Endostatin inhibits androgen-independent prostate cancer growth by suppressing nuclear receptor-mediated oxidative stress

Abstract

Androgen-deprivation therapy has been identified to induce oxidative stress in prostate cancer (PCa), leading to reactivation of androgen receptor (AR) signaling in a hormone-refractory manner. Thus, antioxidant therapies have gained attention as adjuvants for castration-resistant PCa. Here, we report for the first time that human endostatin (ES) prevents androgen-independent growth phenotype in PCa cells through its molecular targeting of AR and glucocorticoid receptor (GR) and downstream pro-oxidant signaling. This reversal after ES treatment significantly decreased PCa cell proliferation through down-regulation of GR and up-regulation of manganese superoxide dismutase and reduced glutathione levels. Proteome and biochemical analyses of ES-treated PCa cells further indicated a significant up-regulation of enzymes in the major reactive oxygen species (ROS) scavenging machinery, including catalase, glutathione synthetase, glutathione reductase, NADPH-cytochrome P450 reductase, biliverdin reductase, and thioredoxin reductase, resulting in a concomitant reduction of intracellular ROS. ES further augmented the antioxidant system through up-regulation of glucose influx, the pentose phosphate pathway, and NAD salvaging pathways. This shift in cancer cell redox homeostasis by ES significantly decreased the effect of protumorigenic oxidative machinery on androgen-independent PCa growth, suggesting that ES can suppress GR-induced resistant phenotype upon AR antagonism and that the dual targeting action of ES on AR and GR can be further translated to PCa therapy.—Lee, J. H., Kang, M., Wang, H., Naik, G., Mobley, J. A., Sonpavde, G., Garvey, W. T., Darley-Usmar, V. M., Ponnazhagan, S. Endostatin inhibits androgen-independent prostate cancer growth by suppressing nuclear receptor-mediated oxidative stress.

Castration-resistant prostate cancer (CRPC) is a lethal form of advanced disease for which treatment options are limited (1, 2). It is now accepted that CRPC retains its dependency on androgen receptor (AR) function; thus, AR targeting remains a mainstay for therapeutic intervention (2–8). However, the molecular mechanism behind resistance to the current AR-directed therapies is yet to be determined.

The castration-resistant phenotype is derived from an adaptive response of prostate cancer (PCa) to the frontline androgen deprivation therapy (ADT) and the ensuing hormone-refractory tumor relapse with aberrant AR activity (1, 4). Relevant to this, a growing body of evidence indicates that oxidative stress induced by ADT contributes to CRPC development via reactivation of AR signaling (9–11). Because androgens have a protective role for oxidative stress (10, 12), chemical or surgical castration is often reported to increase the levels of reactive oxygen species (ROS) by up-regulating NADPH oxidase and down-regulating antioxidant gene expression of manganese superoxide dismutase (SOD2), thioredoxin, and peroxiredoxins (9, 11). In addition, age-dependent reduction of androgens is known to induce redox imbalance by decreasing expression of SOD2, catalase, and glutathione peroxidase (12). SOD2 gene repression and the resultant ROS augmentation are well documented for their role in reactivation of AR in CRPC (13). Suppression of SOD2 is known to promote intratumoral repletion of androgens through de novo steroidogenesis, up-regulation of AR coactivator expression, and induction of IL-6R expression that promotes androgen-independent AR activation (9, 11–13). Conversely, synthetic antioxidant N-acetyl-cysteine and SOD mimetics have been shown to suppress ROS-induced CRPC growth by inhibiting AR activation (14, 15). In addition, antioxidants derived from natural products have been shown to decrease the risk of PCa mortality, suggesting the therapeutic potential of preventing PCa progression (5, 9, 15). Collectively, these studies strongly suggest that oxidative stress induction is a potential mechanism driving the emergence of the castration-resistant phenotype and may be responsible for CRPC progression. In addition to ADT, clinical management of PCa with taxane and/or radiotherapy is known to induce oxidative stress (10, 16), which can lead to the onset of ROS-induced resistance during disease progression.

The adverse effects of ADT and its association with oxidative stress also include osteoporosis, obesity, and a negative cognitive effect (9). Notably, the underlying pathophysiological mechanisms of these disorders are unequivocally overlapped with their implications in glucocorticoid signaling. The glucocorticoid-induced oxidative stress in neuronal cells is almost completely blocked by RU486, a glucocorticoid receptor (GR) antagonist, indicating that glucocorticoid-GR axis mediates oxidative stress (17, 18). In the context of CRPC, recent studies identified that GR up-regulation confers a resistance to second-generation AR-targeted therapies through alternative AR-target gene transcription (19, 20). Therefore, GR appears to play an important role for oxidative stress-induced CRPC development and acquired resistance to PCa therapies.

Recent studies have demonstrated various cellular functions of endostatin (ES) apart from its known suppressive role in angiogenic cascade. Wang et al. (21) recently reported a neuroprotective role of ES in modulating synaptic homeostasis. Recently identified ATPase activity and the complex gene regulations indicated that the antitumor effect of ES includes depletion of intracellular ATP in tumor-associated endothelium (22). In addition, ES can modulate approximately 12% of the human genome, further supporting the pleiotropic effects beyond inhibiting angiogenesis (23). In line with these findings, our previous study identified a novel interaction of ES with AR, leading to inhibition of PCa cell proliferation and AR-target gene expression (24).

Given the GR as a possible upstream stress inducer and the binding interface for ES on AR is conserved in GR, we hypothesized that impairment of AR and GR functions by ES would have antagonistic effects on protumorigenic oxidative stress and redox imbalance. In this study, we demonstrate that GR mediates androgen-independent AR activation through oxidative stress induction as a survival mechanism and progression to androgen-independent growth. Further, we provide evidence for a molecular mechanism of ES in the regulation of GR-mediated oxidative response by reversing the cancer cell homeostasis from a protumorigenic function to an antitumor response.

MATERIALS AND METHODS

Cell lines

Human PCa lines (LNCaP and DU145 cells) and the mouse PCa cell line TRAMP-C1 were cultured as previously described (24, 25). The 22Rv1 cell line was grown in RPMI 1640 (Thermo Fisher Scientific, Waltham, MA, USA) supplemented with 10% fetal bovine serum (Gemini Bio-Products, Sacramento, CA, USA) and 1% penicillin/streptomycin (Thermo Fisher Scientific).

Expression of recombinant human ES and GR

The recombinant human ES was prepared using bacterial expression system as previously described (24). For the recombinant GR expression, the complete cDNA of human GR, obtained from the cDNA Resource Center (Bloomsburg University, Bloomsburg, PA, USA; http://www.cdna.org), was subcloned into pET24a Escherichia coli expression vector (EMD Millipore, Billerica, MA, USA) using 5′-BamHI and 3′-XhoI restriction sites for gene insertion. DH5α cells (Bioline, Taunton, MA, USA) were used for plasmid propagation, and the identified positive clones were transformed into Origami 2(DE3) strain (EMD Millipore) for recombinant protein expression. For coexpression of ES and GR, pET21a-ES and pET24a-GR plasmids were cotransformed into Origami 2(DE3) strain and selected under the 2-antibiotic pressure of kanamycin (Sigma-Aldrich, St. Louis, MO, USA) and ampicillin (Sigma-Aldrich). The positive clones were grown overnight at 37°C, and expression of ES and GR was induced with 1 mM isopropyl-d-thiogalactopyranoside (American Bioanalytical, Natick, MA, USA) at 25°C for 14–16 h. Induced cells were harvested and resuspended in PBS containing EDTA-free protease inhibitor cocktail (Thermo Fisher Scientific), and the soluble cell lysate was prepared by sonication. Expression of recombinant ES and GR was confirmed by immunoblots using anti-ES (Leinco Technologies, Fenton, MO, USA), anti-GR (Santa Cruz Biotechnology, Santa Cruz, CA, USA), anti–T7-tag (GeneTex, Irvine, CA, USA), and anti-His-tag (GeneTex) antibodies.

Proteome analysis

LNCaP cells were treated with 1 μM recombinant human ES for 72 h and harvested for proteomic analysis (UAB Mass Spectrometry/Proteomics Shared Facility) using the GeLC (1D-PAGE-LC-MS) approach as previously described (26). Briefly, the soluble cell lysates were separated on a 1D gel and digested with trypsin, and the peptide digests were subject to LC-MS/MS. A list of protein IDs was generated based on Sequest (Scripps Research Institute, La Jolla, CA, USA; http://fields.scripps.edu/sequest/) search results, and the matching peptides were filtered and quantified using ProteoIQ (Premier Biosoft, Palo Alto, CA, USA). Quantification was based on normalized spectral counting for each assigned spectrum. The results are listed in Supplemental Tables 1–4.

Glutathione assay

Intracellular levels of reduced form of glutathione (GSH) were analyzed using a cell-based fluorometric assay kit (Cayman Chemical, Ann Arbor, MI, USA) according to the manufacturer’s protocol. For GSH quantitation, LNCaP, 22Rv1, and DU145 cells from all experimental conditions were lysed, and the soluble cell lysates were transferred to a corresponding well in the black 96-well plate. After 1–2 h of incubation with monochlorobimane substrate solution, fluorescence intensity was measured at excitation and emission at wavelengths of 380 and 460 nm, respectively. The protein concentration of each lysate was measured by bicinchoninic acid protein assay (Thermo Fisher Scientific) for normalization of GSH levels of each sample. Total GSH levels of all experimental groups were expressed as relative values (means ± SEM) of controls. For in situ staining, LNCaP cells (5 × 10³ cells per well) were incubated with recombinant ES for 48 h. The monochlorobimane was added to cell cultures with a final ratio of 1:10. After incubation for 30 min, GSH levels of cells were visualized at ×100 original magnification with a fluorescence microscope (DMI 4000B; Leica Microsystems, Wetzlar, Germany).

ROS imaging

For in situ staining, LNCaP cells (1 × 10⁴ cells per chamber) were plated on poly-l-lysine-coated chamber slides (BD Biosciences, San Jose, CA, USA) and incubated with 1 μM ES for 48 h. The cells were then incubated with dihydroethidium (Thermo Fisher Scientific) for 30 min and imaged at ×100 original magnification in Leica DMI 4000B fluorescence microscope to qualitatively determine ROS levels.

Glucose uptake assay

Glucose transport rates of LNCaP cells were measured as previously described (27). Briefly, LNCaP cells were treated with 1 μM ES for 72 h and incubated with PBS containing 0.1 mM 2-deoxy-d-glucose (Sigma-Aldrich) and 1 μCi ³H-2-deoxy-d-glucose (American Radiolabeled, St. Louis, MO, USA) for 10 min at 37°C. After washing the remaining glucose with cold PBS, the cell lysate was collected with 1% SDS (Sigma-Aldrich). The ³H levels were counted using UniversolEsliquid Scintillation Cocktail (MP Bio, Santa Ana, CA, USA) and a Beckman LS6500 multipurpose scintillation counter (Beckman Coulter, Brea, CA, USA). Glucose uptake level is presented as intensity normalized by protein concentration of each sample (cpm/mg).

NAD and NADP assays

Total NAD and NADP levels were determined using colorimetric assay kits (Sigma-Aldrich). LNCaP cells were treated with 1 μM ES for 72 h, and cell lysates were prepared according to the manufacturer’s protocol. Cell lysates were subjected to cycling enzyme reaction followed by colorimetric detection at 450 nm. Total NAD and NADP levels, normalized by lysate concentration, were presented as relative values (means ± sem) of control LNCaP cells (100%).

Cell proliferation assay

To determine the effect of dexamethasone (DEX) (Sigma-Aldrich) on androgen-independent LNCaP cell growth, LNCaP cells (5 × 10³ cells per well) were cultured in 24-well plates for 6 d using a charcoal-stripped serum medium supplemented with 100 nM DEX, and cell proliferation was monitored every 24 h. To test the effect of ES, the LNCaP and 22Rv1 cell lines (3 × 10³ cells per well) were grown in 96-well plates for 24 h and replenished with medium containing 1 μM recombinant ES. After incubation for 72 h, phase-contrast images of cells were taken at ×50 original magnification in a Leica DMI 4000B fluorescence microscope. Cell proliferation was enumerated by cell counting using ImageJ (National Institutes of Health, Bethesda, MD, USA; https://imagej.nih.gov/) and presented as a relative value (means ± SEM) compared with control (100%).

AR and GR gene silencing

Validated small interfering RNA (siRNA) oligonucleotides specific for AR and GR and control siRNA were purchased from Santa Cruz Biotechnology. The siRNA transfections in LNCaP, 22Rv1, and DU145 cells were carried out either by using Lipofectamine (Thermo Fisher Scientific) or LipoJet (SignaGen, Rockville, MD, USA) according to the manufacturers’ protocols. Transient knockdown of AR and GR genes was confirmed by immunoblot analysis.

Forced GR-gene expression

The pcDNA3.1 expression vector carrying GR open reading frame was obtained from the cDNA Resource Center. Transient transfection of the expression vector in LNCaP cells was performed using Lipofectamine (Thermo Fisher Scientific) according to the manufacturer’s protocol. After 48 h, transfected cells were harvested, and GR expression was confirmed by immunoblot analysis of cell lysate.

Immunoblotting

The cell lines from all experimental conditions were lysed with RIPA (Thermo Fisher Scientific) after sonication, and protein concentrations were measured by bicinchoninic acid protein assay (Thermo Fisher Scientific). Denatured protein samples (30–40 μg) were resolved on 10–15% SDS-PAGE gels, transferred onto nitrocellulose membrane (Bio-Rad, Hercules, CA, USA), and immunodetected using antibodies against ES (Leinco), GR, AR, SOD2, prostate-specific antigen, and β-actin (all from Santa Cruz Biotechnology). Blots were developed using the corresponding horseradish peroxidase-conjugated secondary antibodies with chemiluminescence reagent (Millipore), and the images were obtained with a Molecular Imager Gel Doc XR system (Bio-Rad) and PXi gel imaging system (Syngene, Frederick, MD, USA). Densitometric analysis was carried out using ImageJ. Protein expression levels of all experimental conditions were represented as relative values (means ± sem) of controls.

Immunocytochemistry

To present the GR-directed effect of ES on androgen-independent PCa cells, 22Rv1 cells (1 × 10⁴ cells per chamber) were plated on poly-l-lysine-coated chamber slides (BD Biosciences) and incubated with 1–2 μM ES for 72 h. The cells were then stained with GR antibody, conjugated with Alexa Fluor 488 (Cell Signaling Technology, Danvers, MA, USA), and imaged at ×100 original magnification with a Leica DMI 4000B fluorescence microscope to qualitatively determine GR levels.

Real-time quantitative PCR

LNCaP cells were treated with 1 μM ES for 48 h, and total RNA was isolated using Trizol (Thermo Fisher Scientific). cDNA synthesis was carried out using iScript Reverse Transcription Supermix (Bio-Rad) as previously described (24). GLUT1 gene expression was determined by real-time PCR with cDNA using iQSYBR green (Bio-Rad). The GLUT1 gene was amplified using the forward primer 5′-GGCATCAACGCTGTCTTCTA-3′ and the reverse primer 5′-CTCGGGTGTCTTGTCACTTT-3′. For GAPDH, the forward primer 5′-GATCCGGTATGGAGCCTTCT-3′ and the reverse primer 5′-TGTAGAAGCCAGAGGCT GGT-3′ were used for PCR. Amplification was initiated at 95°C for 2 min followed by 45 cycles of 95°C for 15 s and 52°C for 2 min. The relative gene expression levels were calculated by the ΔΔC_t method (28). GLUT1 gene expression was normalized to GAPDH (as an internal control), and the fold change was derived from the 2^−ΔΔC_t values relative to the control.

Pull-down assay

For the binding experiment, recombinant human ES and GR were coexpressed in the Origami 2(DE3) cells under 1 mM isopropyl-d-thiogalactopyranoside induction as described previously. The soluble lysate in PBS containing 20 mM imidazole (Sigma-Aldrich) was applied to a Ni-sepharose affinity column (GE Healthcare), where GR was pulled down with His-tagged ES. After washes with PBS containing 50–100 mM imidazole, the bound ES-GR complex was eluted with PBS containing 300–600 mM Imidazole and confirmed by immunoblotting.

Coimmunoprecipitation

The ES-GR complex was immunoprecipitated by using the ImmunoCruz IP matrix system (Santa Cruz Biotechnology). Soluble lysate (0.5–1 mg), obtained from DU145 cells, was incubated with 1–2 µg of recombinant ES at 4°C overnight. The reaction mixture was then incubated with 2 µg of anti-His tag antibody (GeneTex) and 40 µl of agarose beads at 4°C for 6 h. The beads were washed 4 times with 1x PBS, and the bound fraction was eluted with 40 µl of Laemmli sample buffer (Bio-Rad) for SDS/PAGE and immunoblotting. In a separate experiment, soluble lysate containing both recombinant human ES and GR was incubated with 5 µg of anti-T7 tag antibody (GeneTex). Immunoprecipitation was carried out in the same manner, where N-terminal T7-tagged GR complexed with ES was coimmunoprecipitated.

Homology modeling

To generate structural models of ES and GR-ligand-binding domain (LBD) complex, homology modeling was performed by superimposing the GR-LBD (PDB: 1M2Z) structure onto the AR-LBD (PDB: 1XOW) in the reference model of the ES/AR-LBD complex that was generated previously (24). Rigid body movement was made by the least squares approximation fitting function in Coot (29). The figure was generated using the PyMol program (Schrödinger, Cambridge, MA, USA; http://www.pymol.org).

Statistical analysis

The statistical significance of data, obtained from cell-based experiments, biochemical assays, and quantitative analyses of both mRNA and protein expression was determined by Student’s t test. Values of P < 0.05 were considered significant.

RESULTS

AR antagonism induces GR-mediated oxidative stress in PCa cells

Despite a strong association between oxidative stress and AR reactivation in the hormone-refractory PCa, the upstream mechanism triggering this stress-mediated prosurvival signaling remains to be elucidated. To identify molecular cues to the redox imbalance and the resultant CRPC development against AR antagonism, we first abrogated AR function in PCa cells by using siRNA or androgen depletion. The transient AR gene silencing in androgen-responsive LNCaP cells showed a significant decrease in SOD2 and the reduced form of GSH levels (Fig. 1A, B), indicating that AR antagonism induces oxidative stress by suppressing major antioxidant systems in PCa cells. Notably, results of our study indicated that AR gene knockdown significantly up-regulates GR expression (Fig. 1A), suggesting a reciprocal regulation between AR and GR in LNCaP cells. To test whether GR compensation on AR down-regulation can induce redox imbalance, LNCaP cells were transfected with an expression vector encoding GR and then followed by biochemical assays analyzing changes in the levels of ROS scavengers. Indeed, GR overexpression in LNCaP cells led to significant down-regulation of SOD2 expression (Fig. 1A) and to a decrease in the total GSH level (Fig. 1B), indicating that GR mediates oxidative stress in AR-down-regulated PCa cells. Corroborating these results, abrogating GR expression in 22Rv1 CRPC cells resulted in significant up-regulation of SOD2 and increased intracellular GSH, confirming that the antioxidant machinery in PCa cells is negatively regulated by GR (Fig. 1C, D). Similarly, DU145 cells with a repressed AR expression also showed an inverse correlation between GR and SOD2 expression levels (Fig. 1E, F). Collectively, these results demonstrate that GR plays a direct role in elevating oxidative stress by regulating antioxidant gene expression.

Figure 1.

GR contributes to oxidative stress in PCa cells. A, B) GR up-regulation either by AR-gene silencing or forced GR expression negatively regulates the levels of SOD2 and GSH in LNCaP cells. A) Whole cell lysates were isolated from LNCaP cells and transiently transfected with AR siRNA or pcDNA3.1-GR for 48 h, and the protein expression of GR and SOD2 was analyzed by immunoblotting. The levels of GR and SOD2, normalized to β-actin levels, are presented as relative values (means ± sem) of control. Experiments were repeated 3 times. PSA, prostate-specific antigen. *P < 0.05; **P < 0.01 vs. control. B) In a replicate experiment, GSH levels in the cell lysates were analyzed by the reaction with monochlorobimane for 1 h at room temperature. Fluorescence intensity of GSH was measured at 380 nm (excitation) and 460 nm (emission), and the total GSH levels, normalized by the concentration of lysates, are expressed as relative values compared with control as 100%. Triplicate experiments were repeated twice. C–F) GR knockdown in both cell lines significantly increases SOD2 and GSH levels. Following the same procedures, SOD2 and GSH levels in 22Rv1 and DU145 cell lines under transient GR gene silencing were analyzed by immunoblotting (C, E) and GSH assay (D, F), respectively. The levels of GR and SOD2 in each cell line are presented as relative values of controls from 3 independent experiments. G, H) DHT depletion up-regulates GR expression and down-regulates SOD2 expression in androgen-responsive PCa cell lines. LNCaP (G) and TRAMP-C1 (H) cells were cultured in medium supplemented with charcoal-stripped fetal bovine serum for 7 d and harvested for Western blot analysis to determine the expression levels of GR and SOD2. I) DEX treatment increases androgen-independent LNCaP cell growth. LNCaP cells were cultured in the absence of DHT (AD) for 6 d, and cell proliferation was monitored every 24 h by cell counting. In parallel, LNCaP (AD) cells were incubated with 100 nM DEX to compare cell proliferation. The growth of LNCaP (AD) and the LNCaP (AD) cells cultured with DEX were compared with LNCaP cell growth in the presence of 10 nM DHT for 72 h (as 100%). **P < 0.01 vs. LNCaP (AD).

Next, we determined whether ADT-induced oxidative stress and the resultant development of androgen-independent growth phenotype are promoted by GR expression and interdependently up-regulated upon AR suppression. To mimic the emergence of androgen-independent growth in vitro, androgen depletion (AD) condition was established by culturing the cells in dihydrotestosterone(DHT)-free medium. Consistent with the results from transient AR knockdown, AD significantly increased GR activation and SOD2 down-regulation in LNCaP cells (Fig. 1G). To verify that the observed compensatory action of GR upon AD was not restricted to LNCaP-specific molecular event, TRAMP-C1, another androgen-responsive murine PCa cell line, was included in the study. Similar to LNCaP cells, GR induction was observed in TRAMP-C1 cells under AD-mediated AR suppression, which was accompanied by the suppression of SOD2 (Fig. 1H). To study further whether GR activation can simulate androgen-independent growth progression at a cellular level, LNCaP cells under AD condition were stimulated with DEX. The proliferation rate of LNCaP cells was significantly reduced within 72 h under AD condition, whereas DEX stimulation significantly delayed the AD-mediated growth suppression in LNCaP cells (Fig. 1I). These results indicate that GR activation upon AR antagonism plays an important role in promoting androgen-independent growth.

ES differentially regulates the redox state in PCa cells

Androgen signaling is known to promote antioxidant gene expression; thus, the AR-targeted effect of ES along with other AR antagonists presumably contributes to oxidative stress induction in PCa. Surprisingly, unlike the redox imbalance to the pro-oxidant state observed in AR-suppressed PCa cells (Fig. 1), we found that ES treatment up-regulated the antioxidant system. GeLC-based proteome analysis of ES-treated LNCaP cells indicated a significant increase in the levels of ROS scavenging machinery, including major antioxidant enzymes: SOD2, catalase, glutathione synthetase, glutathione reductase, and thioredoxin reductase (Fig. 2A and Supplemental Table 1). In addition, up-regulation of NADPH-cytochrome P450 reductase with a concomitant increase in downstream biliverdin reductase expression suggested the activation of the biliverdin/bilirubin redox cycle (Fig. 2A). The activation of the glutathione system with up-regulation of glutathione synthetase and reductase levels was confirmed by a corresponding increase of intracellular GSH in ES-treated LNCaP cells (Fig. 2B, C). Correlative analysis by in situ dihydroethidium staining further confirmed that ES treatment substantially decreased the basal ROS level in LNCaP cells (Fig. 2D), and immunoblot analysis indicated a 2.5-fold increase in SOD2 level (Fig. 2E). Together, these results indicate that ES differentially up-regulates the antioxidant system in AR-expressing PCa cells. Discrepancies between ES and other AR-directed treatments in modulating the oxidoreductive state in PCa (Fig. 1) prompted us to speculate that ES exerts an antioxidant effect beyond its targeting of AR. Given the requirement of NADPH as a universal cofactor for glutathione reductase, thioredoxin reductase, catalase, and NADPH-cytochrome P450 reductase, we further investigated whether the effect of ES affects the downstream metabolic processes, particularly regulating the intracellular NADPH levels.

Figure 2.

ES up-regulates ROS-scavenging machinery in LNCaP cells. A) Global proteome analysis on LNCaP cells after treatment with 1 µM ES for 72 h. Trypsin-digested whole cell lysates were subject to peptide identification, and the levels of protein quantitation were analyzed by the actual spectral counts of peptides obtained from the triplicate LC/MS analyses. The expression levels of antioxidant enzymes regulated by ES were compared with control and are presented as means ± sd (see also Supplemental Table 1). BLVR, biliruvidin reductase; CAT, catalase; GSR, glutathione reductase; GSS, glutathione synthase; NPR, NADPH-cytochrome P450 reductase; TrxR: thioredoxin reductase. B, C) LNCaP cells were treated with 1 µM ES for 72 h, and the intracellular levels of GSH were analyzed by in situ GSH staining (B) and fluorometric quantitation (C). Fluorescence intensity of GSH was measured exploiting excitation at 380 nm and emission at 460 nm, and the total GSH level affected by ES is expressed as means ± sem. **P < 0.01 vs. control (n = 3). D) In the same treatment condition, intracellular ROS levels were determined by in situ ROS staining. Total ROS levels in control and ES-treated LNCaP cells were visualized using the ROS probe dihydroethidium (DHE). E) From replicate experiments, total cell lysates were isolated, and the protein expression levels of SOD2 were analyzed by immunoblotting. The changes in SOD2 level upon ES treatment ares presented as a relative value compared with control.

ES up-regulates NAD and NADP biosynthetic pathways

Our study next identified that ES treatment promotes glucose uptake in LNCaP cells with a significant increase in gene expression of glucose transporter isoform 1, GLUT1 (Fig. 3A, B). Surprisingly, the indicated glucose influx had no influence on the levels of major metabolic enzymes either in the glycolytic pathway or in the tricarboxylic acid cycle (Supplemental Figs. 1 and 2 and Supplemental Tables 2 and 3). Instead, ES-treated LNCaP cells showed a 5-fold increase in the levels of glucose-6-phosphate dehydrogenase (G6PD) (Fig. 3C and Supplemental Table 4), a rate-limiting enzyme of the pentose phosphate pathway (PPP). Given that G6PD maintains the intracellular levels of NADPH and thus allows conversion of GSH from GSSG, ES-mediated GSH augmentation may be directly attributed to G6PD up-regulation and the resultant increase in NADPH levels. Indeed, biochemical analysis indicated that ES treatment significantly increased the total level of NADP in LNCaP cells (Fig. 3D).

Figure 3.

ES up-regulates PPP and NAD/NADP biosynthetic pathways. A, B) ES treatment stimulates glucose transport and causes an increased expression of GLUT1 in LNCaP cells. After incubation with 1 µM ES for 72 h, GLUT1 gene expression and glucose influx were analyzed by quantitative RT-PCR (A) and ³H-2-deoxy-d-glucose uptake assay (B), respectively. The GLUT1 mRNA level, normalized to GAPDH level, is compared with control. **P < 0.01 vs. control (n = 3). Glucose uptake levels, normalized by concentration of total cell lysate, are shown as means cpm/mg ± sem from 3 independent experiments. *P < 0.05 vs. control. C) Proteome analysis shows ES-treated LNCaP cells show 5-fold increase in the protein level of G6PD in pentose phosphate pathway. The G6PD expression regulated by ES, relative to control, is presented as means ± sd (also see Supplemental Table 4). D) From the replicate experiment, intracellular NADP levels in LNCaP cells were analyzed by cycling enzyme reaction, followed by colorimetric detection at 450 nm. Total NADP level affected by ES was compared with a baseline level in control cells. **P < 0.01 vs. control (n = 3). E) ES up-regulates expression of NAMPT and NAPRT, the critical enzymes for NAD salvaging, and down-regulates poly (ADP-ribose) polymerase 1 (PARP), a NAD-degrading enzyme (also see Supplemental Table 4). F) To confirm the proteome data above, intracellular NAD levels in LNCaP cells were analyzed by cycling enzyme reaction, followed by colorimetric detection at 450 nm. Total NAD level regulated by ES was compared with control. **P < 0.01 vs. control (n = 3). G) Overall scheme shows that ES modulates intracellular ROS levels by up-regulating glucose uptake, NAD biosynthesis, and shunting metabolic pathway to PPP. The enzymes, indicated in a solid square, are significantly up-regulated by ES.

After ES treatment, LNCaP cells also showed a marked increase in nicotinamide phosphoribosyltransferase (NAMPT) and nicotinate phosphoribosyltransferase (NAPRT), critical enzymes for NAD generation (Fig. 3E and Supplemental Table 4). Correspondingly, the level of poly(ADP-ribose) polymerase, an NAD-degrading enzyme, was significantly decreased. In line with the proteomic analysis, biochemical studies indicated that total NAD levels were significantly increased in LNCaP cells after ES treatment (Fig. 3F). Additionally, ES up-regulated enzymes of PPP, promoting de novo synthesis of nucleotide, a precursor molecule for NAD, which in turn increases the total level of NADPH, a reducing power for GSH system. Overall, these results suggest that increases in both NAD and NADP levels may lead to augmentation of intracellular NADPH coenzyme, resulting in enhanced ROS scavenging (Fig. 3G).

ES prevents pro-oxidant signaling by targeting reciprocal GR activation in PCa cells

Given that AR antagonism induces the pro-oxidant signaling mechanism in PCa cells, pleiotropic function of ES regulating the redox state appears to extend its recently identified mode of action on impairing AR function. To understand the plausible mechanisms underlying this action, we first confirmed the reciprocal up­-regulation of GR upon AR antagonism, in which GR compensation may act as a driving force for elevating oxidative stress in PCa cells (Fig. 1). To determine whether the antioxidant effect of ES is directly attributed to targeting of GR, androgen-dependent LNCaP cells and androgen-independent 22Rv1 CRPC cells were tested. Simulating androgen-independent cell proliferation in vitro, LNCaP cells were cultured for 28 d under AD, and the expression levels of AR, GR, and SOD2 were analyzed. Our result indicates that AD-induced LNCaP cells, adapted to androgen-independent growth, showed a reciprocal regulation between AR and GR expression as androgen-independent CRPC signatures with aberrant AR overexpression and concurrent decrease in SOD2 expression (Fig. 4A). In contrast, ES treatment in LNCaP cells under AD down-regulated AR expression and inhibited reciprocal GR up-regulation at the early time point (Fig. 4B). A corresponding increase in SOD2 expression further supported the finding that ES targeting of GR suppressed oxidative stress induction (Fig. 4B). More importantly, this combined effect of ES and AD significantly decreased androgen-independent growth of LNCaP cells (Fig. 4C), suggesting that dual targeting of ES on both AR and GR can improve the therapeutic impact of ADT at the early onset on androgen-dependent PCa.

Figure 4.

The GR-targeted effect of ES on PCa cells. A) Simulating development of androgen-independent growth phenotype in vitro, LNCaP cells were cultured for 28 d under AD using a medium supplemented with charcoal-stripped fetal bovine serum, and the expression levels of AR, GR, and SOD2 were monitored by immunoblotting. AD-induced reciprocal regulation between AR and GR contributes to androgen-independent LNCaP cell phenotypes with aberrant AR overexpression and substantial decrease in SOD2 expression. B, C) Inhibitory effect of ES on androgen-independent LNCaP cell growth. LNCaP cells under AD were incubated with 1 µM ES for 7 d, and cells were harvested for analyzing expression levels of AR, GR, and SOD2 (B) and cell proliferation (C). Data indicate that ES treatment down-regulates AR and prevents GR up-regulation in LNCaP cells with corresponding increase in SOD2 expression. The cell proliferation affected by ES is presented relative to control (as 100%). Representative phase-contrast images of LNCaP cells in all experimental conditions are provided. D–G) The AR- and GR-targeted effect of ES on 22Rv1 CRPC cells. 22Rv1 cells were treated with 1 µM ES for 72 h. D, E) Reduction of GR is shown in immunocytochemistry (D), and down-regulation of AR, AR-splice variant (AR-v7), and GR are shown by immunoblotting (E). E, F) The corresponding increases in SOD2 and GSH levels are shown in immunoblotting (E) and GSH assays (F), respectively. G) Cell proliferation assay shows that ES significantly decreases 22Rv1 cell growth. The cell proliferation affected by ES is presented as a relative value of control (as 100%) from 2 independent experiments in triplicates. **P < 0.01 vs. control.

Next, to test the effect of ES directly on CRPC cells, we used 22Rv1 cells displaying androgen-independent growth with abundant expression of AR and GR (Fig. 4D–G). Upon incubation with ES for 72 h, 22Rv1 cell proliferation was decreased by 40% (Fig. 4G). Immunoblotting and immunofluorescence staining of 22Rv1 cells further indicated that ES treatment substantially reduced the level of GR (Fig. D). As observed in LNCaP cells, suppression of GR reversed SOD2 down-regulation and GSH reduction (Fig. 4E, F). Molecular analysis further identified that ES treatment decreased the expression levels of AR and AR-splice variant (AR-v7), indicating that ES can target AR and GR in CRPC cells (Fig. 4E).

Direct interaction of ES and GR as a potential mechanism antagonizing GR

In conjunction with a significant reduction in GR downstream functions by ES, based on the structural similarity between GR and AR, we speculated that a direct molecular interaction of ES with GR may interfere with GR-mediated transcriptional activity. To test this hypothesis, we first sought structural solutions, showing the potential protein-protein interaction. Structural homology present in 3-ketosteroid nuclear receptors, GR (NR3C1), and AR (NR3C4) enabled us to generate a feasible complex model of GR-LBD and ES (Fig. 5A), adopting a similar binding mode of ES targeting activation function-2 in AR-LBD (24). Next, to verify this protein-protein interaction, an array of binding experiments was performed. The GR extracted from the DU145 PCa cell line, which is devoid of AR expression, was used to avoid possible interference of ES to AR. Coimmunoprecipitation using a His-tag antibody indicated that GR interacts with His-tagged ES (Fig. 5B). The equivalent levels of β-actin in the input lysate and unbound fractions demonstrated a predominant GR binding to ES (Fig. 5B). The direct protein-protein interaction was further confirmed by coexpression of recombinant ES and full-length GR (Fig. 5C). After successful expression of both proteins in a bacterial expression system under 2 different antibiotic selection conditions, a pull-down assay was carried out harnessing His-tagged ES as a bait protein to trap recombinant GR (Fig. 5D), and, inversely, T7-tagged GR was used for coimmunoprecipitation of recombinant ES as a prey protein (Fig. 5E). Results of this assay indicated that ES was coimmunoprecipitated with GR (Fig. 5E) and that the ES-GR complex formation enabled efficient protein purification using Ni-affinity column chromatography (Fig. 5D). Taken together, our structural and functional analyses indicated a direct interaction of ES and GR, which may be a part of the GR-antagonizing mechanism.

Figure 5.

Molecular interaction of ES and GR. A) A structural model of the ES-GR complex was built by homology modeling. A crystal structure of human GR-LBD (shown as a green ribbon, PDB ID code 1M2Z) was superimposed onto the structural model of the human ES/AR-LBD complex (24) by least squares approximation fitting, using human AR-LBD (shown as a gray ribbon, PDB ID code 1T73) as a reference structure. Pymol was used for displaying the figure. B) Coimmunoprecipitation of ES and GR. The reaction mixture of recombinant ES and DU145 cell lysate containing GR was immunoprecipitated with His-tag antibody. After washing, the bound proteins were extracted from the agarose beads using Laemmli sample buffer for SDS-PAGE and immunoblot analysis. Input indicates the concentration of GR in DU145 cells. Detection of both ES and GR from the bound fraction and the equivalent levels of β-actin in the input lysate and unbound fractions indicate the predominant GR binding to ES. C–E) Direct interaction of ES with GR was confirmed by binding experiments using recombinant human ES and GR proteins. C) The pET21a-ES and pET24a-GR plasmids were cotransformed into Origami 2(DE3) strain, and 1 mM IPTG was used to induce the overexpression of ES with the C-terminal His-tag and of GR with the N-terminal T7-tag. D) The soluble lysate containing the recombinant ES and GR was loaded onto a nickel-sepharose column, and the bound fraction was eluted with 30–60% of 1 M imidazole buffer. SDS-PAGE and immunoblots indicate coelution of ES and GR. E) In a replicate experiment, the lysate was immunoprecipitated with T7-tag antibody. The T7-tagged GR complexed with ES was resolved in SDS-PAGE and proved with ES and GR antibodies.

DISCUSSION

Abrogation of androgen signaling in frontline PCa therapy is associated with elevated oxidative stress, leading to AR reactivation and disease progression to CRPC (9–11, 15). Despite the causal role of oxidative stress in the activation of the protumorigenic signaling cascade, minimal effort has been given to understand or target this molecular axis. In the present study, we provide experimental evidence that reciprocal activation of GR upon AR antagonism directly promotes oxidative stress in PCa cells. The implication of glucocorticoid signaling in oxidative stress induction has been reported in various pathologic conditions (18, 30–33) but not in cancer. Our study indicates that AD and AR gene knockdown significantly up-regulated GR expression, which negatively regulated SOD2 and glutathione antioxidant systems in androgen-dependent human and murine PCa cell lines. At a molecular level, GR is known to interact with other transcriptional factors through “protein tethering,” resulting in transrepression of genes influenced by other transcription factors (34). Because the identified binding partners of GR include activator protein-1, activating transcriptional factor-2, and NF-κB (34), it is conceivable that GR transrepression of such ubiquitous transcription factors consequently down-regulates downstream gene expression of ROS scavengers, including SOD2 and glutathione reductase. Indeed, GR augmentation via transient GR-gene transfection in LNCaP cells induced significant down-regulation of both SOD2 expression and GSH levels. Conversely, GR gene silencing was shown to reverse down-regulation of SOD2 and GSH in 22Rv1 and DU145 PCa cells, suggesting that GR plays a direct role in modulating the redox state.

Our study further identified that GR-mediated oxidative stress positively regulates androgen-independent PCa proliferation, a hallmark of the CRPC phenotype. LNCaP cells, conditioned to grow in ligand-depleted conditions, showed increased GR expression, and subsequent glucocorticoid stimulation promoted cell proliferation under androgen blockade. This is corroborated by previous findings that GR up-regulation after AR inhibition activates prosurvival signaling mechanisms and resistance to AR-targeted therapies in PCa (19, 20). However, it is interesting to note that the expression levels of GR in LNCaP cells were gradually decreased while developing androgen-independent growth (Fig. 4A). This observation is in line with the recent presentation at 2015 American Society of Clinical Oncology Annual Meeting that GR bypass may occur in earlier stages of disease (2). In addition, clinical assessment of prostate tissue specimens supports grade-dependent GR down-regulation (35, 36). These findings suggest that GR activation appears to be a part of the process of stage-dependent cellular adaptation toward CRPC, and therefore it is crucial to develop new generation therapies that can both impair AR activity and minimize the onset of oxidative stress-induced disease progression mediated by GR.

In this context, the present study identified, to our knowledge, for the first time that ES inhibits androgen-independent PCa growth by suppressing GR-mediated oxidative stress. The results of our study indicate that ES treatment significantly down-regulated GR levels in LNCaP and 22Rv1 cells, followed by an increase in SOD2 expression and intracellular GSH levels. Our proteomic and biochemical analyses further confirmed that treatment of PCa cells with ES induced significant up-regulation of major ROS scavenging machinery, including SOD2, the glutathione system, and the biliverdin/bilirubin redox cycle, resulting in a concomitant reduction of intracellular ROS. These results are in line with earlier observations that antioxidant SOD mimetics suppress CRPC growth (14). It is also noteworthy that the emergence of the AR-V7 splice variant in patients with CRPC is a newly identified resistance phenotype against enzalutamide and abiraterone therapies (8, 37). Our data demonstrate that ES treatment reduced both AR and AR-V7 in 22Rv1 cells, suggesting that ES not only prevents conversion to androgen-independent growth but that it also directly suppresses proliferation of cells that have already established androgen-independent phenotypes.

Regarding ES-mediated augmentation of antioxidant signaling in PCa cells, it is worth highlighting that ES induces significant up-regulation of glucose influx and NAD/NADP production with a concomitant increase in GSH level, the major endogenous antioxidant thiol. Glucose is a fundamental molecule for the maintenance of antioxidant systems by increasing NADPH production through the PPP. The PPP also serves as a major metabolic cascade for de novo synthesis of nucleotide, a precursor molecule for NAD, which is converted to NADP/NADPH. Our proteome and biochemical analyses indicated that ES treatment significantly increased the level of G6PD and a concomitant increase in NADP level, indicating up-regulation of PPP. In addition, we found that ES directly augmented NAD salvaging pathways by up-regulating NAMPT/NAPRT expression, providing NAD for NADPH production. Given that mammalian cells largely depend on the NAD salvage pathway particularly using NAMPT, it is also worthy to note that ES treatment increased NAPRT expression by 5-fold in LNCaP cells where very limited expression of NAPRT has been reported (38). Collectively, these data suggest that ES regulates intracellular ROS levels by augmenting the GSH system through glucose influx, shunting metabolic pathways to PPP and NAD biosynthesis (Fig. 3G).

Given the GR-mediated transrepression of ROS scavenger genes, we speculate that ES binding to GR may interfere with GR interaction with other transcription factors. Our previous study showed that ES directly interacts with AR-LBD and disrupts AR-target gene transcription in PCa cells (24). Based on the conserved structural similarity in GR-LBD, we performed homology modeling of a complex of GR-LBD and ES using the ES-AR complex (24) as a reference structure. Interestingly, the structural model suggested that ES binding to GR may adopt a similar binding mode of targeting activation function-2 in AR-LBD. An array of binding experiments further confirmed direct molecular interaction between ES and GR, suggesting the potential inhibitory mechanism of action of ES on GR function. However, because the present study did not focus on detailed characterization of the binding interfaces that reside in ES and GR, further structural studies in this line may shed more light on the protein-protein interaction.

Although it is accepted that reciprocal activation of GR is a survival mechanism of PCa, the controversial effects of glucocorticoids have also been reported. Glucocorticoid therapy is known to suppress the adrenal supply of androgens and to reduce the resultant suppression of AR activity, partly benefiting patients with PCa (34). Clinical studies using mifepristone (RU486) also demonstrated that GR antagonism could up-regulate androgen signaling in patients with CRPC (2, 39). Indeed, our study also indicated this interdependent regulation between AR and GR, showing that a forced GR expression down-regulated expression of AR and the downstream AR-target gene prostate-specific antigen in LNCaP cells and that the GR-gene knockdown induced AR up-regulation in 22Rv1 cells (Fig. 1A, C). However, there are rising concerns that corticosteroid therapy might promote PCa progression in the same context. Watson et al. (2) and Arora et al. (19) reported that the androgen-lowering effect of glucocorticoid could paradoxically promote PCa progression in patients developing the CRPC phenotype with aberrant GR activation. Further, the compensatory GR signaling has been reported to confer resistance to enzalutamide, where GR blockade using small-molecule inhibitors has been shown to delay disease progression (19). Collectively, these results suggest that stage-dependent interplay between AR and GR may be a key mechanism of PCa to evade existing therapies, and therefore, combined inhibition of AR and GR could be a more effective therapeutic strategy for targeting CRPC.

Overall, combined with our published results (24, 40), the present study demonstrates that ES can simultaneously target AR and GR functions in PCa cells, providing an innovative angle to suppress AR signaling and the development of the GR-induced resistance phenotype in PCa. Our study suggests that the potential therapeutic application of ES may include combination with the frontline ADT that targets PCa at early stages. Based on the known antiangiogenic properties of ES and on more interesting evidence that human prostate endothelial cells also express AR (41, 42), the application of ES in combination therapies could synergize tumoristatic and tumoricidal effects with minimal resistance. Regarding the recent concerns about the acquired resistance to second-line therapy using enzalutamide and abiraterone, due to GR up-regulation and the emergence of AR splice variants in patients with CRPC (2, 8, 37), our study further provides a rationale to clinically validate the antitumor activity of ES overcoming resistance to current AR-directed therapies, especially in a disease signature associated with GR up-regulation as a compensatory mechanism.

Glossary

AD

androgen deprivation

ADT

androgen deprivation therapy

CRPC

castration-resistant prostate cancer

DHT

dihydrotestosterone

ES

endostatin

G6PD

glucose-6-phosphate dehydrogenase

GR

glucocorticoid receptor

GSH

glutathione

LBD

ligand-binding domain

NAMPT

nicotinamide phosphoribosyltransferase

NAPRT

nicotinate phosphoribosyltransferase

PCa

prostate cancer

PPP

pentose phosphate pathway

ROS

reactive oxygen species

siRNA

small interfering RNA

SOD2

manganese superoxide dismutase

AUTHOR CONTRIBUTIONS

J. Lee and S. Ponnazhagan designed the research; J. Lee, M. Kang, H. Wang, and J. A. Mobley performed the research; J. Lee, M. Kang, H. Wang, J. A. Mobley, G. Naik, G. Sonpavde, W. T. Garvey, V. M. Darley-Usmar, and S. Ponnazhagan analyzed the data; and J. Lee and S. Ponnazhagan wrote the paper.