C4-2 cells (Fig. 5B). These results indicated that P2 element is able to exhibit high activity only in PSA-positive prostate cancer cells. Therefore, we hypothesized that P2 element is associated with a cell-specific transcription factor.
In order to precisely map the binding site of this cell-specific transcription factor, we then performed DNase I footprinting assay with the pN/H fragment. In Fig. 5C, two regions (RI and RII) are distinctly protected by a protein factor present in LNCaP (lane 3) and C4-2 (lane 2), but not in PC3 (lane 1) nuclear extracts. The specificity of the protection was confirmed by successful competition with cold P2 element (lane 7), whereas other nonspecific DNA (lanes 4 -6) was not able to compete away the protections observed. RI region coincides with the P2 element over a 11-bp sequence (agcagggcagg), so we have identified a core sequence within the RI region that is crucial for its activity in prostate cancer cells. RII coincides with the region covered by P5 mutant; as mentioned before, this mutation did not affect the pN/H activity significantly. It is unclear at this point why both footprints could be competed away by P2 element. The RI binding factor may complex with transcription factor that binds to RII, so by squelching RI binding factor with cold P2 element, the RII binding factor was also prevented from binding to RII. Together, these results showed that RI site is occupied by a cell-specific transcription factor present in PSA-positive prostate cancer cells, and it regulates PSA promoter activity in an androgen-and AR-independent manner.
To further characterize the transcription factor that binds RI site, EMSA was performed to investigate the mobility pattern of the RI-binding protein. Two protein-DNA complexes were observed to associate with RI site (Fig. 6A); only one appeared to be specific (complex A), for it was competed away by specific competitor RI, but not affected when nonspecific competitor AP-1 was used. Moreover, complex A seems to be more abundant in C4-2 nuclear extracts (lane 5) than in LNCaP nuclear extracts (lane 2), and it was not observed when PC3 nuclear extracts were used (lane 8). Since a single copy of RI site (P2/TATA) shows higher activity in C4-2 than in LNCaP, and it  has no significant activity in PC3 (Fig. 5C), we concluded that complex A is a strong candidate for regulating the RI activity in LNCaP and C4-2 cells. A data base consensus site search indicated that RI site shares high homology with the SP-1 transcription factor family binding site. Therefore, we compared the mobility patterns of complex A to the SP-1 binding factors (Fig. 6B). With either LNCaP or C4-2 nuclear extracts, the SP-1 consensus site consistently yielded three distinctive bands, which corresponded to SP-1, SP-2, and SP-3, protein factors (lanes 2 and 8). They appeared to have a completely different migration pattern than complex A. Since complex A migrated closely to where the SP-3 factor is on the gel, supershift experiments were carried out with SP-3 antibody to determine whether complex A is SP-3. In the positive control (lanes 4 and 10), SP-3 protein was completely supershifted by the antibody, while complex A was not. Furthermore, unlabeled SP-1 competitor cannot compete away the complex A band (lanes 6 and 12). Therefore, it appears that the factor that binds to RI may not belong to the SP-1 transcription factor family.
Next, we determined the molecular weight of RI binding factor by covalently linking it to a radiolabeled RI site in UV cross-linking experiments. In Fig. 6C, an intense band was observed at 60 kDa; by subtracting the mass of the probe (15 kDa), RI binding factor has an apparent mass of 45 kDa. The specificity of the photoadduct was rigorously determined by competition with both specific and nonspecific competitors, nonspecific competitors like AP-1 and SP-1 were not able to compete away the RI-protein complex, while specific competitor RI could successfully compete away the DNA-protein complex.
In conclusion, we have identified a novel regulatory element (RI site), which is associated with a 45-kDa cell-specific transcription factor (p45) in prostate cancer cells. The increased association of p45 to RI site in C4-2 cells could account for the higher basal PSA promoter activity in AI prostate cancer cells in the absence of androgen. FIG. 6. A, a cell-specific factor binds RI in PSA positive prostate cancer cells. RI was used as the probe, and the competitors used are AP-1 and RI. Lane 1 contains no nuclear extract, lanes 2-4 contain LNCaP nuclear extracts, lanes 5-7 contain C4-2 nuclear extracts, and lanes 8 -10 contain PC3 nuclear extracts. B, complex A could not be supershifted by SP-3 antibody. LNCaP nuclear extracts were used from lanes 2-7, and C4-2 nuclear extracts were used from lanes 8 -13. In lanes 2-4 and 8 -10, SP-1 was used as the probe. In lanes 5-7 and 11-13, RI probe was used. C, UV cross-linking experiment indicated a 45-kDa protein (p45) binds to RI specifically. The apparent molecular mass of the protein-DNA complex is about 60 kDa; after subtracting the molecular mass of the probe, the RI-binding protein is about 45 kDa in size.

DISCUSSION
Hormone-refractory prostate cancer is one of the most detrimental diseases affecting men in the United States. Until now, the progression of prostate cancer from an androgen-dependent to an androgen-independent status has been poorly understood. Rebound of serum PSA is most consistently observed in prostate cancer patients with evidence of androgen-independent progression. Among the gene products specifically expressed by the human prostate, the transcription regulation of the PSA gene has been widely studied (6 -8, 10). However, in most of the studies conducted, investigators have focused mainly on the androgen regulation of the PSA promoter in an androgen-dependent/responsive prostate cancer cell line, LN-CaP (2,4). By doing so, they have identified several AREs in various regions of the PSA promoter (6 -8, 28). In the present report, we describe for the first time the identification of ciselements in the PSA promoter whose up-regulation is associated with AI progression of prostate cancer.
In this study, we provide evidence to indicate that the androgen-independent activation of PSA promoter in androgenindependent prostate cancer cells (C4-2) involves two distinct regions, AREc and pN/H (Figs. 2 and 4). The requirement of these two regions suggested there are two different pathways involved in the up-regulation of PSA promoter activity in C4-2 cells. One pathway clearly requires AR because the binding of AR to the AREs within the AREc appears to be a prerequisite for the high activity of the AREc in C4-2 cells (Fig. 3). It is possible that AR is activated through a ligand-independent pathway where growth factors might be involved. It has been reported that in DU145 prostate cancer cells (with co-transfection of AR), insulin-like growth factor could stimulate ARmediated reporter gene transcription to the same extent as a synthetic androgen, methyltrienolone (R1881). In the same study, Culig et al. (20,21) demonstrated that keratinocyte growth factor and epidermal growth factor could also activate artificial promoter with two AREs. Moreover, activation of the protein kinase C and/or the PKA pathway has also been reported to stimulate AR activity in the absence of androgen (22)(23)(24). AR was shown to be activated by a PKA activator (forskolin) in the absence of androgen. This activation can be blocked by a PKA-specific inhibitor or anti-androgen (Casodex, flutamide). The two pathways just mentioned are not necessarily mutually exclusive. It is conceivable that growth factors like insulin-like growth factor would signal through the protein kinase cascade, which in turn could increase the transactivating activity of AR, either by modification of the protein itself (25,41) or by enhancing the interactions between AR and its co-activators (42,43) (Fig. 7). The fact that AREc is highly tissue-specific (7,8) suggests that in addition to AR, other prostate-specific co-activators are also involved in the androgen-independent regulation of AREc in C4-2 cells. Activated AR is one of the key factor that interacts with prostate-specific transcription factor(s), and together they associate with the AREc and assemble into a highly active AREc enhanceosome complex (10,40) in C4-2 cells. Thus, the aberrant activation of AR and/or its co-activators may be one of the mechanisms that contribute to the re-elevated PSA promoter activity in androgen-refractory prostate cancer cells.
The second pathway mediated by pN/H element appears to be androgen-and AR-independent, for pN/H activity is not affected by the addition of androgen at all (Fig. 4B). Through DNase I footprinting (Fig. 5C) and linker-scanning mutagenesis (Fig. 5A), we have mapped the location of the first AI regulatory element (RI) in the PSA promoter. We believe that RI regulates the basal PSA promoter activity by the binding of a 45-kDa (p45) unknown transcription factor in prostate cancer cells. The absence of p45 in PC3 cells (Figs. 5C and 6A) indicates that its expression is restricted to PSA-positive cells like LNCaP and C4-2. The fact that C4-2 nuclear extracts consistently showed a higher level of RI-p45 complex in EMSA suggested that increased association of p45 to RI site is the mechanism by which PSA promoter is activated in C4-2 cells independent of androgen. The increased association observed in C4-2 cells could be explained as follows: 1) increased gene transcription of p45 in C4-2 cells, resulting in a higher level of p45 in the cells, or 2) p45 is activated in C4-2 by unknown growth signals, so it could bind the RI site better. This activation could be contributed by any modification (e.g. phosphorylation) or mutation of the protein that allows it to associate with DNA with higher affinity, or translocate into the nucleus more efficiently. In conclusion, p45 may represent a new class of androgen-independent prostate-specific transcription factor that regulates PSA expression in prostate cancer cells.
Previously, our laboratory demonstrated that a PSAF secreted by C4-2 cells could up-regulate PSA production in LN-CaP cells (26). We observed that PSA synthesis and secretion were induced in LNCaP cells upon the addition of C4-2 conditioned media. It is possible that autocrine factor(s) like PSAF secreted by C4-2 cells might have similar effects to the growth factors in activating AR and its interaction with co-activators (42)(43)(44), hence creating a highly transcriptionally active AREc even in the absence of exogenous androgen and growth factors. At the same time, the autocrine factor(s) could also enhance the activity of p45 in C4-2 cells (Fig. 6A). The synergistic effect observed in the chimeric sPSA promoter (Fig. 4C) indicates that AR and p45 could work cooperatively in activating the PSA promoter in an androgen-independent manner in C4-2 cells (Fig. 7). In addition to PSA promoter regulation, ligandindependent activation of AR has long been implicated as a culprit in the androgen-independent prostate cancer progression (45); therefore, it is conceivable that activated AR together with p45 in AI prostate cancer cells might give them growth advantages in an androgen-deprived condition. Hence, prostate cancer cells through the production of autocrine factors could have bypassed the requirement of androgen for growth and survival. This then allows them to progress into a hormone refractory stage and become androgen-independent. Further efforts to identify p45 may provide additional insights into FIG. 7. A proposed mechanism of androgen-independent activation of PSA promoter in androgen-independent prostate cancer cells.
prostate cancer progression and PSA regulation in men with hormone refractory prostate cancer.