Cyclooxygenase-2 inhibitor celecoxib augments chemotherapeutic drug-induced apoptosis by enhancing activation of caspase-3 and -9 in prostate cancer cells


  • Devendra S. Dandekar,

    1. Department of Urology, University of Miami School of Medicine, Miami, FL, USA
    Search for more papers by this author
  • Monica Lopez,

    1. Department of Urology, University of Miami School of Medicine, Miami, FL, USA
    Search for more papers by this author
  • Robert I. Carey,

    1. Department of Urology, University of Miami School of Medicine, Miami, FL, USA
    Search for more papers by this author
  • Bal L. Lokeshwar

    Corresponding author
    1. Department of Urology, University of Miami School of Medicine, Miami, FL, USA
    2. Department of Radiation Oncology, University of Miami School of Medicine, Miami, FL, USA
    3. Sylvester Comprehensive Cancer Center, University of Miami School of Medicine, Miami, FL, USA
    • Department of Urology Research Laboratories, University of Miami School of Medicine, P.O. Box 016960 (M-800), Miami, FL 33101
    Search for more papers by this author
    • Fax: +305-243-6893


Many tumors constitutively express high levels of the inducible form of proinflammatory enzyme, cyclooxygenase-2 (COX-2). Increased COX-2 expression is associated with tumor cell resistance to many cytotoxic chemotherapy drugs. Furthermore, increased resistance to cytotoxic antitumor drugs is also known to be dependent on associated stromal cells in many tumors. We investigated whether prostate tumor-associated stromal cells, marrow-derived osteoblasts, affect cytotoxicity of 2 antitumor drugs, COL-3 and docetaxel (TXTR), and whether it is dependent on COX-2 activity. We further examined whether inhibiting the activity of COX-2 negate the stroma-induced decrease in drug sensitivity in tumor cells. COX-2-specific inhibitor celecoxib (CXB) was used to inhibit COX-2 activity and associated alteration in cell death signaling was investigated. Coculturing PC-3ML cells with osteoblasts decreased the cytotoxicity of the tested antitumor drugs and was associated with increased COX-2 activity in PC-3ML cells. A significant decrease in drug-induced PGE2 increase and an increase in cytotoxicity were observed when cells were treated with COL-3 or TXTR combined with CXB. Cytotoxicity of single or combination treatment increased apoptosis, which was associated with caspase-3 and -9 activation, PARP cleavage, increased BAD protein, but decreased protein levels of XIAP and BCL-xL. Oral administration of CXB (40 mg/kg) to mice with PC-3ML tumors for 42 days increased tumor latency, decreased tumor growth and enhanced tumor control with COL-3 or TXTR. Overall, a synergistic enhancement of antitumor activity in combination treatment was observed in vitro and an additive effect in vivo. These observations suggest a potential clinical use of combined dosing of COX-2 inhibitors and cytotoxic drugs at lower, nontoxic dose than currently used to treat advanced prostate cancer. © 2005 Wiley-Liss, Inc.

Prostate cancer ranks second among all malignant cancers in men.1 Hormone-refractory metastatic prostate cancer (HRPC) accounts for most of the morbidity and mortality associated with this disease, killing an estimated 30,000 U.S. men in 2003.1 Treatment for HRPC is mostly palliative with a median survival of 18 months, primarily due to the availability of a very small repertoire of chemotherapy drugs. Multiple physiologic events such as emergence of multidrug-resistant tumor cells, resistance to drug-induced apoptosis and supportive tumor microenvironment are attributed to the failure of cytotoxic chemotherapy.2, 3, 4 Unlike hematologic tumors, HRPC cells seem to overcome the assault of cytotoxic chemotherapy due to a small growth fraction and a decreased rate of apoptosis.5 In addition, many factors produced by both tumor and stromal cells may allow the tumor cells to resist cytotoxicity of anticancer drugs. In this respect, the increased activity of an enzyme such as cyclooxygenase-2 (COX-2) that provides an increased pool of eicosinoids (for instance, prostaglandin E2) may help tumor cells resist the assault of cytotoxic drugs and allow their survival.6, 7

COX-1 and COX-2 are the 2 isoforms of cyclooxygenase, which convert arachidonic acid into several eicosanoids such as prostaglandin, thromboxins and prostacyclin, which participate in several normal physiologic processes and inflammation.8, 9 Whereas COX-1 is constitutively expressed in most tissues, COX-2 is an inducible enzyme, stimulated by cytokines, growth factors, oncogenes, or tumor promoters during inflammation or malignancy.10, 11, 12 At present, the overexpression pattern of COX-2 in CaP,13 in contrast to other cancers such as colorectal malignancies,14 is unresolved. While some reports suggest no significant increase in COX-2 expression in prostate tumor tissues when compared to normal tissues,15, 16 others have reported a direct association of prostate tumor grade and poorly differentiated prostate cancer with COX-2 levels.17, 18, 19, 20 In addition, a significant inhibition of tumor growth by inhibitors of COX-2 in both high COX-2-expressing as well as low COX-2-expressing CaP models has introduced a significant amount of ambiguity on the role of COX-2 in CaP.21 Furthermore, it is not known whether therapeutic intervention in CaP causes changes in COX-2 activity.

Most of the established cytotoxic antiproliferative drugs (chemodrugs) such as docetaxel, doxorubicin, or a novel anticancer drug, COL-3,22 are cytotoxic to CaP cells, although their efficacy varies with their androgen sensitivity and culture conditions.22, 23 HRPC cells such as PC-3ML and DU145 are more resistant to these drugs than the androgen-sensitive LNCaP cells.23 Furthermore, the cytotoxic response also varies with conditions in which the cells are cultured. For example, CaP cells exhibit higher resistance to drug-induced cytotoxicity when cultured in the presence of stromal cells such as prostate fibroblasts, lung fibroblasts and osteoclasts or their culture-conditioned media.24, 25, 26, 27 We reported previously that the expression of both proinflammatory cytokines and chemokines is elevated in CaP cells when they are cocultured with stromal cells, and this may be responsible for the increased resistance of tumor cells to chemodrugs.22, 28, 29 We hypothesized that upregulation of inflammatory response is a mechanism by which tumor cells reduce the cytotoxic effects of anticancer drugs, and the attendant stromal microenvironment further increases the inflammatory response. We further hypothesized that simultaneous or preinhibition of inflammatory response by inhibiting COX-2 activity or expression will increase the efficacy of chemotherapy drugs in vitro and in vivo.

In the present report, we attempted to establish that induced expression of COX-2 is at least partly responsible for the increased resistance to chemotherapeutic drugs. Furthermore, systemic inhibition of COX-2 activity may abrogate this resistance, thus opening a new avenue for increasing the efficacy of antitumor drugs against HRPC.


CaP, carcinoma of the prostate; CCM, culture-conditioned medium; COL-3, chemically modified tetracycline-3 (6-demethyl-6-deoxy-4-dedimethylamino-tetracycline); COX, cyclooxygenase; CXB, celecoxib; HFOB, human fetal osteoblast; HRPC, hormone-refractory prostate cancer; PARP, poly(ADP-ribose) polymerase; p.o., oral gavage; TXTR, Taxotere (Docetaxel).

Material and methods

Chemicals and antibodies

COL-3 was a gift from Collagenex Pharmaceuticals (Newtown, PA). Celecoxib (CXB) was prepared from commercially available Celebrex capsules by solvent extraction and recrystallization.30 Stock solutions of docetaxel (Taxotere, or TXTR) was prepared as per the supplier's instructions (Aventis Pharmaceuticals, Collegeville, PA), made fresh daily for in vivo studies, and dilutions of frozen aliquots were used for cell culture experiments. Indomethacin (a nonselective COX inhibitor), SC560 (a selective COX-1 inhibitor) and NS-398 (a selective COX-2 inhibitor) were purchased from Calbiochem (EMC, La Jolla, CA). Antibodies and assay kits to cytokines, chemokines and apoptosis-related assays were purchased from various sources, as indicated next to their first mention. All other drugs used in the study were from Sigma (St. Louis, MO).

Cells and tumor lines

Established human prostate cancer cell lines LNCaP and DU-145, and PC-3ML were purchased from American Type Culture Collection (Rockville, MD) and maintained in the laboratory in a complete medium (RPMI-1640 medium with L-glutamine; 10% fetal bovine serum, Atlanta Biologicals, Norcross, GA; and gentamicin, 10 μg/ml) as described before.31, 32 Dr. Mark E. Stearns of Medical College of Pennsylvania/Heinemann (Philadelphia, PA) kindly provided PC-3ML cells, a metastatic variant of PC-3 cell line.33 A human fetal osteoblast line (HFOB; ATCC catalog number CRL-11372) was maintained in DME/Ham's F-12 medium with 10% fetal bovine serum (Atlanta Biologicals). HFOB cells are temperature-sensitive immortalized cells, which proliferate rapidly at 33.5°C and differentiate into osteoblasts at 37°C.34 We collected culture-conditioned medium (CCM) from this cell line by growing the cells at 34°C and allowing the confluent-cultures to differentiate at 37°C for 72 hr. The CCM was assayed for alkaline phosphatase to ensure the phenotype of the cultures. CCM from HFOB and PC-3ML cells were stored at −20°C to be used for assays and culture supplements.

Stroma tumor cell cocultures

HFOB and PC-3ML cells (1 × 104 cells/well each) were cultured for 2 days in Costar Transwells (Corning-Costar, Boston, MA) with a 3 μ polycarbonate filter as the diffusion barrier between PC-3ML (bottom well) and HFOB cells (top well). Drugs to be tested were added to both bottom and top wells and the cytotoxicity was assayed by the Thiazolyl blue tetrazolium bromide (MTT) reduction assay32 48 hr later. PC-3ML cells were also cultured in the presence of HFOB-CCM (mixed 1:1 with culture medium) for 2 days before adding the drugs. Drug-induced cytotoxicity was determined 48 hr later by MTT assay.

Determination of PGE2 levels by ELISA

PGE2 released by CaP cells in CCM was measured using the PGE2 Competitive Immunoassay kit (R&D Systems, Minneapolis, MN). Briefly, CCM from CaP cells (5 × 104 cells/well; 12-well plate) cultured with or without drugs for 24–48 hr was collected and assayed for PGE2 as per the supplier's instructions. Levels of PGE2 were normalized to viable cells (e.g., pg/5 × 104 cells).

Determination of cytotoxicity

Drug-induced cytotoxicity was estimated using the MTT colorimetric assay or by cell counting.35 All assays were repeated at least 3 times. Clonogenic survival assays (colony assays) were also undertaken to determine colony-forming potential of CaP cells treated with various drugs (48-hr exposure).

Quantitative determination of apoptotic activity

We used a colorimetric ELISA-like assay (Cell Death ELISA-Plus kit; Roche Molecular Diagnostics, Indianapolis, IN) to quantitate the drug-induced programmed cell death (apoptosis) in CaP cells. The assay measured the amount of free mono- and oligonucleosomes in cell lysates of drug-treated cells undergoing apoptosis. Cell lysates prepared from cell cultures (1 × 104 cells/well in a 48-well plate) incubated with various drugs for 24 hr in duplicate wells were analyzed using triplicate samples according to the instructions provided with the kits.

Immunoblotting (Western blotting)

The expression of COX-1, COX-2, Akt, phosphorylated Akt (P-Akt) and various apoptosis-related proteins were detected by Western blotting using specific antibodies with standard procedures, as suggested by the suppliers, as follows: COX-1: rabbit anti-COX-; COX-2: mouse anti-COX-2 antibody (both 1 μg/ml; Cayman Chem, Ann Arbor, MI), rabbit anti-P-AKT, mouse antitotal Akt IgG (both 1 μg/ml; BD Pharmingen), PARP, Bcl-2, Bad and Bax antibodies (all mouse IgG; 1 μg/ml; BD Biosciences, Bedford MA), procaspase-9, Bcl-xL and caspase-3 (1 μg/ml; all rabbit antibodies; BD Biosciences), rabbit anti-XIAP antibody (0.5 μg/ml; Trevigen, Gaithersburg, MD), rabbit antisurvivin antibody (1 μg/ml; Santa Cruz Biotechnology, Santa Cruz, CA), or cytokeratin8/18 (mouse monoclonal antibody 5D3; 2.5 μg/ml; Novacastra, Newcastle upon Tyne, U.K.). Briefly, the procedure entailed preparing cell lysates in laemmli SDS sample buffer containing 0.1 M dithiotreitol from cultures treated with various drugs for 24 hr, fractionating on an SDS-polyacrylamide gel and electrically blotting onto PVDF membrane and sequentially incubating the blotted membrane with a blocking buffer (5% nonfat milk in 50 mM Tris-HCL buffer with 0.15 M NaCl, 0.25% polysorbate-20, i.e., Tween-20), primary antibody, secondary antibody and detection of the bound protein with diluted appropriate second antibodies conjugated to peroxidase. Bound antibody was made visible by the enhanced chemiluminescence technique using an ECL kit (ECL Plus; Amersham) and exposing the membrane to X-ray film. Band densities were normalized to that of cytokeratin-8 bands. The relative band densities in the blot were digitized by scanning and quantified using Kodak 1D-Image Analysis Software (Eastman Kodak, Rochester, NY). To determine the relative changes in the protein expression in treated samples, the immunoblot (band) intensities were normalized to that of cytokeratin.

Caspase-3 activation assay

Cell death caspase-3 activity in drug-treated cell lysates was determined using a colorimetric assay kit (R&D Systems, Minneapolis, MN). Briefly, CaP cells (1 × 106 cells/60 mm dish) were treated with COL-3, TXTR, or CXB alone or in combination for 24 hr. The amount of chromophore p-NA released from DEVD-pNA substrate peptide was read in a plate reader as per the supplier's instructions.

Cell cycle phase fraction analysis

Cultures of PC-3ML cells (1 × 105/dish) were incubated with or without COL-3 (10 μM), CXB (10 μM), or both for 24 hr. Treated and untreated cultures were harvested; nuclei were labeled with propidium iodide (50 μg/ml) in a single step as described before35 except that suspensions of nuclei were incubated in the dye for 2 hr at 4°C before analyzing on a flow cytometer. The amount of propidium iodide bound to cellular DNA was analyzed using a Beckman-Coulter EPICS XL flow cytometer. The fraction of cell population in G0/G1, S- and G2/M phase was calculated by determining area under the DNA histogram using the Modfit FCM data analysis program (Verity Software, Inc., Topsham, ME).35

Tumor generation and drug administration

Xenografts of PC-3ML cells were generated in 6- to 8-week-old athymic mice (Harlan-Sprague Dawley, Indianapolis, IN) by subcutaneous (s.c.) injection of 0.2 ml suspension of 1 × 106 cells in 50% Matrigel (BD Biosciences). All manipulations were performed according to a protocol approved by the University of Miami Animal Use and Care Committee and as stipulated in the National Institutes of Health guide to the humane care and use of laboratory animals. Mice were housed in a room with 12-hr light-and-dark cycles, and food and water supply was provided ad libitum throughout the experiment. Following implantation of tumor cells, mice were randomly divided into various groups comprising of 6–12 animals each as described in Table I. Tumor growth was examined by palpating the skin around the site of injection. After the tumors became palpable (∼day 7), tumor volumes were measured 3 times a week with a vernier caliper and the volumes were approximated to an ellipsoid (length × height × width × 0.524). Mice were euthanized after 45 days when the mean tumor volume in untreated group measured ∼1,000 mm3.

Table I. Number of Animals per Group and Treatment Regimen for In Vivo Study
Group and number of animals/groupTreatment
1 (n = 6)Control: vehicle-treated 0.5 mL 1% carboxy methyl cellulose (CMC)/mouse
2 (n = 6)COL-3: 40 mg/kg body weight p.o. 0.5 mL/mouse
3 (n = 6)CXB: 40 mg/kg body weight p.o. 0.5 mL/mouse
4 (n = 6)COL-3 + CXB: 40 mg/kg body weight p.o. each drug (CXB was administered 1hr after COL-3 administration)
5 (n = 6)TXTR: 2.3 mg/kg body weight 3 × week i.p.
6 (n = 6)TXTR + CXB: TXTR 2.3 mg/kg body weight 3 × week/mouse i.p. + CXB 40 mg/kg body weight p.o. was administered 1hr after TXTR

Statistical analysis

Triplicate samples were assayed in all in vitro experiments in 3 independent assays. Interaction of combined-drug treatment vs. single-drug treatment in vitro and in vivo was determined using the Calcusyn program (Biosoft, Cambridge, U.K.). Tukey-Kramer multiple-comparison test was used to test the significance of the tumor growth differences between various treatment groups (Instat Program; GraphPad, San Diego, CA). Other data were analyzed using parametric and nonparametric Student's t-tests as appropriate.


Increased PGE2 production by PC-3ML cells in the presence of stromal cells

Increase in COX-2 expression in various malignancies, but not as yet for CaP, is attributed as one of the causes for reduced efficacy of cytotoxic drugs.36, 37, 38 Moreover, stromal cells such as osteoblasts that surround the prostate tumor metastatic to bone are also known to decrease the cytotoxic effects of chemodrugs.25 We investigated whether exposure of CaP cells to chemodrugs with or without coculturing the cells with osteoblastic cells (HFOB) affect COX-2 activity (secreted PGE2 levels).

As shown in Figure 1(a), PGE2 levels in PC-3ML and HFOB monoculture (after 48 hr) were 1,039 pg/5 × 104 cells and 290 pg/5 × 104 cells, respectively. We also observed 25% ± 4.2% increase in PGE2 levels in CCM of PC-3ML cells cocultured with HFOB cells for 24 hr. This increase was not observed in the CCM of either PC-3ML/PC-3ML cocultures or HFOB/HFOB cocultures. Similar increase in PGE2 level was also observed when PC-3ML cells were cultured in 50% HFOB-CCM (data not shown). Next, we investigated whether incubation with cytotoxic drugs would alter COX-2 activity in the cocultures. Incubation with COL-3 or TXTR in similar set of culture wells resulted in further increase in PGE2 levels (63% ± 1.2% and 74% ± 1.5%, respectively) when compared to the PGE2 level in untreated PC-3ML/HFOB coculture (Fig. 1a).

Figure 1.

Increased PGE2 release when PC-3ML cells are cocultured with HFOB or HFOB-CCM (a), protein levels of COX-2 and COX-1 in PC-3ML cells exposed to chemodrugs alone or in combination (b) and decreased cytotoxicity of chemodrugs on PC-3ML cells cocultured with HFOB or with HFOB-CCM (c). (a) PGE2 levels in CCM were measured by an ELISA, following treatment with CXB (10 μM), COL-3 (10 μM), TXTR (30 nM), CXB (10 μM) + COL-3 (10 μM), or CXB (10 μM) + TXTR (30 nM) for 24 hr. (b) Western blot analysis of protein levels of COX-2 and COX-1 was determined in cell lysates of PC-3ML exposed to CXB 10 μM (lane 2), TXTR 30 nM (lane 3), CXB 10 μM + TXTR 30 nM (lane 4), COL-3 10 μM (lane 5) and CXB 10 μM + COL-3 10 μM (lane 6). Lane 1: untreated PC-3ML. Variation in total amount of proteins loaded in each lane was normalized with respect to the levels of cytokeratin-8 in each lane. (c) Viability (%) of PC-3ML cells was determined by MTT assay following exposure to CXB (10 μM), COL-3 (10 μM), TXTR (30 nM), CXB (10 μM) + COL-3 (10 μM), or CXB (10 μM) + TXTR (30 nM) for 48 hr in complete medium or CCM from HFOB cells. Data represents mean ± SEM from 3 independent assays. Asterisk, p < 0.05, t-test treatment vs. control (C).

Since we observed increases in PGE2 release in response to chemodrug treatment, we evaluated the effects of selective COX-2 inhibitor CXB alone or in combination with chemodrugs (i.e., CXB + COL-3 or CXB + TXTR) on PGE2 release in PC-3ML/HFOB coculture. As shown in Figure 1(a), treatment with 10 μM CXB alone resulted in 64% ± 3.1% decrease in PGE2 levels. Furthermore, COL-3 + CXB or TXTR + CXB treatment also resulted in similar decreases (52% ± 4.2% and 53% ± 3.1%, respectively) in PGE2 levels compared to that of untreated cultures.

We also determined the protein levels of COX-1 and COX-2 in PC-3ML cells treated with each of the drugs alone or in combination (Fig. 1b) by Western blot analysis. Protein levels of COX-2 in PC-3ML cells treated with various drugs were comparable to the activity (PGE2 production) as determined by ELISA. The digitized protein band densities of COX-1 and COX-2 were normalized to cytokeratin band intensities of corresponding samples. The results as presented in Figure 1(b) show that the protein levels of COX-2 remained relatively unchanged (less than 10% increase or decrease) irrespective of treatment, except that of a 16% decrease in CXB-treated samples (Fig. 1b, lane 2). COX-1 levels, however, did not change due to any treatment (Fig. 1b). Therefore, it is likely that the observed increase in PGE2 is solely contributed by increased COX-2 activity, not protein content, in drug-treated cells.

Decreased efficacy of COL-3 and TXTR on PC-3ML cells cocultured with HFOB cells or HFOB-CCM

Since we observed increased PGE2 secretion by PC-3ML cells in the presence of HFOB cells and cytotoxic drugs, we investigated whether the coculture conditions affect the response of PC-3ML cells to cytotoxic drugs. PC-3ML cells cultured in the presence of HFOB-CCM (50%) for 48 hr were exposed to COL-3 or TXTR and cell viability was estimated after an additional 48 hr. As shown in Figure 1(c), we observed a significant decrease in efficacy (increased viability) of COL-3 and TXTR on PC-3ML cells cultured in 50% HFOB-CCM by 42% ± 2.2% (p < 0.05) and 32% ± 1.3% (p < 0.05), respectively, when compared to the viability of PC-3ML cells cultured in 50% CCM of PC-3ML with or without the drugs. Due to the observed association of increased PGE2 but decreased cytotoxicity of cells cultured with HFOB cells or their CCM, we next determined involvement of COX-2 in decreasing the cytotoxicity of chemodrugs.

CXB causes maximum cytotoxicity on CaP cell lines among various COX inhibitors

As shown in Table II, androgen-responsive (LNCaP) and androgen-unresponsive (PC-3ML and DU-145) CaP cells were exposed to selective COX-1 inhibitor SC-560, selective COX-2 inhibitors CXB or NS-398, or a nonselective COX inhibitor, indomethacin, for 48 hr. Cell viability was quantified at the end of 48 hr by the MTT assay. As listed in Table II, CXB was the most effective cytotoxic agent among various COX inhibitors tested. PC-3ML was the most resistant CaP cell line; the 50% inhibition dose (IC50) of CXB was 10 μM.

Table II. Percentage Cell Viability
  Percentage cell viability (48h treatment)
  1. Viability of CaP cells exposed to various NSAIDs was determined by MTT assay as described in text. Data shown are mean ± SEM obtained from 3 independent experiments.

10 μMCXB22.78 ± 0.5438.14 ± 0.1456.41 ± 4.17
 NS-39884.55 ± 1.6487.14 ± 1.4091.09 ± 4.93
 SC-56049.60 ± 2.7474.72 ± 2.6573.32 ± 4.30
 IND67.57 ± 1.5492.14 ± 2.1489.98 ± 3.21
50 μMCXB4.62 ± 0.195.14 ± 0.852.05 ± 0.27
 NS-39846.11 ± 2.1961.71 ± 1.7171.94 ± 4.30
 SC-56028.90 ± 3.7438.96 ± 1.3229.22 ± 1.72
 IND55.54 ± 0.6464.57 ± 7.1479.43 ± 0.60
100 μMCXB3.64 ± 0.294.04 ± 0.715.00 ± 0.11
 NS-39817.54 ± 0.4143.85 ± 0.8948.36 ± 4.30
 SC-56027.56 ± 4.6317.28 ± 4.1417.54 ± 0.27
 IND21.51 ± 0.5941.00 ± 7.1452.32 ± 4.77

COL-3 significantly downregulates AKT phosphorylation

Some investigators have reported PDP-kinase-1/AKT signaling pathway as one of the COX-2-independent target of CXB and antiproliferative effect of CXB (50 μM) is mediated through downregulation of AKT.21, 39 Therefore, we determined the effect of CXB alone (10 μM) or in combination with chemodrugs on P-AKT levels when the cells are exposed to CXB in the culture medium containing 10% FBS. As shown in Figure 2, the levels of P-AKT following CXB (10 μM) treatment (lane 2) remained unchanged; however, CaP cells exposed to TXTR alone (lane 3) or TXTR + CXB (lane 4) showed modest decrease in the levels of phosphorylated AKT, whereas exposure to COL-3 (lane 5) alone or COL-3 + CXB (lane 6) caused significant downregulation of P-AKT in CaP cells as compared to the control (lane 1). However, we did observe downregulation of P-AKT when PC-3ML cells were treated with 50 μM CXB and in similar culture conditions used by the above investigators (when the cells are exposed to CXB 50 μM in the serum-free condition medium; data not shown). These results suggest that increased cell kill following COL-3 treatment is partly associated with downregulation of AKT pathway, and that CXB at a lower concentration (10 μM) has no effect on the AKT signaling pathway.

Figure 2.

Protein levels of total and phosphorylated AKT. Protein levels of phosphorylated AKT and total AKT were determined by Western blotting of cell lysates of unexposed PC-3ML cells (lane 1) or cells exposed to CXB (10 μM; lane 2), TXTR (30 nM; lane 3), CXB (10 μM) + TXTR (30 nM; lane 4), COL-3 (10 μM; lane 5), or CXB (10 μM) + COL-3 (10 μM; lane 6) for 24 hr.

Combined administration of CXB with chemodrugs increases cytotoxicity of chemodrugs

Since chemodrug-induced increase in PGE2 and AKT phosphorylation in PC-3ML culture was not observed when the chemodrugs were coadministered with CXB (Figs. 1a and 2), we investigated whether the suppression of chemodrugs induced PGE2 or AKT phosphorylation by CXB in combined treatment has any influence on the efficacy of chemodrugs. As shown in Figure 3, the viability of PC-3ML cells treated for 48 hr with COL-3 or CXB was ≤ 50% and TXTR was ≤ 70%; more importantly, cell viability in combined treatment (CXB + COL-3 or CXB + TXTR) was < 20%, i.e., a 2-fold increase in cytotoxicity when compared to the treatment with single agent (p < 0.05, Tukey-Kramer multiple-comparison test). We further analyzed the cytotoxicity of combined treatment for additive, synergistic, or antagonistic effects by determining combination index (CI) generated by the Calcusyn program. CI values for CXB + COL-3 was 0.801 (slight synergism) and for CXB + TXTR was 0.310 (synergism). These results indicate that a combination of cytotoxic chemodrug COL-3 or TXTR with COX-2 inhibitor (CXB) has a synergistic effect and therefore improves the cytotoxicity of either drug. We next investigated the mechanism of enhanced cytotoxicity following a combination treatment.

Figure 3.

Cytotoxicity of various drugs on PC-3ML cells. Viability of PC-3ML cells (%) was determined by MTT assay following exposure to CXB (10 μM), COL-3 (10 μM), TXTR (30 nM), CXB (10 μM) + COL-3 (10 μM), or CXB (10 μM) + TXTR (30 nM) for 48 hr in complete medium. Inset depicts cell count (Trypan blue exclusion) of PC-3ML cells exposed to various concentrations of CXB for 24 or 48 hr. Data represent mean ± SEM from 3 independent assays.

Increased apoptosis of PC-3ML cells treated with cytotoxic drugs

PC-3ML cells treated with CXB, COL-3, or TXTR and their combination were investigated for apoptotic activity following drug treatment for 24 hr. Results of cell death ELISA (Fig. 4) revealed a complementary pattern of changes to that observed for cell viability (Fig. 3). PC-3ML cells treated with CXB, TXTR, or COL-3 alone for 24 hr showed an increase in apoptotic activity by 1.9-, 2.5- and 2-fold, respectively, when compared to the apoptotic activity in untreated PC-3ML cells. Furthermore, the combination of CXB + COL-3 or CXB + TXTR treatment induced significantly more apoptosis than either drugs alone (4.3- and 4.9-fold, respectively).

Figure 4.

Apoptotic activities in drug-treated PC-3ML cells. Apoptotic activities in PC-3ML cells (1 × 104 cells/well; 48-well plates) exposed to CXB (10 μM), COL-3 (10 μM), TXTR (30 nM), CXB (10 μM) + COL-3 (10 μM), or CXB (10 μM) + TXTR (30 nM) for 24 hr were assayed. Levels of soluble nucleosomes in the cell lysates were determined using the Cell Death Plus ELISA. Data shown are mean ± SEM obtained from 3 independent experiments.

COL-3-, TXTR- and CXB-induced apoptosis in CaP cells is caspase-mediated

We next investigated whether increased apoptosis (Fig. 4) in drug-treated cells is associated with increased activation of cell death-associated caspases, active caspase-3 and caspase-9. As shown in Figures 5 and 6(a), increased cytotoxicity and induction of apoptosis were associated with increased caspase-3 and caspase-9 activity (as observed by decrease in procaspase-9 protein levels) in PC-3ML cells treated with COL-3 or TXTR, more pronounced in COL-3 + CXB or TXTR + CXB combination. Quantitative analysis of caspase-3 showed a 38% ± 1.2% increase in activated caspase-3 in PC-3ML cultures treated with COL-3 (10 μM) + CXB (10 μM) and 35% ± 1.1% increase in TXTR + CXB combination (Fig. 5).

Figure 5.

Caspase-3 activity in PC-3ML cells. PC-3ML cells were exposed to CXB (10 μM), COL-3 (10 μM), TXTR (30 nM), CXB (10 μM) + COL-3 (10 μM), or CXB (10 μM) + TXTR (30 nM) for 24 hr. Caspase-3 activity in various samples was determined using an assay kit with DEVD as the substrate. Data shown are mean ± SEM obtained from 3 independent experiments.

Figure 6.

(a) Expression of proapoptotic and antiapoptotic proteins in drug-treated PC-3ML cells. Protein levels of X-linked inhibitor of apoptotic protein (XIAP), surviving, PARP (cleaved PARP ∼85 kDa), active caspase-3 and procaspase-9 were determined in whole cell lysates of PC-3ML cells after 24-hr exposure to CXB 10 μM (lane 2), TXTR 30 nM (lane 3), CXB_10 μM + TXTR_30 nM (lane 4), COL-3_(lane 5) and CXB_10 μM + COL-3 10 μM (lane 6). Lane 1: caspase-3 activity in untreated PC-3ML cells. (b) Western blot analysis of protein levels of Bcl-2, Bcl-xL, Bad and Bax. Whole cell lysates of PC-3ML cells exposed to CXB 10 μM (lane 2), TXTR 30 nM (lane 3), CXB 10 μM + TXTR 30 nM (lane 4), COL-3 10 μM (lane 5) and CXB 10 μM + COL-3 10 μM (lane 6). Lane 1: untreated PC-3ML cells.

COL-3 or TXTR alone or in combination with CXB causes differential alterations in the levels of pro- or antiapoptotic proteins

To delineate the mechanism underlying increased apoptosis following combined treatment, we determined the levels of inhibitors of apoptotic proteins (IAPs; XIAP and survivin), members of pro- and antiapoptotic Bcl-2 family proteins (Bcl-2, BclxL and Bax, respectively), in CaP cells treated for 24 hr with CXB, COL-3, TXTR, COL-3 + CXB, or TXTR + CXB.

Western blot analysis of protein levels of XIAP and survivin, both capable of inhibiting caspase activity,40 showed no significant alteration in their expression in CXB-treated cells (Fig. 6a, lane 2) as compared to control (Fig. 6a, lane 1). Both TXTR (Fig. 6a, lane 3) and COL-3 (Fig. 6a, lane 5) treatment caused decrease in XIAP levels, whereas COL-3 caused decrease in XIAP as well as survivin levels. However, combination of COL-3 + CXB (Fig. 6a, lane 6) or TXTR + CXB (Fig. 6a, lane 4) did not cause further decrease in XIAP or survivin levels.

In addition to the quantitative determination of apoptotic and caspase-3 activity, we determined protein levels of active caspase-3 and cleavage of poly(ADP-ribose) polymerase (PARP) in CaP cells following CXB, TXTR, COL-3, and TXTR + CXB or COL-3 + CXB treatment. We observed a significant increase in protein levels of active caspase-3 and cleaved PARP in CaP cells exposed to CXB (Fig. 6a, lane 2), TXTR (Fig. 6a, lane 3), or COL-3 (Fig. 6a, lane 5). However, TXTR + CXB (Fig. 6a, lane 4) or COL-3 + CXB (Fig. 6a, lane 6) combinations showed further increase in protein levels of active caspase-3 and cleaved PARP, more pronounced in the TXTR + CXB combination.

We next determined the protein levels of antiapoptotic proteins Bcl-2 and Bcl-xL and proapoptotic proteins Bax and Bad. We observed increased levels of proapoptotic proteins and a concomitant decrease in protein levels of antiapoptotic proteins. Levels of Bcl-2 protein in CaP cells treated with CXB (Fig. 6b, lane 2) remained unchanged, whereas levels of Bcl-2 in CaP cells treated with TXTR (Fig. 6b, lane 3), COL-3 (Fig. 6b, lane 5), TXTR + CXB (Fig. 6b, lane 4), or COL-3 + CXB (Fig. 6b, lane 6) showed a small increase in protein levels of Bcl-2 as compared to the control (Fig. 6b, lane 1). Although we observed a small increase in protein levels of Bcl-2 in CaP cells treated with either drugs alone or in combination, we observed significantly decreased levels of Bcl-xL in CaP cells treated with either drug alone (Fig. 6b, lane 2, CXB; lane 3, TXTR; lane 5, COL-3) or their combination (Fig. 6b, lane 4, TXTR 30 μM + CXB 10 μM; lane 6, COL-3 10 μM + CXB 10 μM) as compared to the control (Fig. 6b, lane 1). The protein levels of Bax were significantly increased in CaP cells exposed to CXB (Fig. 6b, lane 2), TXTR (Fig. 6b, lane 3), TXTR + CXB (Fig. 6b, lane 4), COL-3 (Fig. 6b, lane 5), or COL-3 + CXB (Fig. 6b, lane 6) as compared to untreated CaP cells (Fig. 6b, lane 1). Moreover, CaP cells exposed to COL-3 (Fig. 6b, lane 5) showed the maximum increase in protein levels of Bad compared to other treatments.

Lack of cell cycle arrest (G0/G1 arrest) in CXB treated cells

Some investigators have reported that a possible consequence of COX-2 inhibition by its inhibitors, such as CXB, is the blocking of cell cycle progression at G0/G1 phase.41, 42 As shown in Table III, treatment with CXB (10 μM) for 24 hr did not significantly affect the cell cycle progression in PC-3ML cells. However, treatment with COL-3 alone resulted in an increase in G0/G1 cell population (increase by 50.4% ± 0.12%) and a decrease in S-phase accumulation (decrease by 57.9% ± 0.05%) as reported before.32 The cell population treated with CXB + COL-3 combination did not show a significant increase in G0/G1 fraction over and above that observed in cells exposed to COL-3 alone.

Table III. Treatment Data
PhaseControlCXB (10 μM)COL-3 (10 μM)CXB (10 μM) + COL-3 (10 μM)
  • Cell cyclephase fractionation analysis was performed in untreated PC-3ML or PC-3ML cells exposed to CXB (10 μM), COL-3 (10 μM), or CXB (10 μM) + COL-3 (10 μM) for 24 hr. Values are mean and ± SE from 3 independent experiments.

  • 1

    Percent difference was calculated as [(% of cells in respective phase of untreated sample − % cells in respective phase in cytotoxic drug(s)-treated samples) (% of cells in respective phase in untreated control) × 100)].

G0 – G147.83 ± 3.0556.03 ± 4.3071.92 ± 5.1673.70 ± 6.30
 % difference117.1450.3654.06
S32.06 ± 5.2532.75 ± 4.5413.51 ± 4.3011.72 ± 4.35
 % difference12.1−57.9−63.36
G2 – M19.96 ± 4.0211.22 ± 4.1914.56 ± 2.5014.58 ± 3.32
 % difference1−43.78−27.05−27.02

Increased inhibition of PC-3ML tumor xenograft growth by COL-3 or TXTR in combination with CXB

Encouraged by the data from in vitro studies on the growth-inhibitory potential of CXB, COL-3, or TXTR alone and their combination on PC-3ML cells, we investigated the antitumor potential of single-drug or combination treatment in tumor-bearing mice. Groups of 6 mice each were treated with COL-3, CXB, or both (each 40 mg/kg) daily by gavage or with TXTR (2.3 mg/kg i.p., 3 times per week) for 42 days, beginning 24 hr after tumor implantation. There was no observable systemic toxicity, including little change in body weights, of treated animals. As shown in Figure 7, at the end of 42 days, the tumor size in control group (vehicle alone) reached 800 mm3; however, tumor volumes in CXB-, COL-3-, or TXTR-treated mice were much smaller: 50% ± 1.5%, 57% ± 2.6% and 75% ± 2.2% of the control group, respectively. Tumor volumes in mice treated with CXB + COL-3 or TXTR + CXB were 72% ± 2.1% and 83% ± 1.5%, respectively, less than that of control. Terminal tumor volume of each group was analyzed by the Calcusyn program to determine possible additive or synergy following combined treatment. CI values of CXB + COL-3 was 1.07, indicating “nearly additive” effect, and CI value of CXB + TXTR was 0.788, showing a moderate synergism.

Figure 7.

Growth of PC-3ML tumors in Athymic nude mice with or without treatment with CXB, TXTR, or COL-3 alone or in combination with CXB. Tumor volume in mice bearing xenograft of PC-3ML cells (1 × 106 cells; 0.2 ml). Tumor-bearing animals were divided into 6 groups and were treated with vehicle or various drugs as described in Table I. Tumor volume was significantly decreased in TXTR + CXB combination treatment regimen as compared to vehicle control group.


Activation of prosurvival pathways mediated through upregulation of inducible enzyme COX-2 due to inflammation or stress appears to be one of the limiting factors in reducing the efficacy of chemodrugs and increasing metastatic potentials of established tumors.36, 37, 38, 43 In this regard, we hypothesized that by regulating chemodrug-induced COX-2 activity following combined administration with selective COX-2 inhibitor should improve the efficacy of chemodrugs on hormone-refractory prostate cancer. We determined the efficacy of docetaxel (a taxane commercially known as Taxotere), a known chemodrug widely used for the treatment of various cancers,44 which causes impairment of mitosis by interfering with microtubule disassembly, and COL-3, an antimetastatic drug with antitumor properties on prostate cancer model in combination with specific COX-2 inhibitor, celecoxib. We reported in the past that COL-3 has a potent cytotoxic and antimetastatic action on CaP tumor models, primarily by inducing the CaP cells to undergo cell cycle arrest, caspase-mediated apoptosis and inhibition of matrix metalloproteinases.22, 32 In the present study, we aimed to investigate whether combined application of CXB with COL-3 or TXTR, 2 chemodrugs with distinct mechanism of action, is significantly more effective than either agent alone.

Data presented in this report demonstrate that the chemodrug or stromal factor induced overexpression of COX-2 and thereby increased release of cytoprotective prostacyclin PGE2 in CaP cells, maybe one of the factors responsible for limiting the efficacy of chemodrugs on CaP cells (Figs. 1a and b). Stromal cells and factors derived from them have been known to influence therapeutic response in vivo and reduce cytotoxicity in vitro in a variety of tumors and cell types.45, 46, 47 Some reports documented that stromal cells alter expression of cell adhesion molecules, integrins, cytokines, chemokines and also cell cycle-related events. In addition, stromal cells and their CCM are known to cause accumulation of cells at G0/G1 compartment, in which cells are least affected by common chemodrugs, which mostly affect cells at S- and G2/M phase of cell cycle.25 While all these factors may contribute to acquired (reversible with change in culture conditions) drug resistance in tumor cells, our observation on overproduction of PGE2 by elevated COX-2 activity highlights the importance of cytoprotective PGE2 function in acquired drug resistance.

Interestingly, we did not find PC-3ML cell cultures blocked in G0/G1 phase of the cell cycle either due to coculture with HFOB or as a consequence of COX-2 inhibition by CXB at 10 μM. Thus, it is likely that inhibition of COX-2 activity rather than rescue from G0/G1 block, if any, reduces the stroma-induced resistance to cytotoxic agents such as COL-3 and TXTR. We postulate from this that the simultaneous administration of COX-2 inhibitor and a cytotoxic drug causes significantly increased tumor cell death in situ, which may affect tumor growth at metastatic sites, where most chemodrugs are less effective.48 Improved efficacy of chemodrugs observed in human lung and colon cancer cell lines following combined administration of nonsteroidal antiinflammatory drugs (NSAIDs) and chemodrugs also corroborates our observation.49, 50 Nevertheless, similar combination therapies for CaP are not yet reported. In this regard, although at present 2 clinical trials are ongoing (; NLM identifiers NCT00022399 and NCT00073970), this is the first report to document the efficacy of combination therapy for the treatment of hormone-refractory prostate cancer using COX-2 inhibitor and chemodrugs in a model system. Furthermore, we observed a synergistic enhancement of chemotherapeutic efficacy by including COX-2 inhibitor in vitro and moderate additive to mild synergism in vivo when we used the isoeffect dose calculation analysis of Chou and Talaley.51 This method of analysis has been used to determine combination index in combined treatment protocols52 and by other investigators.53

In our search to find an effective COX-2 inhibitor against CaP cells, we found CXB to be a potent COX-2 inhibitor and caused both cytotoxicity and decreased PGE2 release (> 50%) at 10 μM concentration (IC50 = ∼10 μM) in CaP cells. Our results are in agreement with other published reports,54, 55 which also stated that CXB is the most effective COX-2 inhibitor among several existing COX-2 inhibitors such as NS-398, rofecoxib and DuP697, all of which induce apoptosis in CaP cells. However, the effective concentration reported by earlier investigators has been in excess of 50 μM and some in cultures under reduced serum conditions. Since the objective of our investigation was to identify an effective combination therapy at a significantly low dose of COX-2 inhibitor or chemodrugs, we focused on identifying therapeutic enhancement at 10 μM CXB (IC50 dose in the present study). Moreover, PC-3ML cells exposed to CXB at 10 μM in combination with COL-3 or TXTR caused a significant decrease in COL-3- or TXTR-induced PGE2 release (Fig. 1a). The cytotoxicity of both cytotoxic drugs, COL-3 and TXTR, was synergistically enhanced when cells were simultaneously treated with CXB. In this regard, the combined treatment of TXTR with CXB was more effective than that observed with COL-3 + CXB (Fig. 3).

Interestingly, we observed higher cytotoxicity by the COX-1-specific inhibitor SC-560 than by the COX-2-specific inhibitor NS-398 at all 3 concentrations tested (10 to 100 μM). Moreover, as reported by Johnson et al.,55 cytotoxic activity of NS-398 is modest, as it is a slow-acting inhibitor compared to CXB. At the tested concentrations, both drugs are known to induce apoptosis in tumor cells,54, 55 but NS-398 is less effective. Although SC-560 is an established COX-1 inhibitor and NS-398 is a specific COX-2 inhibitor, these drugs also inhibit both COX-1 and COX-2 at high concentrations currently used. It is reported that SC-560 inhibits both COX-1 and COX-2 at the tested concentrations (IC50 of 9 nM and 6.3 μM for COX-1 and COX-2, respectively). Similarly, NS-398 is a stronger inhibitor of COX-2 (IC50 = 1.77 μM) but a poor inhibitor of COX-1 (IC50 = 75 μM)56, 57 and therefore it is likely that higher cytotoxicity of SC-560 is due to the complete COX inhibitions. In this context, although SC-560 appears to be more potent than NS-398, it is unlikely to be applicable in vivo, as it is a stronger inhibitor of the housekeeping enzyme COX-1, which is essential for normal physiologic functions.8, 58

Earlier studies have stated that exposure of CaP cells to CXB causes downregulation of AKT phosphorylation, a major survival pathway, known to be one of the significant COX-2 independent target of CXB.21, 39 However, in our studies, protein levels of phosphorylated AKT remained unchanged in PC-3ML cells exposed to IC50 concentration of CXB (10 μM). Based on these observations, we propose that at lower concentration of CXB, together when tumor cells show robust COX-2 activity (e.g., PGE2 production), the cytotoxic action of CXB is mediated via its ability to inhibit COX-2 activity rather than apoptosis induction (via the downregulation of AKT phosphorylation). Furthermore, CXB (10 μM) in combination with TXTR (Fig. 2, lane 4) or COL-3 (Fig. 2, lane 6) did not enhance the downregulation of AKT phosphorylation in addition to that observed with TXTR (Fig. 2, lane 3) or COL-3 alone (Fig. 2, lane 5).

Our results show that the increased cytotoxicity in the combination (COL-3 + CXB or TXTR + CXB) treatment (Fig. 3) was due to increased apoptosis of CaP cells (Fig. 4), following increased activation of caspase-3, -9 and cleaved PARP (Figs. 5 and 6a). Although Narayanan et al.41 reported that CXB treatment does increase G1 phase fraction in rat prostate cancer cells treated at 40 μM CXB, we did not observe a cell cycle arrest at G1 phase in cells treated with low concentration of CXB. However, in the samples treated with COL-3 + CXB, there was an increase in G1 phase fraction, but it was not significantly more than that observed when cells were treated with COL-3 alone (Table III).

In the present study, we have shown that the increased efficacy of chemodrugs on PC-3ML cells exposed to chemodrugs in combination treatment (CXB + TXTR or CXB + COL-3) is mainly due to increased apoptosis following increased caspase-3 and -9 activation (Fig. 6a). These observations are in agreement with the reported increased apoptosis due to enhanced caspase-3 and -9 activation in rat C611B cholangiocarcinoma and WB-F344 rat liver cells exposed to celecoxib (35 μM) in combination with a tyrosine kinase inhibitor, Emodin (30 μM).59 Protein levels of dominant IAPs, XIAP and survivin, which are highly expressed in prostate cancer,40 remained unchanged in PC-3ML cells exposed to CXB (10 μM). However, protein levels of XIAP and survivin were significantly downregulated in CaP cells treated with TXTR or COL-3 alone (Fig. 6b), whereas TXTR + CXB or COL-3 + CXB combination did not cause further downregulation of XIAP or survivin. This may be due to the fact that CXB at 10 μM did not alter the levels of phosphorylated AKT, an upstream regulator of XIAP and survivin,60 and therefore combination of CXB with chemodrug did not cause further downregulation of these IAPs. It is noteworthy that although PC-3ML cells exposed to CXB (10 μM) showed significant increase in levels of proapoptotic proteins Bad and Bax and decrease in levels of antiapoptotic protein Bcl-xL (Fig. 6b), combined administration of CXB with chemodrugs did not cause further alteration in levels of pro- or antiapoptotic proteins.

Besides conventional chemodrugs for the treatment of cancer, NSAIDs and COX-2 inhibitors are now showing promising evidence as anticancer therapeutics. However, the use of COX-2 inhibitors as a sole chemotherapeutic agent for cancer treatment is not yet fully established, probably due to varied response of tumor cells in vitro and in vivo to COX-2 inhibitors; more importantly, some COX-2 inhibitors are also known to cause gastrointestinal, cardiac and nephrotoxicity.61, 62, 63, 64 There are reports that CXB alone causes decrease in tumor growth in colon, breast and lung cancer models.65, 66 In addition to this, Liu et al.67 have also shown that i.p. administration of another COX-2 inhibitor, NS398 (3 mg/kg), causes up to 93% growth retardation in mice bearing PC-3 xenograft. However, to our knowledge, there is no report on the efficacy of low-dose oral CXB administration on prostate cancer. This is the first report showing significant tumor growth retardation by oral CXB treatment at a significantly lower dose and enhancement in the efficacy of cytotoxic drug when combined with CXB. Our results establish that combining CXB at a low dose with a cytotoxic drug could potentially be an effective treatment for advanced androgen-refractory prostate cancer.


The authors are grateful to Dr. Vinata Lokeshwar for critically reading the manuscript and suggestions for improvement and Mr. Eluet Hernandez for help with animal care.