Regression of Mouse Prostatic Intraepithelial Neoplasia by Nonsteroidal Anti-Inflammatory Drugs in the Transgenic Adenocarcinoma Mouse Prostate Model

Vol. 10, 7727–7737, November 15, 2004 Clinical Cancer Research 7727

Regression of Mouse Prostatic Intraepithelial Neoplasia by Nonsteroidal Anti-inflammatory Drugs in the Transgenic Adenocarcinoma Mouse Prostate Model

Bhagavathi A. Narayanan,1,2 p53, cyclin-dependent kinase inhibitor p21WAF1/CIP1, p27, Narayanan K. Narayanan,1,2 Brian Pittman,3 and BAX, and caspase-3 to demonstrate the COX-2–independ- Bandaru S. Reddy2 ent mechanism involved in the inhibition of PIN lesions of 1 2 the dorsolateral prostate by both celecoxib and exisulind. Cancer Genomics Laboratory, Chemoprevention and Nutritional Results: We found for the first time that (a) both cele- Carcinogenesis Program, and 3Statistics and Data Management, Institute for Cancer Prevention, Valhalla, New York coxib and exisulind as dietary supplements induce strong inhibitory effects against prostate cancer at doses of 800 and 500 ppm, respectively, after 16 weeks; (b) the histologic ABSTRACT analysis of the dorsolateral prostate after 2 weeks of treat- Purpose: Epidemiologic studies have revealed a de- ment indicated a reduction of PIN lesions from 75% to 19% creased risk of colon cancer among people who have regu- with celecoxib and to 16% with exisulind; (c) more impor- larly taken cyclooxygenase (COX)-2 inhibitors such as tantly, those few PINs and adenocarcinomas in the groups aspirin or other nonsteroidal anti-inflammatory drugs treated with celecoxib or exisulind showed more apoptotic (NSAIDs). Whereas the selective COX-2 inhibitor celecoxib cells, lower levels of proliferating cell nuclear antigen, and a and exisulind, a metabolic product of sulindac, have gained lower number of mitotic cells. To understand the molecular increasing attention as efficacious chemopreventive agents mechanisms involved in the inhibition of PIN lesions, first, against colon and prostate cancer, not much is known about we examined the expression of molecular targets involved in the underlying molecular targets and mechanisms. More- angiogenesis and inflammatory processes. It was clearly over, the side effects of NSAIDs are a major obstacle for evident from Western blot analysis of the total protein lysate large-scale application to the prevention of cancer in hu- derived from the dorsolateral prostate tissues with PIN mans; for example, in the United States in 1998, there were lesions that expression of androgen receptor, vascular endo- 16,550 deaths from NSAID-induced gastrointestinal compli- thelial growth factor, nuclear factor-␬B p65, and BclII is cations. The toxicity associated with these compounds is down-regulated more effectively by celecoxib. Down-regula- raising concerns, and more needs to be known about their tion of AKT protein (total and phosphorylated at Ser473) mode of action and molecular targets. signaling by celecoxib clearly indicates an inhibition of the Experimental Design: We used the transgenic mouse survival gene and the pathological process that could other- prostate (TRAMP) model, which exhibits similarities with wise lead to adenocarcinoma. human prostate cancer, including epithelial origin, progres- Conclusions: Overall, the findings from this study sion from the PIN stage to adenocarcinoma, and metastasis clearly show the effectiveness of celecoxib and exisulind in by a transgene that is hormonally regulated by androgens. reducing the PIN lesions by modulating a cascade of molec- In addition to histologically analyzing the PIN lesions of the ular targets involved in COX-2–dependent and –independ- dorsolateral prostate from TRAMP mice, we delineated the ent mechanisms. Whereas these agents are already in clini- molecular targets and mechanisms of celecoxib and exisu- cal trial or in use as chemopreventive agents, findings from lind against mouse PIN lesions. We performed Western blot this study demonstrate the difference in their mode of ac- analysis of the total protein lysate from the tissues of mouse tion, thus helping us to understand the side effects. PIN lesions to measure the level of expression of androgen receptor, vascular endothelial growth factor, nuclear fac- INTRODUCTION tor-␬B p65, BclII, AKT (total and phosphorylated Ser473), The incidence of prostate cancer, one of the most common malignancies in men in the United States and in other Western countries, is gradually showing an upward trend (1). Recent epidemiologic reports indicate an increase in the number of men Received 4/19/04; revised 6/21/04; accepted 8/5/04. being diagnosed with prostate cancer worldwide (2–6). Prostate Grant support: USPHS grant CA-17613 from the National Cancer cancer growth in humans has been viewed as a multistage Institute and grant 01A015 from AICR to B. Narayanan. process with an early onset of minor histologic changes pro- The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked gressing slowly to metastatic lesions of higher grade (7). A advertisement in accordance with 18 U.S.C. Section 1734 solely to population-based case control study (8), as well as several indicate this fact. related reports (9, 10) on nonsteroidal anti-inflammatory drugs Requests for reprints: Bhagavathi A. Narayanan, Department of En- (NSAIDs), revealed a trend toward reduced prostate cancer risk vironmental Medicine, New York University School of Medicine, 57 Old Forge Road, Tuxedo, NY 10987. Phone: 845-731-3500; Fax: associated with regular intake of NSAIDs. Moreover, increased 845-351-4500. insight into the biology of prostate cancer and the emergence of ©2004 American Association for Cancer Research. new chemopreventive agents offer significant means to test new

prevention strategies (11–17). In this connection, it is interesting MATERIALS AND METHODS to note that findings from human studies and preclinical model Animals and Diets. Hemizygous 6-week–old male assays have suggested a potential role for selective cyclooxy- TRAMP [C57BL/6-TgN(TRAMP)8247Ng] and control genase (COX)-2 inhibitors in the prevention of colon cancer (C57BL/6) mice were purchased from JAX Mice and Services (18–27), but less information exists regarding the role of (Bar Harbor, ME). The mice were maintained in quarantine for COX-2 inhibitors against prostate cancer. Whereas the selective 2 weeks in a holding room in the Research Animal Facility at COX-2 inhibitor celecoxib induced a dose-dependent cytotoxic the Institute for Cancer Prevention (Valhalla, NY). They were effect in colon and breast cancer cells (20, 28), exisulind (su- housed in cages with wood chip bedding in a temperature- lindac sulfone), a weak inhibitor of COX-2, induced apoptosis controlled room (68°F–72°F) with a 12-h light/dark cycle, at a in mammary cancer models and thus also inhibited cancer relative humidity of 45% to 55%. All ingredients of the semi- growth (29, 30). Although COX-2 inhibitors play an important purified AIN-76A diet were purchased from Diets Inc. (Bethle- role in tumor growth inhibition, the mechanisms underlying hem, PA). The modified AIN-76A diet consisted of 20% casein, selective COX-2 inhibitors are not limited to the COX-2 target 0.3% D,L-methionine, 52% corn starch, 13% dextrose, 5% corn but involve COX-2–independent pathways that modulate AKT oil, 5% alphacel, 3.5% AIN mineral mixture, 1% AIN vitamin expression levels as seen in hepatocarcinoma cells (31–33). The mixture, and 0.2% choline bitartrate. The ingredients were effectiveness of these two agents against prostatic intraepithelial stored at 4°C before preparation of the experimental diets. neoplasia (PIN) that expresses a low level of COX-2 has not During the study, mice were permitted free access to the basal been sufficiently investigated. To address this issue, and also to diet (AIN-76A diet) and drinking water. The AIN-76A diet was evaluate the efficacy of celecoxib and exisulind individually against prostate cancer, we used the transgenic mouse prostate fed to the mice as described in previous studies (42, 43). All (TRAMP) model, which exhibits many similarities to human mice were inspected at least once daily to monitor their general prostate cancer, including epithelial origin, progression from the health status, and they were more thoroughly examined and PIN stage to adenocarcinoma, and metastasis by a transgene that weighed once weekly. The identity of the transgenic mice was is hormonally regulated by androgens (34–37), so that it serves established by polymerase chain reaction-based DNA screening as a suitable model. The advantage of the TRAMP model is that as described previously (35). well-differentiated neoplasia is generally observed in 100% of Chemopreventive Agents and Dose Selection. We used these mice between 8 and 12 weeks of age; the metastases to celecoxib (SC-58635), supplied by Searle Research and Devel- distant sites develop between 18 and 24 weeks of age, and all of opment (Pharmacia, St. Louis, MO), and sulindac-sulfone (ex- the mice display primary tumors and metastases to distant sites isulind-CP461) provided by Cell Pathways, Inc. (Horsham, PA). (38, 39). This model has been used extensively to evaluate the Experimental diets were prepared weekly by mixing celecoxib chemopreventive and therapeutic efficacy of several agents (800 ppm) or exisulind (500 ppm) with modified AIN-76A diet. against prostate cancer, including difluoromethylornithine and The diets were stored in a cold room. In this study, celecoxib at R-flurbiprofen (40, 41). In this report, for the first time, we 800 ppm and exisulind at 500 ppm were evaluated for their present results on the efficacy of celecoxib and exisulind against chemopreventive effect. The above-mentioned doses of cele- PIN lesions of prostate cancer, which was assessed by meas- coxib and exisulind were selected on the basis of a maximum uring the total number of PIN lesions, the level of proliferating tolerated dose (80% maximum tolerated dose) assay. cell nuclear antigen (PCNA), and the rate of apoptosis. More Study Design and Experimental Procedure. After the importantly, we have elucidated the effects of celecoxib and quarantine, 6-week–old TRAMP mice were randomly distrib- exisulind on several molecular targets (including AKT signal- uted by weight into experimental and control groups (n ϭ 10), ing) that could play an essential role in PIN development via as shown in Table 1, and fed the experimental diet for a period COX-2–independent mechanisms. The overall findings from of 16 weeks. A cohort of TRAMP mice was used as control this study clearly indicate that celecoxib and exisulind play a group. Nontransgenic mice (C57BL/6) were also used as an significant role in reducing the number of PIN lesions by down- overall control group. The animals in both control groups were regulating the expression of key molecular targets involved in fed AIN-76A diet only. Body weights were recorded weekly, angiogenesis and inflammatory processes. Whereas these agents and the animals were monitored daily for their general health.

are already in clinical trial or in use, findings from this study The mice were sacrificed by CO2 euthanasia at 8 or 24 weeks clearly demonstrate their mode of action at the molecular level, of age. which is vital to understand the side effects. More importantly, Histopathology. Before sacrifice, each mouse was anes- our findings indicate early molecular targets of PIN lesions to be thetized and weighed. The genitourinary tract (GUT) consisting examined in the human clinical samples and thus will enable us of the bladder, urethra, seminal vesicles, ampler gland, and to design more effective prevention strategies. prostate was excised, weighed, and dissected under a micro-

Table 1 Efficacy study design using the TRAMP (C57BL/6-TgN 8247Ng) mouse model Group Treatment No. of animals Period of treatment (wks) Control Fed with AIN-76A diet only 10 16 Celecoxib treatment Fed with 800 ppm of celecoxib in AIN-76A diet 10 16 Exisulind treatment Fed with 500 ppm of exisulind in AIN-76A diet 10 16

scope. The dorsal prostate, lateral prostate, ventral prostate, and developed with chemiluminescence detection reagents [en- anterior prostate lobes of the prostate were microdissected into hanced chemiluminescence (ECL); Amersham Biosciences, Pis- individual lobes whenever possible and fixed overnight in 10% cataway, NJ]. formalin and then transferred into 70% EtOH. Paraffin-embed- Western blot analysis was extended to measure the level of ded tissue sections of 5-␮m thickness were used for histology. expression of androgen receptor (AR), nuclear factor (NF)-␬B Hematoxylin and eosin (H&E)-stained sections were examined p65, vascular endothelial growth factor (VEGF), AKT (total and for neoplastic changes. Histologic evaluations for the presence phosphorylated Ser473), BclII, p53, cyclin-dependent kinase in- of epithelial stratification and related changes indicative of PIN hibitor p21WAF1/CIP1, p27, BAX, and caspase-3 using specific and adenocarcinoma of the dorsolateral prostate were conducted antibodies (Santa Cruz Biotechnology, Santa Cruz, CA) to dem- by assessing the changes that are characteristic for TRAMP onstrate the COX-2–independent mechanism involved in the prostate histology as described previously (36, 44). All tissues inhibition of PIN lesions of the dorsolateral prostate by both from our study were evaluated and reviewed in collaboration celecoxib and exisulind. Densitometric analysis of the protein with the help of a designated pathologist. For the histologic, bands was performed with Gel-Pro Analyzer software (Media immunohistochemical, and Western blot analyses, we used dor- Cybernetics) as described previously (28). solateral prostate tissues derived from treatment and control Measuring Prostaglandin E2 Levels. We used the As- groups sacrificed after 2 weeks (for PIN lesions) or 16 weeks say Designs (Ann Arbor, MI) Correlate-EIA-prostaglandin (PG) (for adenocarcinoma). E2 enzyme immunoassay to measure the PGE2 level in the Scoring of Apoptotic Cells. Apoptosis was assessed by serum and dorsolateral prostate tissues from TRAMP mice. At the identification of cells showing morphologic changes asso- the time of sacrifice, serum and dorsolateral prostate tissue with ciated with condensation of nuclear material by examination of PIN from TRAMP control and experimental groups of cele- H&E-stained sections under a light microscope at high-power coxib- and exisulind-treated mice was snap-frozen and stored at fields. The apoptotic index was estimated by the percentage of Ϫ80°C. Frozen tissues were ground in liquid nitrogen using a apoptotic cells scored under ϫ40 magnification. The percentage mortar and pestle, weighed, and extracted for PGE2 as per the was calculated by dividing the total number of apoptotic cells by manufacturer’s recommendation. PGE2 standard with a stock of the total number of cells per section in 10 different fields. The 50,000 pg/mL PGE2 was used for subsequent lower dilutions. results presented are for the dorsolateral prostate tissues derived The assay involves the use of a monoclonal antibody to PGE2 to from each mouse in the experimental and control groups. bind in a competitive manner the PGE2 in the sample or an Immunohistochemical Detection of Proliferating Cell alkaline phosphatase molecule, which has PGE2 covalently at- Nuclear Antigen. We used paraffin-embedded dorsolateral tached. After a short incubation time, the enzyme reaction was prostate tissue sections of 5-␮m thickness for all immunohisto- stopped, and the yellow color generated was read on a micro- chemical analyses. After rehydration, antigens were retrieved by plate reader at 405 nm. The intensity of bound yellow color is a process involving microwaving with antigen-unmasking fluid inversely proportional to the concentration of PGE2 in the (Vector Laboratories, Burlingame, CA) twice for 5 minutes, sample. The results presented here are based on data from three with a 3-minute interval. After 15 minutes at room temperature, independent assays. the sections were washed and blocked with 10% normal horse Statistical Analysis. Measures of COX-2, PGE2, PCNA, serum. The sections were then incubated for 1 hour with primary GUT weight, the number of PIN lesions, and so forth were mouse anti-PCNA antibody (clone PC10; diluted 1:200 in se- compared among the groups using one-way analysis of variance rum; Lab Vision, Fremont, CA) at room temperature. Overall (ANOVA) followed by Tukey’s multiple comparisons proce- expression levels of PCNA were detected using a universal dure (45). Differences in body weight measured over time were labeling kit (horseradish peroxidase/3,3Ј-diaminobenzidine) evaluated through graphic summaries and repeated-measure from Ventana Medical Systems (Tucson, AZ). The Image Pro ANOVA models. All tests were considered statistically signif- software program (Media Cybernetics, Silver Spring, MD) was icant at P Ͻ 0.01. used to quantitate the total number of PCNA-positive cells in a minimum of 10 fields at ϫ40 magnification. Western Blots for Molecular Targets. Total protein RESULTS from control TRAMP dorsolateral prostate and tumor tissues General Observations. An overall weight gain was ob- from the group of mice fed experimental and control diets for 2 served in those TRAMP mice that received the control diet weeks and 16 weeks, respectively, were isolated. Briefly, 100 AIN-76A and experimental diets containing celecoxib and ex- mg of tissue was extracted with extraction buffer containing 150 isulind. The experimental mice that consumed the diet with mmol/L NaCl, 10 mmol/L Tris (pH 7.2), 5 mmol/L EDTA, either celecoxib (800 ppm) or exisulind (500 ppm) for 16 weeks 0.1% Triton X-100, 5% glycerol, and 2% SDS in addition to a showed no signs of toxicity but had a small decrease in total mixture of protease inhibitors (Boehringer Mannheim, Mann- weight gain as compared with the control group (Fig. 1). At the heim, Germany). Aliquots of protein (20 ␮g/lane) were frac- time of sacrifice, the lower GUT (including testes, bladder, tionated on 10% SDS-PAGE gels and transferred onto polyvi- prostate, and seminal vesicles) was removed, and the total wet nylidene difluoride membranes. The Western blot procedure weight was recorded in grams and is presented in Fig. 2. A was carried out as described previously (28). The level of difference in the total body weight and GUT weight appeared to ␤-actin expression was used as the internal control for equal be attributable to the inhibition of cell proliferation and tumor loading. The antibodies for COX-1 and COX-2 were purchased growth in the experimental mice compared with the weight gain from (Cayman, Ann Arbor, MI). Reactive protein bands were in the TRAMP control animals.

Effect of Celecoxib and Exisulind on Proliferating Cell Nuclear Antigen. Proliferating cell nuclear antigen is an aux- iliary protein for DNA polymerase. It reaches maximal expression during the S phase of the cell cycle. Hence, PCNA has been widely used as an index of the proliferative activity of a tumor. In this study, we measured PCNA labeling only in the adenocarcinoma. The level of PCNA-positive cells was higher in TRAMP control adenocarcinoma than was the level of expression in adenocarcinoma of the mice treated with celecoxib or exisulind after 16 weeks. Lower PCNA labeling indicates less proliferation and hence the regression of prostate cancer (Fig. 6AϪC). Quantitative aspects of the PCNA expression in different groups are presented in Fig. 6D. Effect of Celecoxib or Exisulind on COX-1, COX-2, and

PGE2 Expression. Protein extracted from prostate tissue sam- Fig. 1 Effect of dietary celecoxib and exisulind on body weight. ple at the early stage of PIN lesions, i.e., after 2 weeks of TRAMP mice were given either celecoxib (800 ppm) or exisulind (500 treatment, did not show either COX-1 or COX-2 expression. ppm), along with the modified AIN-76A diet for 16 weeks. TRAMP controls were fed only with the modified AIN-76A diet. Body weights However, dietary intake of celecoxib for 16 weeks reduced the in all groups were monitored over time using repeated measures of expression of COX-2 in the dorsolateral prostate adenocarci- pairwise and time-line comparisons performed with ANOVA. P Ͻ 0. noma tissue. Western blot analysis of the total protein from 001 for the differences observed between control and celecoxib or adenocarcinoma after 16 weeks of treatment indicated a marked exisulind treatment. inhibition of COX-2 by celecoxib by comparison with that in the exisulind-treated or control TRAMP mice, as shown in Fig. 7A. However, we also observed a minor impact on the COX-1 protein in the celecoxib-treated as well as the exisulind-treated Histopathological Evaluation of the Dorsolateral Pros- group of mice. The bar graph represents the level of COX-2 tate. Histologic evaluation of the H&E-stained dorsolateral expression as determined by densitometric analysis (Fig. 7B).

prostate tumor tissues revealed that the tissue sections from the Although the level of PGE2 is higher in the prostate tissue than control TRAMP mice examined between 8 and 10 weeks of age in the serum, both agents inhibited PGE2 very effectively, with had developed PIN with characteristic changes as described by a small difference between celecoxib and exisulind (P Ͻ 0.005) Shappell et al. (44), in addition to information from other as shown in Fig. 7C. studies in identifying the presence of papillary projections of Effects of Celecoxib or Exisulind on AR, NF-␬B p65, epithelial cells in 75% of the total acini counted (Fig. 3A and B). VEGF, BclII, and AKT Expression Levels. To examine the All of the TRAMP mice in each group (100%) had developed molecular targets that are involved in NSAID-mediated regres- prostate cancer after 24 weeks of age. Adenocarcinoma with sion of PIN lesions, we investigated the level of expression of multiple layers of stromal fibroblasts of the dorsolateral prostate was seen in all TRAMP mice (Fig. 3C and D). Regression of Murine PIN Lesions by Celecoxib or Exisulind Treatment. Feeding supplements of celecoxib or exisulind to TRAMP mice for 2 weeks at 800 and 500 ppm, respectively, reduced the number of histologically identifiable PIN lesions. Both celecoxib and exisulind effectively reduced the total numbers of PIN lesions in comparison with controls. The total number of PIN lesions determined in the treatment groups versus the control group of mice is shown in Fig. 4; the reduction of lesions was from 75% to 19% (SE, Ϯ4.26) with celecoxib (P Ͻ 0.0001) and from 75% to 16% with exisulind (SE, Ϯ6.63; P Ͻ 0.0001). Celecoxib- and/or Exisulind-Induced Apoptosis. Nu- merous cells identified in the H&E-stained sections showed condensed nuclei associated with higher nuclear to cytoplasmic ratio (confirmed with light microscope), reflecting apoptotic cell Fig. 2 GUT weight. At the time of sacrifice after 16 weeks of treat- death in the tumors of the dorsolateral prostate observed after ment, the lower GUT components including testes, bladder, prostate, feeding of celecoxib or exisulind feeding for 16 weeks. The and seminal vesicles were removed from TRAMP control mice and well-defined and characteristic apoptotic cells in the PIN and from the experimental groups treated with either celecoxib (800 ppm) or adenocarcinoma of the TRAMP are shown in Fig. 5AϪD. exisulind (500 ppm). Difference in GUT weight for TRAMP control versus celecoxib treatment, P Ͻ 0.05; difference in GUT weight for Quantification of the rate of apoptosis induced by these agents TRAMP control versus exisulind treatment, P Ͻ 0.01 (shown in the bar is presented in Fig. 5E. graph).

Fig. 3 Histologic evaluation of PIN and adenocarcinoma. A, H&E staining of the TRAMP dorsolateral prostate showing histologic changes associated with mouse PIN lesions after 8 weeks of age. B, higher magnification (ϫ40) of the section showing the lesions with papillary projections of proliferating epithelial cells characteristic of PIN. C, a section through the adenocarcinoma of the prostate after 24 weeks of age. D, adenocarcinoma at higher magnification (ϫ40) showing penetra- tion of highly proliferating cells through the basement mem- brane of PIN-involved glands (invasive carcinoma)

AR, NF-␬B p65, VEGF, BclII, and AKT. These molecular levels of p53 and p21WAF1/CIP1 were very clearly evident in the targets are linked to the androgenic, angiogenic, and inflamma- celecoxib- and exisulind-treated groups of mice, suggesting a tory processes of tumor progression. Western blot analysis of significant role of a p53-dependent apoptosis mechanism. The the total protein of prostate tissue with PIN lesions indicated bar graph represents the level of expression of the above- reduced expression of AR, NF-␬B p65, VEGF, total and phos- mentioned markers as determined by densitometric analysis 473 phorylated AKT (Ser ), and BclII in the celecoxib-treated (Fig. 9B). mice compared with that seen in the exisulind-treated or control TRAMP mice, as shown in Fig. 8A. The bar graph represents the level of expression as determined by the densitometric analysis (Fig. 8B). Down-regulation of AKT protein (total and phosphorylated at Ser473) signaling by celecoxib clearly indicates an inhibition of the survival gene and the associated pathological process that could otherwise lead to adenocarcinoma. Although exisulind also inhibited AKT phosphorylated at Ser473, a small apparent increase in the total AKT level was unexpected, and this could be related to several other factors, which need to be examined further in the TRAMP model when tumor growth is neuroendocrine related. Our findings emphasize that the level of AKT and NF-␬B p65 expression in the exisulind-treated TRAMP mice is independent of its total and significant effect in inducing apoptosis and thus reducing the PIN lesions. Celecoxib- or Exisulind-Mediated COX-2–Independent Molecular Targets. Histologic evaluation of both celecoxib- and exisulind-treated mice showed reduced PIN lesions of the dorsolateral prostate and was associated with a higher number of Fig. 4 Effect of celecoxib and exisulind against PIN lesions. Dorso- apoptotic cells. Therefore, we examined the molecular targets lateral prostate from the TRAMP mice sacrificed after 2 weeks of that enhanced the mechanism of apoptosis. Interestingly ele- treatment with celecoxib (800 ppm) or exisulind (500 ppm) as described vated expression levels of tumor suppressor gene p53; cyclin- in Materials and Methods was examined for PIN lesions. The bar graph WAF1/CIP1 represents the total percentage of PIN lesions identified in tissue sec- dependent kinase inhibitors such as p21 , p27, and tions from a total of five mice in each group. The total number of lesions proapoptotic BAX; and caspase-3 were evident from Western counted in 10 high-power fields are presented for control versus cele- blot analysis; these are shown in Fig. 9A. Higher expression coxib or exisulind, where P Ͻ 0.0001.

Fig. 5 Apoptosis induced by celecoxib and exisulind. Histologic demonstration of the presence of apoptotic cells in the TRAMP prostate after administration of either celecoxib (800 ppm) or exisulind (500 ppm), along with the modified AIN-76A diet for 16 weeks. TRAMP control mice were fed only the modified AIN-76A diet. H&E-stained sections showing the condensed and fragmented nuclear material characteristic of apoptotic cells are shown: A, control; B, celecoxib treatment; C, exisulind treatment; and D, higher magnification of apoptotic cells (arrow indicates apoptotic cells).

Fig. 6 Effect of celecoxib and exisulind on the proliferation marker PCNA. Immunohistochemical determination of PCNA levels in PIN lesions of the dorsolateral prostate. A, PCNA expression in control TRAMP. B, PCNA expression in celecoxib (800 ppm)- treated TRAMP. C, PCNA expression in exisulind (500 ppm)- treated TRAMP. The difference in the PCNA expression in the control versus experimental groups is presented in D as a bar graph. Celecoxib, P Ͻ 0.001; exisulind, P Ͻ 0. 0001.

adenocarcinoma of the dorsolateral prostate showed more apoptotic and fewer PCNA-positive cells compared with the control, reflecting that celecoxib and exisulind inhibit cell proliferation and induce apoptosis. Our earlier studies with celecoxib

Fig. 5 Continued. E, apoptotic cells counted in 10 high-power fields were quantified and presented (bar graph) as a percentage of apoptotic cells.

DISCUSSION The present study is part of an ongoing investigation of the chemopreventive efficacy of NSAIDs against prostate cancer in preclinical assays. Recent reports on NSAIDs against prostate cancer strongly support the need to understand the molecular mechanisms behind their action in a dose- and time-dependent manner (46, 47). First of all, we confirm the suitability of the TRAMP model for prostate cancer chemoprevention studies because it enables examination of the effects of the agents on PIN and subsequent stages, including adenocarcinoma of the dorsolateral prostate. Whereas the TRAMP model assays are widely used to foster the understanding of several molecular events related to prostate cancer (34–36, 48, 49), thus far only a few investigators have reported using this assay for studies on the chemopreventive effects against prostate cancer. These in- clude Gupta et al. (40), who tested the efficacy of difluoromethylornithine, and Wechter et al. (41), who examined the efficacy of R-flurbiprofen against prostate cancer. However, these studies provide insufficient information regarding the effect of chemopreventive agents on the early lesions in the dorsolateral prostate that are precursors for prostate cancer. Transgenic mice spontaneously develop prostate cancer through a series of well-defined stages, including PIN lesions that already show the characteristic alteration leading to markers of cell proliferation (34, 44). Therefore, we used the TRAMP assay to investigate the chemopreventive effects of celecoxib and exisulind at specific doses based on maximum tolerated dose Fig. 7 Effect of celecoxib and exisulind on COX-1, COX-2, and PGE2 expression in TRAMP dorsolateral prostate. A, Western blot analysis to assays. detect changes in the expression of COX-1 and COX-2 in dorsolateral Celecoxib is a cancer-preventive agent that has been highly prostate tissue samples from control and experimental groups after effective as an inhibitor of colon cancer in preclinical models administration of either celecoxib (800 ppm) or exisulind (500 ppm) for (20, 24, 25, 50). The data presented here clearly demonstrate 16 weeks. Reactive protein bands for COX-1 and COX-2 were analyzed that administration of either celecoxib or exisulind at 800 or 500 using ECL detection with specific antibodies as described in Materials and Methods. B. Expressed proteins were quantified as the ratio to ppm, respectively, in the diet for 16 weeks is nontoxic. An control ␤-actin. Each bar represents the approximate amount of protein important observation of this study is that mice sacrificed after corresponding to the banding pattern that was measured by densitomet- only 2 weeks of dietary administration of celecoxib or exisulind ric analysis performed with Gel-Pro Analyzer software (Media Cyber- showed a lower number of PIN lesions compared with the netics). C. Serum and dorsolateral prostate tissue from control TRAMP, celecoxib (800 ppm)-treated, and exisulind (500 ppm)-treated mice were control, suggesting that these doses are very effective for pre- analyzed for PGE using correlate-EIA-PGE enzyme immunoassay. venting the occurrence of early lesions of the dorsolateral pros- 2 2 Levels of PGE2 (measured in pg/mL) in the control were compared for tate. More importantly, these few PIN lesions and, later, the treatment with celecoxib and or exisulind.

present study, that COX-2 inhibitors operate by inducing apoptosis and by inhibiting cell proliferation irrespective of their effects on COX-2 function. However, it is also very clear from this study that a decrease in COX-2 by celecoxib is associated

with a more significant decline in PGE2 levels in the prostate tissue as compared with the serum (as shown in Fig. 7C). PGE2 is a major downstream mediator of COX-2 that promotes cel- lular proliferation and angiogenesis, makes cells resistant to apoptosis, enhances invasiveness, and modulates immunosup-

pression. COX-2 and COX-2–derived PGE2 may be involved in

Fig. 8 Expression of AR, NF-␬B p65, VEGF, AKT, and BclII in the dorsolateral prostate with PIN. A. The results are presented for the tissue samples analyzed in the experimental groups after administration of either celecoxib (800 ppm) or exisulind (500 ppm) for 2 weeks. Reactive protein bands were analyzed using ECL detection with specific antibodies as described in Materials and Methods. B. Expressed proteins were quantified as the ratio to control ␤-actin. Each bar represents the approximate amount of protein corresponding to the banding pattern that was measured by densitometric analysis performed with Gel-Pro Ana- lyzer software (Media Cybernetics). The significant difference between treatment and control, determined from pairwise comparisons, resulted in a P of Ͻ0.001.

and exisulind (28), along with reports from the literature (30), clearly support the effectiveness of these two agents in inducing apoptosis and inhibiting prostate cancer cell growth at very low doses. The histologic analysis of apoptotic cells and PCNA Fig. 9 Effect of celecoxib and exisulind on p53 expression and Bax levels clearly correlate with the chemopreventive efficacy of activation in TRAMP dorsolateral prostate. A. The results are presented these agents. for the tissue samples analyzed in the experimental groups after administration of either celecoxib (800 ppm) or exisulind (500 ppm) for 16 In recent years, there has been great interest in the question weeks. Reactive protein bands were analyzed using ECL detection with of whether NSAIDs, including COX-2 inhibitors such as cele- specific antibodies as described in Materials and Methods. Expressed coxib, affect carcinogenesis solely through COX-2 inhibition or proteins were quantified as the ratio to control ␤-actin. B. Each bar also via COX-2–independent mechanism(s). Although the pre- represents the approximate amount of protein corresponding to the banding pattern that was measured by densitometric analysis performed cise mechanisms by which celecoxib inhibits prostate carcino- with Gel-Pro Analyzer software (Media Cybernetics). The significant genesis are not fully known, we propose, on the basis of our difference between treatment and control was determined from a pair- earlier findings with in vitro models and the observations of the wise comparisons resulting in a P of Ͻ0.001.

Fig. 10 Molecular targets and mechanisms underlying the development of mouse PIN lesions and inhibition by NSAIDs. On the basis of our findings from this study using the TRAMP model assay, we derived a possible mechanism underlying the cascade of molecular events that resulted in the regression of PIN lesions and adenocarcinoma in mice fed with celecoxib and/or exisulind for a period of 2 and 16 weeks, respectively. As depicted in this figure, we conclude that the mechanisms involved in the regression of PIN lesions by celecoxib and/or exisulind are the result of several molecular changes in the PIN lesions of the dorsolateral prostate, mediated by an antiandrogenic effect resulting in inhibition of the expression of AR and nuclear transcription factor NF-␬B p65. The cascade of events initiated at this point reduced the expression of VEGF, causing an antiangiogenic effect, followed by a reduced inflammatory response that was evident from the lower expression of COX-2 in the dorsolateral prostate adenocarcinoma. The observation that the expression of survival genes AKT (total and phosphorylated at Ser473) and BclII was down-regulated effectively by celecoxib suggests a close link between the reduced number of PIN lesions and lower expression of survival genes or proteins. Interestingly, both celecoxib and exisulind activated the downstream targets of p53 and apoptosis activators such as BAX and caspase-3, indicating a convergence in the mechanism of the two agents as they inhibit PIN lesions. The NSAID-induced sequence of alterations of key molecular targets involved in the development of PIN lesions to adenocarcinoma is clearly demonstrated in this study.

colon and mammary carcinogenesis (51). Therefore, COX-2– in the exisulind-treated TRAMP mice. Our current ongoing selective inhibitors may have a role in prostate cancer preven- studies are focused more on AKT regulation in exisulind-treated tion, in which expression of COX-2 and PGE2 appears to be a prostate cancer cells. However, studies also indicate constitutive late event in prostate carcinogenesis. It is noteworthy that we activation of AKT in insulin-like growth factor II-overexpress- have shown COX-2 inhibition through celecoxib treatment but ing rhabdomyosarcoma cells (55). Exploring the role of AKT in less of an effect on COX-2 by exisulind, yet both agents reduced the activation of NF-␬B p65 in prostate cancer cells is very PIN lesions and caused more apoptotic cells in the adenocarci- important to understand its role in the initiation of PIN. Al- noma of the dorsolateral prostate than are present in spontane- though AKT expression is an early event in colon carcinogen- ous tumors in the control mice in a COX-2–independent man- esis (56), the exact transition point of AKT activation in the ner. These observations suggest that inhibition of prostate prostate that leads to tumor growth is not yet clear. This suggests cancer by celecoxib and exisulind is mediated through both that a better understanding of AKT signaling, specifically by COX-2–dependent and –independent mechanisms, strengthen- exisulind against prostate cancer, is needed. ing the concept of several investigators that more than one We conclude that the mechanisms involved in the regres- molecular mechanism is involved in tumor inhibition by these sion of PIN lesions by celecoxib and/or exisulind are the result chemopreventive agents (20, 21, 52–54). of several molecular changes in the PIN lesions of the dorso- Phosphatidylinositol 3Ј-kinase/AKT is an important sur- lateral prostate, mediated by an antiandrogenic effect resulting vival signal pathway that has been shown to be crucial in the in inhibition of the expression of AR and nuclear transcription regulation of balance between the proapoptotic and survival factor NF-␬B p65 as illustrated in Fig. 10. The cascade of events signal. initiated at this point reduced the expression of VEGF, causing Our results showed a significant decrease in total and an antiangiogenic effect, followed by a reduced inflammatory phosphorylated AKT at Ser473 with celecoxib compared with response that was evident from the lower expression of COX-2 the control. There is no significant correlation among the num- in the dorsolateral prostate adenocarcinoma. The observation ber of PIN lesions, the rate of apoptosis, and the total AKT level that the expression of survival genes AKT (total and phospho-

rylated) and BclII was down-regulated effectively by celecoxib 17. Badawi AF, El-Sohemy A. Non-steroidal anti-inflammatory drugs suggests a close link between reduced number of PIN lesions in chemoprevention of breast and prostate cancer. Med Hypotheses and lower expression of survival genes or proteins. Interest- 2001;57:167–8. ingly, both celecoxib and exisulind activated the downstream 18. DuBois RN. Nonsteroidal anti-inflammatory drug use and sporadic colorectal adenomas. Gastroenterology 1995;108:1310–4. targets of p53 and apoptosis activators such as BAX and 19. DuBois RN, Abramson SB, Crofford L, et al. Cyclooxygenase in caspase-3, indicating a convergence in the mechanism of the biology and disease. FASEB J 1998;12:1063–73. two agents as they inhibit PIN lesions. The NSAID-induced 20. Williams CS, Watson AJM, Sheng HM, et al. Celecoxib prevents sequence of alterations of key molecular targets involved in the tumor growth in vivo without toxicity to normal gut: lack of correlation development of PIN lesions to adenocarcinoma is clearly dem- between in vitro and in vivo models. Cancer Res 2000;60:6045–51. onstrated in this study. Further investigation on the use of low 21. Lim JT, Piazza GA, Han EK, et al. Sulindac derivatives inhibit growth and induce apoptosis in human prostate cancer cell lines. Bio- doses of these two agents with their different modes of action in chem Pharmacol 1999;58:1097–107. combination appears to be a highly promising approach that 22. Han EK, Arber N, Yamamoto H, et al. Effects of sulindac and its may evolve into a better chemoprevention strategy for future metabolites on growth and apoptosis in human mammary epithelial and efficacy studies in human clinical trials. breast carcinoma cell lines. Breast Cancer Res Treat 1998;48:195–203. 23. Reddy BS, Maruyama H, Kelloff G. Dose related inhibition of colon carcinogenesis by dietary piroxicam, a nonsteroidal anti-inflam- ACKNOWLEDGMENTS matory drug, during different stages of rat colon tumor development. Cancer Res 1987;47:5340–6. We thank Cesar Aliaga and Jeff Rigotty at the Research Animal 24. Kawamori T, Rao CV, Seibert K, Reddy BS. Chemopreventive Facility (Valhalla, NY) for their support in completing the animal study. activity of celecoxib, a specific cyclooxygenase-2 inhibitor, against We are grateful to Dr. Joseph A. Connett (Department of Comparative colon carcinogenesis. Cancer Res 1998;58:409–12. Medicine, New York Medical College, Valhalla, NY) for his assistance 25. Reddy BS, Hirose Y, Lubet R, et al. Chemoprevention of colon in the histology. We also acknowledge Dominic Nargi for his assistance cancer by specific cyclooxygenase-2 inhibitor, Celecoxib, administered with laboratory analysis, and Ilse Hoffmann for editing the manuscript during different stages of carcinogenesis. Cancer Res 2000;60:293–7. (Institute for Cancer Prevention, Valhalla, NY). 26. Masferrer JL, Leahy KM, Koki AT, et al. Antiangiogenic and antitumor activities of cyclooxygenases-2 inhibitors. Cancer Res 2000; 60:1306–11. REFERENCES 27. Badawi AF. The role of prostaglandin synthesis in prostate cancer. 1. Templeton H. The management of prostate cancer. Nurs Stand 2003; BJU Int 2000;85:451–62. 17:45–53. 28. Narayanan BA, Condon MS, Bosland MC, et al. Suppression of 2. Powell IJ, Carpten J, Dunston G, et al. African-American heredity N-methyl-N-nitrosourea (MNU)/testosterone-induced rat prostate cancer prostate cancer study: a model for genetic research. J Natl Med Assoc growth by celecoxib: effects on COX-2, cell cycle regulation and 2001;93:120–3. apoptosis mechanism(s). Clin Cancer Res 2003;9:3503–13. 3. Hirao Y, Cho M. Statistical analysis of prostate cancer in Japan. 29. Thompson HJ, Jiang C, Lu J, et al. Sulfone metabolite of sulindac Nippon Rinsho 2002;60(Suppl 11):53–8. inhibits mammary carcinogenesis. Cancer Res 1997;57:267–71. 4. Nakata S, Ohtake N, Yamanaka H. Epidemiological trends of pros- 30. Piazza GA, Rahm AK, Finn TS, et al. Apoptosis primarily accounts tate cancer in Japan and international comparisons. Nippon Rinsho for the growth-inhibitory properties of sulindac metabolites and involves 2002;60(Suppl 11):44–8. a mechanism that is independent of cyclooxygenase inhibition, cell cycle arrest, and p53 induction. Cancer Res 1997;57:2452–9. 5. Hoffman RM, Hunt WC, Gilliland FD, Stephenson RA, Potosky AL. Patient satisfaction with treatment decisions for clinically localized 31. Leng J, Han C, Demetris AJ, Michalopoulos GK, Wu T. Cyclooxy- prostate carcinoma. Results from the Prostate Cancer Outcomes Study. genase-2 promotes hepatocellular carcinoma cell growth through Akt Cancer (Phila) 2003;97:1653–62. activation: evidence for Akt inhibition in celecoxib-induced apoptosis. Hepatology 2003;38:756–68. 6. Gronberg H. Prostate cancer epidemiology. Lancet 2003;361:859–64. 32. Han C, Leng J, Demetris AJ, Wu T. Cyclooxygenase-2 promotes 7. Long RJ, Roberts KP, Wilson MJ, Ercole CJ, Pryor JL. Prostate human cholangiocarcinoma growth: evidence for cyclooxygenase-2- cancer: a clinical and basic science review. J Androl 1997;18:15–20. independent mechanism in celecoxib-mediated induction of p21waf1/cip1 8. Norrish AE, Jackson RT, McRae CU. Non-steroidal anti-inflamma- and p27kip1 and cell cycle arrest. Cancer Res 2004;64:1369–76. tory drugs and prostate cancer progression. Int J Cancer 1998;77:511–5. 33. Wu T, Leng J, Han C, Demetris AJ. The cyclooxygenase-2 inhibitor 9. Sabichi AL, Lippman SM. COX-2 inhibitors and other NSAIDs in celecoxib blocks phosphorylation of Akt and induces apoptosis in hu- bladder and prostate cancer. Prog Exp Tumor Res 2003;37:163–78. man cholangiocarcinoma cells. Mol Cancer Ther 2004;3:299–307. 10. Sabichi AL, Demierre MF, Hawk ET, Lerman CE, Lippman SM. 34. Gingrich JR, Barrios RJ, Morton RA, et al. Metastatic prostate Frontiers in cancer prevention research. Cancer Res 2003;63:5649–55. cancer in a transgenic mouse. Cancer Res 1996;56:4096–102. 11. Sporn MB. Chemoprevention of cancer. Lancet 1993;342:1211–3. 35. Greenberg NM, DeMayo FJ, Sheppard PC, et al. The rat probasin 12. Wattenberg LW. An overview of chemoprevention: current status gene promoter directs hormonally and developmentally regulated ex- and future prospects. Proc Soc Exp Biol Med 1997;216:133–41. pression of a heterologous gene specifically to the prostate in transgenic 13. Meyskens FL Jr. Re: Cancer chemoprevention: progress and prom- mice. Mol Endocrinol 1994;8:230–9. ise. J Natl Cancer Inst (Bethesda) 1999;91:563–5. 36. Greenberg NM, DeMayo F, Finegold MJ, et al. Prostate cancer in a 14. KelloffGJ, Hawk ET, Crowell JA, et al. Strategies for identification transgenic mouse. Proc Natl Acad Sci USA 1995;92:3439–43. and clinical evaluation of promising chemopreventive agents. Oncology 37. Gingrich JR, Barrios RJ, Kattan MW, et al. Androgen-independent (Huntingt) 1996;10:1471–8; discussion 1484–8. prostate cancer progression in the TRAMP model. Cancer Res 1997; 15. Kelloff GJ, Hawk ET, Karp JE, et al. Progress in clinical chemo- 57:4687–91. prevention. Semin Oncol 1997;24:241–52. 38. Moore RA. Benign hypertrophy and carcinoma of the prostate. Occur- 16. Kelloff GJ, Lieberman R, Steele VE, et al. Agents, biomarkers, and rence and experimental production in animals. Surgery 1994;16:152–5. cohorts for chemopreventive agent development in prostate cancer. 39. Eng MH, Charles LG, Ross BD, et al. Early castration reduces Urology 2001;57(Suppl 1):46–51. prostatic carcinogenesis in transgenic mice. Urology 1999;54:1112–9.

40. Gupta S, Ahmad N, Marengo SR, et al. Chemoprevention of pros- 49. Hernandez I, Maddison LA, Wei Y, et al. Prostate-specific expres- tate carcinogenesis by a-difluoromethylornithine in TRAMP mice. Can- sion of p53(R172L) differentially regulates p21, Bax, and mdm2 to cer Res 2000;60:5125–33. inhibit prostate cancer progression and prolong survival. Mol Cancer 41. Wechter WJ, Leipold DD, Murray ED Jr, et al. E-7869 (R-flurbi- Res 2003;1:1036–47. profen) inhibits progression of prostate cancer in the TRAMP mouse. 50. Lanza-Jacoby S, Miller S, Flynn J, et al. The cyclooxygenase-2 Cancer Res 2000;60:2203–8. inhibitor, celecoxib, prevents the development of mammary tumors 42. Reddy BS, Rao CV. Colon cancer-a role for cyclo-oxygenase-2-spe- in Her-2/neu mice. Cancer Epidemiol Biomark Prev 2003;12: cific nonsteroidal anti-inflammatory drugs. Drugs Aging 2000;16:329–34. 1486–91. 43. Howe LR, Subbaramaiah K, Patel J, et al. Celecoxib, a selective 51. Wang D, Dubois RN. Cyclooxygenase-2: a potential target in breast cyclooxygenase 2 inhibitor, protects against human epidermal growth cancer. Semin Oncol 2004;31:64–73. factor receptor 2 (HER-2)/neu-induced breast cancer. Cancer Res 2002; 62:5405–7. 52. Hsu AL, Ching TT, Wang DS, et al. The cyclooxygenase-2 inhibitor celecoxib induces apoptosis by blocking Akt activation in human 44. Shappell SB, Thomas GV, Roberts RL, et al. Prostate pathology of prostate cancer cells independently of Bcl2. J Biol Chem 2000;275: genetically engineered mice: definitions and classification. The consen- 11397–403. sus report from the Bar Harbor meeting of the Mouse Models of Human Cancer Consortium Prostate Pathology Committee. Cancer Res 2004; 53. Liu XH, Kirschenbaum A, Yao S, et al. Inhibition of cyclooxyge- 64:2270–305. nase-2 suppresses angiogenesis and the growth of prostate cancer in 45. Miller RG. Simultaneous statistical inference, 2nd ed. New York: vivo. J Urol 2000;164:820–5. Springer-Verlag; 1981. p. 37. 54. Song X, Lin HP, Johnson AJ, et al. Cyclooxygenase-2, player or 46. Sabichi AL, Lippman SM. COX-2 inhibitors and other nonsteroidal spectator in cyclooxygenase-2 inhibitor-induced apoptosis in prostate anti-inflammatory drugs in genitourinary cancer. Semin Oncol 2004;31: cancer cells. J Natl Cancer Inst (Bethesda) 2002;94:585–91. 36–44. 55. Wan X, Helman LJ. Effect of insulin-like growth factor II on 47. Terry MB, Gammon MD, Zhang FF, et al. Association of frequency protecting myoblast cells against cisplatin-induced apoptosis through and duration of aspirin use and hormone receptor status with breast p70 S6 kinase pathway. Neoplasia 2002;4:400–8. cancer risk. J Am Med Assoc 2004;291:2433–40. 56. Roy HK, Olusola BF, Clemens DL, et al. AKT proto-oncogene 48. Gingrich JR, Greenberg NM. A transgenic mouse prostate cancer overexpression is an early event during sporadic colon carcinogenesis model. Toxicol Pathol 1996;24:502–4. Carcinogenesis (Lond) 2002;23:201–5.

Downloaded from clincancerres.aacrjournals.org on September 27, 2021. © 2004 American Association for Cancer Research. Regression of Mouse Prostatic Intraepithelial Neoplasia by Nonsteroidal Anti-inflammatory Drugs in the Transgenic Adenocarcinoma Mouse Prostate Model

Bhagavathi A. Narayanan, Narayanan K. Narayanan, Brian Pittman, et al.

Clin Cancer Res 2004;10:7727-7737.

Updated version Access the most recent version of this article at: http://clincancerres.aacrjournals.org/content/10/22/7727

Cited articles This article cites 49 articles, 21 of which you can access for free at: http://clincancerres.aacrjournals.org/content/10/22/7727.full#ref-list-1

Citing articles This article has been cited by 22 HighWire-hosted articles. Access the articles at: http://clincancerres.aacrjournals.org/content/10/22/7727.full#related-urls

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://clincancerres.aacrjournals.org/content/10/22/7727. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.