Suppression of N-Methyl-N-nitrosourea/Testosterone-induced Rat Prostate Cancer Growth by Celecoxib: Effects on Cyclooxygenase-2, Cell Cycle Regulation, and Apoptosis Mechanism(s)¹

Prostate cancer is the second leading cause of cancer-related deaths in American men. African Americans have the highest rate of prostate cancer incidence (1, 2). A long-term goal of our research is to develop innovative strategies for prostate cancer prevention. A number of studies indicate a strong correlation between the levels of arachidonic acid metabolites and accumulation of prostaglandins in prostate carcinogenesis (3, 4, 5, 6, 7). COX-2,³ one of the key rate-limiting enzymes involved in arachidonic acid metabolism, has been shown to be involved in prostate cancer, as well as in several other human cancers and inflammatory diseases (8, 9, 10, 11, 12, 13, 14, 15). Epidemiological studies have revealed a decreased risk of colon cancer among people who have regularly taken COX-2 inhibitors, like aspirin or other NSAIDs (14, 16). Both human studies and preclinical model assays testing COX-2 inhibitors support the role of NSAIDs in the prevention of cancer of the colon and other organs (8, 11, 14, 17, 18, 19, 20, 21, 22, 23). In this context, it is noteworthy that a cohort study suggested that regular use of NSAIDs could protect against prostate cancer (24). Although selective COX-2 inhibitors are known to induce tumor growth inhibition through COX-2 down-regulation, not much is known about COX-2-mediated tumor growth inhibition in prostate cancer, mainly because there is little expression of COX-2 in the initial stages of this cancer, and also because of a lack of proper animal models. Because the MNU/testosterone-induced prostate tumor is a fully developed tumor model, we have chosen it for investigating the role of celecoxib in tumor growth inhibition and related mechanism(s). Expression of COX-2 at the mRNA and protein level in human prostate tumors has demonstrated a direct association between COX-2 levels and prostate cancer (25, 26, 27, 28, 29, 30, 31, 32). Despite considerable enthusiasm about the use of celecoxib against cancer of the colon and of the breast, studies on its use against prostate cancer are few (33, 34, 35, 36, 37). Rats treated with MNU/testosterone develop a high incidence of prostate tumors (38), which is an advantage for testing the efficacy of celecoxib and elucidating its molecular mechanism(s). Earlier studies by Conden et al. (39) had suggested multiple pathways for MNU/testosterone-induced prostate carcinogenesis in the rat model. Even though studies by Pollard and Luckert and Pollard et al. (38, 40, 41, 42, 43) with NSAIDs, such as peroxicum and indomethacine in MNU-treated rats, indicated effects on both intestinal and prostate tumors, effects of celecoxib on COX-2 regulation or apoptosis mechanisms have not been delineated. Interestingly, this rat model was used for several chemoprevention studies, such as (a) the evaluation of the activity of 9-cis-retinoic acid against prostate carcinogenesis in male (HsdCpb:Wu) rats (44); (b) very effective dose-related tumor growth-inhibition studies with dehydroepiandrosterone (DHEA) in MNU-induced rat prostate tumors (45); and (c) treatment with difluoromethyl ornithine or finasteride of MNU-induced rat prostate tumors showed chemopreventive efficacy by inhibiting prostate polyamine synthesis (46).

We investigated the effect of celecoxib on COX-2 expression, cell growth, DNA synthesis, cell cycle distribution, and apoptosis as well as on expression of cyclin D1, cyclin E, and cyclin kinase inhibitors, such as p27^KIP1 and p21^WAF1/CIP1 to verify cancer growth inhibition and induction of apoptosis via cell cycle regulatory mechanisms in dorsolateral prostate cancer cells. Taken together, our observations strongly indicate a correlation in the down-regulation of COX-2 with targets such as NF-κBp65 that makes cells susceptible to the actions of apoptosis-inducing mechanisms that are believed to be simultaneously activated by celecoxib.

The Searle Research and Development, Pharmacia, St. Louis, MO, kindly provided celecoxib. Solutions of 0.5- to 10-, 20-, and 40-μm concentrations of celecoxib were prepared in DMSO. The cell cultures used in the present study were established as described previously (39, 47) from rat prostatic adenocarcinomas induced by MNU plus testosterone in Wistar WU rats (48, 49, 50). Examination of H&E-stained sections of the original tumors indicated a dorsolateral prostatic origin. In vitro, the cells appear polyhedral in shape and grow in a “cobblestone” pattern, suggesting their epithelial nature, which was confirmed by positive immunohistochemical staining for cytokeratin 7. The cell cultures used in this study (I-26) are tumorigenic when injected s.c. into syngeneic hosts. They consistently form metastases in the lung and regional lymph nodes. The in vitro growth rate (doubling-time) of these cultures is within 12 to 24 h, and cells from passage 8 were used in all of these studies. The cells were propagated in F-12K medium containing 50 nm testosterone (Steraloids, Wilton, NH), 5% fetal bovine serum, 100 μg/ml penicillin and 100 mg/ml streptomycin. A uniform number of cells plated in T-25 or 35-mm cell-culture dishes were placed in a humidified incubator at 37°C with 5% CO₂ before treatment with celecoxib. Controls received 1% DMSO only. The experiments were repeated three times for each dose and time period after exposure to the groups, celecoxib, and control.

Cell viability after celecoxib treatment was determined by the trypan blue (0.2%) exclusion assay. After 6, 12, 24, and 48 h of every experiment, adherent and floating cells were harvested by trypsinization and recovered by centrifugation. Trypan blue staining of cells enables easy identification of dead cells because these take up the dye and appear blue with uneven cell membranes. By contrast, living cells repel the dye and appear refractile and colorless.

The soft-agar (Life Technologies, Inc.) clonogenic assay was performed to determine the efficacy of celecoxib at very low doses, ranging from 0.5 to 10 μm. Briefly, rat prostate cancer cells were grown on soft-agar plates (∼50,000 cells/plate) made with F-12K medium containing 1.5% agar. Celecoxib solutions of 0.5-, 2.0-, 5.0-, or 10.0-μm concentrations were added to the culture medium containing the agar. For each dose, triplicate plates were used along with the control. The plates were incubated for 10 days at 37°C in 5% CO₂. Colonies identified by crystal violet staining that showed more than 25 cells/colony were precisely counted to determine the efficacy; the results are presented as a percentage of the control (with DMSO).

Control cells and cells treated with celecoxib as described above, were harvested after 48 h by trypsinization and fixed in ice-cold 80% ethanol for up to 24 h. The cells were then washed twice with PBS, suspended in 1 ml of 0.25% Triton X-100 in PBS, kept on ice for 5 min, and centrifuged. The resulting pellet was suspended in 100 μl of PBS. After a washing with PBS, cells were resuspended in 5 μg/ml PI (Molecular Probes, Eugene, OR) with 0.1% RNase A (Sigma) in PBS. After incubation for 20 min at room temperature in the dark, the cells were analyzed by flow cytometry with the Coulter Epic Elite ESP. Cell cycle analysis was performed as described by Darzynkiewicz et al. (51) and Gong et al. (52).

The effect of celecoxib on DNA synthesis was measured by assessing BrdUrd incorporation into DNA during the S phase as described earlier (53), with some modifications. Briefly, rat prostate cancer cells grown in 35-mm dishes were incubated with BrdUrd (30 μg/ml) along with different concentrations of celecoxib (10, 20, or 40 μm) in the cell culture medium; after 48 h, the experiments were terminated by immediately fixing the cells in 10% formalin. The PBS-washed cells were pretreated with 0.1% Triton X-100, and 2 n HCl at 37°C for 10 min, and were then treated with 0.1 m sodium borate for 5 min. The cells were incubated with mouse monoclonal antibody for BrdUrd (NCL-BrdUrd; Novacastra) for 1 h and, thereafter, with FITC-conjugated secondary antibody (goat antimouse IgG) for 30 min to detect the incorporation of BrdUrd in the S phase of the cells. Cells were viewed for BrdUrd incorporation with a fluorescence microscope (AX-70 Epi-fluorescence scope; Olympus) equipped with a computer-controlled digital camera (Spot) for imaging. Cells showing the FITC signal for BrdUrd incorporation were quantified from three independent experiments.

The rate of apoptosis induced by celecoxib in rat prostate cancer cells was assessed first with DAPI staining of the nuclear material. Briefly, cells grown in 35-mm dishes were treated with celecoxib for 48 h; after termination of the experiments, the floating and adherent cells were first fixed in 10% formalin for 15 min. After a washing with PBS, cells were treated with 0.1% Triton X-100, 4 m HCl, and sodium tetraborate; each treatment was performed for 15 min and was followed by a PBS wash. Cells were then stained with DAPI in 80% methanol for 30 min and again washed with PBS. The cells were viewed under a fluorescence microscope as described previously (54). In addition, annexin V staining of the membrane for PS in the apoptotic cells was performed in the control and celecoxib-treated cells. Cells with the characteristic morphological changes of apoptosis were quantified among 100 cells per field. Representative cells of more than three fields were selected for quantification.

To determine whether rat prostate tumor, as well as the cancer cells, expresses COX-2, we used immunofluorescence detection techniques for both paraffin-embedded tumor tissue sections and the cancer cells derived from the tumor. First, deparaffinized tumor tissue sections were incubated with FITC-conjugated anti-COX-2 primary antibody for 6 h (1:1000) after a brief heat treatment, COX-2-positive cells in the nontumor and tumor regions were detected by comparison with the H&E-stained sections. Incubations of cells treated with 10, 20, or 40 μm celecoxib were performed after 48 h. PBS-washed cells, fixed in 10% formalin and pretreated with 0.1% Triton X-100 and 2 n HCl at 37°C for 10 min, were then treated with 0.1 m sodium borate for 5 min. COX-2-positive cells were localized using human monoclonal COX-2-specific antibody (Cayman, Ann Arbor, MI) covalently linked to FITC. The cells were viewed under a fluorescence microscope as described elsewhere (54).

Rat prostate cancer cells treated with celecoxib for 48 h were harvested by trypsinization. Cellular protein was isolated with protein extraction buffer containing 150 mm NaCl, 10 mm Tris (pH 7.2), 5 mm EDTA, 0.1% Triton X-100, 5% glycerol, and 2% SDS, in addition to a mixture of protease inhibitors (Boehringer Mannheim, GmbH, Mannheim, Germany). Equal amounts of protein (50 μg/lane) were fractionated on 10% SDS-PAGE gels and transferred to polyvinylidene difluoride membranes. The Western blot procedure was carried out as described earlier (55). The antibody used for Western blotting was from Santa Cruz Biotechnology, Santa Cruz, CA (cyclin D1, pRb, cyclin E, p27, and p21). The antibodies for COX-1 and COX-2 were from Cayman, Ann Arbor, MI. Reactive protein bands were developed using chemiluminescence detection reagents (ECL, Amersham). Densitometric analysis of the protein bands was performed with the software Gel-Pro Analyzer (Media Cybernetics, Silver Spring, MD).

Total RNA extracted from cells treated with celecoxib was subjected to RT-PCR analysis using gene-specific primer sequences by adopting protocols described earlier (54) to determine the expression of COX-2, PPARγ, Apaf-1, and caspase 3. All of the templates were initially denatured for 2 min at 94°C, and the amplification of the amplicon were extended at a final temperature of 72°C for 7 min. PCR amplification with glyceraldehyde-3-phosphate dehydrogenase was used as the internal control.

Clonogenic cell survival assays were performed with very low doses of celecoxib as shown in Fig. 1,A. There were remarkably fewer viable colonies with 10 μm compared with the effects with 2.0 and 5.0 μm. Viable cells detected by the trypan blue exclusion assay among the rat prostate cancer cells exposed to 10 μm, 20 μm, or 40 μm celecoxib after 6, 12, 24, and 48 h are presented in Fig. 1 B. More than 50% of the cells were viable after 24 h at 40 μm, but more or less 50% of cell deaths occurred at the 48-h time point. Overall, at the tested levels, celecoxib induced a time- and dose-dependent growth inhibition in rat prostate cancer cells in a nontoxic manner. The IC₅₀ was ∼25–30 μm for this cell type, although the results from the clonogenic assay clearly demonstrated cell growth inhibition at very low doses.

To determine whether the observed cell growth inhibition was caused by cell cycle arrest, the distribution of cells in different phases of the cell cycle was assessed at the 48-h point after treatment with 10, 20, or 40 μm celecoxib. Flow cytometric analysis for DNA content revealed a dose-dependent G₁ arrest accompanied by a decrease of cells in the S and G₂-M phases (Fig. 2, A1–A4). Consistent with this observation, Fig. 2, B1–B4, depicts a dual parameter plot for cell count and PI uptake. Cells with high PI fluorescence have a disrupted surface membrane indicating a greater chance for the presence of apoptotic cells in the same cluster. Findings from the trypan blue exclusion assay and flow cytometric analysis were consistent with distinct morphological changes characteristic of apoptosis that occurred in celecoxib-treated cells but not in the control cells as shown in Fig. 2, C1–C4.

Apoptosis induced by celecoxib was confirmed with DAPI staining for chromatin material and annexin V for PS. Increases in the DAPI uptake and apoptotic cells with specific morphological and nuclear material changes characteristic of apoptotic cells were quantified and are presented in Fig. 3, A1–A3 . PS is redistributed to the external surface of the villous plasma membrane. The level of PS expression increased at up to 48 h of treatment with 40 μm of celecoxib with the binding of annexin V. Similarly, the membrane localization of annexin-V apoptotic cells was quantified (Fig. 3, B1–B3 ). Results from DAPI and annexin-V localization demonstrated that more than 30% of apoptosis was induced by celecoxib at this time period.

A reduced peak for the S phase, observed by flow cytometric analysis, was consistent with less BrdUrd incorporation in cells exposed to different concentrations of celecoxib after 48 h. A gradual decline in the incorporation of BrdUrd at 20- and 40-μm concentrations after 48 h clearly revealed a difference between the control and the celecoxib-treated cells that showed only 8% BrdUrd-positive cells with 40 μm, indicating a dose-related, significantly decreased rate of DNA synthesis (Fig. 4, A–C).

The product of the Rb gene prevents S-phase entry during the cell cycle, and inactivation of this growth-suppressive function is presumed to result from Rb hyperphosphorylation during the late G₁ phase. To examine the effect of celecoxib on Rb, we used flow cytometric analysis to assess the expression of total Rb protein in rat prostate cancer cells exposed to 40 μm celecoxib for 48 h. We demonstrated an increase in the Rb/FITC incorporation in G₁-S phase-arrested prostate cancer cells (Fig. 5, A and B) that was consistent with an increase in the total Rb and pRb protein measured by Western blot analysis (Fig. 5 C).

To examine the molecular mechanism that may be underlying changes in cell cycle profiles and apoptosis, we investigated the upstream events linked to Rb and pRb. We approached this study by measuring the effects of 10, 20, and 40 μm celecoxib on the expression of cyclin D1 and -E and the cdk inhibitors, p27^KIP1 and p21^WAF1/CIP1 in rat prostate cancer cells. Western blots of these proteins indicated a 2-fold decrease in the level of cyclin D1 at 20 μm and a 3-fold decrease at 40 μm after 48 h (Fig. 6, A and B). In separate experiments, celecoxib had no effect on the expression of cyclin E in these cells, although a very minor change in the protein expression was observed with the 40-μm dose. By contrast, Western blot results showed an increase in the level of the cyclin kinase inhibitors p27^KIP1 and p21^WAF1/CIP1 protein with different concentrations of celecoxib after 48 h. Although we were focusing on some of the downstream events related to the COX-2 inhibition and induction of apoptosis, we also observed that celecoxib down-regulated the levels of expression of the transcription factors NF-κB, and PPARγ very effectively within 48 h at the mRNA level (RT-PCR) in addition to affecting the activation of Apaf-1 and caspase 3 as shown in Fig. 6 C.

COX-2 expression was identified using FITC-immunofluorescence in the MNU/testosterone-induced rat prostate tumor, and the preexisting normal glandular regions indicated a significant level in the dorsolateral prostate tumor area (T) compared with the preexisting normal glandular regions (N), as indicated in Fig. 7, A1–A2. The H&E-stained sections were compared with sections used for COX-2 detection. The rat prostate cancer cells derived from these tumors showed a higher level of COX-2 expression (Fig. 7, B1–B2) than the cells exposed to celecoxib. Quantification of the COX-2-expressing cells as depicted in Fig. 7,B3, showed a declining trend in the celecoxib-treated cells. In addition to the immunolocalization of COX-2 expression in the cells, we confirmed the presence of COX-2 protein by Western blot analysis (Fig. 7,C1). A 2-fold decrease in the expression of COX-2 was observed with 20 μm celecoxib, and 40 μm caused complete inhibition after 48 h of treatment (Fig. 7 C2). This clearly indicated the COX-2-inhibitory effect of celecoxib in carcinogen-induced rat prostate cancer cells. No significant changes were observed in the level of COX-1.

In recent years, there has been great interest in the question as to whether NSAIDs, including COX-2 inhibitors, affect carcinogenesis through COX-2 inhibition and/or via some other means. More than one molecular mechanism has been proposed by earlier investigators (17, 33, 34, 56, 57, 58, 59) for the antiproliferative effects of COX-2 inhibitors. Moreover, there appear to be multiple molecular targets through which celecoxib could exert its anticancer effects in addition to its primary function as a COX-2 inhibitor. Efficacy studies, both in vitro and in vivo, conducted with NSAIDs in both colon cancer and breast cancer models have demonstrated their anticancer effects; the few studies done on the chemoprevention of prostate cancer did not contribute much with regard to cellular and molecular mechanisms for prostate cancer prevention other than the modulation of Akt and BclII by celecoxib (59). To identify molecular targets other than COX-2 for prostate cancer prevention, we have investigated the functional mechanism of celecoxib focusing on COX-2-independent pathways involved in cell growth inhibition. We have demonstrated for the first time, a time- and dose-dependent effect of celecoxib on cell cycle arrest and growth inhibition of rat prostate cancer cells that were derived from chemically induced prostate cancer. Cell growth arrest at the G₁-S-phase transition point, associated with an inhibitory effect on the rate of DNA synthesis within 48 h at different doses of celecoxib, was evident from flow cytometric analysis and BrdUrd incorporation and Rb/pRb expression.

Additional attempts to detect and quantify cell death induced by celecoxib, via DAPI and annexin-V staining for apoptotic cells, revealed the presence of more than 30% apoptotic cells. Our observation was consistent with earlier studies, which indicated that colon cancer cell growth inhibition by celecoxib is associated with the induction of apoptosis (17, 60). In addition, as observed in the present study, cell cycle arrest associated with apoptosis is clearly one of the mechanisms by which celecoxib blocks cell cycle progression (59, 61), and it does so in a dose-dependent manner. Results from clonogenic survival assays further indicated that the lowest level of celecoxib that could be effective in preventing cancer cell growth in nude mice, in which the plasma concentrations of celecoxib were in the range of 3–4 μm (100 mg/kg), were similar to those found in humans after a single oral dose of 800 mg (3–5 μm). [Two × 400 mg is approved by the FDA for familial adenomatus polyposis patients as reported by Grosch et al. (60).] The results from the present in vitro study show that doses of 10 to 40 μm celecoxib over 48 h can still induce cellular and molecular changes that lead to apoptosis in a time-dependent manner.

Flow cytometric analysis of cell cycle distribution revealed that, in rat prostate cancer cells, celecoxib at 10, 20, and 40 μm caused an increase in the number of cells at the G₁-S-phase transition point with a trend toward accumulation of more cells in the G₁ phase. Although there was only a small apoptotic peak at the very lowest concentration (10 μm), the results suggest a possibly impaired progression through the S phase. At 40 μm of celecoxib, the cell cycle distribution was affected in two ways: (a) virtually all of the cells accumulated in the G₁-S phase, although it was not possible to rule out a few cells in the G2-M phase; and (b) at higher concentrations of celecoxib, the peak for apoptotic cells appeared small when compared with untreated control cells. Results from our present study on an inhibitory effect on DNA synthesis, determined by BrdUrd incorporation and induction of apoptosis by celecoxib, support the results of Leahy et al. (61). However, the observed time- and dose-dependent growth inhibition of carcinogen-induced dorsolateral prostate cancer cells by celecoxib seems to be caused by its effect on cell cycle perturbation.

In mammalian cells the Rb protein acts as a critical switch, regulating entry into the DNA synthetic phase of the cell cycle. Results from the present study suggest a reversal of the stimulation on the Rb, associated with a reduced expression of cyclin D in celecoxib-treated cells. This was demonstrated by the dose-dependent down-regulation of cyclin D proteins that is consistent with the higher expression of total Rb in G₁-arrested cells. The inhibitory effect of celecoxib on cell cycle progression was directly related to an increase in the proportion of cells with total Rb labeled with FITC. At 30 μm celecoxib, the percentage of Rb-positive cells increased to 50%. This suggests that the suppression of cell cycle progression through the G₁-S transition phase that is induced by celecoxib seems to be mediated by the maintenance of Rb in its normal functional state. Rb is the master switch regulating cell cycle progression, and its status of phosphorylation parallels cell transit through G₁ into the S phase (62, 63). One of the pathways that regulates Rb phosphorylation and controls the cell cycle from G₁ to S phase progression is through cyclin D and the cdk4/6 complex. The activated cyclin D/CDK4 complex can phosphorylate Rb and prevent its binding to E2F, resulting in the entry from G₁ into S phase (64). Several earlier studies indicated the potential role of cyclins D and E in growth-arrested prostate cancer cells and pointed to the distinct class of genes involved in prostate cancer cells (65, 66, 67, 68, 69, 70, 71, 72). However, the role of COX-2 targets and cyclins in the progression of prostate cancer requires further investigation.

To better understand the inhibitory effect of celecoxib on prostate cancer growth via cell cycle regulatory mechanisms, we studied the expression of the direct inhibitors p27^KIP1 and p21^WAF1/CIP1 on cdk activity. The results of this study demonstrate that celecoxib increased the expression of p27^KIP1 protein by comparison with control. Interestingly, we observed an increase in the rate of apoptosis only in cells with a higher activation of p27^KIP1 and p21^WAF1/CIP1, which suggests that the up-regulation of p27^KIP1 may contribute to the induction of cell cycle arrest and apoptosis (73, 74).

Prostate cancer cells used in this study consistently express elevated levels of COX-2 both in the tumor and in the cells derived from the tumor, detected by immunofluorescence and Western blot analysis when compared with celecoxib-treated cells. These results suggest that the inhibitory role of celecoxib in prostate cancer chemoprevention could be mediated via COX-2-independent mechanisms in addition to COX-2-related signaling pathways. A close link between the cell cycle regulatory mechanisms and COX-2 signaling pathways in the presence of celecoxib in carcinogen-induced prostate cancer needs to be further investigated. Nevertheless, overall data suggest a synergism in the mechanisms of action of celecoxib, at least in prostate cancer cells, through the modulation of COX-2 activity and cell cycle regulatory proteins at nontoxic doses. A scheme of predicted mechanism(s) illustrating the cascade of events exerted by celecoxib in rat prostate cancer growth prevention is presented in Fig. 8. At the cellular level, the COX-2 inhibitor celecoxib is shown to induce cell cycle arrest and apoptosis and to inhibit COX-2 expression as a function of dose. This illustration also explains the significance of non-COX targets such as modulation of cell cycle regulatory genes and Bax, Apaf-1, and caspase activation; modulation of NF-κB and PPARγ further supports the non-COX-2 hypothesis to induce antineoplastic effects in vitro. A synergism in the mechanism(s) of COX-2 and non-COX-2 pathways at the point of cytochrome-c activation-mediated apoptosis is also implicated in this illustration. Recent results on the differential expression of genes by microarray analysis, from our ongoing studies with celecoxib in rat prostate cancer,⁴ strongly support the events predicted in the mechanistic pathways shown here. To complement these results from in vitro studies, we are currently testing the anticancer effects of celecoxib at different stages of prostate cancer growth in the transgenic adenocarcinoma of the mouse prostate model. These findings are expected to demonstrate the chemopreventive effects of celecoxib in prostate cancer at clinically achievable doses, in addition to revealing the level of COX-2 expression in prostate tumors and other cell types associated with the tumor milieu at different stages of the disease.

We thank Dominic Nargi for his help in laboratory analysis. We thank Ilse Hoffmann and Joanne Braley for editing the manuscript.