Progression to androgen independence is the lethal end stage of prostate cancer. We used expression of androgen receptor (AR)-targeted short hairpin RNAs (shRNA) to directly test the requirement for AR in ligand-independent activation of androgen-regulated genes and hormone-independent tumor progression. Transient transfection of LNCaP human prostate cancer cells showed that AR shRNA decreased R1881 induction of the prostate-specific antigen (PSA)-luciferase reporter by 96%, whereas activation by forskolin, interleukin-6, or epidermal growth factor was inhibited 48% to 75%. Whereas the antiandrogen bicalutamide provided no further suppression, treatment with the mitogen-activated protein kinase (MAPK) inhibitor U0126 completely abrogated the residual activity, indicating a MAPK-dependent, AR-independent pathway for regulating the PSA promoter. Expression of doxycycline-inducible AR shRNA expression in LNCaP cells resulted in decreased levels of AR and PSA as well as reduced proliferation in vitro. When these cells were grown as xenografts in immunocompromised mice, induction of AR shRNA decreased serum PSA to below castration nadir levels and significantly retarded tumor growth over the entire 55-day experimental period. This is the first demonstration that, by inducibly suppressing AR expression in vivo, there is an extensive delay in progression to androgen independence as well as a dramatic inhibition of tumor growth and decrease in serum PSA, which exceeds that seen with castration alone. Based on these findings, we propose that suppressing AR expression may provide superior therapeutic benefit in reducing tumor growth rate than castration and may additionally be very effective in delaying progression to androgen independence. (Cancer Res 2006; 66(21): 10613-20)
- androgen receptor
- prostate cancer
- tumor progression
- androgen independence
Prostate cancer is the most commonly diagnosed nonskin cancer in males and one of the leading causes of cancer deaths in men ( 1). Although organ-confined prostate cancer is often curable by surgery or radiation, treatment for locally advanced, recurrent, or metastatic prostate cancer is primarily some form of androgen withdrawal therapy ( 2, 3). Unfortunately, the effects of androgen withdrawal therapy are temporary due to progression of surviving tumor cells to the androgen-independent state ( 4, 5). Androgen-independent progression does not seem to involve loss of androgen receptor (AR) but instead requires inappropriate activation of AR ( 6, 7). Amplification ( 8) or mutation ( 9) of the AR gene, altered expression of coregulators ( 10), and ligand-independent activation of AR through convergence of cell signaling pathways ( 6) can all contribute to androgen-independent activation of AR.
Ligand-independent activation of the AR most likely involves a complex interplay of multiple nonsteroidal signaling pathways that converge on AR activation. The cytokine interleukin-6 (IL-6), epidermal growth factor (EGF), and the adenylate cyclase agonist forskolin are examples of agents capable of cross-talk activation of AR. Increased circulating IL-6 levels are found in patients with metastasis and are associated with poor prognosis ( 11). However, the effects of IL-6 on cell growth ( 12) and transcriptional activity of AR ( 13, 14) in LNCaP cells are contradictory possibly due to cell passage number or culture conditions used. Similarly, EGF levels are elevated in prostate cancer patients, but the effect on AR transcriptional activity continues to be debated ( 15, 16). Forskolin-mediated increased intracellular cyclic AMP levels have also been reported to activate AR in the absence of androgen presumably through cross-talk between the protein kinase A and AR pathways ( 17). However, it is unresolved whether a functional AR is required for ligand-independent activation of androgen target genes, such as prostate-specific antigen (PSA).
Because AR is central in the development of androgen independence, AR knockdown has been proposed as additional therapy after failure of androgen ablation ( 18). In this study, we have designed AR-targeted short hairpin RNAs (shRNA) and examined their ability to inhibit ligand-independent activation of a PSA promoter reporter. Using LNCaP cells transfected with shRNAs, we show for the first time that knockdown of AR inhibited most, but not all, activation of the PSA promoter by nonandrogenic compounds and that suppressing AR expression was far more effective than castration in blocking tumor growth and decreasing serum PSA levels in LNCaP xenografts.
Materials and Methods
shRNA and lentiviral constructs. The pSHAG-1-tet plasmid (gift from Dr. Alice Mui, University of British Columbia, Vancouver, British Columbia, Canada) used to clone the shRNA was modified from pSHAG-1 (gift from Dr. Greg Hannon, Cold Spring Harbor Laboratory, Cold Spring Harbor, NY) to include a cassette containing two tetracycline-responsive elements (TRE) in tandem and upstream of the RNA polymerase III (U6) promoter, cloning sites for the shRNA, and recombination sites compatible with the Gateway system (Invitrogen-Life Technologies, Inc., Burlington, Ontario, Canada). Target sequences were chosen following guidelines provided by Invitrogen, and each one was subjected to BLAST search to ensure that there was no sequence homology to other genes. Core sequences for the shRNAs are listed below, and their positions in the coding region of the AR gene are given in parenthesis (AR accession no. NM_000044): (1) 5′-AAGCTCAAGGATGGAAGTGCAGTTAGGGCTG-3′ (1,106-1,136), (2) 5′-AGCAGCAGGAAGCAGTATCCGAAGGCAGCAG-3′ (1,705-1,735), (3) 5′-CACAGCCGAAGAAGGCCAGTTGTATGGACCG-3′ (2,432-2,462), and (4) 5′-GAAAGCACTGCTACTCTTCAGCATTATTCCA-3′ (3,539-3,569). A scrambled sequence (5′-CCGTACCTACACGCAGCGCTGACAACAGTTT-3′) was used as a negative control ( 19). Using an off-target search algorithm for short interfering RNA (siRNA), 1 shRNA 4 was one chosen as an ideal candidate. Complimentary oligonucleotides containing 31 bp of antisense and sense strands of the target sequence separated by an eight-nucleotide spacer for generation of a hairpin loop, a HindIII site for rapid identification of the hairpin, and restriction enzyme cleavage sites for cloning were synthesized. Equal molar amounts of both forward and reverse complimentary oligonucleotides were annealed, kinase treated, and cloned into pSHAG-1-tet plasmid. Recombination reactions using LR Clonase (Invitrogen) were done to transfer the TRE-U6-shRNA cassette into Gateway-modified pHR-CMV-EGFP lentiviral vector (gift from Dr. Alice Mui). The cassette was also transferred into the pLenti6/BLOCK-iT-DEST (Invitrogen), which carries a blasticidin selection marker.
Transactivation assays. Transient transfections of LNCaP cells were done as described previously ( 17). Briefly, LNCaP cells were seeded at a density of 3 × 105 per well in a six-well plate and cotransfected the following day using Lipofectin reagent (Invitrogen) with PSA-luciferase (−6,100/+12; gift from Dr. J. Trapman, Erasmus University, Rotterdam, the Netherlands) or ARR3tk-luciferase ( 20) and either shRNA to AR or scrambled sequence in pSHAG-1-tet. After 24 hours, the medium was replaced with RPMI 1640 (Invitrogen) containing 5% charcoal-stripped serum (CSS; HyClone, VWR, West Chester, PA) supplemented with either 1 nmol/L synthetic androgen (R1881), 50 μmol/L forskolin (Sigma Chemicals, St. Louis, MO), 100 ng/mL IL-6 (Calbiochem-Novabiochem Corp., San Diego, CA), or 100 ng/mL EGF (Sigma). For antiandrogen studies, cells were pretreated for 2 hours with bicalutamide (50 μmol/L; gift from AstraZeneca, Cheshire, England) before addition of stimulating agents. For mitogen-activated protein kinase (MAPK) and phosphatidylinositol 3-kinase (PI3K) inhibition studies, cells were pretreated with 10 μmol/L U0126 (Promega, Madison, WI) or 10 μmol/L LY294002 (Calbiochem-Novabiochem) before addition of stimulating agents. Cells were collected after 24 hours of incubation, lysed, and assayed for luciferase activity using the Dual-Luciferase Assay System (Promega) normalized for protein concentrations of the respective samples as measured by bicinchoninic acid (BCA) protein assay (Pierce Biotechnology, Inc., Rockford, IL).
LNCaP cells stably expressing AR shRNA. A plasmid encoding constitutively expressed tetracycline repressor (TR) and puromycin resistance, together with retroviral packaging plasmids, was transfected into 293T cells using the method described previously ( 21). Resulting retroviral particles were used to generate a stably transduced LNCaP line, LNCaP-TR, by puromycin selection (Sigma). LNCaP-TR cells were then transduced with lentiviral vectors pHR-CMV-EGFP or pLenti6/BLOCK-iT-DEST carrying either AR-targeted or scrambled shRNAs. Stably transformed cells were flow sorted for enhanced green fluorescent protein (EGFP) by fluorescence-activated cell sorting (FACS) and maintained in RPMI 1640 plus 10% fetal bovine serum containing 1 μg/mL puromycin. LNCaP-TR cells expressing pLenti6/BLOCK-iT-DEST AR or scrambled shRNA were selected by maintaining the virally infected cells in puromycin and blasticidin (Invitrogen).
Northern blot analysis. LNCaP cells stably infected with AR shRNAs, scrambled control, or empty vector virus were plated in 10-cm plates (2 × 106 per plate). When the cells reached 60% confluence, the tetracycline analogue, doxycycline hyclate (DOX; Sigma), was added (2 μg/mL) and cells were harvested after 2 days. Total RNA was prepared using Trizol (Invitrogen), and 20 μg of total RNA were used for Northern blot analysis as described ( 22). The human AR probe used was a 980-bp fragment (+1,059 to +2,039) and the PSA was full-length cDNA. Bands were quantified on a Bio-Rad Gel Doc 2000 (Hercules, CA; ref. 23).
Western blot analysis. For in vitro analysis, cells were plated in six-well plates (3 × 105 per well) and treated with DOX as described above and immunoblotted as reported previously ( 23). For xenograft studies, tumor tissue was removed from hosts at various times after treatment, flash frozen in liquid nitrogen, lysed in radioimmunoprecipitation assay (RIPA) buffer, and Dounce homogenized. Total protein (50 μg), quantified by the BCA assay, was used for immunoblotting. Antibodies to AR and PSA were purchased from Santa Cruz Biotechnology (Santa Cruz, CA). Bands were quantified on a Bio-Rad Gel Doc 2000 ( 23).
Cell proliferation assay. Virally transduced LNCaP cells were plated at 3,000 per well in RPMI 1640 containing 5% CSS in a 96-well plate. After 3 days, medium was changed to ±DOX ± 0.1 nmol/L R1881 and cells were replenished with fresh medium every 3 days thereafter. At days 0, 2, 4, 6, and 8 of DOX/R1881 treatment, cell density was measured using the 3-(4,5-dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2H-tetrazolium (MTS; Promega) assay according to the manufacturers' protocol.
Bromodeoxyuridine labeling. Bromodeoxyuridine (BrdUrd; Sigma) labeling was done by adding 100 μmol/L BrdUrd to cells cultured ±DOX during the last 2.5 hours of incubation and subsequently assessed by immunofluorescence microscopy of methanol-fixed cells using anti-BrdUrd-FITC antibody (Boehringer Mannheim, Indianapolis, IN) as described previously ( 24). The mitotic activity of each treatment population was calculated as percentage BrdUrd incorporation for three independent experiments ± SE.
Assessment of shRNA efficacy in xenografts. LNCaP cells transduced with the inducible shRNA lentiviral vector 4 (5 × 106) were inoculated s.c. with 0.1 mL Matrigel (Becton Dickinson Labware, Mississauga, Ontario, Canada) in two flank regions of 6- to 8-week-old male athymic nude mice (Harlan Sprague Dawley, Inc., Indianapolis, IN) under halothane anesthesia using a 27-gauge needle. When the tumors became palpable, their volumes were measured weekly. Blood was also collected from the tail vein at weekly intervals to monitor PSA. Tumor size was calculated by the formula length × width × depth × 0.5236 ( 25), and serum PSA was measured by ELISA (ClinPro International, Union City, CA). When PSA values reached 50 to 75 ng/mL, half of the mice were castrated and given DOX-treated water (200 μg/mL), whereas the remaining animals received DOX-treated water alone. All animal procedures were done according to the guidelines of the Canadian Council of Animal Care and with appropriate institutional certification.
Statistical analysis. Student's t test (two sided) was used to evaluate statistically significant differences. Analysis of cell proliferation differences was done by one-way ANOVA using the Tukey-Kramer multiple comparisons test and done with GraphPad InStat version 3.06 (San Diego, CA).
Inhibition of AR transactivation. Four shRNA constructs targeting AR were designed to evaluate the requirement for AR in androgen-independent progression and in ligand-independent transactivation of androgen-responsive genes. To determine which of the four AR shRNAs was most effective, each was screened for inhibition of activation of the androgen-responsive ARR3tk-luciferase reporter in transient transfection assays ( Fig. 1A ). When compared with empty vector (pSHAG-1), shRNA 4 was found to be the most potent construct, inhibiting AR-induced luciferase expression by 97%. The AR shRNA 1 was the next most effective at 85% inhibition, whereas shRNAs 2 and 3 were the least effective at 56% and 37% inhibition, respectively.
To assess the possibility of off-target effects, AR rescue experiments were done. We first used a constitutively active AR construct (amino acids 1-646) that lacks the ligand-binding domain and thus is not a target of the AR shRNA 4. When expressed in LNCaP cells, activation of the PSA-luciferase reporter in cells transfected with scrambled shRNA or AR shRNA 4 was indistinguishable (data not shown). Secondly, we tested the ability of transfected rat AR to rescue PSA-luciferase reporter expression in cells transfected with scrambled and AR shRNA constructs 1, 2, and 4 ( Fig. 1B). We found that rat AR could rescue AR shRNAs 1 and 2 inhibition of PSA-luciferase expression, which contain 2- and 3-base mismatches with human AR, respectively, but was unable to rescue AR shRNA 4, which is 100% homologous to human AR. Additionally, AR shRNA 4 did not decrease glucocorticoid receptor activation of ARR3tk-luciferase (data not shown), further confirming the specificity of the shRNA. Together, these results are consistent with the ability of AR shRNAs to selectively target AR to varying degrees.
Role of AR in ligand-dependent and ligand-independent PSA promoter activation. We next tested the effects of these shRNAs on the activity of the 6.1-kb PSA promoter. PSA reporter activation by R1881, forskolin, and IL-6 was compared in cells transfected with scrambled or AR shRNA 1, 2, and 4 constructs. The AR shRNAs could effectively inhibit R1881 activation of PSA from 85% for shRNA 1 to 94% for shRNAs 2 and 4 ( Fig. 2A ). Similarly, the antiandrogen bicalutamide inhibited R1881 induction by 93 ± 3% ( Fig. 2B). Thus, AR shRNAs were as effective as antiandrogen at inhibiting ligand-dependent PSA promoter and this was not due to off-target effects. Conversely, all three AR shRNAs only partially suppressed forskolin or IL-6 activation of PSA ( Fig. 2A). Because none could completely eliminate PSA promoter activation by nonandrogenic agonists, androgen-independent activation of androgen response genes, such as PSA, seems to involve both AR-dependent and AR-independent mechanisms.
To further examine the role of AR in ligand-independent activation, LNCaP cells were cotransfected with the PSA-luciferase and the scrambled or shRNA 4 plasmid and treated with forskolin, IL-6, or EGF ( Fig. 2B). Forskolin induced a 98-fold activation of the PSA reporter that could be inhibited by 75 ± 5% with AR shRNA 4 and was resistant to further reduction by bicalutamide. Alone, bicalutamide was less effective, only inhibiting forskolin-induced PSA promoter activation by 44 ± 4%. Similarly, IL-6 caused a 14-fold induction in PSA-luciferase activity, which was inhibited by 58 ± 6% by AR shRNA 4 cotransfection. EGF caused a 3-fold increase in PSA promoter activity that was inhibited by 48 ± 10% by AR shRNA 4. Bicalutamide did not significantly affect PSA reporter expression by IL-6 ( Fig. 2B) or EGF (data not shown).
The MAPK pathway has been previously implicated in forskolin activation of the AR ( 26). Because AKT is constitutively activated in LNCaP cells, both the MAPK/extracellular signal-regulated kinase kinase 1/2 inhibitor, U0126, and the PI3K inhibitor, LY294002, were assessed for their abilities to inhibit PSA promoter activation ( Fig. 2B). U0126 substantially inhibited PSA promoter activation by forskolin (96 ± 14%) and IL-6 (86 ± 44%), whereas LY294002 could only partially inhibit PSA promoter activation by forskolin (29 ± 1%) and IL-6 (59 ± 3%). As reported by others ( 27), we found that U0126 also inhibited (92 ± 7%) R1881-induced PSA promoter activation (data not shown). Our results suggest that androgen-independent activation of the PSA promoter involves both AR-dependent and AR-independent mechanisms. The ability of U0126 to effectively block forskolin and IL-6-induced PSA promoter activation in the presence and absence of the AR shRNA indicates that both the AR-dependent and AR-independent pathways of ligand-independent PSA promoter activation require MAPK activation.
Generation of LNCaP cell lines stably expressing DOX-inducible shRNA to the AR. Transient transfection experiments are not useful in long-term studies because of variability in transfection efficiencies between experiments. Furthermore, knockdown effects on endogenous levels of AR cannot be accurately assessed on a per cell basis because only a fraction of the cells will contain the shRNA plasmid. Because several groups have reported that siRNA to the AR results in apoptosis of LNCaP cells ( 28– 30), we sought to establish a cell line that stably and inducibly expresses AR shRNAs. We chose to create an inducible system that allows one to turn on shRNA production with addition of the tetracycline analogue DOX. LNCaP cell lines stably expressing DOX-inducible AR shRNA, scrambled shRNA, or empty vector were created using the lentiviral approach as described in Materials and Methods. The cDNA encoding EGFP was cis-linked to the shRNA cassette so that EGFP could be coexpressed with the shRNA. This allows monitoring of lentivirus transduction, and EGFP-positive cells can then be readily sorted to enrich for shRNA-containing clones.
The effects of DOX treatment on expression of AR and its target gene PSA were assessed by Northern and Western blotting ( Fig. 3A and B ). Whereas DOX treatment of empty vector and scrambled shRNA-expressing clones had no effect, DOX-treated AR shRNA-expressing cells exhibited reduced AR mRNA (73 ± 10%; P < 0.01) and protein (67 ± 20%; P < 0.05). Furthermore, DOX treatment of AR shRNA-expressing cells exhibited reduced PSA mRNA (64 ± 17%; P < 0.01) and protein expression (60 ± 15%; P < 0.005). This decrease in AR mRNA and protein as well as a reduction in PSA confirm that the response is “classic” RNAi ( 31). However, after long-term passage, there was a drift toward cells that no longer express the AR shRNA possibly because of the absence of a selection pressure. We therefore transferred the shRNA cassette into pLenti6/BLOCK-iT-DEST, which contains the blasticidin selection marker, and maintained the cells in the presence of blasticidin. The antibiotic-selected cells exhibited similar patterns of AR shRNA inhibition of AR and PSA but did not readily lose the capacity to express AR shRNA.
Effects of AR shRNA on in vitro growth of LNCaP cells. Scrambled and AR shRNA-containing blasticidin-selected cell lines were tested for effect of induced AR shRNA expression on mitotic activity of LNCaP cells using BrdUrd incorporation ( Fig. 4A ). DOX treatment had no effect on scrambled shRNA-expressing cells but resulted in a 51 ± 6% (P < 0.05) decrease in mitotic index in AR shRNA-expressing cells. To assess the role of AR in the growth of LNCaP cells in tissue culture, blasticidin-selected cell lines containing DOX-inducible AR shRNA or scrambled shRNA were cultured, deprived of androgens for 3 days, and then treated ±DOX and ±0.1 nmol/L R1881. The MTS cell proliferation assay was used to assess the effects of these shRNAs on cell growth. In the absence of R1881, both cell lines showed very little growth over the 6-day period irrespective of whether shRNAs were expressed ( Fig. 4B). However, in the presence of R1881, the cells expressing AR shRNA grew at a significantly slower rate (P < 0.001; Fig. 4B). Similar results were obtained with LNCaP cells stably expressing AR shRNAs 1 and 2 (data not shown). Thus, in both BrdUrd and MTS assays, suppressing AR levels resulted in a decrease in the rate of cell growth of LNCaP cells.
Effects of AR shRNA on growth of LNCaP tumors. The LNCaP xenograft model generated in athymic nude mice provides an excellent and reproducible in vivo experimental system for monitoring molecular and genetic events associated with growth, regression, and progression to androgen independence ( 32– 34). Although castration does not generally result in significant tumor regression ( 22), it does lead to a drop in serum PSA, which can be monitored over time. In the LNCaP model, time to progression is defined as the duration of time required after castration for serum PSA levels to return to or increase above precastrate levels ( 35). To determine the effects of suppressing AR levels in this model, antibiotic-selected DOX-inducible AR or scrambled shRNA LNCaP cell lines were s.c. injected into male athymic mice. When serum PSA levels reached 50 to 75 ng/mL, the mice were treated with either (a) DOX in their drinking water (+DOX) or (b) castrated and then given DOX in their drinking water (Cx + DOX).
Figure 5A shows serum PSA levels for castrate or noncastrate groups of mice following treatment with DOX to induce AR shRNA or scrambled shRNA. As anticipated, serum PSA for the scrambled shRNA xenograft group continued to increase with just DOX treatment and, by day 26, had increased to 288 ± 110% of pretreatment levels. Castration of this group showed an initial decrease in serum PSA at day 12 (89 ± 34% of pretreatment) before rising above pretreatment levels at day 26 (119 ± 45%). For the AR shRNA xenografts, reduction of AR alone (+DOX) or in combination with castration resulted in a rapid and sustained decrease in serum PSA, which remained low up to 55 days observed in these experiments. For the AR shRNA xenograft group treated with DOX, the drop in serum PSA levels as early as 12 days after treatment was significantly greater (P < 0.05) than that seen by castration alone. Serum PSA reached a nadir at day 26 with an average PSA level of 20 ± 6% of that measured before treatment. After this point, there was a slow and very gradual increase in PSA levels such that, by day 55, the levels were 49 ± 25% of pretreatment. For the AR shRNA plus castration treatment group, the serum PSA nadir was at day 33 with serum PSA levels bottoming out at 8 ± 2% of pretreatment level followed by a slow increase to 13 ± 6% of pretreatment by day 55. None of the AR shRNA xenograft groups had reached PSA androgen independence by this time. These results indicate that suppressing AR expression can dramatically prevent or delay progression to androgen independence.
In addition to effects on serum PSA levels, knocking down AR had a major effect on tumor volume. In agreement with previous studies ( 22), castration alone (scrambled + Cx) did not induce substantial regression of the LNCaP tumors but did initially reduce the rate at which tumor volume increased for ∼20 days ( Fig. 5B). However, castration did not prevent regrowth of the LNCaP tumors, and by day 26, the growth rate was similar to the noncastrate group. By day 55, tumors expressing the scrambled shRNA had volumes that were 451 ± 14% and 312 ± 19% of their pretreatment volumes in noncastrated and in castrated animals, respectively. In contrast, the AR shRNA groups had volumes averaging only 119 ± 16% (castrated plus AR shRNA) and 142 ± 33% (noncastrated plus AR shRNA) of pretreatment sizes ( Fig. 5B), significantly smaller (P < 0.05) than the scrambled group. These results indicate that the growth of LNCaP xenografts is far more sensitive to AR knockdown than to androgen withdrawal.
To confirm that the AR was indeed knocked down in the AR shRNA-expressing xenografts, tissue levels of AR and PSA proteins were assessed by Western blot analyses. Figure 6 shows that the tumors expressing scrambled shRNA in both noncastrated and castrated hosts ( Figure 6, lanes 1 and 2) had considerable amounts of AR and PSA proteins. By comparison, AR shRNA-expressing tumors had barely detectable AR and PSA at the time of their serum PSA nadir ( Figure 6, lanes 3 and 4). However, by day 55, both AR and PSA protein were detected in AR shRNA-expressing xenografts ( Figure 6, lanes 5 and 6), reaching levels comparable with that seen in tumors expressing the scrambled shRNA ( Figure 6, lanes 1 and 2). Thus, there seemed to be an eventual escape from complete AR suppression in these tumors.
These xenograft studies show that knockdown of AR was far more effective and of longer duration than castration alone in suppressing the expression of PSA and tumor growth as well as in delaying progression to androgen independence.
The AR seems to play a pivotal role in hormone-refractory prostate cancer ( 36), and gene expression profiles have shown that the AR is the only gene that is consistently up-regulated when prostate xenograft tumors become androgen independent ( 37). One possible mechanism proposed for maintaining AR activity is ligand-independent activation of AR. To address the relative importance of the AR in progression to androgen independence, we used a shRNA strategy to knock down AR and then examine the consequences on ligand-independent activation of the androgen-responsive PSA gene.
The most potent AR shRNA ( Fig. 1) was directed to the ligand-binding domain and is similar to that reported by Chen et al. ( 37). Transfecting the AR shRNA into LNCaP cells completely inhibits the activation of a PSA-luciferase reporter by androgens (96%), indicating that AR shRNA expression is capable of functionally inactivating the AR. Because AR shRNA did not cause a complete inhibition of PSA induction by forskolin, IL-6, and EGF ( Fig. 2), there seems to be alternative or supplementary AR-independent pathways.
One pathway shared by all three nonsteroidal agents is the MAPK pathway ( 10, 26, 38). Inhibition of the MAPK pathway significantly decreased forskolin (96%), IL-6 (86%; Fig. 2B), and androgen-stimulated PSA promoter activation. This implies that MAPK contributes to both AR-dependent and AR-independent induction of PSA. AKT is constitutively activated in LNCaP cells due to their PTEN-null status ( 39). Our results indicate that this pathway may also be involved in ligand-independent activation of the PSA promoter because blocking PI3K could partially suppress forskolin and IL-6 activation of PSA ( Fig. 2B). Ligand-independent activation of AR-regulated genes may also occur by other means, such as HER2 ( 40) or unknown indirect effects ( 41). However, overall, the MAPK pathway seems to play a crucial role in PSA promoter activation.
Several groups have used siRNAs to silence the AR ( 28, 29, 37, 42– 45). In accordance, we have found that knockdown of AR using shRNA results in decreased PSA and AR expression ( Fig. 3) as well as reduced proliferation of LNCaP cells ( Fig. 4). In vitro, the significant decrease in mitotic activity as measured by BrdUrd incorporation was consistent with decreased rates of growth measured by MTS assays ( Fig. 4). In vivo tumor volume did not decrease with AR shRNA induction ( Fig. 5B), and there was a lack of Ki67 staining in the xenografts at the PSA nadir. Similar to Eder et al. ( 46), we also failed to show a correlation between tumor retardation and apoptotic cells in xenografts. There was no evidence of apoptosis induction, as the markers M30 CytoDEATH and cleaved poly(ADP-ribose) polymerase were not detected in Western blots of protein extracts derived from xenograft or cell lines, and FACS analysis of the stable cell lines indicated that the cells had undergone cell cycle arrest on addition of DOX (data not shown), suggesting that the primary mechanism for growth suppression by AR shRNA is cell cycle arrest.
Xenografts of hormone-refractory LNCaP sublines carrying a lentivirus constitutively expressing AR-targeted shRNA have been reported to grow more slowly than vector-transduced controls in castrated hosts ( 37). The tumors that did grow still expressed AR protein, indicating selection for cells that had escaped AR knockdown. Accordingly, we created a DOX-inducible AR shRNA-expressing LNCaP cell line that enabled us to investigate the in vivo role of AR in tumor progression. The only other reports specifically targeting AR in xenografts used antisense therapy ( 46, 47); however, this treatment was not as effective as castration ( 46). Our results with AR shRNA LNCaP xenografts strongly support a continuing role for AR in maintaining PSA production and cell growth after castration ( Fig. 5). Furthermore, knockdown of AR following DOX-induced expression of AR shRNA was very effective in delaying onset of androgen-independent expression of PSA and tumor growth.
Rapid increases in serum PSA and tumor growth 26 days after castration ( Fig. 5) indicate that LNCaP cells can recover quickly from reduced androgen levels if AR is still present. By comparison, reduction of AR itself results in a longer lag time to androgen independence. The observation that knockdown of AR in combination with castration is slightly more effective than reduction of AR alone implies that there may still be low levels of AR present in these cells capable of responding to androgen. Indeed, AR levels are suppressed initially but increase over an extended period and are accompanied by an increase in PSA ( Fig. 6). These results imply that there may be an escape from AR knockdown likely due to outgrowth of revertent or residual wild-type LNCaP cells, which do not express AR shRNA as a consequence of prolonged growth in the absence of the antibiotic used for selection.
In conclusion, we believe that this is the first demonstration that knocking down AR does not completely eliminate forskolin or IL-6 androgen-independent activation of the PSA promoter. In addition, we developed a unique inducible AR shRNA in a lentiviral vector. Although AR knockdown has been suggested as a means of therapy ( 18), our inducible AR shRNA system enabled us to provide definitive evidence that AR knockdown in established tumors results in impeded tumor growth, decreased serum PSA, and, more importantly, delayed progression to androgen independence. We further conclude that adaptive responses to androgen withdrawal that bypass the AR are insufficient to facilitate androgen-independent tumor growth.
The implications of our studies are that targeting AR for knockdown is extremely beneficial and superior to androgen ablation for treatment of advanced prostate cancer. However, it may be difficult to completely eliminate AR from all prostate cancer cells because there is strong selective pressure to maintain AR in these cells. Nevertheless, under conditions of low tumor burden, targeting the AR for knockdown may be a viable strategy for delaying or preventing progression of prostate cancer to androgen independence and thereby extending the survival of patients with advanced prostate cancer.
Grant support: Terry Fox Foundation, Health Canada, and Prostate Cancer Research Foundation of Canada.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
We thank Dr. Maxim Signaevsky for the scrambled shRNA in pSHAG-1-tet, Dr. Greg Hannon for pSHAG-1, Dr. Alice Mui for modified pSHAG-1-tet and pHR-EGFP plasmids, Mary Bowden for technical assistance, and Robert Bell (Head Bioinformatician of the Gene Array Facility at the Prostate Center, Vancouver, British Columbia, Canada) for statistical analyses.
- Received January 6, 2006.
- Revision received August 4, 2006.
- Accepted August 23, 2006.
- ©2006 American Association for Cancer Research.