|Home | About | Journals | Submit | Contact Us | Français|
The clinical utility of estrogens for treating prostate cancer (CaP) was established in the 1940s by Huggins. The classic model of the anti-CaP activity of estrogens postulates an indirect mechanism involving the suppression of androgen production. However, clinical and preclinical studies have shown that estrogens exert growth-inhibitory effects on CaP under low-androgen conditions, suggesting additional modes whereby estrogens affect CaP cells and/or the microenvironment. Here we have investigated the activity of 17β estradiol (E2) against androgen-independent CaP and identified molecular alterations in tumors exposed to E2. E2 treatment inhibited the growth of all four androgen-independent CaP xenografts studied (LuCaP 35V, LuCaP 23.1AI, LuCaP 49, and LuCaP 58) in castrated male mice. The molecular basis of growth suppression was studied by cDNA microarray analysis, which indicated that multiple pathways are altered by E2 treatment. Of particular interest are changes in transcripts encoding proteins that mediate immune responses and regulate androgen receptor signaling. In conclusion, our data show that estrogens have powerful inhibitory effects on CaP in vivo in androgen-depleted environments and suggest novel mechanisms of estrogen-mediated antitumor activity. These results indicate that incorporating estrogens into CaP treatment protocols could enhance therapeutic efficacy even in cases of advanced disease.
Despite substantial attention, the development of androgen-independent prostate cancer (CaP) is not well understood. Progression to an androgen-independent state represents resistance to suppression of the primary signaling pathway used to control recurrent CaP. Accordingly, an evaluation of the activities and mechanisms of new therapeutics that specifically target androgen-independent CaP growth is of special therapeutic interest.
For some 30 years, estrogens, particularly diethylstilbestrol (DES), were commonly used in the initial treatment of advanced CaP [1–6]. Originally, it was believed that the responses of CaP to estrogen therapy were mediated primarily by the suppression of the hypothalamic-hypophyseal axis and the consequent reduction in testosterone levels [7–10]. However, DES treatment was associated with significant side effects, and the Veterans Administration Cooperative Urological Research Group (VACURG), in 1967, recommended that hormonal therapy with DES be withheld until symptoms of metastatic disease appeared and that administration of DES at a level of 5 mg/day was associated with an excessive risk of cardiovascular mortality [11,12]. In a further study, VACURGII compared various dosages of DES and concluded that 1 mg/day is as effective as 5 mg/day in controlling T3 M+ CaP . In 1988, however, even this level of DES was found to be associated with a high risk for cardiovascular problems, mainly in patients over 75 years of age . The use of DES in the treatment of CaP ended with the advent of luteinizing hormone-releasing hormone analogs, which are now mainly used as a means of chemical castration.
Nevertheless, published studies suggest that: 1) estrogens inhibited the growth of CaP by mechanisms unrelated to androgen suppression; 2) patients treated with estrogen appeared to have survived somewhat longer than patients who had undergone surgical castration ; 3) administration of DES to patients with hormone-independent CaP suppressed prostate-specific antigen (PSA) and prolonged survival more effectively than administration of the antiandrogen flutamide ; and 4) Byar and Corle  commented that no form of endocrine therapy had been proven to be superior to 1 mg of DES daily. The hypothesis of direct inhibitory effects of estrogen on CaP is supported by observations that estrogen receptors are expressed in normal and neoplastic prostate epithelia [16–18], by observations that estrogens exhibit direct cytotoxic effects on CaP cells in vitro [19–23], and by our own demonstration of growth inhibition of CaP by 17β estradiol (E2) in the androgen-free environment of ovariectomized female mice .
The discovery of a second estrogen receptor, estrogen receptor β (ERβ), renewed interest in basic research involving estrogen pathways. Several reports have shown that ERβ is present in normal prostate epithelial cells as well as in CaP, and levels of ERβ messages and/or proteins appear to be downregulated during disease progression [16–18,25]. A straightforward hypothesis holds that ERβ transduces a growth-inhibitory effect of estrogen on CaP cells. In support of this hypothesis, a lower rate of cancer-related deaths was observed in CaP patients with ERβ versus CaP patients without ERβ , and an estrogenic compound operating through the ERβ receptor suppressed the growth of DU145 CaP cells [22,23]. In contrast to decreasing levels of ERβ with CaP progression, we have recently demonstrated that ERβ is expressed in a majority of CaP bone and soft-tissue metastases , as in another report on ERβ expression in a small number of CaP metastases . Together, these studies suggest that estrogen action against prostate carcinoma could involve ERβ or potentially other direct modes of action such that CaP growth may be restrained even in an androgen-independent state.
The current study was undertaken to determine whether estrogenic compounds can inhibit the growth of androgen-independent CaP and to investigate phenotypic changes associated with antitumor effects. Using human CaP xenografts, our results show that estrogenic compounds clearly suppress androgen-independent growth of CaP in castrated hosts, calling into question the traditional view that estrogen's activity against CaP depends solely on androgen suppression. The results indicate that estrogens may be especially useful in the treatment of androgen-independent CaP. We identified several novel molecular alterations resulting from tumor exposure to E2 that may contribute to E2-mediated tumor inhibition. Further studies are warranted to exploit the antitumor effects of E2 treatment in the context of advanced CaP.
Xenografts Androgen-sensitive PSA-producing CaP xenografts LuCaP 35 , LuCaP 23.1 [29,30], and LuCaP 58  (which all originated from lymph node metastases), and androgen-insensitive neuroendocrine-type CaP xenograft LuCaP 49 (which originated from omental fat metastasis)  were used. The xenografts were maintained and propagated in Balb/c nu/nu intact male mice. The androgen-independent variants of LuCaP 35V and LuCaP 23.1 were developed from parental tumors on regrowth after castration [28,31] and were maintained and propagated in castrated B17 Fox Chase SCID male mice (Charles River, Wilmington, MA).
Effects of E2 on recurrent LuCaP 35 after castration All animal procedures were performed in compliance with the University of Washington Institutional Animal Care and Use Committee and National Institutes of Health guidelines. In our first study, LuCaP 35 tissue bits were implanted subcutaneously into SCID male mice. Tumor growth was monitored by measuring tumor volume twice a week. Serum was collected weekly for PSA determination. Animals were castrated when the tumors reached 200 to 400 mm3. Animals with recurrent tumors (determined as two rising serum PSA values) were randomized into three groups of 10 animals each. Group 1 animals received placebo pellets.
Group 2 animals were supplemented with E2 by the subcutaneous implantation of slow-release Trocar pellets (90-day-release E2, 100–125 pg/ml; Innovative Research of America, Sarasota, FL), and group 3 animals were supplemented with DES pellets by the subcutaneous implantation of slow-release Trocar pellets (90-day-release DES, 0.01 mg; Innovative Research of America). Animals were sacrificed when tumors exceeded 1000 mm3 at 90 days postimplantation or when the animals became compromised. Student's unpaired two-tailed t-test was used to analyze the differences between groups.
Effects of E2 on LuCaP 35V in castrated male mice In additional experiments performed to determine the effects of E2 on proliferation and gene expression, we used the androgen-independent xenograft LuCaP 35V . SCID male mice were castrated at 8 weeks of age and implanted with LuCaP 35V tumor bits at least 2 weeks after surgery. Tumor growth was monitored by tumor measurements twice a week using calipers, and tumor volume was calculated as 0.5236LHW. Blood samples were collected weekly for the determination of serum PSA levels (IMx Total PSA Assay; Abbott Laboratories, Abbott Park, IL). When tumors reached 200 to 400 mm3, the animals were randomized into two groups. Group 1 was supplemented with E2 by the subcutaneous implantation of slow-release Trocar pellets (60-day release, 0.05 mg; Innovative Research of America). Group 2, which received placebo pellets, was the control group. Five animals from each group were sacrificed on days 1, 3, and 7 post-implantation of E2 pellets. One hour before sacrifice, the animals were injected intraperitoneally with 80 mg/kg body weight 5-bromo-2-deoxyuridine (BrdU; Sigma-Aldrich Co., St. Louis, MO) for evaluation of tumor cell proliferation. Tumors were fixed in formalin and embedded in paraffin. The 10 remaining animals in each group were monitored for long-term assessment of tumor growth and PSA production after E2 treatment. Animals were sacrificed when tumors exceeded 1000 mm3 at 60 days post-implantation or when the animals became compromised. Tumors were frozen in liquid nitrogen and stored at -80°C and/or fixed with formalin and embedded in paraffin, and serum was collected for determination of E2 levels (IMx Estradiol Immunoassay; Abbott Laboratories). Student's unpaired two-tailed t-test was used to analyze the differences between groups, and a log-rank test was used to evaluate differences in survival.
Effects of E2 on the growth of LuCaP 23.1AI, LuCaP 49, and LuCaP 58 in castrated male mice To investigate whether the E2 inhibition of androgen-independent growth occurs with other CaP cells (not just LuCaP 35 lines), we set up similar experiments with three additional xenografts: LuCaP 35AI, LuCaP 49, and LuCaP 58. The experimental design was the same as for the study with LuCaP 35V. Tumor bits were implanted in castrated male mice (aiming for n = 10 per group) at least 2 weeks after surgery, and tumor growth and PSA levels were monitored. Animals bearing each particular xenograft were randomized into two groups (tumors 200–400 mm3). Group 1 was supplemented with E2 by the subcutaneous implantation of slow-release Trocar pellets (60-day release, 0.05 mg; Innovative Research of America). Group 2, which received placebo pellets, was the control group. Animals were sacrificed when tumors exceeded 1000 mm3 at 60 days postimplantation or when the animals became compromised. Tumors were frozen in liquid nitrogen and stored at -80°C and/or fixed with formalin and embedded in paraffin. Student's unpaired two-tailed t-test was used to analyze differences between groups.
Samples of LuCaP 35V tumors treated with E2 for 1, 3, and 7 days, and control tumors were fixed in formalin and embedded in paraffin. An anti-BrdU immunohistochemistry kit was used to assess the number of proliferating cells (Zymed, San Francisco, CA). Five-micrometer sections of paraffin-embedded tissues were used for the analysis, as recommended by the manufacturer. Apoptosis in tumors was assessed with a FragEL DNA fragmentation detection kit from Oncogene (La Jolla, CA), as recommended by the manufacturer. Positive nuclei or apoptotic cells were counted in five representative fields containing ~1000 cells in three samples of treated and untreated tumors from each time point. Statistical analysis was performed using Student's t test.
Seven hundred to 900 mm3 of LuCaP 35V tumors grown and passaged in castrated SCID mice were harvested for the isolation of epithelial cells . Isolated cells were rinsed thrice and plated overnight in 10% charcoal-stripped fetal bovine serum (Hyclone, Logan, UT) in phenol red-free RPMI 1640 (Invitrogen, Carlsbad, CA). LuCaP 35V cells were treated with 10-8 M E2 or vehicle (0.01% EtOH) for 4 hours.
Following treatment with E2 or vehicle, nuclear and cytoplasmic fractions were prepared as previously published . Proteins (25 µg/well) were separated by 12.5% sodium dodecyl sulfate-polyacrylamide gel electrophoresis and transferred to polyvinylidene difluoride membrane (PVDF) membranes. Blots were blocked in a 1:1 solution of NaP-Sure blocker (Geno Technology, Inc., St. Louis, MO) and Tris-buffered saline + 0.1% Tween-20 for 2 hours, then probed with a rabbit polyclonal antibody against ERβ (Affinity BioReagents, Golden, CO) for 1 hour at room temperature. ERβ immunoreactivity was detected using a goat anti-rabbit secondary antibody conjugated with horseradish peroxidase (1:2000; Amersham, Piscataway, NJ). Blots were developed using the Amersham ECL.
Nuclear extracts fromLuCaP 35V treated with 10-8 M E2 or vehicle (0.01% EtOH) for 4 hours (25 µg) were incubated with 50 fmol of dsDNA probes for 30 minutes at 37°C in a buffer containing: 20 mM Tris (pH 8), 10 mM NaCl, 3 mM EDTA, 0.05% Nonidet P-40, 2 mM DTT, 4% glycerol, 1 mM MgCl2, and 1 µg of poly dI-dC (Amersham). The binding consensus sequences used were an estrogen response element (ERE; GGATCTAGGTCACTGTGACCCCGGATC) and a mutated form of ERE (GGATCTAGTACACTGTGACCCCGGATC; Santa Cruz Biotechnology, Inc., Santa Cruz, CA). Double-stranded DNA were end-labeled with [γ-32P]ATP (Amersham) using T4 polynucleotide kinase (Promega, Madison, WI). For competition studies, 50 fmol of unlabeled probe was added to the reaction. Protein-DNA complexes were separated in 4% nondenaturing polyacrylamide gels.
Tumors from animals treated with E2 for 60 days and control tumors were homogenized using an Omni TH homogenizer (Omni International, Warrenton, VA), and RNA was extracted using TriPure Isolation Reagent (Roche, Indianapolis, IN), according to the manufacturer's instructions. RNA quantity was determined based on A260, and the integrity of RNA was confirmed by agarose gel.
PEDB cDNA microarrays containing ~7000 human prostate-derived cDNA clones were prepared on poly-l-lysine-coated glass microscope slides using a robotic spotting tool, as previously described [34–36]. Equal amounts of total RNA from five tumors of LuCaP 35V (control) and E2-treated LuCaP 35V (treatment) were pooled, and cDNA array experiments and analysis were performed as previously described . For individual experiments, every cDNA was represented twice on each slide, and the experiments were performed in triplicate with a switch in fluorescent labels to account for dye effects, producing six data points per cDNA clone per hybridization probe. Data were filtered to exclude poor-quality spots, were normalized, and included clones whose expression was measurable in at least two of three arrays, reducing the initial list of 6720 clones to 5163 clones.
To compare the overall expression patterns of replicate LuCaP 35V (control) and E2-treated LuCaP 35V (treatment) arrays, log2 ratio measurements were analyzed using the SAM procedure  (http://www-stat.stanford.edu/_tibs/SAM/). A one-sample t-test was used to determine whether the mean gene expression of E2-treated LuCaP 35V versus LuCaP 35V (control) differed significantly from zero. A false discovery rate (FDR) of < 1% was considered significant. Clones differentially expressed with an FDR < 1% were stratified based on fold change, and we chose to further evaluate only those with an average log2 (E2-treated/control) > 0.58 or < -0.58, corresponding to a differential expression effect of 1.5-fold or greater. We assigned differentially expressed genes to the following functional categories based on their annotations in the Gene Ontology database : metabolism, immune/inflammatory response, proliferation/differentiation/apoptosis, signal transduction, structure/adhesion/motility, transcription regulation, translation protein synthesis, transport, or other/unknown.
To determine whether phenotypic changes observed in E2-treated tumors were enriched for genes in certain pathways, cDNA array results were subjected to Gene Set Enrichment Analysis (GSEA) . For this analysis, interferon (IFN)-regulated, androgen-regulated, and estrogen-regulated gene sets were tested against our data. IFN-regulated and estrogen-regulated gene sets were generated from Super-Array Bioscience Corporation GEArray pathway-focused gene lists (http://www.superarray.com), and the androgen-regulated gene set was generated based on the results of DePrimo et al. . To assess the statistical significance of the enrichment score observed in the data set for the three gene sets, we used permutation testing of phenotype labels (e.g., E2-treated versus controls), generating a nominal (NOM) P value. An FDR statistic was computed to adjust for gene set size and multiple hypothesis testing, with an FDR of < 25% considered significant.
Quantitative reverse transcription-polymerase chain reaction (qRT-PCR) First-strand cDNA synthesis was performed with 1.0 µg of pooled RNA from five animals of the E2 and control groups using oligo-dT18 primers according to the manufacturer's instructions (Clontech, Palo Alto, CA). Real-time PCR was carried out on cDNA samples using Platinum Quantitative PCR SuperMix-UDG reagent (Invitrogen) and performed on a Rotor-Gene 2000 (Corbett Research, New South Wales, Australia). PCR primers were designed to span an intron-exon boundary and to avoid amplification of any known pseudogene. Primers for the messages evaluated are listed in Table 1. Two microliters of cDNA was used per reaction with 200 nM primers, 0.5 x Syber Green 1 (Molecular Probes, Eugene, OR), and 5.5 mM MgCl2. The PCR reaction parameters were as follows: 50°C for 2 minutes and 95°C for 2 minutes (one cycle), followed by 35 cycles at 95°C for 10 seconds and annealing/extension at either 65°C or 69°C for 30 seconds; the final extension was 72°C for 7 minutes. PCR reaction products were confirmed by agarose gel electrophoresis. Standard curves for each amplicon were generated from a four-fold dilution series of LNCaP cDNA run in duplicate (all standard curves had r > 0.99). Reactions were carried out in duplicate, and expression levels were calculated from a standard curve.
Normalization strategyThe normalization scheme applied to real-time PCR results was based on the method of Vandesompele et al. . This method employs multiple internal control genes to identify the most stably expressed control genes in samples of interest. The following messages were evaluated for use as internal controls: epithelial glycoprotein (EGP), glyceraldehyde-3-phosphate dehydrogenase (GAPDH), hydroxymethylbilane synthase (HMBS), hypoxanthine phosphoribosyltransferase 1 (HPRT1), and proteasome (prosome, macropain) subunit, β type, 6 (PSMB6). Real-time PCRon pooled samples was performed in duplicate, and expression levels were calculated based on standard curves, as above. The average expression levels were imported into the geNorm program (http://allserv.rug.ac.be/~jvdesomp/genorm/) to determine the two most stably expressed internal control genes. Briefly, geNorm determines the gene stability measure M as the average pairwise variation between a particular internal control gene and all other control genes. The stepwise exclusion of endogenous control genes with the highest M values resulted in the selection of GAPDH and EGP as the most stably expressed control genes. The normalization of the real-time PCR data of the gene of interest was accomplished by dividing raw expression levels by the geometric mean of the most stable endogenous control.
LuCaP 35 is an androgen-sensitive CaP xenograft, expressing PSA and wild-type androgen receptors (ARs), which recapitulates a response to androgen ablation and the development of androgen-independent CaP similar to that observed in humans . Its growth in intact female mice is suppressed in comparison to that in ovariectomized female mice . Therefore, we have chosen this xenograft for initial evaluation of the effects of estrogenic compounds in male mice. Surgical castration of intact male mice bearing LuCaP 35 CaP xenografts resulted in a reproducible time-dependent reduction in tumor volume and PSA serum levels. Recapitulating human disease, 88% of the tumors eventually recurred in the androgen-depleted environment, with a range in time to recurrence of 32 to 91 days (median = 61.5 days; Figure 1, A and B). Tumor recurrence was defined as two consecutive rising values of serum PSA. Without treatment, these androgen-independent tumors continued to grow and reached a size of ~1000 mm3 by days 24 to 31 post-castration. Administration of E2 or DES inhibited the growth of recurrent LuCaP 35 tumors; at 104 days after castration, the tumor volumes were 134.3 ± 16.4 mm3 (mean ± SEM) for E2 (with PSA levels of 1.82 ± 0.66 ng/ml) and 49.8 ± 12.1 mm3 for DES (with PSA levels of 3.20 ± 1.86 ng/ml). Tumor volumes and PSA levels decreased, and none of the tumors reached an estrogen-resistant state during the course of the study (90 days of treatment). PSA values closely followed tumor volume. Three animals from the E2-treated and DES-treated groups were monitored for an additional 60 days after expiration of the estrogen pellets. Tumor volumes and PSA serum levels in these animals started to increase during this period (Figure 1). The tumors in animals that were treated with E2 reached 587.6 ± 194.0 mm3 (P = .0008 from 90 days after pellet expiration), with concordant rises in PSA serum levels to 55.33 ± 21.18 (P = .003; to the levels when pellets expired). Tumors in DES-treated animals started to increase in volume more slowly than E2-treated tumors after pellet expiration; the tumor volumes increased 1.5-fold (79.43 ± 32.5 mm3) but did not reach significance (P = .3075), and PSA serum levels began to rise (17.23 ± 11.20 ng/ml; P = .0533). As observed in our previous study in female mice, administration of E2 inhibited the growth of androgen-independent LuCaP 35V xenografts in castrated male mice as well. The tumor volume of LuCaP 35V-bearing animals treated with E2 increased minimally over the original volume during the 60-day treatment period (Figure 2A). However, the tumor size of LuCaP 35V in the control group increased from the time of enrollment up to the time of sacrifice (days 25–35; tumor volume = 1000 mm3; Figure 2A) (on day 32, P < .0001). PSA serum levels closely paralleled tumor volumes (on day 28, P = .0021) (Figure 2B). Levels of E2 in the control group of castrated animals with LuCaP 35V (untreated) were below the limit of assay detection (< 25 pg/ml). The level of E2 at the time of sacrifice (60 days postimplantation of E2 pellets) was 127.1 ± 22.5 pg/ml in treated LuCaP 35V animals. Survival analysis, using tumor size (≥ 1000 mm3) as a death criterion, showed that E2 dramatically prolonged the survival of LuCaP 35V-bearing animals, as determined by log-rank test (P < .0001; Figure 2C).
The growth of the three additional CaP xenografts LuCaP 23.1AI, LuCaP 49, and LuCaP 58 in an androgen-free environment was inhibited by E2 administration to varying degrees (Figure 3). The tumor volume of LuCaP 23.1AI treated with E2 decreased, with significant differences from untreated tumors after 7 days of treatment (P = .00089), resulting in the near-disappearance of the tumors by day 35. PSA serum levels closely followed the tumor volume. LuCaP 58 growth was also inhibited by E2 treatment, but to a lesser extent; the tumor volume increased minimally over the original volume during the 60-day treatment period (Figure 2A), reaching significant inhibition versus untreated tumors on day 7 (P = .0137). LuCaP 49, a neuroendocrine CaP xenograft in which ARs are absent, was also inhibited by E2 administration, but the pattern of inhibition was different from those of the other three xenografts. No significant inhibition was observed for the first 10 days of treatment, after which significant inhibition was reached (14 days, P = .0289). E2-treated LuCaP 49 tumors continued growing, but at a rate slower than that of untreated tumors.
To evaluate the mechanisms mediating LuCaP 35V tumor reduction after E2 treatment, we measured the incorporation of BrdU in untreated LuCaP 35V tumors versus tumors from mice that received E2 for 1, 3, and 7 days. The number of proliferating tumor cells decreased to 82.7 ± 7.3% of untreated tumors after 1 day (mean ± SEM), to 65.7 ± 4.2% (P = .0063) after 3 days, and to 65.4 ± 10.1% (P = .0105) after 7 days of E2 treatment (Figure 3). The rate of apoptosis in E2-treated and untreated tumors, as measured by the terminal deoxynucleotidyl transferase (TdT) FragEL DNA fragmentation detection, was not significantly different (data not shown).
Comparative analyses of cDNA microarray gene expression profiles derived from LuCaP 35V xenografts treated with E2 and untreated controls identified 300 cDNA whose expression levels were significantly associated with E2 treatment (FDR < 1%) and exhibited a > 1.5-fold difference in expression level. Consolidation of redundant clones resulted in 233 unique genes, of which 129 were downregulated and 104 were upregulated following E2 treatment (Tables 2 and and3).3). E2 treatment resulted in significant increases in the expression of several genes that are involved in immune responses (Table 2). These include major histocompatibility complex (MHC) class I/II proteins, IFN-induced transmembrane protein 1 (IFITM1), IFN-induced transmembrane protein 3 (IFITM3), IFN-induced protein with tetratricopeptide repeats 1 (IFIT1), IFNα-inducible protein 27 (IFI27), and IFN-regulatory factor 1 (IRF1).
We have used GSEA to evaluate whether phenotypic changes caused by E2 treatment in LuCaP 35V were associated with enrichment for IFN-regulated, androgen-regulated, and estrogen-regulated genes. Our analysis showed a significant enrichment of IFN-regulated genes in E2-treated LuCaP 35V tumors (NOM P < .001), which remained significant when adjusted for gene set size and multiple hypothesis testing (FDR = 11.0%) (Figure 4A). Significant enrichment was also detected when the androgen deprivation-downregulated gene set was compared to our results (NOM P < .001); this enrichment also remained significant when adjusted for gene set size and multiple hypothesis testing (FDR = 21.3%) (Figure 4B). Estrogen-regulated genes were also enriched in phenotypic alterations after E2 treatment (NOM P < .001); however, these changes were not significant when adjusted for gene set size and multiple hypothesis testing (FDR = 54.5%). We hypothesize that this is due to the fact that changes in the expression of these genes occur in both up and down directions, and also due to inclusion in the list of genes that are altered in breast cancer, which may not be relevant to this study (Figure 4C).
ERβ (55 kDa) was detected by Western blot analysis in nuclear extracts from—but not in the cytoplasm of—LuCaP 35V and E2-treated LuCaP 35V (Figure 5A). E2 treatment increased levels of ERβ in the nucleus by approximately 30%. Using EMSA, we showed that ERβ in the nucleus is able to bind to DNA. E2 treatment slightly increased levels of ERβ/DNA complexes (Figure 5B). The specificity of the interaction was demonstrated by the disappearance of the specific band in control reactions with a mutated ERE (xERE).
We performed qRT-PCR analysis to confirm the cDNA microarray results for selected genes of potential biologic importance. All messages whose expression was determined to be upregulated by cDNA array analysis were also increased by qRT-PCR in E2-treated LuCaP 35V (Figure 6). We next examined whether immune response-related genes found to be upregulated by E2 treatment of LuCaP 35V xenografts were also altered by E2 treatment in other CaP xenografts. In LuCaP 58, the patterns of E2 alteration in the expression of these genes were similar to those in LuCaP 35V. In contrast, in LuCaP 49 (a neuroendocrine CaP xenograft whose growth suppression was less pronounced), the expression of evaluated genes was minimally altered (Figure 6). LuCaP 23.1 regressed almost completely after E2 treatment, and, unfortunately, there was insufficient tissue remaining for analysis. Gene expression changes in LuCaP 35 tumors treated with E2 or DES after castration were also evaluated. We found that the expression of genes related to immune regulation was altered by E2 and DES treatment, as in LuCaP 35V tumors. We continued to examine tumor gene expression levels after expiration of the E2 pellets and found that levels of E2-induced messages decreased, indicating dependence on the presence of E2 (Figure 7).
Several studies dating back to the 1980s have suggested that mechanisms other than androgen suppression may be involved in the estrogen-mediated inhibition of CaP growth. Estrogens appear to be slightly more effective in treating CaP than other means of androgen suppression . Compounds with estrogenic activity are capable of exerting direct cytotoxic effects on androgen-independent CaP cells in vitro [19–23]. Our data, obtained from the androgen-deficient environment of female mice  and from the present work, show that estrogens have powerful growth-inhibitory effects on CaP in vivo.
In the present study, we have shown that E2 and DES both inhibit the growth of androgen-independent CaP tumors in the androgen-depleted environment of castrated male mice. These data clearly demonstrate that E2 exhibits effects on CaP cells that are unrelated to the suppression of the hypothalamic-hypophyseal axis and the subsequent decrease in testosterone. This novel observation prompted us to characterize the effects of E2 on androgen-independent CaP at the molecular level by profiling transcript alterations. Although many of the genes differentially regulated by estrogen in this system are of unclear significance, others have quite plausible roles in the observed growth inhibition on the basis of their established functions. Among these are genes involved in signal transduction, cellular metabolism, and the control of transcription and translation. We also observed substantial changes in genes that function to regulate immune responses—a mechanism that may contribute to tumor growth-inhibitory effects resulting from estrogen treatment.
Among immune response-related genes altered by E2 treatment in CaP are those modulating cellular responses to IFNs. This group was found to be significantly enriched in the set of genes upregulated by E2 when tested by GSEA using an independently generated list of IFN-regulated genes. The increased expression of IFN-regulated genes is of particular interest due to the direct antitumor activities reported for these cytokines [43–50]. Our results are in keeping with the results on the upregulation of IFN-regulated genes in LNCaP CaP cells following exposure to the estrogenic herbal preparation PC-SPES  and the induction of IFNγ-regulated genes after E2 treatment in other tissues . In addition, tamoxifen has been shown to enhance IFN-regulated gene expression in breast cancer cells . Specifically, IRF1, whose expression was increased three-fold by E2 (qRT-PCR data), has been described as a negative regulator of proliferation  and has exhibited tumor-suppressor activities in breast cancer cells . These published observations and our results are consistent with a model in which IFN and genes regulated by IFN modulate a component of the growth-inhibitory activity of E2 toward androgen-independent CaP cells.
E2 treatment significantly increased the expression of several MHC class I/II transcripts in the androgen-independent LuCaP 35V xenograft. Similarly, the upregulation of MHC class I transcripts has been observed in LNCaP cells on PC-SPES exposure . MHC class I molecules are expressed in most human cells and play a pivotal role in the immune response to viruses and tumor cells. Tumor cells often evolve mechanisms to modulate or escape immune surveillance through the downregulation of MHC class I molecules [56–60]. IFNγ treatment, like E2 treatment in our studies, has been reported to upregulate the expression of MHC class I/II molecules in CaP cell lines [44,58,59]. According to this evidence, the treatment of advanced CaP patients with E2 might result not only in direct inhibitory effects but also in the stimulation of T-cell attack on tumors by the upregulation of MHC proteins. Such a mechanism could not be directly tested in our study, which employed immune-compromised SCID mice, but it represents an independent potential benefit of E2 treatment that could be exploited in the context of clinical therapies employing vaccine or other immunomodulatory treatment strategies.
DES has been reported to be ineffective in inhibiting LuCaP 35 growth in intact male mice . We also observed that E2 did not inhibit LuCaP 35 growth in intact male mice (data not shown). These results suggest that phenotypic changes caused by E2 treatment are specific to an androgen-depleted environment. In contrast to our E2 data, raloxifene, an estrogen receptor antagonist, has been reported to inhibit the growth of both androgen-sensitive and androgen-independent CaP in vitro [20,21]. Raloxifene has also been reported to delay CaP development in probasin/SV40 Tantigen transgenic rats  and to inhibit the growth of both androgen-sensitive and androgen-independent variants of the CWR22 CaP xenograft . Thus, the emerging picture of estrogenic effects on androgen-independent CaP is complex, possibly involving multiple mechanisms, some of which may involve signal transduction by estrogen receptors. Additional preclinical studies are clearly warranted to deconvolute these effects.
A potential mechanism whereby E2 may cause alterations of the gene expression profile we have observed in CaP cells is signal transduction through ERβ expressed by CaP cells. It has been reported that ERβ expression declines as CaP develops in the prostate gland, but we and others have shown that it reappears in lymph node and bone metastases . This apparent discrepancy is probably explained by the recent findings of the reversible epigenetic regulation of ERβ in CaP metastases . We have shown previously that the xenografts used in this study express ERβ . In the present study, we have shown that the androgen-independent LuCaP 35V xenograft expresses ERβ protein in a form that is capable of DNA binding, and that ERβ levels in nuclei and DNA-binding activities are increased on E2 treatment. Together, these results suggest the possibility that E2-mediated inhibition is, at least in part, transduced by ERβ signaling, but further studies are required to demonstrate direct involvement of ERβ with these phenomena. One important aspect of preclinical testing involves the use of models that mimic the disease in patients. If it is eventually found that E2 is beneficial in advanced CaP and that the effects are mediated by ERβ, then evaluation of the expression of ERβ in patient tumors could prove to be valuable in treatment decisions, as is the case with HER2/Neu and herceptin treatment today.
The E2-inhibitory effects observed cannot be caused by suppression of the hypothalamic-hypophyseal axis reduction in testosterone levels because the tumors were grown in castrated male mice. However, our data do suggest that AR signaling may be at least partially involved in the inhibitory effects observed. All of the xenografts, except LuCaP 49, express AR (data not shown), and the inhibition of LuCaP 49 by E2 was less pronounced than in other xenografts. Moreover, GSEA showed that genes in an independently generated list of genes downregulated by androgen deprivation were significantly enriched in the phenotype of E2-treated LuCaP 35V, with about half of the genes downregulated by E2 and half upregulated by E2. For example, the expression of heat shock protein 70, which is downregulated after castration , was upregulated by E2 treatment (Table 2). These results illustrate the complexity of these signaling networks. Further studies are needed to delineate the action of E2 on AR signaling in CaP cells.
The results reported here support the multifaceted roles of estrogen in the inhibition of androgen-independent CaP growth. These observations extend the traditional view of estrogen activity beyond the suppression of circulating concentrations of androgens. Direct cellular effects and the modulation of immune responses represent additional potential mechanisms that could be further exploited through combination therapies. Given that estrogens also decrease bone lysis caused by androgen suppression  and may ameliorate cognitive side effects associated with low testosterone , the use of estrogens should be considered as a viable first-line treatment strategy for androgenindependent CaP.
PSA reagents were kindly provided by Abbott Laboratories. The authors would like to thank Janna Quinn and Austin Odman for excellent technical assistance, Stacy Moore for help with cDNA arrays, Michael Corey for editorial assistance, and Bruce Montgomery, Bernd Stein, and Shuk-Mei Ho for helpful discussions.
1This research was supported by grants DAMD17-01-1-0114 (E.C.) and W81XWH-04-1-0198 (E.C.) from the US Army Medical Research Material Command Prostate Cancer Research Program, by grants CA97186 and CA85859 (P.S.N.) from the National Institutes of Health, and by a grant from the Signal Pharmaceutical Research Division of Celgene.