Antagonism between PTEN/MMAC1/TEP-1 and Androgen Receptor in Growth and Apoptosis of Prostatic Cancer Cells*

PTEN/MMAC1/TEP-1 (PTEN) tumor suppressor and androgen receptor play important roles in prostatic tumorigenesis by exerting opposite effects on homeostasis of prostatic epithelium. Here, we describe a mutual repression and selective dominance between PTEN and the androgen receptor (AR) in the growth and the apoptosis of prostatic cancer cells. On the one hand, PTEN and an inhibitor of phosphoinositide 3-kinase repressed the transcriptional activity of the AR as well as androgen-induced cell proliferation and production of prostate-specific antigen. On the other hand, androgens protected prostate cancer cells from PTEN-induced apoptosis in an AR-dependent manner. Whereas the repression of the transcriptional activity of the AR by PTEN is likely to involve the down-regulation of AKT, androgens protected prostate cancer cells from PTEN-induced apoptosis without an effect on AKT activity, demonstrating a differential involvement of AKT in the interaction between PTEN and the AR. Our data suggest that the loss of PTEN function may induce tumorigenesis through unopposed activity of the AR as well as contribute to the resistance of prostate cancers to androgen ablation therapy.

PTEN/MMAC1/TEP-1 (PTEN) tumor suppressor and androgen receptor play important roles in prostatic tumorigenesis by exerting opposite effects on homeostasis of prostatic epithelium. Here, we describe a mutual repression and selective dominance between PTEN and the androgen receptor (AR) in the growth and the apoptosis of prostatic cancer cells. On the one hand, PTEN and an inhibitor of phosphoinositide 3-kinase repressed the transcriptional activity of the AR as well as androgen-induced cell proliferation and production of prostate-specific antigen. On the other hand, androgens protected prostate cancer cells from PTEN-induced apoptosis in an AR-dependent manner. Whereas the repression of the transcriptional activity of the AR by PTEN is likely to involve the down-regulation of AKT, androgens protected prostate cancer cells from PTENinduced apoptosis without an effect on AKT activity, demonstrating a differential involvement of AKT in the interaction between PTEN and the AR. Our data suggest that the loss of PTEN function may induce tumorigenesis through unopposed activity of the AR as well as contribute to the resistance of prostate cancers to androgen ablation therapy.
Androgens are responsible for the development, maintenance, and regulation of male phenotype and reproductive physiology. These activities are mediated through the AR 1 (1)(2)(3)(4), which belongs to the steroid/thyroid receptor superfamily, a group of ligand-regulated transcription factors (5). Like other members of the family, the AR protein is modular in nature and composed of an amino-terminal A/B region, a DNAbinding domain and a "hinge" region in the middle, and a hormone-binding domain at the carboxyl terminus. Whereas the amino-terminal A/B region contains the major transcriptional activation function, a weaker activation function in the hormone-binding domain also contributes to the total tran-scriptional activity. The hormone-binding domain represses the activation functions until androgens bind to it and relieve the repression by inducing the formation of a conformation suitable for the interaction with transcriptional cofactors (6 -8). In addition, the polymeric stretches within the amino-terminal region (9) and receptor phosphorylation (10,11) also contribute to the AR transcriptional activities.
Besides their established physiological functions, androgens are implicated in multiple pathological processes including prostate cancer (PCa), which is the most commonly diagnosed malignancy and second only to lung cancer in the mortality rate of American males (12). Chronically maintained androgen level sustains the total prostate cell number by both stimulating the rate of proliferation and inhibiting the rate of death of prostate epithelial cells (13); both lead to an increase in the cell number. To maintain the homeostasis of prostate epithelium, there must exist a "brake" system that opposes these effects of androgens.
PTEN tumor suppressor is a 403-amino acid phosphoprotein/ phospholipid dual-specificity phosphatase (14,15). Somatic mutation of PTEN is a common event in diverse human cancers including PCa (14,15), and heterozygous deletion of PTEN in mice leads to neoplasm in multiple tissues including the prostate (16,17). The significance of the phosphoprotein phosphatase in tumor suppression is largely unknown, although one known substrate is focal adhesion kinase (18). In contrast, multiple findings support a role of the lipid phosphatase activity in PTEN-mediated tumor suppression (19 -23). PTEN lipid phosphatase catalyzes the dephosphorylation of phosphatidylinositol 3,4,5-trisphosphate (PI(3,4,5)P 3 ) (24), resulting in inactivation of the downstream protein kinase B, also named AKT, activity (21,22). Ectopic PTEN expression in PTEN-null PCa cells induced cell cycle arrest and apoptosis (22,25); both activities are opposite to those of androgens, suggesting that PTEN may provide the brake to balance the functions of AR in prostate cells.
In the current study, we provide experimental evidence that the activities of PTEN and AR are antagonistic in PCa cells. Interestingly, PTEN and AR do not simply shut down each others activity but differentially antagonize and selectively dominate each other in PCa cell apoptosis and proliferation.
ELISA and Colorimetric MTT Assay-LNCaP cells were plated in RPMI 1640 containing 10% FBS at 1 ϫ 10 5 cells/well in 96-well plates for MTT assays. After cells attached, they were starved in RPMI 1640 with 1% charcoal-stripped FBS for 48 h and treated with synthetic androgens or PI3K inhibitors. The medium was collected and PSA levels were determined using the Tandem-E PSA ImmunoEnzyMetric Assay Kit (Hybritech Inc., San Diego, CA) following manufacturer's protocols. The absorbance at 405 nm was measured using a UV spectrophotometer. PSA concentration (ng/ml) was determined based on a standard curve generated with PSA controls of known concentrations provided by the kit. MTT assays were performed as described (32). The plates were read on a MRX microplate reader (DYNEX Technologies, Chantilly, VA) using a test wavelength of 595 nm.
Apoptotic Assays-Transfected cells were washed with PBS and fixed in 3.7% formaldehyde/PBS at room temperature for 10 min. The viability of transfected cells in each well was determined by counting the total number of green cells in each well under a fluorescence microscope. For the demonstration of apoptotic cells, fixed cells were stained at room temperature for 15 min with 4Ј,6Ј-diamidino-2-phenylindole (DAPI) at a concentration of 1.5 mg/ml in the VECTASHIELD Mounting Medium (Vector laboratories, Burlingame, CA). Green and blue fluorescence were observed with a Leitz Orthoplan 2 Microscope. Representative micrographs were captured by a CCD camera with the Smart Capture Program (Vysis, Downers Grove, IL). The apoptotic index of GFPpositive cells was determined by scoring 300 GFP-positive cells for chromatin condensation and apoptotic body formation.
Immunoblotting Analysis-To determine the expression of AR, duplicate wells of PC3 cells transfected in parallel to the cells prepared for transcriptional assays were pooled and lysed in the same buffer as described above for kinase assays. The lysate was mixed with one-sixth of 6ϫ SDS-PAGE sample buffer, heated for 10 min at 100°C, and separated on 8% SDS-PAGE. To determine the expression of HA-AKT, half of the cell lysate prepared for kinase assays was processed similarly as for AR and separated on 10% SDS-PAGE. After separation on SDS-PAGE, samples were transferred to a nitrocellulose membrane, probed with the PG-21 anti-AR (Upstate Biotechnology, Lake Placid, NY) or the 12CA5 anti-HA antibodies, and visualized using ECL as described (28).

Repression of AR Transcriptional Activity by PTEN and a PI3K Inhibitor and Relief of the Repression by a Dominantly
Active AKT-To test whether PTEN inhibits AR transcriptional activity, we transfected into AR-negative, PTEN-null PC3 cells (22) an AR expression vector, pLENhAR, together with a PTEN expression vector or a control vector expressing a phosphatase-inactive PTEN. The transcriptional activity of AR was measured by assaying the luciferase activity from a cotransfected AR reporter, AREe1bLuc, in which luciferase expression is under the control of synthetic androgen response elements (AREs) placed in front of the simple adenovirus E1b promoter. The effect of PTEN on the AR activity was measured by comparing the AR activity in cells transfected with PTEN to those transfected with the control vector. Because of the concern that co-transfected PTEN may alter the expression of AR by affecting the activity of the promoter or the enhancer of the AR expression vector, the reporter activity was normalized with ␤-gal activity from co-transfected pLEN␤gal vector in which ␤-gal expression is under the control of human metallothionin-II promoter and SV40 enhancer, the same regulatory sequences used by the AR vector. As shown in Fig. 1A, AR basal activity was low but the activity was significantly induced by treatment with R1881, a stable synthetic androgen agonist. Whereas the basal activity was not affected, R1881-induced AR activity was repressed, but not abolished, by PTEN in a dosagedependent manner.
Because both AR and ␤-gal were expressed using the same vector, the way we normalize the reporter activity should have eliminated the potential interference caused by either variation in transfection efficiency or PTEN-induced alterations in the promoter activity of the AR expression vector. However, the lower AR activity in the presence of PTEN could arguably be because of lower level of AR protein as a result of decreased protein stability or reduced efficiency of protein translation. To test whether the level of AR protein was altered by PTEN, PC3 cells were transfected with pCMVhAR and pCMV␤, which express the receptor and ␤-gal, respectively, under the control of the stronger cytomegalovirus (CMV) promoter, permitting the analysis of AR protein level and transcriptional activity in the same experiment. The AR activity from pCMVhAR was decreased by PTEN to the same degree as demonstrated for pLENhAR (Fig. 1A). Parallel Western blotting demonstrated that PTEN, in the presence of R1881, did not decrease the AR protein level (Fig. 1B). These data demonstrate that PTEN repressed the transcriptional activity of the AR per molecule. In the absence of R1881, PTEN expression caused a decrease in the level of AR protein. The decrease, based on later studies, was evidently because of the death of transfected cells that expressed PTEN to high levels.
To further substantiate the AR repression by PTEN, we performed a time course study as shown in Fig. 2. In cells transfected with mutant PTEN, R1881-induced a dramatic AR activation at times as early as 12 h post-transfection and the continuous treatment proportionally increased the AR activity over time. Whereas the basal AR activity in the absence of R1881 was not affected, the proportional increase in AR activity as a result of prolonged R1881 treatment was almost eliminated by PTEN.
In theory, the AR repression could be mediated through either the protein or lipid phosphatase activity of PTEN. Because the lipid phosphatase activity has a well established role in tumor suppression, we decided to first investigate the role of the lipid phosphatase activity in the AR repression. PTEN lipid phosphatase is known to decrease the level of PI(3,4,5)P 3 by catalyzing its dephosphorylation. Thus, we examined whether the decrease of PI(3,4,5)P 3 by other means such as inhibition of its synthesis will have the same effect on AR activity as PTEN expression. AR was transfected into PC3 cells and treated with either R1881 alone or co-treated with LY294002, a synthetic inhibitor for PI3K. As shown in Fig. 3A, LY294002 repressed the AR activity in a dosage-dependent manner. A time course study showed that the percentage of AR repression by LY294002 was best observed at the shortest time point (Fig.  3B). This is different from the PTEN data (Fig. 2) and probably reflects the fact that LY294002 exerts its effect through PI3Kmediated signaling pathways (which usually takes just minutes), and the effectiveness of synthetic drugs decreases over time because of the issue of stability and active clearance by cells. These studies indicate that the PTEN-induced AR repression is most likely mediated through the PI(3,4,5)P 3 phosphatase activity of PTEN.
Because AKT is the major downstream target of PI(3,4,5)P 3 , we next examined the involvement of AKT in PTEN-induced AR repression. PC3 cells were transfected with PTEN, AR, and a dominantly active AKT of which the activity cannot be re-pressed by PTEN or a kinase-inactive AKT. The AR repression by PTEN was analyzed as shown in Fig. 4. Whereas PTENinduced AR repression occurred in the presence of the kinaseinactive AKT, co-expression of the dominantly active AKT blocked the AR repression by PTEN. These data suggest that the down-regulation of PI(3,4,5)P 3 by PTEN lipid phosphatase, and the subsequent inactivation of AKT kinase mediated the AR repression, establishing the role of the PTEN lipid phosphatase in AR repression.
Impairment of the Biological Activities of Endogenous AR by PTEN and the PI3K Inhibitor-So far, we have demonstrated an inhibitory effect of PTEN and the PI3K inhibitor on the activity of transiently transfected AR with a reporter constructed with synthetic AREs. To determine whether PTEN also represses endogenous AR activity with natural promoters, we transfected PSALuc into PTEN-null but AR-positive LNCaP cells (22) and examined the effect of PTEN and the PI3K inhibitor on endogenous AR activity. PSALuc is an AR reporter in which luciferase expression is under the control of the promoter of human PSA, a PCa marker, which is transcriptionally regulated by AR through complex AREs (29). As shown in Fig. 5A, the transcriptional activity of the endogenous AR in cells transfected with the PSALuc was increased by R1881, and R1881 induction was blocked by PTEN. This demonstrates that PTEN represses endogenous AR activity on natural promoters and thus is not limited to ectopic AR or synthetic AREs.
To rule out the possibility that the AR repression was caused by nonspecific repression of LNCaP cell transcription by PTEN, we transfected into LNCaP cells a Gal-VP16 expression vector and a Gal4 reporter and measured the activity of the fusion activator in the presence or absence of PTEN. As shown in Fig.  5B, the reporter was inactive in the absence of Gal-VP16, and the co-expression of the fusion activator induced the activity dramatically. More importantly, the Gal-VP16 activity was not inhibited by PTEN. These data demonstrate that the AR repression was not caused by a nonspecific effect of PTEN on LNCaP cell transcription.
To determine the biological consequence of the observed repression of AR transcriptional activity, the effect of PI3K inhibitor on androgen-induced PSA production and cell proliferation was examined in LNCaP cells. LNCaP cells were treated with either R1881 alone or co-treated with LY294002 for indicated times. Culture medium from the treated cells was collected and analyzed for PSA level by ELISA. As shown in Fig.  6A, R1881 treatment stimulated PSA production and the stimulation was inhibited by the co-treatment with LY294002. Sim- ilar to PSA production, R1881 induced an increase in cell number as measured by MTT assays, and this increase was blocked by co-treatment with LY294002 (Fig. 6B). LY294002 did not consistently decrease the basal PSA level or LNCaP cell numbers in the absence of R1881. These analyses demonstrate that the repression of AR transcriptional activity by the PI3K inhibitor impaired the biological functions of endogenous AR. by the number of GFP-positive cells in representative micrographic fields (Fig. 7B, panels 1-3).

AR-dependent Protection of PTEN-induced Apoptosis by Androgens in PCa Cells
To confirm that the decreased viability of PTEN-transfected cells is the result of cell apoptosis, cells were fixed after transfection and stained with DAPI, and the nuclear morphology of transfected cells was examined for features of apoptosis under a fluorescence microscope that permits the simultaneous visualization of both blue and green fluorescence. As shown in Fig.  7B, panels 4 -6, as representative micrographs, cells trans-fected with the control vector displayed a normal morphology similar to surrounding non-transfected cells (Fig. 7B, panel 4). Cells transfected with PTEN without R1881 treatment frequently displayed an apoptotic morphology (Fig. 7B, panel 5). Similar to controls, most cells transfected with PTEN but treated with R1881 showed a normal morphology (Fig. 7B,  panel 6). Apoptotic index, as determined by counting apoptotic cells in 300 green cells per sample 24 h after treatment, was 5% for controls, 20% for cells transfected with PTEN without R1881 treatment, and 5% for cells transfected with PTEN but treated with R1881 (Fig. 7C). These analyses show that PTEN induced apoptosis in LNCaP cells, and this PTEN function was blocked by androgen treatment.
Because of the presence of endogenous AR in LNCaP cells, LNCaP cell experiments did not show that the androgen protection of PTEN-induced apoptosis is mediated through the AR. To determine whether the androgen effect on apoptosis is ARdependent, AR-negative PC3 cells were transfected with or without AR, and the effect of androgen on PTEN-induced apoptosis was examined as in LNCaP cells. As shown in Fig. 8A, R1881 had no effect on the viability of PTEN-transfected cells in the absence of ectopic AR expression. After co-transfection with AR, both PTEN-induced decreases in cell viability (Fig.  8A) and increases in apoptotic index (Fig. 8B) were blocked by R1881, demonstrating that the androgen protection of PTENinduced apoptosis was mediated through the AR.
To further analyze the anti-apoptotic function of AR, a time course study was performed (Fig. 8C). In this study, the number of transfected (green) cells peaked at 24 h post-transfection, and R1881 protected PC3 cells from PTEN-induced death similarly at all three tested time points, presumably because of the synchronized expression of AR and PTEN proteins in co-transfected cells.
Lack of an Androgen Effect on AKT Activities in PCa Cells-Our data in Fig. 1 showed that the PTEN repression of AR depended on the down-regulation of AKT activity. In addition, co-expression of the dominantly active AKT blocked PTENinduced apoptosis in LNCaP cells (Fig. 9A), demonstrating that PTEN-induced apoptosis was mediated through AKT downregulation. So we investigated the possibility that androgens might protect apoptosis by regulating AKT activity. Western blotting with anti-[phospho-Ser 437 ]AKT antibody did not detect an androgen effect on endogenous AKT activity in LNCaP cells (data not shown).
LNCaP cells are PTEN-null and contain high levels of endogenous AKT activity. Although androgens do not have a direct effect on endogenous AKT activity, it may affect the negative regulation of AKT by PTEN. So we next examined whether androgen treatment blocked PTEN-induced AKT down-regulation. We transfected a HA-tagged wild-type AKT into LNCaP cells with the PTEN expression vector and analyzed the AKT activity by in vitro immunocomplex kinase assays. As shown in Fig. 9B, in either the presence or absence of R1881, the kinase activity of AKT was dramatically decreased by PTEN expression whereas the level of AKT protein was slightly reduced (Fig. 9B). The data demonstrate that R1881 did not block the down-regulation of AKT activity by PTEN in LNCaP cells. DISCUSSION Our studies demonstrated an antagonistic interaction and selective dominance between PTEN and AR in the proliferation and apoptosis of PCa cells as well as a differential involvement of AKT in the interaction. The lack of an androgen effect on both AKT activity and its down-regulation by PTEN in PCa cells suggests that androgens protect PTEN-induced apoptosis either by activating a survival pathway that is independent of AKT as suggested by a previous study (33) or by activating the same survival pathway at steps downstream of AKT.
It appears paradoxical that PTEN-or LY294002-induced repression of AR transcriptional activity was sufficient to block androgen-induced proliferation and PSA production but unable to override the protective effect of androgens on apoptosis. One possibility is that AR-dependent protection of apoptosis by androgens might be mediated through non-genomic effects of the receptor. Our data clearly showed that the androgen effect on PTEN-induced apoptosis is AR-dependent. Consistent with our data, reported studies demonstrated that the "decoy" of ARE triggered apoptosis in LNCaP cells (34), suggesting that the anti-apoptotic effect of androgens is mediated through the genomic effect of the AR. Because the AR transcriptional activity was repressed but not abolished by PTEN or LY294002 in our experiments, it is likely that the androgen induction of genes involved in promoting cell proliferation, and those in apoptosis protection require different amounts of AR transcriptional activity.
Because our studies indicate that androgen target genes involved in proliferation and apoptosis protection may have a differential sensitivity to cellular status of PI(3,4,5)P 3 signal pathway, it would be interesting to determine whether AR could still mediate the induction of anti-apoptotic genes under conditions when androgen-induced cell proliferation is blocked by the suppression of PI(3,4,5)P 3 signaling. Unfortunately, genes mediating the anti-apoptotic effect of androgens in PCa cells remain to be identified. Under the same conditions when PTEN-induced apoptosis of LNCaP cells was blocked by androgens, we did not detect any androgen effect on Bcl-2 expression (data not shown), although Bcl-2 up-regulation by androgens in LNCaP cells had been described in previous studies (35).
Phosphorylation is known to regulate the transcriptional activity of steroid receptors including AR (10,11,36). Based on the consensus sequence for AKT phosphorylation (37), there are two potential AKT phosphorylation sites in AR: Ser 212 in the A/B region and Ser 782 in the ligand-binding domain. Therefore, it is conceivable that AR activation may need the phosphorylation of these sites by AKT and that AR repression by PTEN might be caused by the inhibition of the AKT phosphorylation. However, transfection of the dominantly active AKT into DU145 cells that contain wild type PTEN and low endogenous AKT did not cause either ligand-independent activation nor enhanced the androgen-induced activity of co-transfected AR (data not shown), suggesting that PTEN repression of AR activity may involve more complex processes than a simple inhibition of AR phosphorylation at the potential AKT sites.
Our data are the first to clearly demonstrate a mutual antagonism between PTEN and AR in PCa cells. The mutual antagonism implies that the balance between the function of PTEN and AR may maintain the homeostasis of prostate epithelium in adult males. The demonstration that AR is more active in the absence of functional PTEN suggests that the loss of PTEN function may induce prostatic tumorigenesis by exposing prostatic epithelial cells to unopposed AR activity. Similarly, excessive androgens may induce prostatic tumorigeneis by blocking the apoptosis-promoting function of PTEN. The induction of apoptosis by the restored PTEN expression in LNCaP cells only occurred in the absence of androgen, implying that PTEN mutation or decreased expression may contribute to the resistance of PCa to androgen ablation and that the combinational inhibition of both PI3K/AKT and androgen signals could be an effective approach for the treatment of ARpositive PCa.