Radiation Protection of the Gastrointestinal Tract and Growth Inhibition of Prostate Cancer Xenografts by a Single Compound Free

Radiotherapy is the most common therapeutic modality across a wide range of malignant diseases, including prostate cancer. However, the delivery of curative radiation doses is hampered by acute or chronic “collateral damage” affecting normal tissues. When treating tumors in the abdominal cavity, toxicity to the intestine and the bladder are often dose limiting (1). Highly targeted methods to deliver radiation specifically to disease sites alleviate radiation toxicity, yet 40% to 50% of patients with locally advanced prostate cancers recur locally following treatment (2). Hence, protection of normal tissue will be a critical element of future dose-escalation trials in patients with locally advanced prostate cancer. Existing radiation protectors, including amifostine (3), are of limited utility in protecting the small and large intestines against radiation effects.

Inflammation is a key element of the radiation response of normal and tumor tissues and is commonly associated with increased activity of NF-κB (4, 5). Previously, we demonstrated that several inhibitors of canonical NF-κB activation improved survival of lethally irradiated zebrafish embryos and preserved gastrointestinal morphology and function (6). Inhibitors of glycogen synthase kinase (GSK)3 similarly protect normal tissues, including the gastrointestinal tract (7, 8). While the role of GSK3β in cell stress responses is complex (for review, see ref. 9), it has been implicated in modifying NF-κB–dependent transcription of genes encoding proinflammatory proteins (10, 11).

Here, we performed a side-by-side comparison of radioprotective properties of five compounds targeting either GSK3 and/or NF-κB with a focus on the gastrointestinal tract. We report that the triterpenoid RTA 408 provides robust radiation protection to the murine gastrointestinal system and markedly improves overall survival of lethally irradiated mice. Importantly, normal tissue protection by RTA 408 is contrasted by inhibition of human prostate cancer xenograft growth in mice.

Compounds were obtained from the following sources: Ethyl pyruvate and lithium chloride (Sigma-Aldrich), SB216763 and SB415286 (Tocris Bioscience), amifostine (Medimmune), and 2-cyano-3,12- dioxooleana-1,9 (11)-dien-28-oic acid (CDDO) derivative RTA 408 (REATA Pharmaceuticals). Prostate cancer cells (PC3, LNCaP/C4-2B, DU145, and CWR22Rv1) were originally obtained from ATCC or from Dr. Thomas Pretlow (Case Western Reserve University, Cleveland, OH) and generously provided by Dr. Marja Nevalainen (Thomas Jefferson University, Philadelphia, PA), and immortalized NHPrE-1 and BHPrE-1 prostate epithelial cells were a gift from Dr. Simon Hayward (Vanderbilt University Medical Center, Nashville, TN). Normal primary prostate epithelial cells (PrEC) were from Lonza. The prostate cancer cell lines were authenticated on a regular basis by monitoring cell morphology, androgen responsiveness, and the expression of cell line–specific markers. Normal primary epidermal keratinocyte cultures were established using standard protocols. Cells were routinely tested for mycoplasma contamination using the MycoSensor PCR Assay Kit (Stratagene). Tumor cells were grown in RPMI1640 supplemented with 10% FBS (Corning Cellgro). Normal and immortalized prostate epithelial cells and primary keratinocytes were grown in specialty media (Lonza). For in vivo imaging, PC3 prostate carcinoma cells were stably transfected with reporter plasmids [pGL4.51(luc2/CMV/Neo) and pNL3.2.NF-κB-RE(NlucP/NF-κB-RE/Hygro), Promega] encoding firefly luciferase (FLuc) and NanoLuc luciferase, respectively, and luciferase reporter activity tested using reporter-specific in vitro assays (Promega).

C57Bl/6 mice (6–8 weeks old) were from Charles River Laboratories. Mice were kept in pathogen-free conditions and handled in accordance with the requirements of the Guidelines for Animal Experiments and after approval of the experimental protocols by the Institutional Animal Care and Use Committee of Thomas Jefferson University (Philadelphia, PA). Ionizing radiation (IR) was administered at doses ranging from 5 to 30 Gy using a 250-kVp X-ray machine (PanTak) with 50-cm source-to-skin distance and a 2-mm copper filter. The dose rate was approximately 1.4 Gy/minute. Drugs were uniformly administered by intraperitoneal injection for up to 2 days before IR treatment, and on days 1, 2, and 3 after IR treatment as indicated. For comparison of RTA 408 and amifostine, mice received one dose (17.5 mg/kg) of RTA 408 24 hours before IR (whole body, 9 Gy), one dose 1 hour before IR, and 2 additional doses 24 and 48 hours after IR; amifostine was injected once (250 mg/kg) 15 minutes before IR. All injections were done intraperitoneally. Animals were euthanized at the end of the observation period, or when weight loss reached or exceeded 20% of the initial weight, or if they showed signs of severe morbidity (lethargy, hunched posture, and/or shivering or severe diarrhea). Kaplan–Meier survival curves were compared by using the log-rank (Mantel–Cox) test.

Prostate carcinoma cells were inoculated by subcutaneous injection (5 × 10⁶ per mouse) into the lower abdominal skin of male Foxn1nu (nude) mice (6–8 weeks old; Charles River Laboratories). Tumor progression was monitored by caliper measurements and by in vivo live imaging (see below). Xenografts were allowed to grow for 2 to 3 weeks before treatment. RTA 408 (17.5 mg/kg) or vehicle control (DMSO) were administered intraperitoneally three times per week until the end of the observation periods. To assess effects of the combination of RTA 408 and IR, radiation (5 Gy) was administered at different time points as indicated. Tumor volumes were calculated by multiplying the two longest planar axes measured by the depth of the tumor (as determined by caliper measurements). Mixed effects regression models were used to determine statistical significance of tumor growth data over time.

For FLuc-based in vivo live imaging, mice were injected intraperitoneally with 200 μL d-luciferin in PBS (15 mg/mL) per 20 g of mouse body weight 15 minutes before imaging, and imaged using an IVIS In Vivo Imaging System (Caliper Life Sciences). For in vivo imaging of the NanoLuc luciferase under the control of the NF-κB response element, 100 μL of the NanoGlo substrate (10 μg; Promega) were injected via tail vein. Image analysis and quantitation was done using Living image 4.2 software (Caliper LifeSciences). Luciferase-positive areas on individual images were selected as regions of interest (ROI) with a 14% threshold. Planar spectral images were automatically analyzed by the software. Total counts for all pixels inside the ROI were recorded. At least 3 animals from each experimental group were used for each time point.

For histologic, immunofluorescent, and direct fluorescent analyses, tissue samples (e.g., small intestines, tumors) were embedded in the optimal cutting temperature compound (Tissue-Tek), and cryosectioned (7 μm). Hematoxylin and eosin staining was done on ethanol/acetic acid–fixed slides. Apoptosis incidence was determined by terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL) using the In Situ Cell Death Detection Kit (Roche Applied Science) and 4′,6-diamidino-2-phenylindole (DAPI) counterstain. Quantitative analysis was done using ImagePro software (Media Cybernetics) on at least 6 different and independent microscopic fields for each treatment condition. Indirect immunofluorescence was performed by incubation with primary antibodies for cleaved PARP (Asp214) Human Specific (Cell Signaling Technology), M30CytoDEATH (Roche Applied Science), or Ki-67 (Abcam) for 1 hour at room temperature or overnight at 4°C followed by secondary antibodies labeled with either Alexafluor 488 or Alexafluor 594 (BD Biosciences). Sections were counterstained with DAPI and slides were mounted using anti-fade Fluorosafe reagent (Calbiochem).

Cells were seeded in 96-well plates at 15,000 cells per well; after 24 hours, RTA 408 or vehicle were added in triplicate. After 72 hours, attached cells were fixed with 70% ethanol, stained with crystal violet solution (0.2% crystal violet in 2% ethanol), and quantitated by measuring absorbance (OD_A595). Metabolic activity was determined after 72 hours by addition of WST-1 reagent (Roche; 10 μL per 100 μL supernatant) for at least 3 hours at 37°C, followed by measuring absorbance at OD_A450 with a reference of OD_A650 and using wells containing media without cells for background subtraction. Statistical differences between treatment groups were determined using one-way ANOVA with Tukey posttest correction (GraphPad Prism).

Prostate carcinoma cell lines were treated for 24 hours with RTA 408 or vehicle at the concentrations indicated. Immunoblots were reacted with (i) PARP-1 primary antibody (C2-10; Santa Cruz Biotechnology) and IRDye 800CW goat anti-mouse IgG1-specific secondary antibody; (ii) cleaved caspase-3 (Asp175; 5A1E) primary antibody (Cell Signaling Technology) and IRDye 800CW donkey anti-rabbit IgG (H + L) secondary antibody; (iii) cleaved caspase-8 (Asp384; 11G10) primary antibody (Cell Signaling Technology) and IRDye 800CW goat anti-mouse IgG1-specific secondary antibody, or (iv) M30CytoDEATH primary antibody (Roche Applied Science) and IRDye 800CW goat anti-mouse IgG2b-specific secondary antibody; all secondary antibodies were from LICOR. Filters were analyzed on a LICOR Odyssey imaging system.

Cells were seeded at clonogenic densities in T-25 flasks and treated with RTA 408 at various concentrations or DMSO, at 24 and 1 hour before radiation exposure. IR was administered at 0, 2, 4, 6, and 8 Gy. All treatments were performed in biologic triplicate. After IR, flasks were incubated for 2 weeks. Colonies were identified by crystal violet staining; those containing ≥50 cells were counted. The data were fit to a linear quadratic model for cell survival by using GraphPad Prism software and the equation Y = exp(−a × x − b × x²) (12). Statistically significant differences between drug and control curves were determined by using two-way ANOVA.

We performed a side-by-side comparison of several NF-κB and GSK3 inhibitors on mice challenged with a lethal whole body radiation dose (8 Gy; Fig. 1). All of the compounds reportedly protect against or mitigate radiation injury in different experimental settings either in vitro or in vivo. They included two inhibitors of canonical NF-κB signaling (ethyl pyruvate; ref. 13) and RTA 408 (ref. 14; Fig. 1A), and three GSK3 inhibitors, including lithium chloride (LiCl; ref. 8), SB415286 (8), and SB216763 (ref. 8; Fig. 1B, C). Ethyl pyruvate interferes with NF-κBp65 signaling by covalently modifying a reactive cysteine residue (Cys36) of the NF-κBp65 subunit (15). RTA 408 is a variant of the triterpenoid CDDO that reversibly and covalently modifies reactive cysteine residues on multiple proteins, including several of potential relevance to radiation protection. Specifically, binding of CDDO to Cys179 in IKKβ leads to inhibition of canonical NF-κB signaling (16) and binding to KEAP1 leads to increased levels of the transcription factor Nrf2 and of multiple antioxidant and phase II defense enzymes (17). RTA 408 was included in the screen because we previously observed robust radioprotection of zebrafish embryos by another variant of CDDO (CDDO-TFEA; ref. 6). SB415286 and SB216763 are ATP-competitive GSK3 inhibitors (18), whereas LiCl increases inhibitory phosphorylation of GSK3 (19). To allow direct side-by-side comparison all drugs were given using a standardized regimen, i.e., for 1 day and 1 hour before radiation and daily for 3 days after. Drug dosages were guided by published results and administration was by intraperitoneal injection. We observed various levels of radiation protection with each of the compounds tested. The CDDO derivative RTA 408 provided robust and consistent levels of radiation protection [100% at 30 days post-IR (8 Gy)] either as a single compound (Fig. 1A) or in combination with the GSK3 inhibitor SB216763 (Fig. 1C). In agreement with an earlier report (13), ethyl pyruvate also markedly increased survival of lethally irradiated mice (Fig. 1A). Interestingly, the survival advantage provided by RTA 408 was compromised when combined with SB415286 but not when combined with SB216763 (Fig. 1C). Similarly, survival of lethally irradiated mice treated with a combination of RTA 408 and LiCl was slightly lower than survival of mice receiving RTA 408 alone (Fig. 1C). Finally, RTA 408 produced levels of radiation protection similar to amifostine, the only currently approved radiation protector (Fig. 1D). These results encouraged us to further investigate tissue protection provided by RTA 408 alone.

Next, we investigated the effect of RTA 408 on the small intestine in C57Bl/6 mice irradiated at a dose (9 Gy) that causes death from gastrointestinal syndrome within 10 to 15 days (20). We observed that RTA 408 preserved the integrity of the mucosal lining of the small intestine of lethally irradiated mice (Fig. 2A). Mice that did not succumb to gastrointestinal syndrome lived beyond 30 days after IR, consistent with radiation protection of multiple organs, including the hemopoietic system. Furthermore, as determined by Ki-67 staining RTA 408–treated mice revealed robust proliferation in the crypt area at 2 and 7 days after IR, whereas radiation alone (9 Gy) markedly reduced proliferation in this tissue compartment concurrent with extensive tissue destruction (Fig. 2A). RTA 408 also significantly reduced radiation-induced apoptosis in both villi and crypts as determined by TUNEL staining (Fig. 2B). This effect extended to the skin, in which RTA 408 similarly reduced the apoptosis incidence caused by radiation exposure (Supplementary Fig. S1).

To address whether the cytoprotective effects of RTA 408 extended to tumor cells, we first tested the effects of RTA 408 on four different prostate cancer cell lines representing advanced, androgen-independent tumor stages (LNCaP/C4-2B, CWR22Rv1, DU145, and PC3). We observed robust tumor growth inhibition by RTA 408 of established xenografts (tumor size > 30 mm³ when treatment commenced) of all four cell lines tested even in the absence of radiation. Tumor growth, as determined by caliper measurements, is shown in Fig. 3A and representative in vivo tumor images at different days after treatment in Supplementary Fig. S2. In marked contrast to the protective effects observed in normal tissues, RTA 408 did not radioprotect PC3 xenotransplants. Rather, when used in combination with radiation, RTA 408 amplified the antitumor effect of radiation alone (P = 0.001; Fig. 3B). In vivo imaging revealed complete tumor growth inhibition in animals that received both radiation and RTA 408 at 17.5 mg/kg (Fig. 3B, see insert) but not in mice treated with 5 mg/kg RTA 408 (not shown). RTA 408 induced high levels of intratumoral apoptosis as determined by detection of fragmented DNA (TUNEL), cleaved PARP, and the caspase-3 cleavage product of cytokeratin18 (Fig. 4A). The antibodies used to detect cleaved PARP and cytokeratin18 do not crossreact with mouse tissues indicating that RTA 408 induced apoptosis of human tumor cells in situ. RTA 408–dependent inhibition of PC3 xenografts was associated with significantly reduced proliferation as determined by Ki-67 staining (Fig. 4B).

Next, we examined dose-dependent effects of RTA 408 on prostate cancer cells in vitro. Within 24 hours of exposure, RTA 408 (1 μmol/L) induced varying degrees of apoptosis in all four prostate cancer cell lines as determined by detection of cleaved caspase-3, cytokeratin18, and PARP-1 in both attached and, more prominently, in cells detached from substrate (Supplementary Fig. S3). In contrast, caspase-8 cleavage was only marginally detected in DU145 cells at the higher doses of RTA 408 (0.5–1 μmol/L) tested but not the other three cell lines. As determined by clonogenic survival assays, RTA 408 radiosensitized all four prostate cancer cell lines under investigation with the strongest effects observed in DU145 and LNCaP/C4-2B cells (Supplementary Fig. S4). As assessed by crystal violet staining and by WST assay, RTA 408 reduced viability of all four prostate cancer cell lines in a dose-dependent fashion (Fig. 5A). The IC₅₀ for inhibition of in vitro growth and survival of these cell lines ranged from 250 to 750 nmol/L. Interestingly, RTA 408 also inhibited in vitro growth and survival of PrEC, the immortalized NHPr-E1 and BHPrE-1 prostate cells and, primary human epidermal keratinocytes (NHEK-1, -2, and -3; Fig. 5B). The IC₅₀ as determined by crystal violet staining for the normal or premalignant cells was in the range of 125 to 250 nmol/L. Compromised cell viability was associated with substrate detachment of normal prostate epithelial cells and keratinocytes, as well as control PC3 cells (Supplementary Fig. S5). Collectively, these results highlight a broad spectrum of inhibitory effects of RTA 408 on benign and malignant prostate cells and on normal prostate epithelial cells and keratinocytes in vitro. The effects on normal epithelial cells in vitro are in marked contrast to tissue protection of normal epithelial tissues in irradiated mice.

The selective antitumor activity of RTA 408 on prostate cancer cell lines in vivo raises the question which molecular target(s) are responsible for this effect. In prostate cancer, deregulated NF-κB signaling is associated with disease progression, contributes to expression of both prostate specific antigen (PSA; 21) and androgen receptor (22), and is prevalent in castrate-resistant and metastatic tumors (23–26). Conversely, disrupting NF-κB signaling by forced expression of a phosphorylation-deficient IκB radiosensitizes PC3 prostate cells in vitro (27). Other NF-κB inhibitors, including curcumin (28), parthenolide (29), and SN52 (30), similarly inhibit prostate cancer growth and survival. On the basis of this prior work, we used NF-κB-NLuc- and control CMV-Fluc -reporter constructs to measure NF-κB activity in transfected PC3 tumors in vivo, before and after treatment with RTA 408 and/or IR (Fig. 6A and B). As expected, radiation induced NF-κB activity in tumor tissue (Fig. 6, panel 4). In the posttreatment group, the ratio of NF-κB-NLuc- to CMV-Fluc activity in RTA 408–treated mice (Fig. 6, panel 6) was significantly lower compared with that observed in mice treated with vehicle alone (Fig. 6, panel 2). In addition, the ratio of NF-κB-NLuc- to CMV-Fluc activity in mice treated with RTA 408 and IR combined (Fig. 6, panel 8) was significantly lower compared with that observed in mice treated with IR alone (Fig. 6, panel 4). Hence, at tumor growth–inhibitory concentrations, RTA 408 effectively inhibited transcription of an NF-κB–responsive reporter construct in PC3 cells in vivo.

These observations extend and confirm previous reports describing in vitro growth inhibition of human prostate cancer cells by CDDO variants. Specifically, Deeb and colleagues described proapoptotic effects of CDDO, CDDO-methyl(ME), and CDDO-imidazole(IM) in cultured human LNCaP, PC3, and DU145 and murine TRAMPC-1 prostate cancer cells in vitro (31, 32). Furthermore, Gao and colleagues described CDDO-dependent chemoprevention of prostate cancer development in transgenic TRAMP mice in which the SV40 T antigen is expressed by prostate epithelial cells (33). Tumor growth inhibition by CDDO derivatives extends to other tumor types ranging from leukemias (34–37) to solid malignancies (38–42). A common denominator of these tumor types is deregulated NF-κB activity, which is effectively inhibited by RTA 408 not only in vitro but also in vivo. It remains to be determined whether inhibition of survival pathways beyond NF-κB plays a role in prostate cancer inhibition by RTA 408. For example, the Akt/mTOR pathway is also reportedly inhibited by CDDO-ME in prostate cancer cells (43).

The mechanistic basis for the dual and opposite effects of RTA 408 on normal and a broad range of malignant tissues remains to be investigated further. Of particular relevance to cytoprotection, CDDO covalently attaches to KEAP1, disrupts KEAP1/Nrf2 interaction, and triggers Nrf2-dependent transcription of a host of genes encoding antioxidant enzymes (16). Nrf-2 activation by the CDDO derivatives CDDO-ethylamide (EA) and CDDO-ME has been proposed to improve survival of irradiated mice (44). We observed that topical application of RTA 408 markedly reduced radiation dermatitis in mice associated with significant increases in Nrf2 target genes and significant decreases in NF-κB target genes (45). Interestingly, radiation protection of normal prostate epithelial cells contrasted by growth inhibition of prostate cancer cells in vivo has been very recently described for dimethylamineparthenolide (DMAPT; ref. 46). DMAPT and its parent compound parthenolide alkylate reactive cysteines on multiple protein targets, including KEAP1, and inhibit canonical NF-κB signaling by interacting with IκB and the NFκBp65 subunit (47–49). In contrast to RTA 408, parthenolide or DMAPT reportedly did not inhibit cultured normal or immortalized prostate cells to the same extent as their malignant counterparts and this difference has been attributed to differential effects of DMAPT on KEAP1-dependent oxidation status in normal and malignant cells (46). This difference between RTA 408 and DMAPT suggests that tissue protection by RTA 408 as seen in vivo is not primarily due to cell-autonomous effects but likely depends on environmental factors provided by the tissue context in vivo. A precedent for “contextual” antitumor effects of CDDO-ME has been established previously (50). Specifically, CDDO-ME inhibited myeloid-derived suppressor cells in the tumor microenvironment associated with improved immune responses. Regardless of the relative contribution of cell-intrinsic or “environmental” antitumor mechanisms, the results obtained for DMAPT (46) and RTA 408 (this study) validate the concept of selective radiosensitization of tumor tissues by thiol-reactive compounds (5).

K. Ward has ownership interest (including patents) in Reata Pharmaceuticals. U. Rodeck received a commercial research grant from REATA Pharmaceuticals. No potential conflicts of interest were disclosed by the other authors.

Conception and design: K. Ward, A.P. Dicker, U. Rodeck

Development of methodology: V. Alexeev, A. Linnenbach

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): V. Alexeev, E. Lash, A. Bitterman, A. Linnenbach, U. Rodeck

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): V. Alexeev, L. Corsini, A. Bitterman, A.P. Dicker, A. Linnenbach

Writing, review, and/or revision of the manuscript: V. Alexeev, K. Ward, A.P. Dicker, A. Linnenbach, U. Rodeck

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): E. Lash, A. Aguillard, L. Corsini, A.P. Dicker, U. Rodeck

Study supervision: U. Rodeck

The authors thank Phyllis Wachsberger (Thomas Jefferson University, Philadelphia, PA) for advice on performing clonogenic assays, Simon Hayward (Vanderbilt University, Nashville, TN) for providing immortalized prostate cells, and Marja Nevalainen (Thomas Jefferson University, Philadelphia, PA) for providing prostate cancer cells.

This work was supported by DoD grant W81XWH-12-1-0477 and a pilot project under NIH grant U19A1091175. Additional support was provided by the Prostate Cancer Foundation and by REATA Pharmaceuticals.

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.