SOD Mimetics: A Novel Class of Androgen Receptor Inhibitors That Suppresses Castration-Resistant Growth of Prostate Cancer

This article is featured in Highlights of This Issue, p. 1

The androgen receptor (AR), a transcription factor that mediates the biological effects of androgens, testosterone, and dihydrotestosterone (DHT), is vital for the development and progression of prostate cancer. After an initial response to androgen ablation therapy, which suppresses AR signaling, the majority of advanced tumors eventually transition to the currently incurable androgen-independent or castration-resistant prostate cancer (CRPC; refs. 1, 2). Importantly, CRPC continues to be highly dependent on the persistent expression and function of AR to survive and progress (3, 4). Studies reporting significant inhibition of in vitro and in vivo growth of CRPC following disruption of AR expression and/or function (5, 6) have generated much interest in the AR as a key therapeutic target and have intensified efforts to uncover potent AR inhibitors.

Elevated levels of cellular reactive oxygen species (ROS) significantly contribute to the initiation and progression of cancer (7, 8), and the degree of ROS generation correlates with the aggressive phenotype of prostate cancer (8). Cellular ROS levels are normally kept in check by a very efficient cellular detoxifying system, which includes the mitochondrial antioxidant enzyme, superoxide dismutase 2 (SOD2), which catalyzes the conversion of superoxide (O₂⁻) to hydrogen peroxide (9). The expression of SOD2 or manganese SOD is commonly downregulated in cancer cells, and restoration of SOD2 activity via SOD2 overexpression significantly inhibits in vitro and in vivo tumor growth, including prostate cancer growth (10–12). SOD2 levels progressively decline during the transition from prostatic intraepithelial neoplasia to androgen-dependent (AD) prostate cancer to CRPC (13–15). Strikingly, SOD2 levels in CRPC are just 11% of that found in AD prostate cancer (15), supporting the notion that there may be selection for decreased SOD2 expression in advanced prostate cancer. SOD2 downregulation increases AR transcriptional activity, and this effect is reversed with the antioxidant, N-acetylcysteine (16). These findings raise the possibility that therapies aimed at specifically augmenting SOD2 activity might offer an effective and feasible means of treating CRPC, by directly targeting the key player, the AR.

Use of SOD mimetics to augment the natural antioxidant defenses of the cell has been beneficial in animal models of a number of neoplastic and nonneoplastic diseases, in which oxidative stress is implicated in disease progression (17, 18). As oxidative stress is an integral component of and contributor to cancer progression (7, 8), use of SOD mimetics not only lowers tumor incidence (19–21) but also markedly inhibits in vitro and in vivo tumor growth (22–26). Although the effects of SOD mimetics on a variety of cancers have been investigated, their effects on AR function and prostate cancer growth was hitherto unknown. Here, we show for the first time that SOD mimetics are effective in suppressing AR activity and in vitro and in vivo CRPC growth.

See Supplementary Information Materials and Methods.

LNCaP, CWR22Rv1, and PC-3 prostate cancer cells [American Type Culture Collection (ATCC)] were maintained in RPMI-1640 medium (Sigma-Aldrich) supplemented with 10% FBS and 1% penicillin/streptomycin. Cells are authenticated by ATCC. LAPC-4AD prostate cancer cells (provided by Dr. Charles Sawyers, Memorial Sloan-Kettering Cancer Center, New York, NY) were maintained in modified IMEM medium (Invitrogen) supplemented with 10% FBS and 1% penicillin/streptomycin.

After the indicated treatments, whole-cell lysates were prepared and subjected to Western blot analysis as described previously (27). After incubation with primary antibody, the blots were probed with an IRDye-labeled secondary antibody (LI-COR Biosciences). Scanning of the blots and densitometric analysis of protein bands was done using the LI-COR Odyssey IR Imaging System. The intensity of a target protein band in each sample was normalized to that of β-actin in the same sample and expressed as a fold change, with expression in the control set at 1.

This assay is based on the principle that hydroethidine is oxidized by O₂⁻ and converted to fluorescent hydroxyethidium (28). After the indicated treatments, cells were washed with PBS and incubated with 15 μmol/L hydroethidine in serum-free medium at 37°C for 2 hours. Fluorescence was read at 510 nm_Ex/580 nm_Em and represented as relative fluorescence units (RFU).

Cell viability was determined using the CellTiter-Fluor Cell Viability Assay (Promega) as per the manufacturer's protocol. Cell viability was assessed by measuring live cell protease activity, which is determined by measuring fluorescence emitted by the fluorogenic, cell-permeant peptide substrate, GF-AFC. The intensity of fluorescence generated is directly proportional to cell viability and was read at 400 nm_Ex/505 nm_Em.

Animal procedures were in accordance with UT Southwestern Medical Center Institutional Animal Care and Use Committee-approved animal protocol. Ten to 12 week old male, nonobese diabetic severe-combined immunodeficient mice were injected subcutaneously in the flank region with 10⁷ LAPC-4AD cells suspended in Matrigel. Tumor volume was calculated by the formula: length × width²/2 (29). Once tumor volume reached 275 ± 75 mm³, the mice were castrated. Following castration, once tumors exhibited the first measurable increase in growth, mice were randomized into 2 groups: (i) Control diet (n = 6) and (ii) Tempol (4-hydroxy-2,2,6,6-tetramethylpiperidine-N-oxyl) diet (n = 4). Tempol treatment was in the form of bacon-flavored mouse chow mixed with powdered Tempol at a concentration of 58 mmol/L (10 mg/g; Bioserv). Control animals received bacon-flavored mouse chow minus Tempol. The animals were individually caged and food intake monitored at regular intervals. Individual relative tumor volume (RTV) at any given time (x) was calculated as follows: Vx/V1, in which Vx is the tumor volume at time x and V1 is the tumor volume at the start of treatment. Mean RTV and SD for control and Tempol groups were calculated at the end of each week of treatment and statistical significance assessed by unpaired t test. Animals were euthanized 9 weeks after commencement of control or Tempol diet and tumors excised and processed for Western blotting.

Data were tested for statistical significance by 1-way ANOVA followed by Tukey post hoc test or by unpaired t test where appropriate (as indicated in the figure legends) using the Graph-Pad Instat software. Data were considered statistically significant only if P < 0.05.

Tempol (Supplementary Diagram S1) is a cell membrane permeable, water soluble compound belonging to the nitroxide class of SOD mimetics (30). Tempol dose dependently attenuated O₂⁻ levels in AR-dependent CWR22Rv1 and LAPC-4AD cells, as well as in AR-negative PC-3 prostate cancer cells (Fig. 1A).

The AR-dependent prostate cancer cell lines used in this study express AR variants that represent the spectrum found in CRPC: LNCaP cells harbor AR with a mutation in the ligand-binding domain (LBD), CWR22Rv1 cells express constitutively active, LBD-lacking, truncated AR and full-length AR with a mutated LBD, and LAPC-4AD cells predominantly express wild-type AR (31, 32). Tempol significantly decreased androgen response element (ARE)-driven luciferase reporter activity, in the presence and absence of DHT, the major AR agonist (Fig. 1B), suggesting that AR transcriptional activity is attenuated by Tempol. To determine how closely inhibition of AR transcriptional activity by Tempol mimics the effect of SOD2, we assessed ARE-driven luciferase reporter activity after SOD2 overexpression. Inhibition of ARE-driven luciferase reporter activity after SOD2 overexpression (Fig. 1C) closely paralleled that following Tempol treatment, both in the presence and absence of DHT (Fig. 1B).

The Tempol-mediated decrease in AR transcriptional activity was reflected in a significant decline in mRNA levels of AR target genes, PSA and FKBP5. Inhibitory effects of Tempol on AR target gene expression were evident in the presence and absence of DHT (Fig. 1D, Supplementary Fig. S1), as well as in a dose-dependent (Fig. 1E, Supplementary Fig. S2) and time-dependent manner (Supplementary Fig. S3). Dose–response studies revealed that Tempol decreased PSA mRNA levels by 83%, 80%, and 60% in LNCaP, CWR22Rv1, and LAPC-4AD cells, respectively, relative to that after vehicle treatment (Fig. 1E). A similar trend was seen with FKBP5 mRNA levels (Supplementary Fig. S2). A 2.5 mmol/L dose of Tempol elicited a significant, time-dependent decline in PSA and FKBP5 mRNA levels in LNCaP, CWR22Rv1, and LAPC-4AD cells at each time point under study, relative to that in the vehicle-treated cells at the same time points (Supplementary Fig. S3).

These data suggest that Tempol is effective in suppressing AR transcriptional activity in prostate cancer cells, and this suppression is closely tied to its ability to lower cellular O₂⁻ levels.

To determine whether the Tempol-mediated inhibition of AR transcription was due to a decrease in AR protein, we next assessed changes in AR protein following treatment with Tempol. Although DHT is known to increase AR protein (33), Tempol markedly decreased AR protein, both in the absence and presence of DHT (Fig. 2A). Remarkably, Tempol not only significantly reduced protein levels of full-length AR (high molecular weight AR band in LNCaP, CWR22Rv1, and LAPC-4AD cells) but also decreased levels of truncated AR (lower molecular weight AR band in CWR22Rv1 cells; Fig. 2A). For all AR Western blot analyses, we used an antibody targeting the N-terminal domain of AR; hence, it can detect full-length AR and truncated ARs that lack the C-terminal LBD (31, 34).

A consistent dose-dependent decline in full-length AR (LNCaP, CWR22Rv1, and LAPC-4AD cells) and truncated AR (CWR22Rv1 cells) was observed following treatment with Tempol (Fig. 2B). A 2.5 mmol/L Tempol elicited a significant time-dependent decline in both full-length and truncated AR over the course of 48 hours, relative to that following vehicle treatment (Fig. 2C, Supplementary Fig. S4). In contrast to its ability to dose dependently decrease AR protein across all prostate cancer cell lines under study, Tempol did not have similar suppressive effects on other nuclear receptors, such as glucocorticoid receptors-α and -β (GR-α and GR-β) or estrogen receptor-β (ER-β; Fig. 2D), indicating that Tempol selectively decreases AR protein in prostate cancer cells.

To dissect the mechanism by which Tempol decreases AR protein, we assessed whether Tempol promotes AR protein degradation. To this end, prostate cancer cells were treated with the protein synthesis inhibitor, cycloheximide (CHX), in the presence or absence of Tempol (Fig. 3A). We reasoned that once new protein synthesis is inhibited with CHX, levels of residual AR protein would reflect the rate of AR degradation; therefore, more rapid degradation of AR would result in lower residual AR protein levels. Concomitant treatment with CHX and Tempol decreased AR protein levels more rapidly than did treatment with either CHX or Tempol alone (Fig. 3A), suggesting that Tempol enhances AR protein degradation.

Next, to explore whether the enhanced degradation of AR by Tempol is due to increased ubiquitination, prostate cancer cells were treated with the proteasomal inhibitor, MG-132, in the presence or absence of Tempol, and AR was immunoprecipitated and probed for ubiquitin (Fig. 3B, Supplementary Fig. S5). Tempol increased AR ubiquitination, relative to that in the control. Also, the level of ubiquitinated AR after cotreatment with MG-132 and Tempol was significantly higher than that with either MG-132 or Tempol alone (Fig. 3B, Supplementary Fig. S5). These findings suggest that Tempol promotes AR degradation via increased ubiquitination of AR.

A close assessment of AR protein levels in the MG-132 ± Tempol input lysates revealed that despite the presence of MG-132, Tempol appreciably decreased AR protein relative to that in the vehicle control (Fig. 3B, Supplementary Fig. S5). This observation pointed to the possibility that Tempol may also be decreasing AR synthesis. Indeed, Tempol dose dependently decreased AR mRNA in LNCaP, CWR22Rv1, and LAPC-4AD cells (Fig. 3C). Collectively, we show that Tempol decreases AR via accelerated degradation of AR protein and a reduction in AR mRNA.

To determine whether the suppressive effects of Tempol on AR is a property also shared by other SOD mimetics, we tested the effects of MnTBAP and MnTMPyP, both of which belong to the class of SOD mimetics termed manganese (III) metalloporphyrins (Supplementary Diagram S1). MnTBAP and MnTMPyP have significant anticancer activities against cancers of the colon, liver, esophagus, and prostate (24–26). Both MnTBAP and MnTMPyP dose dependently decreased full-length and truncated AR protein levels (Fig. 4A). Forty μmol/L MnTBAP and MnTMPyP elicited a significant time-dependent decline in both full-length and truncated AR protein in LNCaP and CWR22Rv1 cells (Fig. 4B). The above results closely paralleled those seen with Tempol (Fig. 2B and C), thereby indicating that the inhibitory effects of Tempol on AR are due to characteristics also shared by other SOD mimetics and are not unique to Tempol alone.

Studies have shown marked in vitro and in vivo declines in CRPC growth following inhibition of AR expression and/or function (5, 6). Becasue our data showed that Tempol significantly decreases AR protein levels and AR transcriptional activity in prostate cancer cells (Figs. 1–3), we next investigated the effects of Tempol on prostate cancer cell survival and proliferation. First, we determined the cell-cycle phases that were altered by Tempol (Fig. 5A). The most consistent trend that emerged across prostate cancer cell lines after treatment with 2.5 mmol/L Tempol was a significant increase in the Sub G₀ (dead cell) fraction, relative to that after vehicle treatment. In addition, Tempol significantly decreased the G₂-M fraction of CWR22Rv1 cells (Fig. 5A).

To test whether the Tempol-mediated suppression of AR leads to a selective decline in viability of AR-dependent over AR-negative cells, we assessed prostate cancer cell viability after treatment with 0 to 2.5 mmol/L Tempol (Fig. 5B). Tempol significantly decreased the viability of AR-expressing prostate cancer cells at all doses. Dosages of 0.25, 0.5, and 2.5 mmol/L Tempol decreased CWR22Rv1 cell viability by 18%, 23%, and 57%, and LAPC-4AD cell viability by 21%, 33%, and 57%, respectively, relative to that after vehicle treatment. Although 0.25 and 0.5 mmol/L Tempol had significant suppressive effects on viability of AR-positive prostate cancer cells, it only elicited a 5% and 16% decrease in viability of AR-negative PC-3 cells, relative to vehicle treatment. However, 2.5 mmol/L Tempol decreased PC-3 cell viability by 46% (Fig. 5B). Our data (Fig. 5A and B) indicate that Tempol significantly decreases prostate cancer cell viability, with a more pronounced effect on viability of AR-dependent cells. In addition, AR overexpression significantly blunted the ability of Tempol to decrease prostate cancer cell viability, suggesting that the Tempol-mediated decline in viability of AR-dependent cells is predominantly via suppression of AR function (Fig. 5C).

Because our in vitro studies revealed a significant Tempol-mediated decrease in AR protein levels and prostate cancer cell viability, we next evaluated the effect of Tempol on CRPC growth in vivo. To this end, subcutaneous LAPC-4AD tumor–bearing mice were surgically castrated and placed on control diet or Tempol-containing diet for 9 weeks. This form of Tempol treatment has been shown to produce an average serum Tempol concentration of 90 to 100 μmol/L and to selectively decrease ROS and oxidative damage in vivo without selectively inducing cytotoxicity (21). The average RTV and cellularity in the Tempol group was consistently and significantly lower than that in the control group (Fig. 6A; Supplementary Fig. S7). After 9 weeks of treatment, 5 of the 6 mice in the control group had relatively large tumors, with individual RTVs of 3.34, 3.18, 1.93, 1.86, and 1.55; 1 control mouse had a regressing tumor with an individual RTV of 0.75. In contrast, all 4 mice in the Tempol group had comparably smaller tumors, with 3 being regressing tumors (individual RTVs of 0.39, 0.51, and 0.66), and 1 slow-growing tumor (individual RTV of 1.35; Supplementary Fig. S6). After 9 weeks of treatment, the average RTV of tumors in the Tempol group was 0.72, versus 1.89 in the control group (Fig. 6A). We did not observe a statistically significant difference in food intake or body weight between mice in the control and Tempol groups (Fig. 6B, left and right panels), suggesting that differential food intake was probably not a factor influencing tumor growth in the Tempol group.

We next assessed the extent to which 9 weeks of Tempol treatment modulated AR protein in the prostate cancer xenografts (Fig. 6C). Tempol treatment elicited a 20% decrease in full-length AR. Interestingly, the LAPC-4AD xenografts also expressed a truncated AR variant that had the same mobility as the truncated AR in the CWR22Rv1 cell lysate, which was run in parallel. Notably, expression of the truncated AR variant was suppressed by more than 50% in the Tempol group, relative to that in the control group (Fig. 6C).

Current androgen deprivation strategies to treat advanced, metastatic prostate cancer have been unable to adequately suppress AR function, resulting in a vast majority of patients failing treatment and progressing to the lethal CRPC (1). Clearly, there is an urgent need for the discovery of AR inhibitors that will remain highly effective, notwithstanding the numerous current-therapy-evading AR molecular alterations that crop up both during and following the development of castration resistance. This study is the first to identify SOD mimetics as effective AR inhibitors in prostate cancer cells and also the first to document significant in vivo tumor growth suppressive effects with the use of a SOD mimetic in castrated, tumor-bearing mice.

Several reports (31, 34, 35) on the intriguing discovery of novel AR splice variants that lack the LBD in CRPC have necessitated a reevaluation of strategies aimed at targeting the AR in advanced prostate cancer. Targeting truncated AR variants is critical because (i) they are constitutively active (31, 34, 35); (ii) their expression is significantly increased during prostate cancer progression, and they confer androgen-independent prostate cancer growth (31, 34, 35), and (iii) their expression correlates with the risk of tumor recurrence after radical prostatectomy (31, 35). We show that SOD mimetics downregulate full-length and truncated AR protein levels, both in vitro and in vivo, without significantly altering levels of other nuclear receptors. In CWR22Rv1 cells, which express high levels of truncated AR, all 3 SOD mimetics under study, that is, Tempol, MnTBAP, and MnTMPyP, significantly decreased both full-length and truncated AR forms in vitro. Truncated AR protein was not readily detectable in LAPC-4AD cells in vitro; however, the LAPC-4AD xenografts accumulated appreciable levels of truncated AR. This finding is in conformity with findings by Dehm and colleagues, which showed that xenograft-based models of CRPC were enriched with truncated AR splice variants (34). Notably, the accumulation of truncated AR in LAPC-4AD xenografts was significantly blunted in mice that were on Tempol-containing diet. Interestingly, Tempol had a more pronounced suppressive effect on truncated AR rather than full-length AR in vivo (Fig. 6C). This finding raises an interesting question about the relative contributions of full-length and truncated AR forms to tumor growth in vivo. Although our study has not specifically addressed this question, Guo and colleagues have previously shown that specific short hairpin RNA–mediated knockdown of truncated AR expression, without altering full-length AR, was sufficient to attenuate prostate cancer growth in xenograft models (31). Furthermore, truncated AR regulates a unique set of genes that are not regulated by full-length AR (31). Together, these observations, as well as ours, imply that expression of the truncated AR is critical for regulating prostate cancer growth.

Our data reveal a Tempol-mediated decline in viability of AR-positive, as well as AR-negative, prostate cancer cells. However, it is noteworthy that AR-positive prostate cancer cells are more susceptible to the growth inhibitory effects of Tempol and exhibit a greater decline in viability when treated with lower doses of Tempol than the AR-negative PC-3 cells at the corresponding dose (Fig. 5B). Furthermore, AR overexpression blunts suppressive effects of Tempol on viability of AR-dependent prostate cancer cells (Fig. 5C). Together, these observations suggest that (i) effect of Tempol on AR-dependent prostate cancer cell viability is predominantly mediated via AR, and (ii) the AR-expressing prostate cancer cells depend on the growth-promoting effects of AR signaling (4–6, 36), and hence, even a minor suppression of AR function causes them to become acutely sensitive. On the other hand, AR-negative prostate cancer cells may have developed alternative AR-independent, but ROS-dependent means of survival and proliferation, and hence they require a higher dose of SOD mimetic to lower ROS levels to an extent that will sufficiently suppress those ROS-dependent cell survival pathways. This notion is in partial agreement with a finding by Venkataraman and colleagues (10) that shows that SOD2 overexpression inhibits growth of AR-negative PC-3 cells, suggesting that reduced cellular ROS levels inhibit tumor growth, independent of the AR. In line with this view, we have shown a significant Tempol-mediated reduction in cellular O₂⁻ levels in AR-positive and AR-negative prostate cancer cells (Fig. 1A). Collectively, our data suggest that Tempol may be exerting its antiproliferative effects via AR-dependent, as well as AR-independent, but ROS-dependent mechanisms (37, 38). Safety and toxicity studies conducted by many research groups have confirmed that concentrations of Tempol needed to suppress tumor growth in vivo do not elicit signs of general or organ toxicity and that Tempol is significantly more effective in inhibiting the growth of a range of neoplastic than nonneoplastic cell lines (20–23). The above findings, together with our demonstration of significant AR suppressive effects by Tempol, provide a mechanistic rationale for the use of Tempol and other SOD mimetics in prostate cancer treatment.

In summary, we have discovered SOD mimetics to be a promising, novel class of AR inhibitors with the ability to suppress CRPC growth. Given that SOD2 expression progressively declines in the spectrum from benign prostatic epithelium to CRPC (13–15), and that ROS and AR play a critical role in every stage of prostate cancer development (4, 8, 10, 16, 39, 40), this new class of AR inhibitors could be beneficial in prostate cancer treatment, with maximal beneficial effects seen in CRPC, in which SOD2 expression is decreased the most. Our demonstration of significant suppression of both full-length and truncated ARs with SOD mimetics could also open up the prospect for the use of SOD mimetics in prostate cancers that are not amenable to treatment with therapeutic modalities that inhibit the synthesis of androgens and those that competitively block access to the AR LBD.

N. Sharifi has been compensated as a consultant for Ortho Biotech.

The authors thank Drs. Jer-Tsong Hsieh, Daxing Xie, Amy Sieve, Katja Schuster, Georgia Konstantinidou, and Andrea Rabellino for technical input, Dr. Charles Sawyers for the LAPC-4AD cells, and Dr. Elaine Hurt for pGL3-TK-3xARE-FLuc and eGFP-SOD2 plasmid constructs.

This work was supported by a US Army Medical Research and Material Command Postdoctoral Prostate Cancer Grant # PC094309 to R. Thomas, a Physician Research Training grant # PC080193, a Prostate Cancer Foundation Young Investigator Award, and a Howard Hughes Medical Institute Physician-Scientist Early Career Award to N. Sharifi.

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.