INTRODUCTION

Top Abstract Introduction Procedures Results References

Prostatic epithelial cells are dependent on androgen for survival and enter programmed cell death following hormone ablation (1-3), resulting in a massive involution of the prostate gland (4). Hormone ablation also results in the death of most primary prostate tumor cells, but this relief is transient, because recurrent tumors usually develop, and the cells of these recurrent tumors are able to proliferate and metastasize in the absence of serum androgen (5).

Despite the importance of androgens for the growth of prostatic epithelia, the mechanisms of their effect on the stimulation of proliferation and inhibition of cell death remain obscure. Similarly, little is known about how recurrent prostate tumor cells are able to proliferate in the absence of serum androgen. It is known that androgen exerts its biological effects through the androgen receptor, a member of the nuclear steroid hormone receptor family (6). However, target genes of the androgen receptor that promote proliferation remain unknown. In this report, we sought to identify downstream targets of the androgen-responsive proliferative pathway by determining the effects of androgen withdrawal on the cell cycle machinery of androgen-dependent cells. In addition, we determined how the requirement for androgen might be circumvented by misregulation of specific cell cycle pathways.

While in G₁, mammalian cells evaluate growth-promoting and/or growth-inhibitory environmental cues to either progress through the mitotic cell cycle or enter into quiescence (7, 8). Extracellular signals ultimately impinge on cyclin-dependent kinases (CDKs),¹ which dictate transitions into and within the proliferative cell cycle (9, 10). The activity of these kinases is regulated at many levels, including (i) both inhibitory and stimulatory phosphorylation events; (ii) binding to cyclins; and (iii) binding to CDK inhibitors (11, 12).

Cell cycle progression in G₁ is regulated by the activity of cyclin D-, E-, and A-associated CDKs. D-type cyclins (cyclins D1, D2, and D3) are the earliest of the cyclins to accumulate in the cell cycle, and once produced, they associate with and promote the activation of their CDK partners, CDK4 or CDK6 (13, 14). Subsequent to CDK4(6)-cyclin D activation, cyclin E accumulates and activates CDK2 for progression through G₁. Cyclin A, whose expression is initiated temporally after cyclin E in late G₁, associates with both CDK2 (S phase) and CDC2 (late S and G₂). Cyclin A-associated kinases are required not only for progression through late G₁ but also for completion of S phase (10). All three G₁ cyclins shorten the G₁ phase when ectopically expressed and are required for progression from G₁ to S phase (15, 16).

In addition to regulation by cyclin association and phosphorylation, mammalian CDKs are regulated by cyclin-dependent kinase inhibitors (CDIs). These molecules attenuate the kinase activity of CDKs by binding directly with active CDK-cyclin complexes or by competing with cyclin for binding to CDKs (12). The CDIs can be divided into two families, the INK4 family (p16^ink4a, p15^ink4b, p18^ink4c, and p19^ink4d), which act early in G₁ to inhibit the activity of CDK4 and CDK6, and the CIP family (p21^cip1, p27^kip1, and p57^kip2), which inhibit a large number of CDK-cyclin complexes (17).

The principal substrates for G₁ cyclin-associated kinase activity appear to be the retinoblastoma tumor suppressor protein, RB, and the related proteins p107 and p130 (collectively deemed the "pocket proteins") (18). Prior to their phosphorylation by G₁ CDKs, RB, p107, and p130 are believed to prevent progression to S phase by sequestering key proteins required for DNA synthesis, such as the E2F family of heterodimeric transcription factors (19). E2F regulates the expression of a number of genes required for cell cycle expression, for example, cyclin E, cyclin A, and DNA polymerase  (20, 21). Binding of RB to specific E2F family members results in repression of these E2F-responsive genes, thus preventing S phase initiation. Upon phosphorylation of RB, E2F and other S phase promoting proteins are released and activated (19-22). Although cyclin E and cyclin A-associated complexes are hypothesized to phosphorylate substrates in addition to RB, a large body of evidence supports the notion that the main function of CDK4-cyclin D or CDK6-cyclin D complexes is to phosphorylate RB (13). Specifically, CDK4-associated kinase activity is no longer required for cell cycle progression in RB-deficient cells (23). Although initiation of RB phosphorylation is initially catalyzed by the G₁ cyclin-associated CDKs, cyclin B-associated complexes (CDC2-cyclin B), which promote G₂/M transition, are thought to also contribute to the post-S phase maintenance of RB phosphorylation. In fact, phosphorylation of RB is maintained until the completion of mitosis, at which time RB is dephosphorylated to allow the cells to enter a new G₁ phase or to exit the cell cycle (22).

Importantly, although phosphorylation of RB by the G₁ CDKs is required to traverse the G₁-S transition, it is not sufficient for S phase progression. This fact is evident in that (i) RB-negative cells still require growth factors for cell cycle progression (24); (ii) premature G₁ cyclin expression shortens G₁ but does not eliminate the requirement of the cell for growth factors (15, 25); and (iii) specific late-G₁ arrested cells contain hyperphosphorylated RB (26). As such, mechanisms separate from the regulation of RB phosphorylation must contribute to the control of cell cycle progression.

Here, we sought to determine the effects of androgens on the regulation of key cell cycle regulatory proteins and to elucidate cell cycle pathways that may be deregulated to achieve androgen-independent growth. We utilized cultured LNCaP human prostatic adenocarcinoma cells (27), which retain most of the features common to androgen-responsive tumor cells, especially: (i) the expression of both the androgen receptor and prostate-specific antigen (28, 29); (ii) stimulation of proliferation by androgens in vitro (30); (iii) formation of tumors in male mice but only weakly in castrated male mice or female mice (27, 31); and (iv) growth arrest upon androgen removal (32). The results presented provide the first evidence that multiple G₁ elements are responsive to the presence of androgen and act to prevent G₁-S progression in its absence. Surprisingly, these regulatory mechanisms cannot be circumvented through the ectopic expression of G₁ cyclins. However, oncoproteins capable of sequestering pocket proteins rendered prostatic carcinoma cells androgen-independent for proliferation. Thus, although androgen acts as a growth stimulus for prostatic cells, it does so in a manner that is distinct from that of many mitogens.

	EXPERIMENTAL PROCEDURES

Top Abstract Introduction Procedures Results References

Cell Culture-- The human prostatic adenocarcinoma cell line, LNCaP, was obtained from the American Type Culture Collection, and passages 25-39 were used in the experiments described. For regular passage, cells were grown in Improved Minimum Essential Medium (Biofluids) containing 10% heat-inactivated fetal bovine serum (Hyclone) supplemented with 100 units/ml penicillin-streptomycin and 2 mM L-glutamine at 37 °C in a humidified atmosphere of 5% CO₂. For growth in androgen-depleted media, cells were propagated in improved minimum essential medium containing 5% charcoal dextran-treated fetal bovine serum (Hyclone). Dihydrotestosterone (Steraloids), when added, was diluted in ethanol and then medium and used at a final concentration of 0.1 nM (ethanol was less than 0.1% in the medium). Medium containing hormone and/or ethanol was changed every 48 h. For growth analysis, cells were seeded at a density of 2-3 × 10⁴ cells/well in poly-L-lysine-coated 6-well dishes, and medium was changed every 48 h. At the times indicated, cells were washed with Dulbecco's phosphate-buffered saline lacking calcium and magnesium (MediaTech) and harvested using 0.05% trypsin-EDTA solution (Life Technologies, Inc.). Cell numbers were each determined using a hemocytometer and trypan blue exclusion. All conditions were tested in triplicate, and the averages/deviations are shown. Poly-L-lysine coating was carried out by incubating freshly diluted 0.001% poly-L-lysine (Sigma) in H₂O on the desired substrate for 20-30 min in a sterile tissue culture hood. After incubation, poly-L-lysine was completely aspirated, and substrates were thoroughly dried. Poly-L-lysine had no effect on the growth or cell cycle profile of LNCaP cells, under all conditions used (data not shown).

Plasmids-- Plasmid expressing the green fluorescent protein, pGreen Lantern, was obtained from Life Technologies, Inc. pRSV-T-antigen was a gift of Dr. Suresh Subramani (University of California at San Diego). Plasmids expressing the SV40 early region and corresponding mutants (WT-T-Ag, PVU1-T-Ag, K1-T-Ag, and 2831-T-Ag) were kindly provided by Dr. Charles Cole (Dartmouth University) and have been previously described (33). Plasmids WT-E1A, CXDL-E1A, and 30-85-E1A, which express E1A constructs from the cytomegalovirus promoter, were obtained from Dr. David Livingston (Dana-Farber Cancer Institute). Plasmids expressing the E1A splice variants 12 S and 13 S from the cytomegalovirus promoter were obtained from Dr. Gilbert Morris (Tulane University). Plasmids expressing WT-LP, PSM.7-LP, cyclin A, and p16^ink4a from the cytomegalovirus promoter were generous gifts of Dr. Jean Wang (University of California at San Diego) and have been previously described (34). Cyclin D1 and cyclin D3 expression constructs (expressed from the Rous sarcoma virus promoter) were obtained from Dr. Charles Sherr (St. Jude's Children's Research Hospital) (35). The cyclin E expression construct (expressed from a long terminal repeat) was obtained from Dr. James Roberts (Fred Hutchison Cancer Research Center) (36).

Transfections-- For transfection, LNCaP cells were seeded at a density of 3 × 10⁵ cells/well in poly-L-lysine-coated wells of 6-well dishes (or poly-L-lysine-coated coverslips resting in 6-well dishes). After 48 h, wells were washed one time with serum-free and antibiotic-free improved minimum essential medium. Transfection substrate (1 ml of serum-free/antibiotic-free improved minimum essential medium, 30 µg of Lipofectin (Life Technologies, Inc.), and 5 µg of DNA) was then applied and allowed to incubate for 6 h, after which the serum-containing medium indicated was added back. To monitor protein expression, cells were harvested 48 h post-transfection. Co-transfection of pCMV-LP and the cyclin expression constructs were performed with a 1:2 ratio of plasmid DNA. To monitor S phase progression, co-transfections with pGreen Lantern (1.0 µg) and a secondary effector expression plasmid (4.0 µg) were carried out. Cells were labeled 48 h after transfection with cell proliferation labeling reagent (Amersham Pharmacia Biotech), according to manufacturer's recommended protocol. Pulse labeling was continued for 14 h, at which time cells were fixed and processed for immunofluorescence.

Flow Cytometry-- LNCaP cells were propagated under the conditions described and harvested at the times shown by trypsinization, washed once with phosphate-buffered saline, fixed with ethanol, and stained with propidium iodide as described previously (37). Analysis of the stained cells was carried out using a FACSort (Becton Dickinson) and the CELLFit software program. At least 10,000 forward scatter gated events were collected for each sample, and each experiment was performed in triplicate.

Immunofluorescence-- To monitor BrdUrd incorporation of both transfected and untransfected cells, cells were fixed and subjected to indirect immunofluorescence as described previously (38). Cells were visualized using a Zeiss Axiophot with a 40× objective, and pictures were captured using a Color Chilled 3CCD Camera (Hamamatsu).

Immunoblotting-- For immunoblotting, 2-4 × 10⁶ cells were harvested at the times indicated by trypsinization and washed with phosphate-buffered saline. Cells were lysed in ice-cold RIPA buffer containing protease inhibitors (10 µg/ml 1, 10 phenanthroline, 10 µg/ml aprotinin, 10 µg/ml leupeptin, and 1 mM phenylmethylsulfonyl fluoride) and phophatase inhibitors (10 mM sodium fluoride, 10 mM sodium pyrophosphate, 0.1 mM sodium vanadate, 15 mM p-nitrophenylphosphate, and 10 mM -glycerophosphate) for 15 min on ice. Following brief sonication, lysates were clarified at 10,000 × g (4 °C for 15 min). Protein concentration of the clarified lysates was determined using the Bio-Rad DC Protein Assay Reagent. Lysates were denatured by boiling in SDS loading buffer, and 20 µg of each sample was resolved by SDS-polyacrylamide gel electrophoresis (PAGE) and transferred to Immobilon-P (Millipore) by standard methods. Blots were probed for the following proteins with polyclonal antibodies: RB and LP (851, gift of Dr. Jean Wang, University of California at San Diego); p130 (C-20, Santa Cruz Biotechnology); CDK4 (H-22, Santa Cruz Biotechnology); cyclin D3 (C-16, Santa Cruz Biotechnology); CDK2 (M2, Santa Cruz Biotechnology); cyclin A (BF683, PharMingen); cyclin E (HE12, PharMingen); p21 (C-19, Santa Cruz Biotechnology); p27 (N-20, Santa Cruz Biotechnology); p16 (G₁75-405, PharMingen); T-Ag (Pab 101, Santa Cruz Biotechnology); and E1A (13 S-5, Santa Cruz Biotechnology). Goat anti-mouse horseradish peroxidase or protein A-horseradish peroxidase (Bio-Rad) was used for antibody visualization via enhanced chemiluminescence (Amersham Pharmacia Biotech).

Immunoprecipitations and Kinase Assays-- For in vitro kinase assays, 2-4 × 10⁶ cells from each condition indicated were harvested by trypsinization and washed in phosphate-buffered saline. To analyze CDK2-associated kinase activity, cells were lysed as in ice-cold IPCDC2 buffer (10 mM Tris-HCl, pH 7.5, 5 mM EDTA, pH 8.0, 130 mM sodium chloride, and 1% Triton X-100) containing the above mentioned protease and phosphatase inhibitors. After sonication and clarification, 200 µg of lysate was incubated for 2 h (4 °C with rotation) with anti-CDK2 antiserum (Santa Cruz Biotechnology) or anti-VEGF antiserum (Santa Cruz Biotechnology) as a negative control. Immunocomplexes were recovered with protein A-Sepharose (Amersham Pharmacia Biotech). Washes and kinase reactions were carried out as described previously (39). Histone H1 substrate was purchased from Amersham Pharmacia Biotech. To analyze CDK4-associated kinase activity, cells were lysed in ice-cold cyclin D lysis buffer (50 mM HEPES, pH 7.5, 150 mM sodium chloride, 1 mM EDTA, pH 8.0, 2.5 mM EGTA, pH 8.0, 80% glycerol, 0.5 mM dithiothreitol, and 0.1% Tween 20) containing the above mentioned protease and phosphatase inhibitors. After brief sonication and clarification, 200 µg of lysate was precleared using normal rabbit serum and protein A-Sepharose. After preclearing, lysates were incubated for 2 h (4 °C with rotation) with anti-CDK4 antiserum (Santa Cruz Biotechnology) or anti-VEGF antiserum (Santa Cruz Biotechnology) as a negative control. Washes and kinase reactions were carried out as described previously (40). A glutathione S-transferase fusion of the C terminus of RB was used as substrate and was generously provided by Dr. Erik Knudsen (University of California at San Diego). Relative kinase activity signals were determined using a PhosphorImager (Molecular Dynamics).

	RESULTS

Top Abstract Introduction Procedures Results References

LNCaP Cells Deprived of Androgen Arrest in G₁-- Although LNCaP cells have been shown to be dependent on androgen for growth (27, 31), the precise mechanisms for this have not been determined. To address this, we first characterized the growth of LNCaP cells in the presence or absence of serum androgen (Fig. 1). Subconfluent, exponentially growing cultures of LNCaP cells propagated in complete serum (FBS) were harvested and equally seeded into media containing one of three serum conditions: (i) 5% FBS; (ii) 5% charcoal dextran-treated FBS (CDT); or (iii) 5% CDT supplemented with 0.1 nM dihydrotestosterone (DHT). As shown in Fig. 1A, cells grown in complete serum exhibited a generation time of approximately 48 h. However, cells cultured in serum stripped of steroid hormone (CDT) demonstrated no increase in cell number up to 8 days post-treatment. This growth arrest was partially reversed by addition of nearly physiological levels (41, 42) of DHT, as expected based on previous reports (29, 32, 43). Also in agreement with previous reports, the proliferation rate of cells propagated in CDT serum supplemented with DHT is not as robust as that of cells propagated in complete serum (32, 43). Similar results are observed when comparing the growth of estrogen-dependent breast carcinoma cells in CDT supplemented with estrogen versus complete serum (44). Although it is possible that optimal cellular proliferation could be achieved upon readdition of multiple steroids or factors, only DHT was added back to allow for clear study of androgen-dependent proliferation. Cells grown in CDT supplemented with DHT demonstrated a doubling time of approximately 72 h when medium changes were carried out every 48 h. Growth of "androgen-independent" human prostate tumor cell lines DU-145, PC-3, and TSU-Pr1 demonstrated no decrease in their growth rates when cultured in the absence of androgen (data not shown), confirming that the growth arrest of LNCaP cells cultured in CDT is unique to prostate cells still dependent on androgen for growth (29, 32, 43).

View larger version (43K): [in this window] [in a new window]   Fig. 1.   Androgen depletion results in G1 arrest. Exponentially growing LNCaP cells were harvested at time 0 and split into one of three serum conditions: (i) complete serum (5% FBS); (ii) androgen-depleted serum (5% CDT); or (iii) CDT supplemented with 0.1 nM DHT. A, cells were harvested at the times indicated and counted to monitor cell growth and doubling times. B, cells were harvested at the times indicated, fixed, treated with RNase A, and stained with propidium iodide. Subsequent to staining, DNA content of the cells was determined using a FACSort (Becton-Dickinson) and the CELLFit analysis program. For cells propagated in FBS, approximately 65-68% of the population demonstrated a 2 N DNA content for each of the times indicated (data not shown). Increases in the percentage of cells demonstrating a 2 N DNA content in CDT or CDT + DHT are plotted. C, after 96 h of propagation in androgen-depleted medium, cells were labeled with BrdUrd for a period of 14 h. Subsequent to the pulse, cells were fixed and processed to monitor BrdUrd incorporation via indirect immunofluorescence. Data shown are the averages of three independent experiments in which at least 500 cells/experiment were analyzed. D, representative immunofluorescence from the experiment described in C.

Pocket Protein Phosphorylation Is Reduced in the Absence of Androgen-- Because phosphorylation of the retinoblastoma tumor suppressor protein, RB, is required in mammalian cells for the G₁-S transition (34, 45), we analyzed the phosphorylation state of RB and the related pocket protein, p130, in LNCaP cells. Hyperphosphorylated RB and p130 migrate at a slower mobility by SDS-PAGE than do the corresponding underphosphorylated forms. As monitored by immunoblot, both RB and p130 existed in all phosphorylation states after 48 h of propagation in androgen-containing and androgen-deprived media (Fig. 2, lanes 1-3). However, by 72 h of treatment, the underphosphorylated forms of RB and p130 predominated in cells grown without androgen (Fig. 2, compare lanes 4 and 5). This trend was even more dramatic by 96 h without androgen (Fig. 2, compare lanes 7 and 8), when G₁ arrest was maximal (Fig. 1B).

View larger version (24K): [in this window] [in a new window]   Fig. 2.   Pocket protein phosphorylation is reduced in the absence of androgen. At the times indicated, cells were harvested, and clarified lysates were prepared. Proteins were resolved by 6.5% SDS-PAGE, and the RB and p130 proteins were detected via immunoblotting with anti-RB and anti-p130 antibodies, respectively. Underphosphorylated species are designated pRB and p130, whereas hyperphosphorylated species are designated ppRB and pp130.

CDK4 Kinase Activity Is Diminished in Cells Deprived of Androgen-- Phosphorylation of RB and p130 is catalyzed by the G₁ cyclin-dependent kinases, CDK4 and CDK2 (18, 46). As such, the increase in underphosphorylated RB and p130 associated with androgen depletion (Fig. 2) indicated that the G₁ arrest observed by fluorescence-activated cell sorting analysis (Fig. 1B) was likely to be correlated with a reduction in G₁ CDK-cyclin kinase activities.

View larger version (43K): [in this window] [in a new window]   Fig. 3.   CDK4 kinase activity is diminished in cells deprived of androgen. A, at the times indicated, cells were harvested, and clarified lysates were prepared. Proteins were resolved by 12% SDS-PAGE, and the CDK4 and cyclin D3 proteins were detected via immunoblotting with anti-CDK4 and anti-cyclin D3 antibodies, respectively. B, at the times indicated, lysates were prepared and subjected to immunoprecipitation with either anti-VEGF (lane 1) antiserum or anti-CDK4 antiserum. Immunoprecipitates were utilized in in vitro kinase assays using the C-pocket of RB as a substrate. Phospho-RB was subjected to 12% SDS-PAGE and visualized by autoradiography. C, at the times indicated, lysates were prepared and subjected to immunoprecipitation with anti-cyclin D3 antiserum. Immunoprecipitates were utilized in in vitro kinase assays using the C-pocket of RB as a substrate. Phospho-RB was subjected to 12% SDS-PAGE and visualized by autoradiography.

CDK2 Kinase Activity Is Also Reduced upon Androgen Depletion-- After initial phosphorylation of RB by the CDK4-cyclin D complexes, G₁ CDK2 kinase complexes are required to maintain RB phosphorylation and to enter S phase (18, 46). We therefore sought to determine the levels and activation state of CDK2-associated complexes in LNCaP cells. Similar to CDK4, CDK2 protein levels did not significantly fluctuate in accordance with the presence or absence of androgen (Fig. 4A, top panel). Likewise, levels of cyclin E, which migrated as a series of species, did not fluctuate with androgen levels (Fig. 4A, middle panel). However, cyclin A protein levels were responsive to the presence of androgen (Fig. 4A, bottom panel). After 48 h of treatment, cyclin A levels were significantly reduced (8-fold) (compare lanes 1-2). This reduction in cyclin A levels continued throughout treatment (Fig. 4A, compare lanes 4 and 5 and compare lanes 7 and 8).

View larger version (46K): [in this window] [in a new window]   Fig. 4.   CDK2 kinase activity is reduced upon androgen depletion. A, at the times indicated, cells were harvested, and clarified lysates were prepared. Proteins were resolved by 10% SDS-PAGE, and the CDK2, cyclin E, and cyclin A proteins were detected via immunoblotting with anti-CDK2, anti-cyclin E, and anti-cyclin A antibodies, respectively. B, at the times indicated, lysates were subjected to immunoprecipitation with either anti-VEGF (lane 1) anti-serum or anti-CDK2 antiserum. Immunoprecipitates were utilized in in vitro kinase assays using the histone H1 as a substrate. Phospho-H1 was subjected to 12% SDS-PAGE and visualized by autoradiography.

p21^cip1 and p27^kip1 Are Alternately Regulated in Response to Androgen Removal-- To determine what role the CDK inhibitors may play in the G₁ arrest induced by androgen withdrawal, p21^cip1 and p27^kip1 were analyzed. The p16^ink4a protein was only weakly expressed in LNCaP cells grown under any condition, and levels of p16^ink4a were not affected by the presence or absence of androgen (data not shown). However, p21^cip1 protein levels were correlated directly with cell growth in LNCaP cells (Fig. 5, top panel). Cells proliferating in complete serum exhibited relatively constant levels of p21^cip1 for all times tested (Fig. 5, compare lanes 1, 4, and 7). By contrast, p21^cip1 levels diminished by 72 h of propagation in the absence of androgen and continued to diminish through 96 h of treatment (compare lanes 2, 5, and 8). These data indicate that in LNCaP cells, p21^cip1 is retained in cells that exhibit high levels of G₁ CDK kinase activity, similar to a variety of other systems (40, 47). They also corroborate the suggestion (40, 47) that p21^cip1 can serve as an assembly factor for G₁ CDKs and, in some instances, promotes G₁ CDK activity.

View larger version (19K): [in this window] [in a new window]   Fig. 5.   p21cip1 and p27kip1 are alternately regulated in response to androgen loss. At the times indicated, cells were harvested, and clarified lysates were prepared. Proteins were resolved by 12% SDS-PAGE, and the p21cip1 and p27kip1 proteins were detected via immunoblotting with anti-p21cip1 and anti-p27kip1 antibodies, respectively.

Ectopic Expression of G₁ Cyclin Does Not Rescue the G₁ Arrest Associated with Androgen Depletion-- The data presented in Figs. 1-5 indicate that the G₁ arrest associated with androgen withdrawal is characterized by underphosphorylated RB and reduced G₁ CDK kinase activities. Because G₁ cyclins are normally limiting for G₁-S progression (15), we hypothesized that their forced overexpression might restore cell cycle progression in the absence of androgen. To test this, we co-transfected plasmids expressing the green fluorescent protein (GFP) and individual G₁ cyclins (at a 1:4 ratio) into LNCaP cells that had been propagated for 48 h in the absence of androgen. The ability of these cells to incorporate BrdUrd was assessed 48 h later by indirect immunofluorescence. Experiments were repeated at least two or three times, and at least 500 transfected (GFP-positive) and untransfected (GFP-negative) cells were counted for each experiment. Only 2-5% of cells transfected with vector alone in the absence of androgen incorporated BrdUrd (Fig. 6, , A and C, left panel), and approximately equal numbers of nontransfected cells on the same coverslip incorporated BrdUrd (Fig. 6A). Surprisingly, cells transfected with expression constructs for the G₁ cyclins cyclin D1, cyclin D3, cyclin E, or cyclin A were not induced to incorporate BrdUrd in the absence of androgen (Fig. 6, A and C). To affirm that functional cyclins were being expressed from these plasmids, cells were co-transfected with plasmids expressing the G₁ cyclin and the large pocket region of RB, deemed LP (48, 49). The LP protein contains the A, B, and C pockets of RB, and as such contains 10 of the 16 consensus CDK phosphorylation sites present in full-length RB (49). Co-transfections were carried out with a 2:1 ratio of cyclin plasmids to LP, and the phosphorylation state of LP was ascertained by immunoblot (Fig. 6B). Cells transfected with vector alone lacked the LP protein, which migrated at an apparent molecular mass of 60 kDa, as opposed to endogenous RB protein, which migrated as a species between 105 and 110 kDa (48). Cells co-transfected with plasmids expressing LP and empty vector demonstrated only minimal phospho-LP because of limiting endogenous G₁ CDK-cyclin activity (Fig. 6B, lane 2). However, cells co-transfected with both LP and plasmids encoding any of the G₁ cyclins (cyclin D1, D3, E, or A) demonstrated substantial increases in phospho-LP (Fig. 6B, lanes 3-6). These data indicate not only that the G₁ cyclins were expressed by transient transfection, but that they formed active complexes with endogenous CDKs. Thus, it appears that overexpression of G₁ cyclins was insufficient to induce S phase progression in the absence of androgen.

View larger version (31K): [in this window] [in a new window]   Fig. 6.   G1 cyclin overexpression does not rescue the G1 arrest. A, cells were propagated on glass coverslips in the absence of androgen for 48 h prior to transfection. Cells were then co-transfected with plasmids expressing (i) GFP and (ii) a plasmid expressing cyclin D1, cyclin D3, cyclin E, cyclin A, or SV40 large T-antigen at a 1:4 ratio. BrdUrd labeling mix was added 48 h post-transfection. After a 14-h labeling period in 5% CDT, cells were fixed, and BrdUrd incorporation was monitored by indirect immunofluorescence. Data shown are the results of at least two or three experiments, and for each experiment at least 500 transfected (GFP-positive) and 500 untransfected (GFP-negative) cells were counted. B, to monitor cyclin function, (cyclin D1, lane 3; cyclin D3, lane 4; cyclin E, lane 5; and cyclin A, lane 6), cyclin constructs were co-transfected at a 2:1 ratio with the "large pocket" fragment of RB, here designated as LP. Lysates were prepared from transfected cells, and underphosphorylated pLP and hyperphosphorylated phospho-LP (ppLP) were monitored by subjecting lysates to 8% SDS-PAGE and immunoblotting with anti-RB antiserum. C, representative immunofluorescence of data shown in A. Transfected cells capable of supporting BrdUrd incorporation are shown with yellow arrows.

RB Phosphorylation Is Required for S Phase Progression in LNCaP Cells-- Oncoproteins such as T-Ag act in part by sequestration of cellular RB, thus eliminating the requirement for CDK4 kinase activity and allowing premature release and activation of S phase promoting genes such as E2F (18, 46). As such, we wanted to determine the requirement of RB phosphorylation for S phase progression in LNCaP cells. To test this, LNCaP cells propagated in complete serum were co-transfected with plasmids expressing GFP and proteins that modulate RB activity at a 1:4 ratio (Fig. 7B). More than 1500 transfected cells and an equal number of untransfected cells were microscopically scored. Approximately 50% of cells transfected with vector alone supported BrdUrd incorporation (Fig. 7, A and C), and a similar proportion of untransfected (GFP-negative) cells also incorporated BrdUrd (Fig. 7A), demonstrating that transfection of GFP bears no influence on the ability of LNCaP cells to incorporate BrdUrd. In contrast, only 4-6% of cells transfected and expressing the CDK inhibitor p16^ink4a were able to incorporate BrdUrd, suggesting that these cells were inhibited from completing the G₁-S transition. Untransfected cells from the same experiment retained the ability to incorporate BrdUrd.

View larger version (41K): [in this window] [in a new window]   Fig. 7.   RB phosphorylation is required for S phase progression. A, cells seeded on glass coverslips in the presence of complete serum were then co-transfected with plasmids expressing (i) GFP and (ii) a plasmid expressing p16ink4a, WT-LP, or PSM.7-LP at a 1:4 ratio. BrdUrd labeling mix was added 48 h post-transfection. After a 14-h labeling period, cells were fixed, and BrdUrd incorporation was monitored by indirect immunofluorescence. Data shown are the results of at least two or three experiments, and for each experiment at least 500 transfected (GFP-positive) and 500 untransfected (GFP-negative) cells were counted. B, to monitor expression, individual constructs were transfected into LNCaP cells, and lysates were harvested 48 h post-transfection. Lysates were subjected to 15% SDS-PAGE to detect p16ink4a via immunoblotting with anti-16ink4a anti-serum (lanes 1, 2) or 8% SDS-PAGE to detect WT-LP and PSM.7-LP via immunoblotting with anti-RB antiserum (lanes 3-5). C, representative immunofluorescence of data shown in A. Transfected cells capable of supporting BrdUrd incorporation are shown with yellow arrows.

Pocket-Protein Binding Is Required for Oncoproteins to Initiate G₁-S Progression in the Absence of Androgen-- Having observed that RB phosphorylation was required in LNCaP cells for the G₁-S transition and that ectopic expression of T-Ag was able to initiate S phase progression in the absence of androgen, we hypothesized that the ability of oncoproteins to initiate androgen-independent cell cycle progression may be linked to their ability to sequester the activity of RB pocket proteins. To test this, we co-transfected plasmids expressing GFP and mutants of viral oncoproteins in the absence of androgen. The ability of cells to complete the G₁-S transition was measured by incorporation of BrdUrd. Cells transfected with DNA encoding the entire early region of SV40 (Figs. 6A and 8A) or the large T-Ag cDNA (data not shown) incorporated BrdUrd in the absence of androgen. By contrast, T-Ag constructs carrying mutations affecting the LXCXE motif (mutants Pvu-1 or K1) lost the ability to promote androgen-independent S phase progression (Fig. 8, A and B); this LXCXE motif has been shown to be required for binding to the pocket proteins (50, 51). A mutant of T-Ag defective in the N-terminal J domain function (mutant 2831) also lacked the ability to promote S phase progression. The J domain is thought to serve several different functions, including binding to Hsc 70, ATPase activity, and cellular transformation (52, 53). Recent evidence suggests that the J domain also plays a role in the ability of T-Ag to modulate the activities of pocket proteins (54). Thus, the ability of T-Ag to promote androgen-independent G₁-S transition appears to be bestowed through its ability to modulate pocket proteins.

View larger version (23K): [in this window] [in a new window]   Fig. 8.   SV40 large T-antigen renders LNCaP cells androgen-independent for cell cycle progression. A, cells were propagated on glass coverslips in the absence of androgen for 48 h prior to transfection. Cells were then co-transfected with plasmids expressing (i) GFP and (ii) a plasmid expressing WT-T-Ag, PVU1-T-Ag, K1-T-Ag, or 2831-T-Ag at a 1:4 ratio. BrdUrd labeling mix was added 48 h post-transfection. After a 14-h labeling period, cells were fixed, and BrdUrd incorporation was monitored by indirect immunofluorescence. Data shown are the result of at least two or three experiments, and for each experiment at least 500 transfected (GFP-positive) and 500 untransfected (GFP-negative) cells were counted. B, to monitor expression, individual T-Ag expression constructs were transfected into LNCaP cells, and lysates were harvested 48 h post-transfection. Lysates were subjected to 10% SDS-PAGE to detect T-Ag proteins via immunoblotting with anti-T-Ag antiserum (lanes 1-5).

View larger version (26K): [in this window] [in a new window]   Fig. 9.   E1A renders LNCaP cells androgen-independent for cell cycle progression. A, cells were propagated on glass coverslips in the absence of androgen for 48 h prior to transfection. Cells were then co-transfected with plasmids expressing (i) GFP and (ii) a plasmid expressing WT-E1A, 12S.E.1A, 13S.E.1A, CXDL-E1A, or 30-85-E1A at a 1:4 ratio. BrdUrd labeling mix was added 48 h post-transfection. After a 14-h labeling period, cells were fixed, and BrdUrd incorporation was monitored by indirect immunofluorescence. Data shown are the results of at least two or three experiments, and for each experiment at least 500 transfected (GFP-positive) and 500 untransfected (GFP-negative) cells were counted. B, to monitor expression, individual E1A expression constructs were transfected into LNCaP cells, and lysates were harvested 48 h post-transfection. Lysates were subjected to 12% SDS-PAGE to detect E1A proteins via immunoblotting with anti-E1A antisera (lanes 1-6).

	ISCUSSION

In this report, we provide the first evidence that androgen-responsive growth signals are mediated through regulation of multiple elements controlling the G₁-S cell cycle transition. We have also devised a reliable system through which specific proteins can be individually introduced into androgen-dependent cells and assayed for their abilities to promote androgen-independent cell cycle progression and have identified pathways that can be disrupted to convert prostatic epithelial cells to androgen independence.

Androgen Withdrawal Causes a Specific Early G₁ Arrest-- Androgen-dependent prostatic cells depleted of hormone arrested in early G₁ phase (Fig. 1), as characterized by a global decrease in hyperphosphorylated RB and p130 (Fig. 2) and a reduction in CDK4-cyclin D3-associated RB kinase activity (Fig. 3). This reduction in kinase activity was at least partially accounted for by a decrease in cyclin D3 protein levels. A reduction in CDK2-associated kinase activity was also observed in response to androgen withdrawal (Fig. 4). Interestingly, cyclin E protein levels were unaffected by the presence or absence of androgen, whereas cyclin A accumulation was drastically diminished in the absence of androgen. These observations suggest that a determination of the mechanisms through which androgen regulates the synthesis or stability of cyclin D3 and cyclin A will be of interest, because androgen regulates their accumulation.

Ectopic G₁ Cyclin Expression Does Not Bypass the Androgen Requirement-- Because G₁ cyclins are normally limiting for progression past the G₁-S transition and because the G₁ arrest associated with androgen depletion was characterized by a reduction of G₁ cyclins and G₁ cyclin-associated kinase activity, we tested whether ectopic expression of individual G₁ cyclins would be sufficient to induce cell cycle progression in the absence of androgen. Strikingly, expression of no single G₁ cyclin was sufficient to induce cell cycle progression (Fig. 6). These data emphasize the strength of the androgen requirement, because ectopic expression of cyclin E has been shown to even rescue other cell types that have been blocked by expression of p16^ink4a or constitutively active RB (45, 57). However, Alevizopoulos et al. (57) have shown that cyclin overexpression cannot bypass a p27^kip1-mediated cell cycle arrest. Because we observed slightly increased levels of p27^kip1 in androgen-depleted cells (Fig. 5), this may partially explain the inability of excess cyclins to promote cellular proliferation in the absence of androgen.

Viral Oncoproteins Capable of Sequestering Pocket Proteins Alleviate Androgen Dependence-- Because inactivation of RB was shown to be required for G₁-S progression in LNCaP cells but overexpression of G₁ cyclins was incapable of reversing the G₁ arrest associated with androgen depletion, we sought to identify additional cell cycle pathways that might be disrupted to allow androgen-independent growth. We chose the viral oncoproteins SV40 T-Ag and adenovirus E1A, because they have been shown in other systems to partially bypass the regulation of the G₁-S transition but not necessarily allow for growth in the absence of serum (51). We observed that wild-type T-Ag and E1A supported androgen-independent cell cycle progression (Fig. 7-9). To determine the mechanisms whereby this occurred, we used specific mutants of T-Ag and E1A. Such mutants disrupted in the conserved LXCXE motif (mutants K1-T-Ag, PvuI-T-Ag, and CXDL-E1A) were unable to promote proliferation. The LXCXE motif is known to be the region through which both T-Ag and E1A bind and sequester the pocket proteins RB, p107, and p130 (50, 51). A mutant of E1A defective in a secondary region of binding to pocket proteins (30-85) was also incapable of driving the S phase transition. Lastly, mutants disrupting the J domain of T-Ag also eliminated proliferation potential. The J domain shares homology with the Escherichia coli DnaJ protein, is responsible for several functions, including binding to Hsc 70 and DNA replication (52, 53, 62), and has recently been shown to be important for inactivation of p107 and p130 (54). Taken together, these data suggest that inactivation of RB (and other pocket proteins) may be required for the transition to an androgen-independent state. This observation is likely the underlying mechanism whereby transgenic mice carrying a probasin-promoter-SV40 large T-antigen fusion develop prostatic carcinomas capable of making the transition to an androgen-independent state upon castration of the animal (63). Additionally, our observations agree well with what has been observed in human tumorigenesis. For example, mutations of the RB1 gene have been shown to correlate with poor prognosis of prostatic adenocarcinomas (64) and have been shown to occur as frequent, early events in prostate cancers (65); between 20-60% of prostate cancers have mutations of the RB1 gene (65-67). Moreover, inactivation of p16^ink4a occurs at a relatively high frequency (40-60%) in advanced prostate carcinomas (68, 69). It has been well documented that the growth inhibitory activity of p16^ink4a is dependent on functional RB, and p16^ink4a inactivation simulates RB phosphorylation and subsequent inactivation by G₁ CDKs (45, 58).